Citation
Medical-Legal patnerships

Material Information

Title:
Medical-Legal patnerships reducing health disparities by disrupting cumulative disadvantages faced by vulnerable Colorado families
Creator:
Matthew, Dayna Bowen ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file : ;

Subjects

Subjects / Keywords:
Public health administration ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
In Colorado, low-income and minority children live shorter and less healthy lives than wealthier and white children. Data on childhood asthma, obesity, and infant mortality disparities are disheartening. For example, although Colorado has the 5th lowest infant mortality rate nationally, Colorado’s black infants die nearly three times more often than the state’s white infants; Latino infants are 70% more likely to die during their first year than whites. Data from 2015 not only show the immunization rate for Colorado toddlers continued a troubling downward trend so that Colorado children rank 30th in the nation, but a research collaborative between the University of Colorado School of Medicine and Children’s Hospital Colorado reported that “[significant disparities exist in vaccination delivery by age group …, race/ethnicity, and socioeconomic status.”
Review:
This dissertation presents a study premised on the understanding that the inequitable distribution of access to important social determinants of health is an underlying cause of these health disparities among Colorado children. The study examines the impact of medical-legal partnerships (MLPs), an intervention designed to equalize access to preventive medical care, decent housing, and other social determinants in order to disrupt the fundamental association between poverty, minority race and ethnicity status, and poor health. MLPs are an innovative health delivery model that integrate lawyers into clinical delivery teams to help equalize vulnerable patients’ access to the social determinants of health that are guaranteed by law.
Review:
The MLP studied here operated between 2009 and 2013 between Children’s Hospital of Colorado, Colorado Legal Services, and a law school program called the Colorado Health Equity Project. This study retrospectively analyzed quantitative medical and qualitative legal data to examine whether MLP services improved children’s health and social outcomes by reducing upstream social risk factors that could decrease health disparities over the life course. Specifically, this study analyzed the MLPs impact on young children’s vaccination compliance and access to better neighborhood resources due to upward residential mobility. Although this MLP did not have a statistically significant impact on either outcome of interest, the methodology developed here suggested a relationship between MLPs and other health outcomes that warrants further study.</DISS_para>
Thesis:
Thesis (Ph.D.)--University of Colorado Denver
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Dayna Bowen Matthew.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
on10611 ( NOTIS )
1061150735 ( OCLC )
on1061150735

Downloads

This item has the following downloads:


Full Text
Medical-Legal Partnerships: Reducing Health Disparities By Disrupting Cumulative
Disadvantages Faced By Vulnerable Colorado Families
by
Dayna Bowen Matthew A.B., Harvard-Radcliffe, 1981 J.D., University of Virginia, 1987
A Dissertation Submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy
Health and Behavioral Sciences 2018


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
11
2018
Dayna Bowen Matthew
All Rights Reserved


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
in
This dissertation for the Doctor of Philosophy degree by Dayna Bowen Matthew has been approved for the Health and Behavioral Sciences Program by
Ronica Rooks Sharon Devine Patrick Krueger Yvonne Keller-Guenthar Tillman Farley
Date: May 12, 2018


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
IV
Matthew, Dayna Bowen (Ph.D., Health and Behavioral Sciences Program)
Medical-Legal Partnerships: Reducing Health Disparities by Disrupting Cumulative Disadvantages Faced by Vulnerable Colorado Families Dissertation directed by Research Assistant Professor Sharon Devine
ABSTRACT
In Colorado, low-income and minority children live shorter and less healthy lives than wealthier and white children. Data on childhood asthma, obesity, and infant mortality disparities are disheartening. For example, although Colorado has the 5th lowest infant mortality rate nationally, Colorados black infants die nearly three times more often than the states white infants; Latino infants are 70% more likely to die during their first year than whites. Data from 2015 not only show the immunization rate for Colorado toddlers continued a troubling downward trend so that Colorado children rank 30th in the nation, but a research collaborative between the University of Colorado School of Medicine and Childrens Hospital Colorado reported that [significant disparities exist in vaccination delivery by age group race/ethnicity, and socioeconomic status.
This dissertation presents a study premised on the understanding that the inequitable distribution of access to important social determinants of health is an underlying cause of these health disparities among Colorado children. The study examines the impact of medical-legal partnerships (MLPs), an intervention designed to equalize access to preventive medical care, decent housing, and other social determinants in order to disrupt the fundamental association between poverty, minority race and ethnicity status, and poor health. MLPs are an innovative


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
v
health delivery model that integrate lawyers into clinical delivery teams to help equalize vulnerable patients access to the social determinants of health that are guaranteed by law.
The MLP studied here operated between 2009 and 2013 between Childrens Hospital of Colorado, Colorado Legal Services, and a law school program called the Colorado Health Equity Project. This study retrospectively analyzed quantitative medical and qualitative legal data to examine whether MLP services improved childrens health and social outcomes by reducing upstream social risk factors that could decrease health disparities over the life course.
Specifically, this study analyzed the MLPs impact on young childrens vaccination compliance and access to better neighborhood resources due to upward residential mobility. Although this MLP did not have a statistically significant impact on either outcome of interest, the methodology developed here suggested a relationship between MLPs and other health outcomes that warrants further study.
The form and content of this abstract are approved. I recommend its publication.
Approved: Sharon Devine


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
vi
DEDICATION
This dissertation is dedicated to my family: My beloved husband Thomas L. Matthew, M.D., my forever brother, Vincent E. Bowen, III, and my sons and daughters Sarah Griffin, Mark Benjamin, Thomas William, Marion Lewis, Erin Joy, and Amelia Cizwala each of whom has chosen to live a life of consequence that profoundly inspires me, and is destined to help many to persevere and triumph over adversity.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES vii
ACKNOWLEDGMENTS
I thank the University of Colorado COMIRB for approval of my study Protocol Number 14-1475. I also thank to my committee members for generously sharing their intellectual guidance and encouragement. I owe a special debt of thanks to Patrick Krueger for tirelessly reviewing prior drafts and research designs. However, my deepest gratitude is reserved for Sharon Devine, who exemplified excellence in legal scholarship, research integrity, and commitment to social justice; without these bright lights I would never have found my way.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES viii
TABLE OF CONTENTS
I. INTRODUCTION................................................................15
Health Disparities: Background and Significance............................. 16
Health Disparities In Colorado.............................................. 19
Colorados Poor And Minority Children .......................................21
Hypothesizing Medical-Legal Partnerships As A Solution To Health Disparities In Colorado.................................................................26
The Role Of Medical-Legal Partnerships...................................32
Research Question And Specific Aims......................................41
II. PUBLIC HEALTH LAW LITERATURE REVIEW.......................................466
Public Health Law Literature Review.......................................466
Medical-Legal Partnership Literature Review...............................500
Observational Studies In MLP Literature...................................544
III. THEORETICAL FRAMEWORK....................................................644
An Ecosocial Theory Of Health Inequality .................................644
Fundamental Social Causes Of Health Inequity...............................68
The Life-Course Cumulative Disadvantage Theory Of Health Inequity.........744
Conceptual Framework For This Study: The MLP As A Disruptive Protective
Resource...................................................................77
IV. Research Design ...........................................................88
Overview...................................................................88
Research Design ...........................................................89
Data ......................................................................90
Methods
912


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES ix
Treatment Group...............................................................922
Control Group.................................................................934
Pre-Treatment Variables .......................................................97
Outcome Variables ......................................................... 10909
Quantitative Data Analysis ...................................................119
Pre- / Post-Analysis Within Treatment Group...................................122
Qualitative and Mixed Methods Outcome Analysis ............................ 12323
V. RESULTS.......................................................................127
The Outcome Variables...................................................... 12727
Quantitative Results ...................................................... 12828
Outcome Means Comparison................................................... 12929
Estimated Average Treatment Effect of MLP.................................. 13232
Linear and Logistic Regression Analysis.................................... 13636
Pre-/Post-MLP Analysis Results ............................................ 14444
Qualitative And Mixed Methods Results........................................1500
VI. DISCUSSION................................................................. 15757
Limitations and Implications For Future Study
16462


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
x
LIST OF TABLES
TABLE
1. Frequency And Mean Comparisons Of Treatment Group And MLP Negative
Patient Groups ...................................................................95
2. Pre-Exposure Covariates For Propensity Score Matching..............................99
3. Comparison Of Matched And Un-Matched Group Of Treated And Control Patients ......106
4. Standardized Mean Differences: Matched And Un-Matched Groups...................... 108
5. Recommended Age Ranges for Vaccine Dose Administration........................... 112
6. Median Family Income Tiers for Patient Census Tracts..............................117
7. Upward Residential Mobility Outcome Variable By Patient Cohort....................118
8. Immunization Dosage Compliance Rate Variable .....................................127
9. Upward Residential Mobility Outcome Variable......................................128
10. Mean Comparison, Vaccine Dose Compliance .........................................130
11. Mean Comparison, Upward Residential Mobility......................................131
12. Means of Pre- and Post-MLP Compliance Rates by Age Groups.........................146
13. Frequency of Pre-MLP Residential Mobility for Treated Patients by Age Group.......147
14. Frequency of Post-MLP Residential Mobility for Treated patients by Age Group..... 147
15. Life Table: Pre- and Post MLP Immunization Compliance Hazard Rate for Treated
Patients .........................................................................149
16. Life Table: Pre- and Post-MLP Upward Residential Mobility Hazard Rate for Treated
Patients .........................................................................150
17. Statistically Significant Estimated Average Treatment Effects of MLP On Inpatient
Hospitalizations and Total Missed Appointments
159


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xi
18. Racial Composition of Patients Under Age 3 at First Visit to CHCO MLP
Clinics during Study Period...................................................... 193
19. Ethnicity of Patients Under Age 3 at First Visit to CHCO MLP Clinics During Study
Period ...........................................................................193
20. Distribution of Children Under Age 3 Included in Study Population................194
21. Age Distribution of Treated and Un-Treated Patients .............................194
22. Age Distribution of Treatment Group Children, by Cohort .........................195
23. Removing Variables with High Missing Values Increased Resulting Matches.......... 198
24. ATT of MLP on Immunization Compliance Rate After Removing Variables With
High Percentage of Missing Data...................................................199
25. ATT of MLP on Upward Residential Mobility After Removing Variables with High
Percentage of Missing Data .......................................................199


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xii
LIST OF FIGURES
1. Self-Reported Health Disparities In Colorado..................................1919
2. Colorado Health By Income, 2013 ...............................................200
3. Colorado Poverty And Deaths Due To Diabetes And Heart Disease .................200
4. Asthma Racial And Ethnic Health Disparities For Colorado Children .............222
5. Colorado Infant Deaths Due To Asthma By Race ...................................23
6. Colorado Infant Mortality By Race..............................................233
7. Colorado Infant Mortality By Race And Income, 2014 ........................... 244
8. Childhood Obesity Among Colorado Children By Income, 2008 .....................255
9. Limited And Inequitable Access To The Social Determinants Of Health Mediate
The Social Gradient Relationship Between Poverty/Race And Inferior Outcomes....30
10. Children's Hospital Of Colorado MLP A Disruptive Moderator...................311
11. Frequency And Type Of Problems Identified By Vulnerable Minnesotans, 2011 .....344
12. Inadequate Housing Among Households With Children By Race and
Annual Income, 2011 ............................................................37
13. Study Specific Aims Summarized......................Error! Bookmark not defined. 1
14. Specific Aims And Research Questions For Each Phase Of Study .............444
15. Summary Of Major Observational Studies Of MLP Impact .........................5454
16. Racism As A Fundamental Cause Of Health Inequality, Phelan & Link (2015)........72
17. Ben-Shlomo & Kuh Biological Psycho-Social Pathways To Disease Disparities.......78
18. Conceptual Framework: MLP As Protective Resource To Disrupt Pathways To
Health .........................................................................80
19. Three Phases Of This Study.....................................................900


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xiii
20. Inclusion Criteria For Children's Hospital Colorado (CHOC) MLP Treatment Group
Patient Population, January 2009 December 2013 ...................................94
21. Colorado And USA Immunization Rates Rising But Remain Below Healthy
People 2020 Objective .............................................................Ill
22. Testing The MLP As Moderator Of The Social Gradient Relationship Between Race,
Low SES, And Inferior Outcomes.....................................................120
23. Linear and Logistic Regression Variables ...................................... 12222
24. Qualitative Data Analysis Approach Based On Miles, Huberman, Saldana (2014).... 12424
25. Low- To High-Intensity Legal Services Within The CLS Case Cohort Of Treatment
Group..............................................................................125
26. Summary of Average Treatment Effect Results (ATT and ATE) on Outcome
Variables .........................................................................136
27. Odds Ratio for High Upward Residential Mobility For MLP Patients............... 14242
28. Logistic Regression Results For High Residential Mobility Outcome Variable..... 14343
29. Linear Regression Results For Association Between Compliance Rate And High
Intensity MLP Services.............................................................143
30. Logistic Regression Results For Association Between High Upward Residential
Mobility And High Intensity MLP Services...........................................144
31. Frequency Of Pre- And Post-MLP Immunization Compliance For Treated Patients.... 14545
32. Comparison Of Pre- And Post-MLP Treatment Frequency Of Upward Residential
Mobility Moves ....................................................................148
33. Race, Ethnicity, And Gender Of 92 CLS Case Patients (B=Black, H=Hispanic,
W/H=White/Hispanic, 0=Other, 0/H=Other/Hispanic, W=White, A=Asian
151


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xiv
34. Types Of Legal Problems, 92 CLS Case Patients (MED=Medical Insurance,
INC+Income/Public Benefits, FAM=Family Law, HSE=Housing Law...................152
35. Types Of Legal Services And Legal Outcomes For 92 CLS Patients (C/E=
Counseling And Education; LTR=Letter On Client's Behalf; APP=Appeal Of Adverse Decision; PER=Personal Court/Administrative Appearance ...................153
36. Intensity of Legal Services and Outcomes Among 32 CLS Case Patients By Race
And Ethnicity..................................................................154
37. Legal Outcomes by Race/Ethnicity for 92 CLS Case Patients.................. 15555
38. Immunization Compliance (Red) And Upward Residential Mobility (Blue) By MLP
Service Intensity For 92 CLS Case Patients ...................................156


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
15
CHAPTER I INTRODUCTION
Eliminating racial, ethnic, and socioeconomic health disparities remains one of the nations top medical, scientific, and political priorities. Yet in Colorado, as in the nation overall, health equity for all regardless of income, race, or ethnicity remains a tragically elusive goal notwithstanding commitment from the highest levels. Existing efforts to combat health disparities are falling short. The United States Department of Health and Human Services Agency for Healthcare Research and Quality (AHRQ) released a report in 2012 showing that the fight against racial health disparities has stalled. Between 2002 and 2010, over 80% of all measures comparing the quality of medical care that whites and minorities receive in this nation have remained unchanged. The data show even less progress in narrowing the gap for measures of disparate access to care. Between 2002 and 2009, some access gaps between white and Asian-Americans narrowed, but for all other groups white patients continued to enjoy consistently better access to health care than minorities (United States Department of Health and Human Services, 2012). Similarly, researchers have shown only limited progress in reducing socioeconomic disparities. For example, Paula Braveman presented comprehensive data related to socioeconomic disparities in the United States in 2010; Braveman reported that for 11 adult and child health indicators, the least educated and lowest income members of American population suffered the worst health outcomes, and improvements in health were generally seen at higher income levels (Braveman, Cubbin, Egerter, Williams, & Pamuk, 2010). This study examines the impact of an intervention designed to disrupt this fundamental association described by the social gradient.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
16
Medical-Legal Partnerships (MLPs) are an innovative health delivery model that integrates lawyers as part of the health delivery team to help equalize access to the social determinants of health. MLP lawyers and health care providers collaboratively deliver medical and legal services in the clinical setting in order to mitigate the legal barriers to the fundamental determinants of healthy living that poor and minority patients face every day. MLP attorneys combat discriminatory practices in housing and employment, argue to reverse erroneous public benefit denials, enforce disability laws that grant access to quality education, and assist patients in a variety of legal actions that enforce existing legal rights to access the social determinants. Because law plays a central role in organizing, regulating, and distributing both the medical and social determinants of health, this study examines whether MLPs can help reduce disparities and increase health equity.
Health Disparities: Background and Significance
Two main reasons underlie the persistence of health inequity in America. Lirst, the United States dedicates few resources to ensuring that all populations have a healthy social environment in which to live, work, and play. Populationsespecially vulnerable onesneed much more than access to good health care to be healthy. Indeed, while social scientists recognize five factors that may contribute to health outcomes (McGovern, Miller, & Hughes-Cromwick, 2014), the social factors are the most important (Tarlov, 1999). These social factors include the social environment (e.g. income, gender, disability, and race discrimination, transportation and food access), the physical environment (neighborhood and housing conditions), (Heiman & Artiga, 2015) and access to health care; they have at least as strong an association with poor health outcomes as genetics, biology, or effects of individual behaviors related to diet, smoking, exercise, or alcohol consumption (Lorchuk, Dickins, & Corring, 2016).


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
17
Yet the United States spends far less on ensuring adequate access to the social needs of populations than most other developed nations. A recent study published by the RAND Corporation importantly confirmed that countries with greater social expenditure have better health outcomes than countries that spend relatively more on medical care. (Rubin, et al., 2016). Moreover, the association between higher social spending and better health outcomes holds within the United States as well. Elizabeth Bradley and colleagues found that states with a higher ratio of social and public health spending (for housing, food and income benefits, etc.) to healthcare spending (for Medicare and Medicaid reimbursement) had significantly better health outcomes for adult obesity, asthma, mentally unhealthy days, days with activity limitations, and mortality rates for lung cancer, heart attack, and Type 2 diabetes (Bradley, et al., 2016). My study proceeds from the understanding that eradicating health disparities depends principally upon ensuring access to the social determinants of health.
The second major contributor to the prevalence and persistence of health disparities is inequality. Specifically, social inequality that distorts access to the social determinants of health may in fact be the most powerful predictor of disparate mortality and morbidity than any other single or combined set of known factors (Marmot & Brunner, 2005). Since the Whitehall Studies (Marmot & Brunner, 2005) convincingly documented evidence of the near uniformly inverse relationship between health, health behaviors, and social class in England, knowledge about the social gradient has shed considerable light on the crucial role that an inequitable distribution of the social determinants of health plays in producing health disparities. The social gradient describes the relationship between the circumstances in which people live, play, and work and their poor health outcomes (Marmot & Brunner, 2005). These social determinants of health are essentially the causes of the causes (Institute of Medicine, 2011) of health disparities. As such,


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
18
they are an indispensable component to improving population health outcomes and reducing health disparities (Davey Smith, 1998). To the extent that the social determinants are inequitably organized and distributed, poor or minority communities suffer decreased opportunities to be healthy.
The Whitehall Studies (Marmot & Brunner, 2005) revealed not only the fact that people with lower socioeconomic status (SES) bear at least twice the risk of shorter life expectancy and disease morbidity than people with higher SES, but also that universal access to health care such as the British have enjoyed since the end of World War II did not disrupt the social gradient. Instead, the Whitehall studies confirmed that conditions at work, home, and in communities primarily account for the social gradient (Marmot & Brunner, 2005). Considerable evidence has further demonstrated that low income and minority populations suffer inferior health outcomes because they disproportionately lack access to decent housing, safe recreational spaces, quality education, fair employment, income supports, and food security (Woolf and Braveman, 2011). Inequality constructs the environments in which poor and minority people live, work, and play. Moreover, these social inequalities also adversely influence health behaviors so that individuals living in poverty and racial segregation are more likely than others to smoke, eat poorly, consume alcohol excessively, develop drug dependence, and lead sedentary lives that result in inferior health outcomes (Pampel & Krueger, 2010). Based on the association between inequitable distribution of the social determinants and poor health outcomes, this study posits that improved distribution of the positive social determinants will also reduce health disparities. It tests this theory using data from a patient population in Colorado, where there is substantial evidence of health inequity.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
19
Health Disparities In Colorado
Low income and minority Coloradans do not have the same access to the social determinants of health as wealthier and white Coloradans. Therefore, the cEsadv ant aged experience inferior health outcomes. The Colorado Commission on Affordable Health Care released a report in March 2016 that identified self-reported health differences among Colorado adults by income, race, and sexual orientation. Figure 1 summarizes the data from that report, which confirms that white, wealthier patients report better health than low-income and minority patients.
Colorado Adults who Report Good to Excellent Health by Household Income, Race/Ethnicity and Sexual Orientation
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Figure 1. Self-Reported Health Disparities In Colorado Income disparities perpetuate health disparities by ensuring an unequal distribution of the social determinants of health Figure 2 shows that household income is closely associated with the quality of physical, mental, and oral health that Coloradans report.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
20
Self-reported fair or poor health status
General Oral Mental1 General Oral Mental1
At or below 200. FPL^H Above 200% FPL
Figure 2. Colorado Health By Income, 2013
Figure 3 shows that place matters. Average death rates due to diabetes on the left and heart disease on the right are shown for Colorados highest (red) and lowest (blue) poverty counties in Colorado. Poor people die at a higher rate than their well-off counterparts who have the same disease. These data show that beyond subjective self-reports, objective data also reveal the inequities that affect all social determinants of health.
Average Death Rate Per 100,000 for Counties with the Highest and Lowest Poverty Rates
Lowest Poverty Counties Diabetes
Highest Poverty Counties Coronary Disease
Figure 3. Colorado Poverty And Deaths Due To Diabetes And Heart Disease


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
21
Thus socioeconomic, racial, and ethnic health disparities in Colorado remain a pervasive and persistent public health problem.
Colorados Poor And Minority Children
Poverty is a key determinant of poor health in children, and Colorado has one of the fastest growing child poverty rates in the nation (Office of Health Disparities, 2009) Between 2000 and 2011, the poverty rate in Colorado increased from 10% to 18%, leaving more than 1 in 6 Colorado children living in poverty (Colorado Children's Campaign, 2013). Denver County has the second highest poverty level in the state with 26.2% of all children living in poverty. In 2016, KIDS COUNT Colorado (Colorado Children's Campaign, 2016) reported that despite the good news that Colorados overall child poverty rate of 15% had declined for two consecutive years, significant racial, ethnic, and economic disparities persist. In Colorado, only 8% of white children live in poverty, while 27% of Latino children and 31 % of black children are poor, according to the Colorado Department of Public Health and Environment. Predictably, low-income and minority children in Colorado are less healthy than wealthier and white children. Black and Latino families in Colorado earn substantially less than white and Asian families. As a result, Colorado social indicators confirm that Colorado children from low-income, black, and Hispanic families suffer disparate access to the social determinants of health. In Colorado and in Denver County, the regional home to the majority of patients in this study, food insecurity and nutrition data further demonstrate health disparities among Coloradans. According to the Colorado Department of Public Health and Environment (CDPHE), 20.4% of Colorados white children ages 1-14 sometimes relied on low cost foods during the past year, while 57.6% of American Indian/Native Alaskan children, 49.0% of Latino children, and 34.4% of African-


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
22
American children consume low-cost foods. These inequities are directly related to vulnerable childrens disparate health outcomes.
Colorados health disparities data on asthma, obesity, and infant mortality provide evidence of a fundamental association between poverty, race, and poor health for Colorados children. During the period from 2006 2007, Figure 4 shows the asthma prevalence was 12.5% for black children, 7.7% for white children, and 7% for Hispanic children. During the same period, death rates among persons with current asthma were higher for blacks (3.4 deaths per 100,000 persons) than whites (1.9 deaths per 100,000) (Jaeobellis et al, 2008).
30% 25% 20% 8 15%
£ 10%
5%
0%
Prevalence of Asthma by Race/Ethnicity, Colorado Children Ages 1-14, 2006-2007
Current
White Non- Black Non- White Hispanic
Hispanic Hispanic
Race/Ethnicity
Source: Child Health Survey, Health Statistics Section, Colorado Department of Public Health and Environment
Figure 4. Asthma Racial And Ethnic Health Disparities For Colorado Children
The incidence of chronic diseases such as asthma disproportionately impacts the health of Coloradans of color as compared to whites in the state, as shown in Figure 5.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
23
** i----------------------------near
Tufl LMtne Mad Vmr. ldiM
Source: Behavior Risk Fortor Surveillance S>em, Health Statistics Section, Colorado Department oj Public Health and Environment
Figure 5. Colorado Infant Deaths Due To Asthma By Race Colorados overall infant mortality data are particularly disturbing. Although Colorado has the 5th lowest infant mortality rate in the nation, Colorados black infants die three times more often than the states white infants. As shown in Figure 6, twelve in every 1,000 live births to African American women end in death. Latino infants are 70% more likely to die than whites. (Wilcox, 2016).
Infant Mortality Rates by Race and Ethnicity, Since 1979
Figure 6. Colorado Infant Mortality By Race
Importantly, socioeconomic status does not fully explain these disparate outcomes as demonstrated by the differences in birth outcomes that black and white mothers at different


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
24
income levels experience in Colorado. Figure 7 shows that unlike other health disparities, African-American infant mortality gaps do not narrow with improved socioeconomic status. Black mothers who earn $50,000 $75,000 per year suffer the same infant mortality rate as black families earning less than $15,000 annually. Middle income blacks have infant mortality rates that are twice the rate of white families below the poverty line and over 4 times as high as white middle-income families.
Infant Mortality Rates by Race and Income per 1,000 Live Births
'5!
13
BLACK 7 LATINO 7 WHITE
less than $i5,ooo/yr
Sfc* Udi (kp 12
BLACK 6
LATINO 4 WHITE
$50,ooo-75.ooo/yr
Rocky Mountain PBS I Rocky Mountain PBS
The infant mortality rate is considered one of the most important indicators of the general health of any community. It tracks the number of infants who die before their irst birthday, per 1,000 live births. The infant mortality rate in Colorado is 5th lowest overall among the states. But within that ranking is a striking racial disparity. The rate of infant death for black babies is triple that of white babies in the state. (Ryan Corely/Rocky Mountain PBS)
Figure 7. Colorado Infant Mortality By Race And Income, 2014
Obesity data similarly confirm that health disparities remain prevalent in Colorado; these data also portend continued health disparities over the life couise for racial and ethnic minorities. As seen in Figure 8, the percent of Colorado children who are obese and therefore exposed to higher rates of disease follows the familiar pattern of the social gradient in which children of low-


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARJTIES
25
income parents suffer greater rates of obesity than children of wealthier, better-educated families (Graham. 2008).
V* of*
i A
J2 (N >0 P
Source: Graham 2008
Figure 8. Childhood Obesity Among Colorado Children By Income. 2008 Morbidity and mortality among poor and minority children outpace wealthier, non-minority children at alarming rates in virtually every leading disease category, providing further evidence of the social gradient.
Sadly, race and economic disparities plague Coloradans access to health care as well. In 2015. the immunization rate for Colorado toddlers continued a troubling downward trend, dropping significantly so that Colorado children rank 30th in the nation (Colorado Health Foundation. 2015). In 2011. 75.8% of parents got recommended immunizations for children between 19 and 35 months old. but in 2013 only 69.2% of toddleis completed recommended immunizations at 35 months of age. down from 80.3% in 2007. This decline does not bode well for the long-term health of Coloradans generally or low-income and minority Coloradans specifically. Indeed, a research collaborative between the University of Colorado School of Medicine and Childrens Hospital Colorado reports that [significant disparities exist in vaccination delivery by age group (e.g.. Child vs. Adolescent vs. Adult), race/ethnicity. and


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
26
socioeconomic status (ACCORDS, 2016). On a national level, evidence of racial and socioeconomic disparities in vaccine coverage led the CDC to implement the Vaccines for Children Program (VFC) in 1994. Colorados VFC program leverages federal funding to provide low- or no-cost vaccines for low-income children. Colorado law requires documented immunization compliance for all children attending kindergarten. Nevertheless, the best evidence is that although racial and ethnic disparities in immunization compliance have narrowed, gaps persist in Colorado despite the success of the VCF program(Colorado Health Institute, 2005); these gaps predictably contribute to health disparities in Colorado as in the nation.
Hypothesizing Medical-Legal Partnerships As A Solution To Health Disparities In Colorado
The MLP approach is founded on the premise that addressing inequities in poor and minority patients incomes, housing, educational attainment, personal safety, and other social determinants will improve health outcomes. Today nearly 300 MLPs operate in 155 hospitals,
139 health centers, and 34 health schools across the country. (National Medical Legal Partnership Center, 2017) Often the inequitable distribution of social determinants is due to legal problems or problems such as housing code violations or improper termination of benefits. In these cases, it is fair to say that unmet legal needs are a social determinant of health. MLPs are designed to bridge the gap between the need to address unmet legal issues that exacerbate patients health problems and assist the medical provider who is treating patient health problems but does not have the time or resources to include solving legal problems in the care model. MLPs train clinicians to regularly screen patients for health-harming legal needs and work through in-clinic attorneys to address those needs preventively before they become serious. In


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
27
contrast, providers in most traditional health care settings that do not screen for unmet legal needs discover a patients health-harming, social problems only once the patient is in crisis (e.g., evicted, without heat, suffering elevated blood lead levels), if at all. Even when the astute physician practicing in a traditional setting suggests that a patient consult an attorney outside the clinic, vulnerable patients are unlikely to get the help they need. This is because without any prior relationship, a low-income patient seeking help from an attorney who is unaware of the health impacts caused by the legal crisis, and quite likely under-resourced, could be slow to resolve the legal issue, and the associated adverse impacts on patient health may be prolonged. MLP attorneys provide three core services. First, MLP lawyers provide legal representation to address adverse social conditions for which there are legal remedies, and which have the potential to improve patient health. Examples include requiring landlords to remove lead paint, toxins or mold, appealing wrongful public benefit terminations, and enforcing educational accommodations for disabled children. Second, MLPs reform health and legal institutional practices by training clinical providers to screen for and identify patients social and legal needs during office visits. The goal is to identify these needs while they may be addressed preventively in the same way that physicians seek to provide preventive rather than crisis medical care. Third, MLPs advocate for structural policy changes at an institutional, local, state, and federal level. MLP attorneys bring a patient-to-policy perspective, identifying needs in the communities they serve and then working to improve policies and laws that impact those communities and ultimately the social determinants of individual and population health.
Colorados first MLP began operating in 2009 at the Childrens Hospital of Colorado (CHCO). This MLP operated in two CHCO clinics (the CHCO MLP). A total of 31,844 pediatric patients visited the two CHCO clinics from April 2009 to December 2013. The CHCO


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
28
MLP attorneys used a brief survey to screen approximately 7,000 children of those 31,844 children from all age groups to identify those who screened positive because they identified an MLP need and requested contact from an MLP representative. The CHCO MLP served all patients who screened positive for need and of this cohort that screened positive, Colorado Legal Services (CLS) opened legal case files on 150 patients. The CHCO MLP provided legal services to patients at Childrens Hospital for five years before it closed in December 2013.1 worked with the Colorado Legal Services attorneys and CHCO MLP staff during its last year before it closed. In the following year, with the help of CLS attorneys, I co-founded the Colorado Health Equity Project (CHEP) to continue building MLPs in Colorado. CHEPs mission is to remove legal barriers to better health outcomes for low-income Coloradans by forming medical-legal partnerships. In 2013 CHEP opened two new MLPs in order to expand this collaborative work and fulfill the three core components of the MLP model and for one year worked with CLS on CHCO patient cases.
CHEP introduced students from the Colorado School of Public Health and the University of Colorado Law School to work with pro bono private attorneys as well as the CLS lawyers in serving vulnerable patients. CHEP opened MLPs at the Colorado Center for Refugee Wellness and the Salud Lamily Health Center in Commerce City. During its first year of operation, CHEP MLPs provided legal representation to six families, introduced training to frontline providers in two safety-net health clinics, and provided public policy advocacy on several legislative initiatives during 2014. A CHEP student again worked on health equity advocacy in the 2016 Colorado General Assembly session. Moreover, MLPs initiated by CHEP served an additional 8 patients in 2015 and screened approximately 200 patients in 2016. In 2017, CHEP represented patients from the Globeville and Elyria-Swansea neighborhoods in anti-pollution litigation


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
29
related to the Interstate 70 highway expansion. CHEP has been approached by at least two other major health providers to form new MLPs in collaboration with public and community hospital systems. The point of recounting CHEPs success is twofold. First, CHEPs good work addresses but a fraction of the need for health-related legal services in Colorados poor and minority patient populations. Second, despite my first-hand experience with the CHCO and CHEP MLPs, neither the real impact nor the optimal mix of MLP services is well understood in Colorado or in the nation. Therefore, this study addresses an academic gap in the knowledge of how well MLPs work and also explores an incomplete pragmatic understanding of what type of legal services have the greatest impact on helping vulnerable populations. This study has the potential to have tangible impact on the health and social outcomes that poor and minority Coloradans experience by improving the knowledge base required to expand and better structure medical-legal partnerships that provide services that demonstrably improve patient outcomes throughout the state.
Figure 9 provides a graphic summary of my conceptual framework for understanding the relationship between poverty, race, and poor outcomes. In this figure, limited and inequitable distribution of the social determinants of health explain (or mediate) the social gradient relationship between poverty and race on the left, and inferior health outcomes on the right (Baron & Kenny, 1986).


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
30
Figure 9. Limited And Inequitable Access To The Social Determinants Of Health Mediate The Social Gradient Relationship Between Poverty/Race And Inferior Outcomes Theoretically, as MLPs improve access to the social determinants, they improve access to healthy food, clean and decent housing, full educational opportunities, violence-free homes and other social determinants (Regenstein et al., 2018). Thus, MLPs may also improve the opportunity to experience good health by disrupting the social gradient. Figure 10 illustrates the MLP role this study analyzes; I ask whether MLPs can weaken the relationship between poverty and race (independent variables) and poor health or social outcomes (dependent variables). Put another way, discussed further in Chapter 3, this study examines the moderator effect MLPs may have on the social gradient.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
31
Predictors:
Poverty, Race and Ethnicity
Moderator:
CHCO Medical Legal Partnership (Disrupting Cumulative disadvantages that limit access to health care, housing, income,)and Family Stability)
Interactive Terms:
Race x MLP Low SES x MLP
Inferior Health and Social Outcomes (Immunization Compliance Rate
Figure 10. Children's Hospital Of Colorado MLP A Disruptive Moderator Some researchers have attempted to identify the precise causal mechanisms that link specific social determinants to poor health outcomes. Several studies have connected poverty and mortality (Brooks-Gunn & Duncan, 1997). Others identify a strong relationship between income inequality and mental as well as physical health disparities (Marmot, 2005). A number of studies have identified poor housing conditions (Krieger & Higgins, 2002), stressful work environments (Siegrist, 1996), and educational disparities (Zimmerman, Woolf, & Haley, 2015) as social mechanisms that have a deleterious impact on health. Importantly, poor social conditions such as low income (Sorensen, Barbeau, Hunt, & Emmons, 2004; Barbeau, Krieger, & Soobader, 2004) and low educational attainment have been tied to resulting poor health behaviors such as


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
32
smoking, alcohol consumption, and drug use. The exact extent of the causal relationship between these social risks is a matter of debate.
In this study, I posit that in order to disrupt the association between higher mortality and morbidity on the one hand, and poverty, race, and ethnicity on the other hand, effective health interventions must address inequality that affects both upstream and downstream social causes of health disparities. The conceptual framework for this study assumes that a comprehensive approach to eradicating disparities addresses the fundamental inequality that characterizes downstream social determinants such as access to preventive medical care and also the inequality that affects upstream social determinants of health like housing and education. Because law is a particularly useful tool for addressing social inequality broadly, this study posits that low-income populations require legal representation to address the inequality that characterizes all social determinants of health, which stand as a barrier to good health itself. In other words, this study identifies un-met legal needs as a social determinant of health that, once addressed, can improve health outcomes for poor and minority families. My mixed-method approach examines the health-related impacts of a legal intervention aimed at equalizing access to two social determinants of health regular preventive medical care and improved neighborhood resources -for vulnerable Colorado families.
The Role Of Medical-Legal Partnerships
Many existing laws guarantee patients access to social determinants. However, often the laws are unenforced or enforced inequitably. Therefore, vulnerable patients can face an impossible situation in which they require legal assistance to gain access to social supports essential for healthy living but cannot find or afford an attorney to represent their cause. The evidence is that millions of Americans are in this position (Legal Services Corporation, 2017).


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
33
Families need legal help fighting unlawful eviction and compelling landlords to fix building code violations that produce unhealthy substandard living conditions. Victims of domestic violence need representation in court to obtain protective orders and to preserve safety for their homes and children. Workers and indeed entire communities may be exposed to excessive toxins or other environmental health hazards unless an attorney intervenes. In these and many other circumstances, limited access to legal services disproportionately harms the health of vulnerable populations that have the greatest social needs and the fewest resources to address them. For these families, the MLP model integrates legal assistance to address their unmet legal needs as a vital component of patient health care.
The need for MLP representation is particularly acute among the poor. Studies consistently show that low income people have significantly more unresolved civil legal problems than higher income people and that low-income people are less likely to obtain legal assistance for their problems (Greene, 2016). As a result, a high proportion of the legal needs related to housing, family, and consumer issues that low-income families face goes unaddressed (Scherer, 2016). While evidence of the relationship between unmet legal needs and health problems might be found in every state, Minnesotas state bar association has gathered exemplary and detailed data. A qualitative study of Minnesotas low-income population shows the intersection between physical and mental disability and these unresolved legal issues in data that are representative of populations around the country. These frequency data have been uniquely collected in Minnesota, though the data are shown in Figure 11 below as representative of what might be seen in Colorado and nationally based on the literature, and experience I have had representing MLP clients. Housing and healthcare problems were the leading problems that poor people with physical or mental disabilities suffered, and the vast majority of all problems


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
34
they identified would likely require legal services to resolve (Minnesota State Bar Association, 2011).
Figure 11. Frequency And Type Of Problems Identified By Vulnerable Minnesotans, 2011
Similarly, a survey of 600 Colorado patients, conducted to determine the need for a medical-legal partnership in Denver, confirmed similar problems are prevalent in this part of the state. A majority of low-income, Colorado patients surveyed (66%) indicated that they had a legal problem over the past 12 months that adversely affected their health (Miller and Ayers, 2013). These patient responses reflect the fact that unmet legal needs are a social determinant of poor health (Tyler, 2012) and potentially a mechanism that influences the cumulative disadvantages that lead to disease risks over the life course in Colorados low-income and minority children. Un-represented (or pro se) litigants in civil matters suffer much worse outcomes than those with legal representation (Engler, 2009) and are therefore most vulnerable to the health effects of unmet legal needs. For these people, the fact that a law exists to protect their rights is meaningless without an attorney to help enforce the law on their behalf. Colloquially, MLPs are often said to improve vulnerable patents health by making the law on the streets conform to the law on the books. More formally, the difference between the civil legal needs of low-


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
35
income people in the state and the availability of attorneys to address those needs is called a justice gap. The justice gap in Colorado is considerable. Recent statistics show that in Colorado domestic relations cases approximately 75% of litigants are unrepresented. In Colorado county courts, which handle cases common to low-income families such as housing evictions and financial collection cases, approximately 98% of people are unrepresented. Colorado Legal Services, the federally funded pro bono legal services organization for the entire state, has only 51 lawyers. Thus, given that approximately 11% of Colorados 5.5 million people live in poverty, the ratio of available Colorado-licensed lawyers to all Coloradans is more than 1:200, while the ratio of available lawyers to low income Coloradans is a dismal 1:11,800 (Colorado Supreme Court Library, 2016). These Coloradans, like most low-income Americans, have a broad range of civil legal problems (Legal Services Corporation, 2017) that have a deleterious effect on health. Limited access to attorneys who can help address them (National Center for Access to Justice, 2016) contributes to their inferior health status.
Lormer HHS Secretary Sylvia Mathews Burwell explained, civil legal aid ensures that more Americans have access to good nutrition, safe housing and other basic human necessities that are essential to overall health and well-being (White House Legal Aid Interagency Roundtable, 2016). Scholars also have begun to recognize that the inequitable enforcement of laws that controls social determinants is a mechanism through which social structures contribute to the health-harming disadvantages experienced by low-income and minority populations (Burris, Kawachi, & Sarat, 2002). Therefore, when an MLP gives patients from vulnerable communities legal representation, the MLP itself can serve as mechanism to equalize inequitable social structures that are then translated into more equitable levels and distributions of health (Burris, 2001). Housing provides an illustrative example.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
36
Numerous studies associate poor housing conditions with a broad range of health problems including asthma, lead poisoning, developmental delays, heart disease, and neurological disorders (Desmond & Bell, 2015). More than one researcher has called the state of inadequate housing in poor American communities a public health crisis (Bashir, 2002). Two million people in the U.S. occupy homes with severe physical problems while 4.8 million occupy homes with moderate physical problems (Krieger & Higgins, 2002). Annually, 13.5 million non-fatal injuries occur in and around U.S. homes; 2,900 people die in house fires, and two million people visit the emergency room for asthma related illness. One million children in the U.S. have blood lead levels high enough to adversely affect their intelligence, development, and behavior. In Colorado, nearly one in three children lives in families who cannot afford decent housing (Colorado Childrens Campaign, 2016). Statewide, the number of Colorado children who were homeless in 2014 increased to over 24,600 (Robles, 2016). Figure 12 demonstrates the fact that low-income and minority children are the populations most vulnerable to the health hazards that result from substandard, unaffordable, or overcrowded housing.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
37
Less Than 20K-40K 40K-S0K 60K-80K 80K-100K AI WMc Black Hispanic
20K Households
Source: United States Census Bureau. 2011. American Housing Survey National and Metropolitan Public Use File.
Figure 12. Inadequate Housing Among Households With Children By Race And Annual Income,
2011
An abundant body of local, state, and federal laws organize the rights vulnerable families have to clean and safe housing, free from mold, pest infestation, toxins such as lead paint or radon, and adequate heat in cold weather. These laws include the Fair Housing Act, which prevents discrimination, state building codes, which require safe and sanitary housing, and landlord-tenant laws, which equalize the power relationships between poor renters and their wealthier landlords. However, laws that regulate housing safety and sanitation issues tend to be under-enforced in low-income communities (Ross, 1993). Substantial evidence shows that legal representation improves housing code enforcement (Ortiz, 2014) and further, that code enforcement can improve health. The majority of tenants in housing courts nationally do not have legal representation, while most landlords have attorneys (Engler, 2009). There is a shortage of attorneys to help a poor asthmatic childs family with housing problems by enforcing building sanitation codes, landlord-tenant laws, anti-discrimination regulations, and improving fair access to public benefits that reduce homelessness. As a result, an asthmatic childs health outcomes may remain poor despite access to medical care simply because the family lacks access to legal


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
38
help that could improve the family home environment, one of the most important social determinants of health.
Still, health care providers seldom regard legal services as an important part of the health delivery model. Moreover, health policy experts rarely regard legal representation as a tool to fight health disparities. MLPs seek to transform fragmented and siloed approaches to health disparities, which separate access to legal services from access to health care delivery. The existing approach works to the detriment of low-income and minority patients who disproportionately suffer inequitable access to health care, legal services, and the social determinants of health. MLPs seek to deliver integrated medical care and legal representation to vulnerable patient populations as a way to break the chain of disadvantage and cumulative inequality that leads to health inequity.
Other researchers have posited an association between the MLP interventions ability to solve patients legal problems and improved patient health. However, generally, these studies have involved small numbers of patients, a single disease or condition, or one type of MLP service. Moreover, previous estimated associations fail to control for the possibility that commonly shared personal, family, and structural circumstances may also confound the relationship between MLP services and improved health and social outcomes. Therefore, the extent to which statistical associations have identified a cause and effect relationship between MLPs and better health outcomes remains unclear. Admittedly, the randomized clinical experiment is could be done prospectively, but the CHCO MLP did not design an experiment to test causation. Therefore, using the observational data collected for treatment purposes, I conducted a retrospective analysis to analyze the MLPs impact. I have chosen a statistical method that reduces confounding bias, compares treated to comparison group populations, and in


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
39
this study involves over 2,000 participants who presented with a broad range of medical conditions and received a spectrum of MLP services. I have used the propensity scoring method to estimate the causal relationship between MLP services as a treatment variable, and two outcomes childrens immunization compliance and upward residential mobility. Both of these outcome variables were refined during the project.
Originally, I considered a dichotomous variable called immunization completion, but on the recommendation of my committee, I changed this to immunization compliance to examine a continuous variable over time. Also, I originally considered residential instability, counting all moves as a measure of instability, but changed to upward residential mobility to minimize confounding bias, conform to the theoretical literature and better reflect an outcome relevant to the very young children in this study. I elected not to analyze residential instability primarily because it is a measure that affects older children whose social networks, school relationships, and established routines are disrupted by frequent moves. These are factors less relevant to children under age three. However, upward residential mobility has the potential to expose younger children to improved resources that could affect their health and social outcomes over time.
This project takes a sequential, Qual-QUANT-Qual-Quant approach. First, I interviewed practitioners and patients who participated in the CHCO MLP in order to understand the structure and scope of the MLP intervention. Next, I collected and retrospectively reviewed data from CHCO medical records to quantitatively analyze the medical-legal partnerships treatment and moderator effects on the social determinants of health and social outcome variables of interest in children between birth and three years of age. Next, I explored the documentary data contained in the legal files for those CHCO patients who received legal representation from


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
40
Colorado Legal Services (CLS) attorneys as a result of the MLP intervention. Finally, I integrated the quantitative health and social outcomes data with the qualitative data for patients who received MLP services to explore whether the intensity of MLP services is associated with the quality or quantity of either the legal, social, or health outcomes of patients in this study.
I selected the two dependent variables of interest in this study because their association with health outcomes is well-established. Each has been shown to contribute to cumulative adversity or advantage over the life course. Additionally, MLPs have the potential to impact these outcomes positively. Existing evidence shows that improving vaccination coverage (Aungst, 2011) and residential stability (Bures, 2003) during childhood can improve physical and mental health into adulthood. At the same time, upward social mobility can constrain social inequality over the life course. As residential changes improve parental socioeconomic position, children may move to neighborhoods where they might experience the cumulative advantages of improved access to resources available to higher socioeconomic groups (Blane, 2006).
I hypothesize that MLPs can improve vaccination compliance by solving legal problems that stand in the way of families taking advantage of available preventive medical care. For example, the CHCO MLP addressed health coverage issues such as wrongful Medicaid or Social Security Disability Insurance (SSDI) termination, financial issues such as wrongful denials of Supplemental Nutrition Assistance Program (SNAP), Temporary Assistance for Needy Families (TANF), or Supplemental Security Income (SSI) benefits, and immigration or asylum issues that limited patients access to public benefits and regular preventive medical care. I further hypothesize that MLPs can improve upward residential mobility by reducing the number of moves that result from legal problems with housing such as eviction, foreclosure, sanitation violations, racial or ethnic discrimination, and overcrowding, while increasing the moves to areas


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
41
of greater economic opportunity by stabilizing employment, improving income, and reducing the stress of unsolved legal issue. Thus, my study assumes that showing improvement in immunization compliance or upward residential mobility variables would suggest medical legal partnerships could have beneficial long-term health impacts and may reduce the cumulative adversity that results in health disparities. In summary, my study tests the theory that medical legal partnerships act as a protective factor to improve access to the social determinants of health.
Research Question And Specific Aims
The overarching research question for this study is: Do MLP services improve the health and social outcomes that could reduce health disparities for low-income and minority children? The study has five specific aims, each shown in Figure 13 below with the analytical approach I took to meet the specific aim.
Specific Aim Analytical Approach
1 To determine whether access to MLP services is associated with better health and social outcomes than those experienced by children who do not receive MLP services Means comparison between treated and untreated cohorts
2 To estimate the average treatment effect of medical legal partnerships on immunization compliance and residential mobility Average treatment effect on treated and untreated cohorts
3 To determine whether introducing the MLP intervention improves the post-intervention measures of outcomes as compared to pre-intervention measures Pre-post intervention survival analysis within treated cohort
4 To examine whether the type and intensity of MLP legal services is related to the quality of legal outcomes that patients who also became MLP law firm clients experienced Mixed-method analysis
5 To determine whether the intensity of legal services that MLPs provide have an impact on either patients housing mobility or immunization compliance Mixed-method analysis
Figure 13 Study Specific Aims Summarized
In Colorado, as in the nation, socioeconomic and racial inequality limits access to the resources that families need to live healthy lives. MLPs across the nation are built on the belief that lawyers can reduce this inequality. But how and by how much specific legal services impact


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
42
health is not well understood. My study contributes to understanding whether the 290 MLPs that now operate nationally are just another well-intentioned but ineffective health reform effort or are, in fact, measurably contributing to improved access to the social determinants that allow vulnerable patient populations to experience improved outcomes that can reduce health disparities. This study proceeded in three phases; the specific aims and research questions associated with each are summarized below in Figure 14:


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
43
Specific Aim Study Phase
Phase I
Aim 1: To determine whether access to MLP services is associated with better health and social outcomes than those experienced by children who do not receive MLP services RQl(a) Are pediatric patients who receive MLP services more compliant with the Advisory Committee on Immunization Practices (ACIP)/CDC recommended immunization schedules than patients who do not receive MLP services?
RQl(b) Are pediatric patients who receive the MLP intervention more likely to experience upward residential mobility than children who do not receive the MLP intervention?
RQl(c) Does the MLP intervention moderate the social gradient relationship between poverty, race, and poor health outcomes?
Aim 2: To estimate the average treatment effect of medical legal partnerships on immunization compliance and residential mobility RQ2(a) What impact does the receipt of MLP services have on the odds of childrens immunization compliance before age 3?
RQ2(b) What impact does the receipt of MLP services have on the ability of a young childs family to move to a neighborhood with higher median family income?
Aim 3: To determine whether introducing the MLP intervention improves the post-intervention measures of outcomes as compared to pre-intervention measures among treated patients. RQ3(a): Do pediatric patients who receive MLP services show greater immunization compliance after the MLP intervention than they showed before the MLP intervention?
RQ3(b): Do pediatric patients who receive MLP services have greater upward residential mobility after the MLP intervention than before the MLP intervention?
Phase II
Aim 4: To explore whether the type and intensity of MLP legal services is related to the quality of legal outcomes that patients who also became MLP law firm clients experienced RQ4(a): What types of legal problems do MLPs address among children whose families receive MLP services as clients of CLS?
RQ4(b): What types of legal services do MLPs provide to children whose families receive MLP services through legal representation as CLS clients?
RQ4(c): What types of legal outcomes do children whose families receive MLP services through legal representation as CLS clients obtain?
RQ4(d): Do the types of legal problems, MLP legal services, or outcome that patients represented by CLS receive vary by race or ethnicity?
Phase III


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
44
Aim 5: To determine whether the intensity of legal services that MLPs provide has an impact on either patients housing mobility or immunization compliance
RQ5: Does immunization compliance or residential mobility vary among children whose families receive CLS representation depending upon the intensity of the MLP legal services provided?
Figure 14 Specific Aims And Research Questions For Each Phase Of Study
A recent literature review conducted by the National Center of Medical Legal Partnership
(NCMLP) concluded, no systematic assessment of the impact of MLP integration has been measured to date (Beeson et al., 2013). This study begins to fill that void. Moreover, this study directly addresses the absence of any qualitative research in the MLP literature. Filling this methodological gap in the literature (Robert Wood Johnson Foundation, 2011) is essential in order to understand the extent to which, and the ways in which law is implemented and enforced so that MLPs around the country can more effectively address the inequality that characterizes low-income and minority patients access to the medical and social determinants of health (Burris, 2001). Thus, this study is designed to make a significant contribution in three
ways:
First, this study contributes to equipping providers, physicians, and clinicians with information they seek to better treat poor and minority patients. In 2011, the Robert Wood Johnson Foundation surveyed a randomly selected, representative sample of 1,000 U.S. primary care physicians. The results showed that 85% of physicians believed that patient social needs are as important to address as their medical conditions. The percent of providers holding this belief climbed to 95% of physicians surveyed when asked about low-income patients. Put another way, physician respondents reported that if they had the power to write prescriptions to address social needs, these would represent approximately 1 out of every 7 prescriptions they write (Robert Wood Johnson Foundation, 2011). Medical-legal partnerships could represent a


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
45
transformative model to equip providers to deliver health care that connects patients to the social resources physicians want to provide and that patients need to thrive.
Second, this study contributes to the long-term sustainability of MLPs in Colorado. Nationally, the empirical record must be strengthened if MLPs are to create lasting, systemic change. But the need for evidence of the MLP modeTs impact in Colorado is particularly acute. Because of the lack of evidence for the MLP performance in Colorado, the first MLP established in Colorado has ended. Childrens Hospital patients will no longer have an integrated partnership with legal aid lawyers addressing their social needs. This MLP closed for lack of funding in December 2013 and was replaced by a small, ad hoc effort sponsored by a private law firm with limited time to dedicate to the documented breadth of childrens need. My experience raising grant funding for CHEP confirms that the long-term financial future of scalable MLPs that can serve all Colorados low-income and minority patients depends upon generating the kind of empirical information that can show the return on investment that MLPs provide in medical, social, and financial terms. This study will contribute information about medical and social impacts that may help sustain the MLP model in Colorado financially.
Linally, this study contributes to the MLP and public health law literature by introducing a new theoretical framework for understanding whether MLPs disrupt patients cumulative adversity (Hatch, 2005). By adding a combined qualitative and quantitative understanding of the way that law generally and MLPs specifically affect low-income patients in Colorado, this study will contribute to understanding and improving the role that MLPs play in addressing health disparities in Colorado and the nation.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
46
CHAPTER II LITERATURE REVIEW
This study seeks to contribute to two related bodies of literature. Public health law research combines empirical legal scholarship, health services research, and public health law scholarship to advance public health goals. A goal of this study is to add to the theory based, quantitative and qualitative approaches that inform new public health solutions. More precisely, this study adds to the emerging body of transdisciplinary public health law research that has been called legal epidemiology (Burris et al., 2016). Next, this study adds to a sparse but growing body of medical-legal partnership scholarship. These studies report evidence of the way that legal services change health care delivery, health outcomes, and health care costs. This chapter summarizes the existing literature in both these areas and concludes with the latest research pertaining specifically to the two outcome variables studied here.
Public Health Law Literature Review
Elizabeth Tobin Tyler explains the inter sectional ity between law and public health by highlighting an increasing need for a fuller understanding of the role of social justice in health .
. Particularly as mounting evidence points to the role of social conditions in health outcomes (Tyler, 2012). Tyler explains that the law plays two essential roles in affecting the social determinants of health. First, laws organize and perpetuate the structures that we describe as social determinants. In the case of housing, for example, laws define the rights of home ownership or occupancy, enforce financing terms, and establish construction standards. Construction laws organize laborers to build and meet safety standards, tax laws govern affordability, zoning laws control density and the neighborhoods physical built-environment and demographic composition, criminal laws seek to ensure neighborhood safety, and public benefit


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
47
laws affect accessibility and affordability. The second role for law is corrective. In this role, law serves as a tool to transform inequitable social structures. Law has been and will remain critical for creating the infrastructure that supports directed and accountable action, as well as for limiting some actions that diminish health or require actions that enhance it (Tyler, 2012). For example, public health laws can be used to penalize unsafe or unsanitary housing conditions. Environmental laws may mandate removal of hazardous toxins from a neighborhood. A landlord can be compelled to remove lead paint or exterminate rodents when housing habitability laws are enforced. Positively, a city can increase recreational spaces or food security by enacting tax and zoning provisions to reward health enhancing neighborhood design.
Professor Scott Burriss model for integrating law and social epidemiology describes a dual role for law that is similar to Tylers description but goes further. Burris introduces a textured and normative understanding of the way a legal system may effectuate better or poorer health of populations by defining social position. He explains that law operates as a pathway through which lower social economic position and poor social cohesion translate into stress and resultant poor health. According to Burris, integrating law, social science, and social epidemiology identifies two measurable relationships between law and health: law may operate as a pathway along which broader social determinants of health have an effect; and laws or legal practices may contribute to the development, and influence the stability, of social conditions that have been associated with population health outcomes (Burris, Kawachi, & Sarat, 2002). For example, negative experiences with law say with law enforcement officers have far-reaching, health-harming psychosocial effects on individuals and neighborhoods. Disparate enforcement of laws may affect self-esteem, mutual respect, and social cohesion. Similarly, much of the effect of race or class difference on health seems to be mediated by the accumulation of small stressors


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
48
over a life course, many of which are communicated through law. Burris explains how law mediates job insecurity, fear of neighborhood crime, and experiences with racial prejudice and can result in wear and tear on the cardiovascular, endocrine, immunologic, and metabolic systems, eventually leading to a host of maladies ranging from hypertension, obesity, diabetes, as well as depression, asthma and susceptibility to infection (Burris, Kawachi, & Sarat, 2002). Professor Lawrence Gostin describes the intersectionality between law and health with a focus on the branch of law that governs public health particularly, and more generally on the role of the state in ensuring public health. Gostin explains that public health law specifically defines a set of legal powers and duties belonging to the state and its partners to ensure the conditions for people to be healthy (to identify, prevent, and ameliorate risks to health in the population), and [define] limitations on the power of the state to constrain for the common good the autonomy, privacy, liberty, proprietary, and other legally protected interests of individuals. The prime objective of public health law is to pursue the highest possible level of physical and mental health in the population, consistent with the values of social justice (Gostin, 2008).
These theories of law and public health are applied practically when law is described as an integral part of the social determinants strategy. Scott Burris translates the large theoretical paradigm that essentially says the law is all over and presses the law into service to conform the lofty and theoretical laws on the books with the systems and institutions and practices that touch peoples daily lives through the laws on the streets (Burris, 2001). In a nutshell, the role of the medical-legal partnership is to change the status of low-income patients from having to accept inferior distribution of the protections, property, opportunity, goods, and services to which the law entitles them. MLPs aim to improve the trajectory of those social conditions that most directly influence a patients life chances and overall health and social outcomes. If our


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
49
theories are correct, [access to legal services will] improve both the level and distribution of health, because they address fundamental causes that find expression in a wide range of ultimate health states reached in a wide range of ultimate health states reached via a plethora of pathways across the life course (Burris, 2001).
The most apt description of the work undertaken in this study only recently appeared in the literature. Scott Burris, Maurice Ashe, and others coined the term legal epidemiology to describe the work using social science methods to conceptualize, implement, and evaluate both laws and legal practices that change unhealthy behaviors and environments. (Burris et ah, 2016) This study includes components of legal prevention and control as it explores MLPs as a way to intervene preventively in the lives of vulnerable patient populations. Moreover, my study adds to legal etiology literature the study of the structural unintended consequences of laws that are inequitably enforced. In both cases, this study fills gaps in the literature identified by Burris and Ashe (2016):
Efforts to theorize social determinants and to prescribe reforms have been impressive and important, but our health research investments remain biased toward the individual risk factors and proximate causes of death. Because law plays such an important and pervasive role in structuring environments and behaviors and beliefs, research under the heading of legal etiology is crucial to charting a course of practical reform. It encompasses laws structural role in shaping the level and distribution of health in a community, laws contribution to cultural beliefs about how health is produced, protected, and distributed, and how we can use legal interventions, such as enforcing healthy housing codes, to improve health and health equity. In this way, my study builds on and extends these analyses by Tyler, Gostin, Burris, and Ashe which describe the reason and methodology that inform my examination of whether the Childrens Hospital of


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
50
Colorado Medical-Legal Partnership moderates immunication compliance and upward residential mobility outcomes.
Medical-Legal Partnership Literature Review
In 2013, the National Center for Medical-Legal Partnership published a white paper collecting the evidence to support the efficacy of MLPs (Beeson et al., 2013). This paper highlighted the limitations of the empirical record. As a result, the NCMLP has focused considerable attention on developing a series of common metrics for the more than 290 MLPs nationwide to use in evaluating their impact (Regenstein, 2014). However, this process has been slow and incomplete. So far, the development process has yielded proposals on six metrics focused primarily on process and MLP structure evaluation but largely ignoring the type of outcome measures necessary to determine whether MLPs are contributing to reducing disparities or improving population health outcome (NCMLP Performance Measures Handbook, 2016). The absence of such outcome measures not only affects the assessment of the quality and effectiveness of MLPs, but this void is also hindering the long-term financial sustainability of MLPs.
The NCMLP reported in 2012 that 23 published articles descriptively present the components and function of the MLP model. These articles generally explain what medical-legal partnerships are and why they are needed to address the social determinants of health. A few contain some empirical evidence that patients identify unmet legal needs as harmful to their health (Sandel, Suther, Brown, Wise, & Hansen, 2014). Practice reports or case studies comprise another large category of MLP literature; they outline the activities and structure of medical legal partnerships in a single setting (Newman, 2012). Observational studies make up the smallest category of published MLP research; the NCMLP identified 13 articles reporting


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
51
observational studies of MLP impacts on patients, providers, or communities. The NCMLP white paper remains the most comprehensive compilation of MLP research literature. However, since its publication several notable studies have been published. Figure 15 describes the NCMLP reported observational studies and those that have been published since the white paper.
Author Title Methods and/or Measures Used or Suggested Results
Atkins, et al., 2014 Medical-Legal Partnership and Healthy Start Practice Report Number of cases opened (200), number legal consultations (150), resolved (70) and legal advocacy training hours (48) over 3 years
Beck et al., 2014 Housing Code V iolation Density Associated with Emergency Department and Hospital Use by Children with Asthma Correlated number of code violations in census tract with population level asthma morbidity Found density of housing code violations associated with and predicted asthma morbidity and hospitalized patients risk of subsequent morbidity
Beck et al., 2012 Identifying and Treating a Substandard Housing Cluster Using a Medical-Legal Partnership Identified number and type of cases, counted housing repairs completed 16 apartments in a 19-building complex treated. Housing and health outcomes such as asthma, developmental delays, and elevated blood levels more likely for 16 of 45 pediatric patients who were in poor quality housing
Cohen et al., 2010 Medical-Legal Partnership: Collaborating with Lawyers to Identify and Address Health Disparities Practice Report -Evaluated four MLPs Found established MLPs improve provider knowledge; reduce provider concerns over making patients nervous with legal questions; increase resident referrals to legal services after training; and one MLP found 67% medical residents uncomfortable addressing social issues


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
52
Conover et al., 2002 Impact of Ancillary Services on Primary Care Use and Outcomes for HIV/AIDS Patients Using primary data from 1997 survey of low income patients in North Carolina, multivariate logit analysis to estimate effect of receiving housing, legal services, and substance abuse treatment on use of primary care, CD-4 counts, viral load, and self-rated health status Legal services one of several ancillary services analyzed. Legal services positively associated w/increased access to primary care but negatively associated with viral load.
Hernandez et al., 2016 Extra Oomph: Addressing Housing Disparities through Medical Legal Partnership Interventions Qualitative interviews with 72 families about affordability, substandard conditions, and stability. 83% of MLP families improved their housing conditions. Legal services helped reinstate or prevent utility shut-off, retain or regain housing subsidies, relocate to better residential environments, and appeal a rent hike. Patients in control group less likely to resolve need for safe, affordable housing; 64% of control group patients and 17% of MLP patients did not resolve housing problem
Fleishman et al., 2006 The Attorney as the Newest Member of the Cancer T reatment Team Practice Report -Observed patients stress, financial situations, and reviewed appointments and clients served. At legal health New Y ork, a fully staffed free legal services program for cancer patients, patients had reduced stress and maintained treatment regimen, families had better financials, and MLP tracked number of clients.
OSullivan et al., 2012 Environmental Improvements Brought by the Legal Interventions in the Homes of Poorly Controlled City Adult Asthmatic Patients: A Proof-of-Concept Retrospective study of 12 patient charts for adult patients with poorly controlled asthma; patients self-reported allergen exposures; pre-post intervention analysis of peak expiratory flow rate (PEFR, asthma severity class, medications, ED Effect of MLP intervention to compel landlords to provide better living conditions for asthma patients improved mean PEFR, reduced ED visits from 22 to 2; reduced hospital admissions from 11 to 1 admission, and all patients had reduced need for medication and improved asthma severity.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
53
visits, hospitalizations and steroid requirements
OToole et al., 2012 Resident Confidence Addressing Social History Survey of 40 residents Survey of 40 residents show those with more legal and social resources have greater confidence, knowledge; screen more; and take longer social histories
Petti gnano et al., 2012 The Health Law Partnership: Adding a Lawyer to the Health Care Team Reduces System Costs and Improves Provider Satisfaction Practice Report Patients served, Medicaid cost savings/claims recovered by coverage cohort, provider satisfaction and CE credits rendered 4-year study increased Medicaid reimbursement payments, saved hospital $10,000 in continuing education costs, and increased MD satisfaction
Pettignano et al., 2011 Medical-Legal Partnership: Impact on Patients with Sickle Cell Disease Practice report number of families referred, number of legal issues, cases closed, cases won Retrospective cohort of 81 SCD parents with 76 children opened 106 cases, 99 closed with 21 measuring legal gains
Rodabaugh et al., 2010 A Medical-Legal Partnership as a Component of a Palliative Care Model 3-yr review of palliative care MLP for cancer patients 297 legal referrals for custody over 3.5 years, 17 benefits denial cases overturned, and MLP demonstrated financial sustainability as institutions recovered $923,188 for 17 benefits advocacy cases. Staffed by fulltime social worker and 0.5 FTE attorney doing advocacy in guardianship, estate, and advance care planning, housing and legal services cases.
Ryan et al, 2012 Pilot Study of Impact of Medical-legal Partnership Services on Patients Perceived Stress and Wellbeing Pre- post study of stress levels using questionnaires Changes in perceived stress strongly related to participants concern over legal issues addressed by MLP. MLP reduced stress and improved overall wellbeing.
Sege et al., 2015 Medical-Legal Strategies to Improve Infant 330 Medicaid families randomly assigned to MLP or safety intervention. MLP associated with improved access to income supports and better quality


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
54
Health Care: A Randomized Clinical Trial Pre- and Post-intervention interviews care. At 6- and 12- month surveys more MLP group had obtained income, utility, housing, and food assistance than control grp
Teufel et al., 2012 Rural Medical-Legal Partnership and Advocacy: A Three-Y ear Follow-up Study Five years review of baseline data and three years follow up data showed benefit relative to cost of MLP increased from 2002-2006 and 2007-2009. Number of people served and cases won increased. Health care recovery dollars showed 319% ROI between 2007-2009, and $4 million in health care debt relieved for patients.
Weintraub et al., 2010 Pilot Study of Medical-legal Partnership to Address Social and Legal Needs of Patients 54 low-income families completed baseline and 6 month follow-up assessments to test hypothesis that integrating legal services into pediatric settings would increase families awareness of and access to legal and social services, decrease barriers to childrens health care, and improve child health. 2/3 reported improved child health and well-being, increased food and income support utilization, and decreased avoidance of health care due to lack of insurance. Percent of patients accessing legal services, and patients with improved satisfaction increased.
Figure 15 Summary Of Major Observational Studies Of MLP Impact Observational Studies In MLP Literature
The observational MLP studies from the NCMLP white paper contain a limited body of evidence that MLPs affect financial, professional, and health outcomes. Financial impact studies typically analyze the return-on-investment that MLPs provide when the cost of volunteered services is measured against the savings to institutions or patients. These studies attempt to prove the financial viability of MLPs by focusing on a traditional cost-benefit analysis. For example, a longitudinal study of an MLP in rural Illinois compared the financial benefit as compared to the initial investment to calculate a financial measure useful for analyzing stock and bond values. This article calculated the MLPs ROI return on investment for the initial period


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
55
from 2002-2006 and the relative benefit of the MLP during 2007-2009. Researchers reported the program produced a 149% return on investment (ROI) of $115,438 for the hospital partner during the first period, and the hospitals Medicaid ROI increased to 319% based on an investment of $116,250.
Another approach in financial studies is to evaluate the impact the programs have on patients who have been denied health benefits or who have fallen into medical debt. For example, one researcher analyzed the return on investment hospitals received from MLP programs that resolve previously denied benefit claims. Rodabaugh and colleagues reported on an MLP that worked with cancer patients and generated nearly $1 million in benefits by resolving denied benefit claims (Rodabaugh, et al., 2010). In a retrospective study of 71 parents or guardians with 76 children diagnosed with sickle cell disease, researchers measured patients gains in benefits due to the MLP intervention (Pettignano, Caley, & Bliss, 2011). This MLP, formed between a Childrens Hospital and Georgia Law School, measured the monthly amount of public benefits obtained or retained including housing, education, and health insurance benefits, as well as the number of patients who received end of life care following MLP legal counsel (Rodabaugh, et al., 2010). In still another approach, researchers at Johns Hopkins conducted a longitudinal study to determine that an Illinois MLP relieved patients of over $4 million in health care debt and obtained a total of $1 million in social security benefits (Teufel et al., 2012).
The second group of observational studies report MLP impact on attorney or physician professional practices. Most of these studies are descriptive or anecdotal. Fleishman, for example, observed that attorneys benefitted from the collaboration by improving their preventive clinical skills, while health institutions experienced fewer missed appointments and treatment


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
56
interruptions when cancer patients social needs were addressed (Fleishman, Retkin, Brandfield, & Braun, 2006). Another study highlighted the benefit to law students of developing interdisciplinary communication and collaborative practice techniques (Wettach, 2008). A survey of 40 medical residents found the MLP training gave physicians greater confidence and knowledge, leading them to screen for patients social and legal needs more frequently and spend more time taking social histories (Wettach, 2008). In addition to recovering previously unreimbursed Medicaid payments and saving continuing education costs, a Georgia MLP surveyed physicians, social workers, and nurses to find that the interdisciplinary MLP collaboration increased provider satisfaction. These respondents perceived that patient emergency department visits and readmissions declined, while the providers ability to reallocate their time to other cases improved (O'Toole et ah, 2012). Another exemplar from this group of observational studies presents descriptive data from an MLP in Chester, Pennsylvania. After reviewing the number of legal cases opened during one year, the average number of legal issues addressed for each client, and the number of legal and advocacy training hours the MLP contributed to public health issues, Atkins concluded, [a] current need exists for longitudinal data showing the impact of MLP services over time on factors such as health birth outcomes, mortality, emergency room utilization treatment compliance, and absenteeism, to provide even more compelling evidence on MLP efficacy (Atkins, 2014).
While the growing evidence confirms that both patients and providers perceive the need for integrating legal services into the health care delivery model, the studies that evince the impact of integrating attorneys into the health care delivery model seldom quantify the impact MLPs have on patient health outcomes and generally do not take a longitudinal perspective. Six studies


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
57
currently comprise the third and smallest group of articles from the NCMLP report that present empirical evidence of MLP impact on patient health and well-being.
Interesting behavioral health outcomes such as stress have received a large share of the attention in the MLP literature. In one study, researchers administered 10-item Perceived Stress Scale (PSS10) and Measure Yourself Concerns and Wellbeing instruments to find that 67 patient participants showed a slight decrease in their mean stress level and an even more slight improvement in their self-rated well-being (Ryan, 2012). In the other study, a survey of 20 clients of a New Y ork City MLP showed that 75% of patients reported reduced stress as a result of the legal help they received; 30% of these patients said the MLP services helped them to maintain their treatment regimen and 25% said the MLP helped them to keep their medical appointments (Pleishman, 2006). This finding that MLP participation positively impacts access to health care is consistent with an earlier study of HIV/AIDS patients. In 2002, researchers measured the impact of a variety of ancillary services on HIV/AIDS patients. In this study, legal services were one of several ancillary services that were positively associated with increased patient access to primary care services. However, counter intuitively the availability of legal services was also negatively associated with patient viral loads. In other words, the health outcomes of patients worsened with increased access to legal services. In this study, improved housing was the only ancillary service associated with improved primary health care access and health outcomes (Conover, 2002).
A Stanford-based medical legal partnership conducted a 36-month, prospective study of the impact their MLP had on patients access to legal and social services as well as access to health care (Weintraub, 2010). A cohort of 54 Childrens Hospital and Health Center patient families, who received free MLP legal services, completed baseline and six-month follow-up


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
58
assessments. This pilot study found that integrating legal services into health care delivery not only decreased barriers to health care access significantly fewer participants reported they avoided going to the doctor for their children after the MLP intervention but the MLP was also associated with increased awareness of other social supports such as Supplemental Nutrition Program for Women, Infants, and Children; Food Stamps; and Social Security Income Benefits. Several MLP studies have examined the MLPs impact on housing conditions. In 2012, Beck et al. reported a study involving an urban, 19-building complex owned by a single housing firm (Beck et al., 2012) .This group studied the health of children receiving outpatient primary care in the neighborhood where housing was affected by known health harming risks such as pest infestation, water damage, and poor ventilation. Physicians identified patients through social risk screening and referred these families to an MLP advocate who identified a cluster of substandard housing conditions, formed a tenant association, and advocated in court, before city council, city planners, and housing inspectors, to cause the building owner to repair the patient housing to meet code standards. This study did not measure childrens health outcomes after the intervention but found an association between the substandard home environments and poor health outcomes. Beck and colleagues studied 45 children living within the identified units; 36% of these children had asthma, 33% had developmental delays, and 9% showed elevated lead content in their systems, leading these researchers to conclude that if attorneys could improve housing conditions, they could also improve health. Beck and colleagues published a second study in which he found that an MLP in Cincinnati, Ohio, successfully intervened to enforce building ordinances requiring a landlord to make improvements in a large cluster of pest-infested apartments (Beck et al, 2014).


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
59
A 2012 study that measured MLP impacts on childrens asthma is particularly instructive for this project. OSullivan and colleagues (2012) looked at the effect an MLP had on the health outcomes of 12 asthma patients with poorly controlled disease. These researchers evaluated patients pre- and post-intervention outcomes by measuring Peak Expiratory Flow Rates (PERF), asthma severity class, medications, ED visits, hospitalizations, and the requirement for steroids before MLP intervention and 6-12 months after the intervention. The results of that study showed that patients all had increased PERF, decreased number of ED visits and hospitalizations, fewer required systemic steroids, and all had reductions in the dose and/or number of medications used. Despite the limitation of this study which include its small number of patients, the lack of a control population, and the limited time period considered, the researchers concluded the medical-legal collaboration is highly effective in improving the control of inner-city asthmatics by effecting improvements in the domestic housing environment. OSullivan measured a 91% reduction in ED visits and hospital admissions for inner city New York asthmatic adults after an MLP intervention (OSullivan et al, 2012).
More recently, Diana Hernandez published an influential qualitative study on an MLPs impact on 72 families with housing issues. Hernandez examined the mechanisms through which MLPs work to resolve housing problems that fell into three categories: affordability, substandard conditions, and stability. Hernandez study found that the interaction of physicians and lawyers in a patient centered medical home allowed MLP participants to resolve housing affordability (e.g., averting utility shut-off), adequacy (e.g., substandard conditions) and stability (e.g., eviction) problems more successfully than families without MLP assistance (Hernandez, 2016). This research team conducted interviews with low-income families who participated in MLPs and with a control group of families. Before the MLP intervention, 53% of these families


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
60
reported living in inadequate housing, 33% were struggling to afford rent or utilities, and 14% reported housing instability because they were at risk of eviction and homelessness. An overwhelming majority83% of families in this study who received the MLP services improved their housing conditions. Legal services helped reinstate or prevent utility shut-off, retain or regain housing subsidies, relocate to better residential environments, and appeal a rent hike. In contrast, the patients in the control group were less likely to resolve their need for safe, affordable housing; 64% of the control group patients and 17% of MLP patients did not resolve their housing problem during the study period (Hernandez, 2016).
In May 2015, researchers at the Boston Medical Center reported a randomized clinical trial testing the effect of an intervention that incorporates the MLP model in care delivered to 330 families of healthy newborns in Pediatrics (National Low Income Housing Coalition, 2016) There, families were randomly assigned a specialist who provided support until the newborns 6-month routine health care visit. The specialist conducted home visits and telephone check-ins, provided age-related information on child development, and when necessary combined elements of an MLP program at Boston Medical Center. At the conclusion of the study, infants who received the support intervention were 14% more likely to have completed their 6-month immunization schedule by 6 months of age. They were also more likely to have five or more routine preventive care visits by age 1 and less likely to have visited the emergency department by 6 months. While it is difficult from this study to isolate the MLPs impact on these improved outcomes, and while the study period followed children for only a short time after birth, this article importantly represents the first randomized clinical trial reported to test the MLP intervention. At the outset, 73% of families reported economic hardships. More than half (61%) reported food insecurity, 28% reported they were unable to pay rent or mortgage during the


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
61
previous 12-months, 42% reported missing a payment for gas, electricity, or water utilities, 12% reported a utility shut-off for failure to pay, and almost half (44%) reported their telephone service had been disconnected for failure to pay sometime during the previous year. The MLP intervention accelerated access to concrete income supports for newborn babies and their families. At six- and 12-month post-intervention surveys, significantly more patients in the MLP group had obtained income, utility, housing, and food assistance than patients in the control group. For example, 12 months after the birth of their children, 21.1% of families had obtained income assistance compared to 18.8% of control families (p value = .029). Because prior research confirms that concrete support in early months of a childs life may protect against child neglect and abuse by reducing parental stress, and may improve primary health care, the results from this study point out the impact that MLP income supports can have on child health outcomes (Shonkoff, 2012). My study builds upon this work to isolate the treatment effect of the MLP intervention itself on childrens immunization completion in order to infer patients participation in regular preventive care.
In 2015, a study reported in the Journal of Health Care for the Poor and Underserved showed how an MLP improved home heating security for poor children and their families (Sege et ah, 2015). In this pre-post study, researchers compared the number of families reporting energy insecurity on a waiting room screening questionnaire before and after MLP intervention. The certification of medical need approvals for local utility companies increased by 65%, preventing utility cut-offs for 396 more families with vulnerable children between the first and second year that the MLP operated. In another study using population-level asthma morbidity data, Cincinnati Childrens Hospital researchers found that the density of housing code violations in a census track could predict a hospitalized childs risk of subsequent emergency department


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
62
visits and hospitalizations. Children who had been hospitalized for asthma had 1.84 greater odds of a subsequent hospital visit due to asthma morbidity within a year if they lived in the areas with the highest number of housing code violations. This study supported the inference that MLPs could reduce asthma related morbidity in children by reducing housing code violations (Taylor et al., 2015)
My work in this project extends the Taylor (2015) and Beck (2014) studies of MLP impacts on internal housing conditions as a social determinant of childrens health. My study uses median family income by census tracts to infer the quality of both internal housing and external neighborhood conditions that may impact child health. By measuring the housing variable based on median family income of the census tract, I have aimed to capture a broader range of housing conditions relevant to health. Second, the housing mobility variable allowed me to connect the legal issues that accompany poor housing conditions and affect childrens health on a population level. Finally, while the Beck study considers the effect of poor housing on children with asthma, my study includes a broad range of childhood diseases as co-variates, and looks not only at housing related outcome variables but also at a variable related to immunization completion as a proxy for overall child health rather than singling out a single disease condition such as asthma.
The current body of MLP literature is characterized by inquiries that have yet to examine the broader implications of the potential impact of the MLP model. On one hand, inquiries about investment returns, financial benefits, and professional practices ignore the long-term impact that an MLP may have on patients biomedical outcomes. On the other hand, existing studies of MLP health impacts narrowly focus on one disease or health outcome and miss the opportunity to evaluate the broader social impact that MLPs are intended to have. My study begins to fill


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
63
these gaps in the MLP literature and does so using data from thousands of pediatric patients, over a five-year period, and employing a methodology designed to identify the treatment effect of the MLP service specifically. The next chapter elaborates the theoretical framework that I have employed.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
64
CHAPTER III
THEORETICAL FRAMEWORK
First, it is well accepted that the most fundamental causes of health disparities are inequities in socioeconomic status (SES) within the United States (Beck, Huang, Chandur, & Kahn, 2014). Moreover, a growing body of research supports the claim that racism is also a fundamental cause of health inequity (Adler & Newman, 2002). Racism, an organized system of categorization and ranking of societal groups into hierarchical races that devalues, disempowers, and differentially allocates desirable opportunities and resources, causes disparate health outcomes for minority patients (Phelan & Link, 2015). My theoretical starting point for this study is an ecosocial understanding of the impact that economic and social inequality have on patterns of health inequality in Colorado children. Second, I rely upon fundamental cause theory to inform my hypotheses about the vulnerable populations studied. Third, the dependent variables chosen to reflect the life-course theoretical approach to health interventions. I take each concept in turn and conclude this chapter with a description of the theories that support my selection of outcome variables.
An Ecosocial Theory Of Health Inequality
Ecosocial theory explains that populations biologically embody adverse exposures from ecological and societal influences. The result of economically and socially skewed influences is a disparate distribution of health (Williams & Mohammed, 2013; Krieger, 2001). Disparate influences begin imposing harmful health effects early in life. Researchers have described a process called biological programming, to explain how socially mediated factors such as poverty and racism can adversely affect human growth even before birth and throughout infancy (Barker ,1998). J.L. Aber and others suggest that poverty is associated with higher neonatal and


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
65
post-natal mortality rates, increased risks of accidental injuries, asthma, and abuse as well as lower cognitive development scores (Aber et al., 1997). Importantly, ecosocial theory describes harmful health impacts that begin in utero, persist during childhood, and reach into adulthood.
Jack Shonkoff, for example, explains that the stress of early childhood adversity leaves a lasting signature on genetic predispositions, brain architecture, and long-term health so that the adverse health effects of inequality in childhood are permanent (Shonkoff, 2011). Moreover, these health effects on children are both direct and indirect. For example, poor mothers cope with stress through higher rates of smoking, excess alcohol intake, drug misuse, and dietary deficits. These factors affect not only the mothers health but also the childs health beginning in the developmental stages of pregnancy, leaving newborns vulnerable to later health insults. In 1999, Hertzman used the term biological embedding to describe this interaction between physical development and the environment in early childhood that reduces weight and height growth and impacts mental functioning for life (Hertzman, 1999).
The vulnerability of a childs physical development before birth continues during the postnatal period, as children and their families interact with their social environment. Children are particularly vulnerable to the psychosocial causes of poor health (Haas, Krueger, and Rohlfsen, 2012). The evidence shows that children whose early lives are characterized by a poor socio-economic environment are at increased risk of low birth weight, smoking, alcohol abuse, and mental health problems. The Adverse Childhood Experience (ACE) Study also convincingly demonstrates a strong link between adverse childhood experiences and higher rates of heart disease, diabetes, lung disease, hepatitis, depression, obesity, and suicide (Felitti & Anda, 1997). The ACE Study linked childhood trauma to increased incidence of risky health behaviors during
childhood and adolescence (Shonkoff et al., 2011). Moreover, adverse effects of the interaction


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
66
between poor social circumstances and poor health and health behaviors have been shown to accumulate. For example, depressed socio-economic circumstances experienced early in life are associated with an increased likelihood of poor educational attainment, which is associated with poor health behaviors that produce overweight adults, who are ultimately susceptible to disease and early death later in life. In addition to explaining socioeconomic disparities, the ecosocial theory also provides a framework for my understanding racial and ethnic differences in health outcomes. I sought to capture environmental and psychosocial variables that may change childrens health outcomes over time. I examined immunization compliance during early childhood as a proxy for the access a child may have had to preventive health care which could improve long term health outcomes as well as for the direct benefits accruing from immunization compliance in avoiding infectious diseases. Also, I analyzed upward residential mobility to examine whether the MLP intervention could change neighborhood factors such as access to resources, material goods, and social influences that might improve long term health outcomes. My research asked whether a disruption that could improve those variables might improve long term health outcomes for participant patients.
My interest in influences on the relationship between race and poor outcomes arises from the understanding that racial inequality simultaneously benefits groups who claim superiority while harming groups deemed inferior. The impact of historic and contemporary racial and ethnic discrimination is to produce inequitable living, working, and environmental conditions. Exposures to toxins, trauma, and violence then are expressed in disparate biological patterns along racial and ethnic lines. Arline Geronimuss (2006) groundbreaking work on the weathering framework links racism and stress to explain health disparities. Geronimus found evidence that racial inequalities in health that manifest in biological systems cannot be explained solely by


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
67
racial differences in poverty. Racial and ethnic minorities navigate the results of individual discrimination and structural racism on a daily basis. These vulnerable populations experience overt and covert racism, concentrated poverty, high crime, and chronic environmental pollution, and over the course of a lifetime the physiological and psychological responses lead to health problems at rates not experienced by majority populations who are free from such life stressors.
By measuring racial differences in the allostatic load scores of her study participants, Geronimus (2006) was able to capture the biological responses to the stress of discrimination and its impact on cardiovascular, metabolic, and immune systems and attribute these health differences to the weathering effects of living in a race-conscious society (Geronimus, 2006). Race, for example, independently predicts low birth weight and infant mortality levels among babies such that babies born to black mothers with high income and education suffer poorer morbidity and mortality than babies born to white mothers with low income and educational attainment (Adelman, 2007). Colorado data are consistent with Geronimus weathering hypothesis. Although Colorado has the fifth lowest infant mortality rate in the country, black babies have nearly three times the infant mortality rate of white babies and Latino babies are now 70% more likely to die before their first birthday than white infants.
Finally, ecosocial theory provides a theoretical basis for questions of accountability and agency that I raise in this study. I ask whether MLP attorneys can disrupt the predicted association between race, poverty, and inferior health outcomes based on a theoretical understanding that societal actors both individual and institutional can be held responsible for changing the stressors that produce unequal population health outcomes. I assume that most patients who are victims of inequality do not, in their own right, have the agency to require the necessary changes. Thus, embodiment theories in social epidemiology counsel the understanding


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
68
that children biologically incorporate the inequities present in the world around them and inform the core concept driving this study: inequitable population patterns of health, disease, and wellbeing can be improved by compelling those responsible for unjust distribution of environmental and economic resources to rectify these inequities (Office of Health Equity, 2013).
Fundamental Social Causes Of Health Inequity
The second theoretical construct important to this study is that social determinants are a fundamental cause of disproportionately poor health outcomes for vulnerable populations (Krieger, 2001). More precisely, this study is based on the understanding that inequities in distribution of the social determinants of health, including the conditions under which people work, live, eat, and the environment in which they live, more directly explain health disparities than any biological explanation alone (Marmot, 2004).
An economic historian is credited with the first systematic assertion that improvements in population health are associated with social, economic, and industrial changes rather than exclusively relying upon improvements in medical interventions (McKinlay, 1977). Later, in the Whitehall Studies Sir Michael Marmot reported a pair of longitudinal studies that analyzed UK health data in which he identified an inverse relationship between the social status of British civil servants and their relative risk of morbidity and mortality (Marmot et ah, 1991). The Whitehall studies prospectively compiled over 15 years of longitudinal data for a cohort of over 10,000 study participants. The data showed that shorter life expectancy and most diseases occur more commonly further down the social ladder, less commonly in the middle-class, and least among upper-class populations (Marmot, Rose, Shipley, & Hamilton, 1978). Access to health care could not explain this inverse relationship called the social gradient that the Whitehall studies revealed since all the study participants obtained health care through the National Health Service


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
69
(Wilkinson & Marmot, eds., 2006). Instead, differences in the social conditions faced by participants were shown as determinants of health outcomes. In 1998, the World Health Organization (WHO) published comprehensive evidence of the association between social determinants and population health globally, based on clear evidence of the inverse relationship between socioeconomic status and poor health (Wilkinson & Marmot, 2006).
In 1995, Link and Phelan introduced evidence to explain the social gradient further.
Their research explained that variable access to resources that affect social conditions, rather than individual and biological risk factors, results in unequal health outcomes and inequitable ability to address deficiencies in the social determinants. In their parlance, a fundamental cause of disease is one that influences multiple disease outcomes, through multiple risk factors, by limiting access to the resources that could avoid risk factors and that is replicated over time through various mechanisms (Link and Phelan, 1995). This paper asserts that a lack of legal services is a fundamental cause of poor health.
Law organizes access to the social determinants of health in American society.
Therefore, I apply fundamental cause theory to postulate that a lack of access to justice through legal services satisfies the three constituent claims of fundamental causality posited by Lutfey and Lreese (Luftey & Lreese, 2005). Lirst, lack of access to legal services is multiply realized to affect diffuse proximate causes of poor health. Examples include lead poisoning suffered due to unenforced housing codes, lack of health care access suffered due to wrongly terminated benefits, and mental health impacts of police brutality and community violence. Second, lack of access to legal services satisfies the holographic claim of a fundamental cause. The relationship between poor health outcomes and the lack of legal services is replicated even when poor health is separated into a wide variety of health outcomes. Examples include inferior treatment of


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
70
chronic diseases, recurring insults to personal safety, acute injury from trauma, environmental health issues, medical disability, and any subclass of poor health outcome imaginable. Finally, lack of access to legal services satisfies the predictive claim of fundamental causality. Changing the structure of how poor health outcomes are realized will have only modest effects on the observed relationship between legal powerlessness and poor health. The best example of this relationship is the limited impact of health policy reforms that substantially increase access to health care for the poor. SCHIP, for example, has broadened access to medical care for poor children but has had limited impact on the population-wide health outcomes of unrepresented children, since they still lack the legal resources to enforce their SCHIP rights to receive health treatment absent advocacy by legal representative. In 2014, the Colorado Health Institute reported that 45,227 children were eligible for Medicaid but not enrolled and 36,380 children were eligible for CHP but not enrolled. While not all 81,607 of these uninsured children require lawyers to enforce their rights, the evidence of pervasive health disparities in Colorado suggests that a significant number of them may.
Poverty and minority status are both characterized by a general lack of resources. In this study I theorize that the inability to address legally enforceable rights is one of the most important resources that low-income and minority children in Colorado lack. I suggest that limited access to legal representation broadly influences myriad resulting resource limitations including a lack of money, knowledge, power, prestige, and the kinds of interpersonal resources embodied in the concepts of social capital, social support, and social networks. MLPs may increase families social capital by giving them a way to increase and take control of their income and allocate funds and time needed to access pediatric preventive health care more easily. MLP representation may also increase available income by enforcing laws that control


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
71
access to social security, Medicaid, TANF, SNAP, and other income benefits. I further theorize that MLPs may improve residential conditions for children, by reducing destabilizing economic and legal stressors such as eviction and foreclosure, while increasing upward residential mobility so that patients families can move to neighborhoods with greater social opportunity in the form of food security, educational opportunity, and transportation. Improved residential stability and upward mobility can lead to improved social networks for children and their families, in communities with improved built environments and access to populations that support healthy behaviors. In future studies, researchers may also find that MLPs increase self-efficacy and mastery, decrease stress, and improve health behaviors that could allow low-income and minority families to better provide childrens health care, food, and shelter.
The lack of legal representation influences multiple risk factors that contribute to a persistent association between limited legal access and poor health. Existing evidence shows that even progressive improvements in the underlying laws that affect low-income and minority families whether large sweeping pronouncements such as the federal Civil Rights Act of 1964 which prohibits all forms of discrimination or minor building code regulations that require habitable housing conditions low income and minority families who lack legal resources will be largely unable to take advantage of their rights guaranteed by the law because they will be unable to enforce them as they pertain to their daily lives and interactions (Seron et ah, 2001).
Low income families have more frequent and more urgent legal needs than other families (Rhode, 2004). Thus the fundamental relationship between their lack of legal services and poor health outcomes will remain undisturbed. I posit, therefore, that the lack of legal services suffered by minority and low-income Coloradans represents a fundamental social cause of poor


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
72
health that limits access to resources that vulnerable populations require to avoid or correct the disproportionate risks for morbidity and mortality to which they are exposed.
I have considered MLPs impacts on racial and ethnic minority patients because Link and Phelan (1995) have argued convincingly that race discrimination is a systemic, fundamental cause of health inequality, independent of SES. Race is associated with a variety of flexible, race-related resources such as occupational prestige and power, beneficial social connections, freedom, and neighborhood segregation, that produce disparate health outcomes (Phelan & Link, 2015).
Moreover, Phelan and Link (2015) assert they find evidence of differences in multiple, flexible resources related to health disparities, which can only be explained by racial disparities, not attributable to differences in SES. Specifically, Phelan and Link (2015) identify neighborhoods as an important mechanism to understand racial differences in health. This, they explain, is because neighborhoods, independent of individual-level SES, account for differences in toxic environmental exposures, nutrition, poorer police and fire protection, inferior recreational spaces, higher levels of tobacco and alcohol advertising, and a technologically inferior quality of hospitals and other health providers. Figure 16 below describes the theory of race as a fundamental cause of health inequality.
Systemic
racism
Multiple
replaceable
mechanisms
Figure 16. Racism As A Fundamental Cause Of Health Inequality, Phelan & Link (2015)


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
73
The evidence that race, through the mechanism of racism, impacts housing is particularly persuasive. Racial segregation in housing is a fundamental cause of racial disparities in health (Williams & Collins, 2001). Segregation is defined as the geographical separation of people, primarily unrelated to personal preferences, based on ethnicity or race. Residential segregation is detrimental to health outcomes for minority populations (Sudano, et al., 2013). This is because when black and Latino populations live in segregated neighborhoods, they are isolated from the resources white populations can access to protect and improve their health. Residential segregation relegates black and Latino populations to areas of lower quality housing, schools, food, employment, and recreational spaces as well as increased exposure to violence, environmental hazards, and disparate law enforcement practices (Smith & Petrocelli, 2001; Briggs & Keimig, 2016). Residential segregation is also associated with inferior access to health care providers and access to lower quality providers such as pharmacies with lower-quality inventories, clinicians with inferior training and experience, hospitals with worse outcomes, older physical plants, and less medical equipment (Skinner et al., 2005; Bach et al., 2004).
Higher rates of racial segregation are associated with higher rates of adult and infant mortality (Laveist, 1993; Laveist, 2003), coronary heart disease (Diez Roux et al., 2001; Fang et al., 1998) infectious disease such as tuberculosis (Acevedo-Garcia, 2001; Acevedo-Garcia, 2000) and with poorer mental health (Aneshensel & Sucoff, 1996), even when researchers control for poverty rates. Residential segregation is also associated with higher homicide rates, one of the key drivers of the gap in black-white life expectancy. While this study does not directly evaluate the impact of racial segregation on health outcomes, the study does evaluate the relationship between race and health outcomes as a way to infer whether moving from more to less segregated


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
74
neighborhoods, which are highly correlated with improved median family incomes, is a mechanism the MLP intervention might impact.
The objective of this study is to evaluate medical legal partnerships as a disruption of the fundamental cause relationship between poverty, race, and poor health outcomes, due to the lack of legal access that vulnerable Colorado families suffer. I theorize that medical legal partnerships change not only the diffuse health consequences of poverty and discrimination, but also change the way in which health care delivery is fundamentally realized. I have chosen a methodology to infer a cause-and-effect relationship between MLPs and outcomes that may improve the outcomes that fundamental causation predicts. Whereas poverty and race predict poor access to health, preventive health care, and residential stability, this study will examine whether MLPs disrupt this relationship to improve access to primary care as evinced by immunization compliance and improved residential stability.
The Life-Course Cumulative Disadvantage Theory Of Health Inequity
The final theoretical construct informing this study is life course theory. I hypothesize that the impact of medical-legal partnerships may redirect life-course trajectories for pediatric patients and their families who receive the intervention. The life course theoretical framework is appropriate to study the MLP influences, first because it is rooted in a contextualist perspective (Elder & Johnson, 2003) and second because it examines changes in individuals and groups as large social forces alter environmental pathways. The life-course theory examines changes by noting first how life trajectories are formed in a historical and social context and then how a substantial change at the individual level through legal representation and more broadly by policy advocacy may represent a turning point in that trajectory.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
75
The life course perspective has its origins in longitudinal studies examining individual lives over changing times. For example, Leonard D. Cain, Jr.s essay, Life Course and Social Structure, introduced the concept that aging represents a systematic series of stages or statuses that individuals occupy over time (Cain, 1964; Marshall & Mueller, 2003). Later, the theory was both contextualized and collectivized as Matilda White Riley and others developed four central premises of life-course theory which included the notions that a cohort of aging individuals could both be affected by social and environmental changes and also could produce social change as a group of individuals in the same cohort shared common experiences (Riley, 1973). Organizing family, work, and other societal roles into trajectories shaped and influenced by changes he called transitions, Glen Elder, Jr. is credited with articulating the five principles of the life course, which have emerged as the dominant conceptualization of life-course theory and that are widely used to study human development and aging. The principles are 1) human development and aging are lifelong processes; 2) individuals construct their life course through human agency; 3) the life-course is shaped by historical times and places; 4) events and transitions vary according to their timing over a persons life; and 5) individual lives are linked through a network of shared relationships (Elder & Johnson, 2003). According to Elder, the life course paradigm represented a major change in the way to think about and study human lives by focusing on the social forces that shape and produce developmental consequences throughout life (Elder, 1994). My study theorizes that all five of Elders principles of development are affected by the level of legal power individuals may exercise to enforce their economic and social rights. Specifically, low-income and minority children are thwarted in their ability to experience good health early in life, because the historic inequities, the environment, and the social relationships that surround them are shaped by inequitable distribution of opportunity and resources.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
76
Importantly for this study, life course theory has provided a method to study changes in health experienced by cohorts of individuals as they age. As a result, the perspective has highlighted the possibility that the roots of health inequalities begin in the biological and social experiences that individuals and groups experience early in life, and further, that the relationship between socio-economic inequality and health increases over time. This conceptualization, called divergence theory or the cumulative advantage/disadvantage hypothesis predicts that socioeconomic differences in health increase with age. Differences in social resources between socioeconomic groups produce inequalities early in life, which increase over time and consequently produce increasingly greater health disadvantages. Taking a life-course approach to inequality, in 1968, Merton argued that inequality results from the unequal distribution of resources that supports productivity with recognition leading to further productivity and increasingly working to the advantage of the few, while also working to disadvantage the most (Merton, 1968).
Lisa F. Berkman and Ichiro Kawachi identify three hypotheses that have been proposed by social epidemiologists as life course theories to explain the early social and biological influences in life that affect the onset of disease in middle and later life. I apply their second hypothesis of cumulative disadvantage in my study. Cumulative disadvantage theory posits that disadvantage in early life sets in motion a series of subsequent experiences that accumulate over time to produce disease after years of disadvantage (Berkman & Kawachi, 2000). Catherine Ross and Chia-ling Wu (1996) convincingly applied this cumulative disadvantage theory to health inequality by demonstrating that the health advantages of high income and educational attainment increases with age (Ross & Wu, 1996).


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
77
In 2005, Stephanie Hatch further elucidated the cumulative advantage hypothesis (Hatch, 2005). Hatch predicted that heterogeneity in trajectories through exposure to adversity and advantage are organized around key transitions that increase relative inequality over the life course. According to Hatch, the cumulative adversity processes can take many forms. A single hardship or loss of a single protective factor might trigger a cumulative disadvantage. Alternatively, a chain of contingencies might occur such that each hardship sequentially surpasses its predecessor. In a third scenario, a cascading sequence of layering hardships may build increasingly intensifying effects that begin small but grow into larger consequences.
Regardless of scenario, intervening adversity mediates the relationship between SES and poor health outcomes. Importantly, Hatch (2005) also posits that protective resources that increase social mobility may affect the causal relationship between low SES and poor health outcomes by making the sources of adversity avoidable and by making it possible for individuals to achieve in and attach to social institutions over their life course. I suggest that MLPs may be such a protective resource. I further suggest that by measuring MLPs impact on early childhood experiences with the outcome variables chosen here immunization compliance and residential mobility my study can shed light on the health and social advantages poor and minority children may enjoy over the life course given equal access to the social determinants of health. Conceptual Framework For This Study: The MLP As A Disruptive Protective Resource
In an article that is particularly important for the conceptual model that informs this study, Yoav Ben-Shlomo and Diana Kug (2002) further elaborated the integrated biological and psychosocial pathways that a life course approach to disease suggests. Their diagram in Figure 17 below illustrates that exposures to adversity might take a predominantly biological pathway


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
78
(Path a), a predominantly social pathway (Path b), a socio-biological pathway (Path c), or a biosocial pathway (Path d) (Ben-Shlomo & Kug, 2002).
Figure 1 Schematic representation of biological and psychosocial exposures acting across the life course that may nfluence lung function and/or respiratory disease
Figure 17 Ben-Shlomo & Kug (2002) Biological Psycho-Social Pathways To Disease Disparities Ben-Shlomo & Kuh (2002) capture the complexity of multiple pathways that combine to produce health disparities. My study asserts that regardless of which of these pathways social adversity travels whether that adversity has taken a biological, social, or mixed pathway to produce poor health outcomes an MLP intervention that re-distributes access to the social determinants of health can disruptively weaken and redirect the association between social adversity and poor health to ultimately reduce disparities. I theorize that access to legal services can increase and equalize social determinants of health that affect all pathways between social adversity and inferior health outcomes. In the Figure 16 below, I condense the multiple pathways suggested by Ben-Shlomo and Kug (2002). My conceptual diagram summarizes the various exposures that affect a child s health over the life course into two cumulative and interactive categories of disadvantage. One category contains social disadvantage (e.g., air pollution, poor educational attainment, and occupational hazards from Ben-Shlomo and Kuhs


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
79
diagram), and the other contains biomedical disadvantages such as poor growth in utero, nutrition, asthma, or childhood chest illness from Figure 17.
The diagram of my theoretical conceptualization for this project focuses on the limited and inequitable access to the social determinants of health as the overarching mechanism that operates along all the various pathways to disease disparities that originate during early childhood. The grey arrows in Figure 18 connect limited and inequitable resources to poverty and race, and then show that each of these two independent, explanatory variables lead to inferior health. I intend to suggest, by this diagram, that poverty and race may overlap but are each independently associated with inferior health and are not individually but systemically generated. Moreover, I intend to illustrate in Figure 18 that I adopt Stephanie Hatchs (2005) view, that the adversities, which result from a misdistribution of the social determinants of health, are interconnected and cumulative. For example, housing and education inequity combine to produce employment and income deficiencies, which exacerbate the inability of the poor and of minority families to equitably access health care and healthy food for their children.
Figure 18 depicts the role I theorize MLPs play as a protective resource to disrupt the pathways of cumulative adversity. The blue arrows point to the role that MLPs can play addressing structural inequality through legal and policy advocacy, inequities at the individual and family level through legal services and representation, and health care delivery disparities through training and education. Moreover, these blue arrows illustrate my conceptual hypothesis that MLPs may disrupt all pathways that connect social adversity to poor health outcomes, whether social, biological, or mixed. MLPs may intervene to improve limited and inequitable access to the social determinants of health that prevent poor and minority populations from coping with the effects of biological as well as social adversity. Figure 18 offers a detailed,


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
80
graphic look at my overarching theory that limited and inequitable access to the social determinants of health explains why poverty and race are associated with poor health through a variety of pathways over the life course and how MLPs may disrupt this association.
MLP Improves Access to Social Determinan ts of Health
'' >
Limited and Inequitable Access to the Soda I Determinants of Health
' '
Cumulative Social Disadvantages < * Cumulative Biomedical Disadvantages
.
MLP Improves Access to Social Determinan ts of Health
Figure 18. Conceptual Framework: MLP As Protective Resource To Disrupt Pathways To Health Disparities
I theorize that medical-legal partnerships may provide legal representation as a protective resource and thus potentially weaken the effect of inequities that characterize the social determinants of health. As with other protective resources, MLPs provide services that act on mediators (e.g., social and biological disadvantages) and that can improve causal pathways; MLPs can intervene in any one of the socio-biological pathways that lead to health disparities, thus creating disruptions in the established relationship between race, poverty, and poor health.
In sum, I have applied ecosocial theory, fundamental cause theory, and a theory of cumulative advantage to achieve the specific aims of this study. I sought to determine whether medical legal partnerships are an intervention that can redirect inequitable access to the social


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
81
determinants of health for poor and minority populations in Colorado, thereby moderating the intervening adversity that connects race and low socio-economic status to disparately adverse health outcomes. Although I could not accurately measure changes over the life-course without more extensive longitudinal data, I expected my analysis to show improvement in immunization compliance and residential mobility that is associated with the CHCO MLP services. I predicted, based on the life-course/cumulative advantage theory, that the likely impact of these improvements in childrens lives will last longer than the five-year period of this study. Theoretical Support For The Choice Of Study Measures: Immunization Compliance And Residential Mobility
First, this study considers whether children whose families have legal representation through MLPs are more likely to adhere to immunization schedules recommended by the Advisory Committee on Immunization Practices (ACIP). Childrens immunization compliance is theorized as a proxy for evidence the child has regular access to preventive health care required to adhere to the recommended schedule and objectively better protection against infectious diseases. Of course, immunization compliance only suggests but does not confirm that a patient had regular preventive care, because a number of non-clinical settings in Colorado can offer vaccinations needed to remain immunization compliant, without providing any accompanying preventive care. Nevertheless, it is not unreasonable to use immunization as a proxy for preventive care access because compliance represents some level of clinical encounter.
The ACIP recommends that children receive routine childhood vaccinations by age 2 years, yet the evidence suggests that fewer than 25% of children receive vaccine doses at the appropriate time. 1 Children who do complete the recommended series of vaccinations between birth and age 2 are protected against 14 childhood diseases. Pediatric immunizations have been


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
82
linked to improved health outcomes and health cost savings over the life course. The CDC reports that among U.S. children born during 1994-2013, vaccination will prevent an estimated 322 million illnesses, 21 million hospitalizations, and 732,000 deaths during their lifetimes (CDC, 2014). Economists have estimated that between 2005-2009, compliance with the ACIP immunization program could have prevented 42,000 early deaths and 20 million cases of disease over a lifetime and could have saved approximately $13.5 billion in direct costs and $68.8 billion in total societal costs in the United States (Kurosky, Davis, & Krishnarajah, 2016).
Estimated vaccine coverage among poor children is lower than among children who live above the poverty level. Median vaccination coverage rates are near 95% of national Healthy People 2020 targets for U.S. children generally, however, low vaccination coverage can cluster within poor communities leaving low-income and minority children more vulnerable to vaccine-preventable diseases. In Colorado, however, vaccination coverage patterns are more complex. Overall, vaccine coverage rates are declining among all children in Colorado. In this state, vaccine exemption rates are higher and completion rates are among the lowest in the nation, yet it is wealthier and typically white families who avail themselves of vaccine exemption in Colorado (CDC, 2014). Moreover, although black and Hispanic children are disproportionately represented among Colorados poor, a group that has lower vaccination coverage than wealthier families nationwide, and although Hispanic children are three times more likely to be uninsured than whites in Colorado, Latino vaccination coverage is slightly higher than for whites in this state. This reversal is possibly because the states Vaccines for Children (VCF) program is an effective outreach to low-income and Hispanic families, while Colorados wealthier, white families are more often the demographic group that uses Colorados exemption from


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
83
vaccination. This study will, in any event, provide additional information about the influences that affect the complex picture of vaccine coverage in Colorado.
This study hypothesizes that MLPs may improve ACIP compliance rates for low income and minority children in Colorado in four ways that could reduce health disparities. First, the MLP intervention may provide a childs household with improved access to health insurance to reduce financial barriers that impede access to regular preventive health care for children. When MLP services improve access to Medicaid or CHIP benefits, children may experience better immunization compliance because their caregivers have insurance coverage to pay for preventive health services. Second, when MLP representation improves a familys access to other income supports such as SNAP, TANF, disability, or unemployment benefits, a childs caregivers may redirect funds from providing immediate needs such as food or shelter to cover the cost of routine health care. Moreover, MLPs may also give caregivers the financial freedom to avail themselves of transportation, childcare, or time off work, making health care for children not only affordable but also accessible. Third, my interviews with former MLP clients suggest that MLPs may reduce the stress caregivers experience as a result of unsolved legal problems, thereby reducing underlying psychosocial factors that may stand in the way of immunization compliance (Zhou et ah, 2014). A primary example is resolving immigration status concerns. CHEP has successfully adjusted legal status for several clients who then explained they had one less barrier to obtaining regular health care for their children (Matthew, 2010). Fourth, CHEP faculty and student attorneys have seen families respond to receiving MLP services by engaging in better health behaviors such as missing fewer clinic appointments, suggesting that resolving legal problems may improve self-mastery and control in a way that increases access of preventive health care services such as childhood immunization.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
84
The second outcome variable that this study examines is upward residential mobility. This variable is derived from the absolute number of residential moves participants made during the study period. The frequency of address changes suggests the level of a childs residential stability. Children who enjoy stable living conditions are more likely to belong to households that have food security, well-developed social networks, and social cohesion that significantly influence childrens mental and physical health (Drukker et ah, 2003). In contrast, the evidence shows that housing instability is associated with food insecurity and limited health care access, affecting routine and acute health care utilization in children. Such residential disruptions in a childs life can produce increased susceptibility to health and health disparities throughout a lifetime (Ma, Gee, & Kushel, 2008). Studies demonstrate that housing instability is independently associated with not having a usual source of care, postponed medical care and medications, and increased hospitalizations and ED use. Moreover, researchers have concluded that policies that decrease housing insecurity can promote the health of young children, thereby affecting their health in later life (Cutts et ah, 2011).
However, the outcome measure here is a refinement on a simple count of the number of residential moves a child may have experienced because frequent residential relocations may not always signal instability. Research based on the Moving to Opportunity experiment showed that children whose families move to lower-poverty areas experience significant improvement in mental and physical health, as well as in family safety (Kushel, 2006). In addition, recent research offers compelling evidence that for younger children, moving to lower-poverty neighborhoods has positive long-term economic impacts as well. Moreover, the best evidence is that family moves from high-poverty to lower-poverty neighborhoods are associated with reduced dependence on public benefits, improved health status, and improved school and


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
85
behavior outcomes for children (Ludwig et al., 2013). Chetty, Hendren and Katz showed that improving neighborhoods has a causal exposure effect on childrens educational achievement and earnings outcomes (Chetty, Hendren, & Katz, 2016). Importantly, the length of time that a child spends in improved neighborhood conditions matters to their long-term economic prospects. Specifically, they found that every year spent in a better area during childhood increase[s] college attendance rates and earnings in adulthood. In other words, the gains from moving to a lower-poverty neighborhood are greater for children who are younger when they move. Indeed, Chetty describes the treatment effects at younger ages as substantial: children who move to better neighborhoods when they are less than 13 years old earn annual incomes that are 31% higher on average than children who do not move, while children whose families move to lower-poverty neighborhoods when they are over age 13 may experience a negative impact on their earnings when compared to children who remain in higher poverty neighborhoods (Chetty & Hendren, 2015). Therefore, I consider the MLPs impact on the neighborhood economics of housing changes that pediatric patients experienced during the study period. I hypothesize that the MLP intervention will encourage upward housing mobility. I expect that treated families will show more upward mobility than untreated families, and families will show more upward residential mobility after the MLP intervention than before.
My study considers housing mobility at an important time in Colorados economic history. Colorado is experiencing a decline in affordable housing, especially in the Denver metropolitan region where many of this studys patients live. Colorado wages have not risen as fast as housing costs throughout the state. Therefore, the lack of affordable housing can spell danger for children whose families spend more than 30% of their income on housing. They may find housing affordable only when conditions are overcrowded or substandard. Frequent moves


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
86
and even homelessness may result. Moreover, children may more frequently live in conditions that lead to illness and school absences. Residential security has been linked not only to disruption in school attendance but also to the quality of school instruction. High-mobility students tend to receive slower, repetitive curricular content. They also may lose the benefit of stable social networks. Kidscount Colorado concluded that [f]orty-one percent of highly mobile students are low achievers, compared with 26 percent of stable students. Mobile students are half as likely to graduate (Chetty, Hendren, & Katz, 2016). These are the long-term consequences that I theorize could be changed over the life course if MLP interventions prove effective.
I theorize that MLPs may improve families financial resources to allow them to provide housing for children in better neighborhoods with access to better social capital and networks. MLPs may help low-income and minority families enforce legal rights that protect against eviction, foreclosure, housing discrimination, and other law-based affronts to residential stability. Social mobility during childhood may narrow social inequality during adulthood as a move may also expose children to better schools, safer streets, and increased food security. Where health outcomes are concerned, social mobility may serve as a gradient constraint so that the size of health inequalities among populations that are socially stable are greater than the size of inequalities experienced by the socially mobile (Colorado Children's Campaign, 2013). In other words, the effect of social mobility can be to constrain long-term inequality (Blane et ah, 1999). To the extent that MLPs improve not only financial resources but also the social networks and resources that children are exposed to, the intervention may affect positive social mobility for children that could last a lifetime. In short, this study conceptualizes medical-legal partnerships as a protective resource that mitigates the adversity low-income and minority children experience


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
87
because of inequality. The next chapter presents the research design and methodology for this study.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
88
CHAPTER IV RESEARCH DESIGN
The goal of this study was to examine whether MLP services can improve health and social outcomes of disadvantaged children.
Overview
As more fully described below, I analyzed data from a group of 1,077 pediatric patients who received MLP services and were all under age 3 during the study period. The two outcomes of interest in this study are immunization compliance rates and upward residential mobility. The treated patients were divided into two groups; 985 patients were treated with MLP services without formally becoming legal aid clients, and 92 patients were treated with MLP services as legal aid clients. I obtained retrospective data from both treated and un-treated patients electronic health records and compared the two groups immunization compliance and upward residential mobility. I also compared the MLPs effect on these outcomes for a subset of 272 patients from the Treatment Group, selected using propensity score matching to form a control group for analysis. Next, I examined how the MLP affected the predicted association between patients race and SES and the outcome variables. In order to compare the outcome variables pre- and post-MLP intervention, I calculated the hazard rate as the conditional probability that each child would experience immunizations or upward residential mobility at a given age group. Lollowing the methodology outlined by Singer and Willett, I defined the beginning of time in my study as 2009, when the MLP program began (Garg, Marino, Vikant, & Solomon, 2012). The metric for clocking time was the age of patient participants in years. Lastly, I conducted a qualitative and mixed method analysis on the 92 patients who received MLP services from legal aid. I explored relationships among those patients legal problems, services, and outcomes by


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
89
race and gender, and then explored whether the intensity of legal aid services they received affected the two outcome variables of interest.
Research Design
I employed an explanatory, sequential, mixed methods, retrospective, cohort design to gain several advantages (Leech, Nancy L. And Anthony J. Onwuegbuzie, 2009). First, the mixed-methods design fits the theoretical framework that informs this study. The quantitative analysis of associations between the MLP intervention and the dependent variables that reflect health and social conditions in early childhood sheds light on the ecosocial theory of health disparities that is the theoretical foundation of this study. The mixed methods design tests not only whether the MLP intervention changes the patients outcomes after receiving services but also examines how those outcomes correlate with specific legal services and outcomes reflected in the legal records (Bartley & Plewis, 2007). A second benefit of this design is that it afforded flexibility in addressing research variables that have been under-explored (e.g., immunization compliance) or wholly ignored (e.g., type of MLP service type and residential mobility) in the MLP literature to date. Third, this study design could help to promote collaboration among health providers, legal professionals, policy-makers, and social scientists by using a mixed method approach that relates data from medical and legal sources. Finally, the study design accounted for the respective weaknesses in quantitative and qualitative analysis by combining both methodologies to study the impact of MLP interventions. For example, while a quantitative study of the CHCO data may show the relative extent to which children received health-related benefits from the MLP intervention, the qualitative study added an enriched understanding of whether some MLP services are more beneficial than others. Together, the qualitative and quantitative aspects of this study provided a more contextualized picture of the MLPs impact.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
90
Once approved by the COMIRB in February 2015, (Protocol Number Id-1475, Attachment A) this study proceeded in three phases as shown in Figure 19 below:
\
Phase I
Quantitative analysis of MLP's effects
/
Phase II
Qualitative analysis of relationship between legal aid services and outcomes shown in documentary legal records
Phase III
Mixed methods analysis of relationship between legal issues, MLP services, legal outcomes and health outcomes
\ /
Figure 19. Three Phases Of This Study
I selected the research design based on my MLP experience described in Chapter 1, the literature reviewed in Chapter 2, the theoretical framework presented in Chapter 3, and my interviews with the pediatricians, attorneys, and paralegals who staffed the CHCO MLP from 2009-2013. These interviews helped me to identify the patient population for this study and to understand the MLP treatment model they employed. I then chose an analytical approach to answer the questions that arose based on these experiences and existing data.
DATA
I compiled the quantitative datasets for my study from the de-identified, EPIC electronic
medical records maintained by CHCO and from data on immunizations from the Colorado
Department of Public Health and the Environment (CDPHE). I collected immunization records
and all other health indicators in two person-period datasets in which each patient had a separate


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
91
row for each visit they made to the CHCO clinic. Most covariates in these datasets are static, however each patients age in days from birth is time-varying. I also collected explanatory, demographic information for each patient in a person level dataset and then merged the three files into a single person-period dataset.
I gathered and coded the documentary data for the qualitative phase of this study from legal files obtained from Colorado Legal Services (CLS), the states legal aid law firm that provides legal assistance and advocacy for low-income Coloradans. CLS provided de-identified, redacted copies of the legal files for the CHCO patients for whom Colorado Legal Services entered a formal retainer agreement; from this group, I analyzed data from the patients who were under age 3 during the study period. Lor these patients, I collected data from the following legal documents:
Legal Services Agreement with Colorado Legal Services, Inc. (CLS) This document states the legal problem CLS is retained to address.
CLS Intake/Eligibility Information Sheet Checklist to establish client meets CLS low income and citizenship criteria.
Case notes and correspondence Containing attorneys notes, letters written and received in the case.
Case Closing Checklist and Certification or Closing Memo Which includes dates case opened and closed, code description of case outcome, and code or description of services provided.
Based on these documents, I extracted data from the legal records and then coded fields on an Excel spreadsheet to identify which legal problem, services, and outcome each participant had. Next, I imported the excel spreadsheet into Dedoose, a qualitative research software program developed by sociocultural Research Consultants, a software design group based at UCLA.
Methods


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
92
I analyzed observational and longitudinal data for all 11,553 pediatric patients who visited the CHCO clinics where the MLP operated and who were under age 3 during the five-year study period. Inclusion criteria were based on the information I received in interviews about screening procedures. All patients who visited the clinics were eligible to be screened and therefore eligible to be included in the study. However, patients who visited the clinic were not uniformly screened during the five-year period. During some years, patients were screened one day per week, but it is not clear they were screened on the same day each week. During other years, patients were screened multiple days per week. On still other occasions patients may have been referred to the MLP for screening based on the ad hoc judgment of an individual physician or social worker. Different staff members took responsibility for screening and each followed record-keeping protocols to differing extents. Therefore, I could not rule out selection bias based on the screening procedures. I divided the 11,553 patients into two groups: 4,898 patients who visited the clinic during the study period and were screened comprised one group. The remaining 6,655 patients who visited the clinic but were not screened comprised the un-screened group. Treatment Group
Next, I identified which of the screened patients received the MLP treatment and which patients remained untreated. All staff interviewed agreed that 1) MLP representatives, or in some cases a hospital representative screened for MLP patients by presenting the instrument shown in Attachment B to the caregiver accompanying the pediatric patient to the clinic visit; 2) any patient whose caregiver responded to the questions on Attachment B were considered screened and therefore entered into a the study database; 3) patients whose caregiver answered "Yes to any of the first four screening questions and also answered Yes to the fifth screening question to indicate they wished to receive a contact from the MLP, were considered to have screened


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
93
positive for MLP need(MLP Positive); 4) all patients who screened MLP Positive patients received some sort of MLP services; 5) patients who did not answer with two Yes responses were considered negative for MLP need (MLP Negative) and did not receive MLP services. This interview process resulted in a sample group of 1,077 treated patients. Staff interviews and legal records further revealed that 92 of the 1,077 treated patients were referred to CLS and became legal aid clients (CLS Case patients). The remaining 985 treated patients received other MLP services. Together, these 1,077 patients comprised the Treatment Group. Pigure 20 provides a flowchart to graphically describe the inclusion criteria for this study, which I have confirmed with the CHCO MLP staff. Attachment C contains demographic information about each patient group including distribution by race, ethnicity, and age group.
Eligible Patients Under Age 3 Who Visited Clinics Where MLP Screened
11,553 Patients Ages 0-2.99 years
Screened v. Unscreeened Patients
4,898 Patients Age 0-2.99 years Screened
6,655 Patients Age 0-2.99 years Not Sceened
MLP Positive v. MLP Negative Patients
1,077 Patients Age 0-2.99 MLP Positive for MLP Need
3,821 Patients Age 0-2.99 Negative for MLP Need
Treatment Group (n=l,077)
985 Patients Age 0-2.99 Positive for MLP Need
92 MLP Positive Patients Age 0-2.99 With CLS Legal Aid Case File
Ligure 20 Inclusion Criteria Lor Children's Hospital Colorado (CHOC) MLP Treatment Group Patient Population, January 2009 December 2013
Control Group


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
94
I experimented with several approaches to constructing the control group. First, I considered comparing the 1,077 Treatment Group patients to a control group comprised of the 3,821 MLP Negative patients. However, when I compared these groups, the data suggested that patients who screened positive for MLP need were generally from lower socio-economic populations than patients who screened negative for MLP need. Indeed, several studies confirmed the family and household income levels among Treatment Group patients would be unlike the control group if selected from patients who screened negative for MLP need. For example, in Development of a Brief Questionnaire to Identify Families in Need of Legal Advocacy to Improve Child Health, Dr. David Keller found, families that identified themselves as Hispanic, with lower incomes, and families seen at community health centers were independently more likely than families without those characteristics to be screened positive and referred for legal counsel (Keller et ah, 2008). Gottieb et al found similar results among patients who screened positive in a randomized trial testing electronic versus face-to-face social screening tools (Gottlieb et ah, 2014). Similarly, a group at Johns Hopkins Childrens Center reported the most prevalent need among 1,059 families studied was for employment (Brookhart, et. ah, 2006). And medical homes have regularly confirmed that income related concerns such as the need for employment are the most frequently identified patient concerns in social needs assessments if low income children (Garg et ah, 2008).
The literature correctly predicted the income differences among patients in the Treatment and MLP Negative groups empirically as well. I found a significant difference between the two groups with respect to insurance sources, which I interpreted as a proxy for income. MLP Negative patients were nearly twice as likely as patients in the Treatment Group to have private health insurance, the type of coverage that is more often associated with steady employment.


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
95
Table 1 shows that while only 7.7% of the MLP Positive cohort had private insurance, 14.7% of the MLP Negative cohort had private insurance. Tablel also shows the Treatment and MLP Negative groups were otherwise quite similar with respect to ethnicity and race; medical debt owed; frequency of in-patient hospitalization; number of ER visits; and APGAR scores (a proxy for the health of these patients at birth).
Table 1 Frequency And Mean Comparisons Of Treatment Group And MLP Negative Patient Groups
Characteristic MLP Positive Patient Group MLP Negative Patient Group P-value
Patients (n) 1,077 3,821 -
First Age (years) .44 .44 .00
Race Black (%) 28.0 27.6
White (%) 27.7 40.0 -
Other (%) 44.4 42.4
Ethnicity Hispanic (%) 49.0 44.7 -
Sex (% Female) 45.7 47.3 -
First Insurance
% Public (income 91.1 84.0
contingent) 7.7 14.8 -
% Private % Uninsured 20.5 19.1
% With Medical Debt to CHCO 27.2 26.3 -
ER Visits, mean (SE) 4.7 (.16) 4.0 (.07) .00
Inpatient Hospitalizations, mean (SE) .6 (.04) .4 (.019) .24
Apgar at 1 Minute, mean (SE) 7.6 (.09) 7.5 (.05) .0001
I concluded from this analysis that the differences between the MLP Positive (Treatment
Group) and MLP Negative patient group were too large to use the latter as a source for creating a


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
96
control group. Instead I used the propensity score matching method to construct a control group from the unscreened group to compare with the Treatment Group outcomes.
Propensity score matching
I used propensity score matching as a method to approximate a randomized experiment from which I could infer causality by creating a control group of 1,231 patients from among the 6,655 unscreened patients who had no prior contact with MLP staff or services. The propensity score matching method uses proportional quota sampling to select control group members that have observable pre-treatment characteristics in approximately the same proportion as patients in the Treatment Group. In 1983, Rosenbaum and Rubin introduced the propensity score, a number to represent the conditional probability that a person might be assigned to receive a particular treatment, given a vector of pre-treatment, observable covariates (Rosenbaum & Rubin, 1983). Propensity score matching is an appropriate sampling method for this study that analyzes observational data using retrospective analysis, where I could not employ a randomized technique to sample patients for comparison. Yet the method allowed me to estimate a cause-effect relationship between the MLP intervention and outcome variables of interest.
According to the propensity matching theory, the bias between the MLP Treatment Group and the control patient group was reduced to the extent that this sampling methodology compares individuals from the two groups who have the same value propensity scores. Thus, any differences that remain may reasonably be attributed to the MLP intervention.
Bingenheimer, Brennan, and Earles used this method to estimate the causal impact of exposure to firearm violence and the subsequent perpetration of serious violence (Bingenheimer, Brennan, & Earles, 2005). In their study, these researchers claimed that comparing individuals with identical propensity scores but different realized exposures is [sic.] analogous to conducting


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
97
a randomized experiment, and therefore provides a valid basis for measuring a cause-effect relationship between exposure and outcome" (Richards, 1998). My overall approach to propensity matching follows the analysis reported by Bingenheimer and colleagues in their estimate of the average treatment effect of exposure to firearm violence on serious violent behavior by adolescents. To apply the Bingenheimer methodology, I used Statu's propensity estimators described by Becker and Ichino to execute the propensity score matching method. A limitation of the propensity matching approach is that it cannot account for unobserved variables which may also confound outcomes.
Another important caveat is that this method depends on having appropriate observed, pre-treatment variables to match the patients in the Treatment Group with patients in the control group. The next section describes the pre-treatment variables I selected for to calculate the propensity scores used to match patients in the Treatment Group with patients from the unscreened group in order to construct a control group for this study.
Pre-Treatment Variables
I selected the pre-treatment variables to create a control group for this study based on the propensity score literature, the theoretical framework underlying this study, and my understanding of the MLP policies and procedures from interviews with CHCO MLP staff. I collected data for 202 pre-treatment variables from patient medical records in CHCCf s Epic Systems Corporation, electronic health record software (Verona, Wisconsin). The original data for the pre-treatment variables in this study appeared in Epic either as text (such as descriptive information about race, gender, or age) or as International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9- CM) codes, a standard list of six-character


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
98
alphanumeric codes to describe diagnoses. (See, excerpt from my codebook shown in Attachment D).
The variables I selected initially were guided in large part by a general, though not uniform, consensus in the literature that all available variables that are related to both the MLP exposure and the outcomes should be included in the propensity score model in order to increase the precision of my estimated exposure effect without increasing bias. The logic behind this approach is to reduce bias the distance of the estimated treatment effect from the true effect, and also increase efficiency the precision of the estimated treatment effect (Garrido et al,
2014). Because I am using observational data, the patient characteristics in my datasets are likely to be associated with both the MLP treatment selection and the outcomes of interest. Without propensity scoring, there is likely to be a different distribution of these covariates in treatment and control groups. However, the overall goal of the propensity matching method is to balance covariates between patients who are treated and those who are untreated so that I can isolate the effect of the MLP.
Due to high percentages of missing data, only the 87 pre-treatment covariates listed in Table 2 below were available to use in calculating propensity scores. I divided these 87 covariates into two domains to help identify pre-treatment variables for the model; the domains served as a way to select the covariates that related to the treatment selection and outcomes as discussed earlier. The first domain, Patient Variables, contains codes and descriptive variables about each patient individually. This domain contains ICD-9-CM codes that are related to patients general and specific health conditions, as well as demographic descriptors. These covariates include indicators of underlying mental and physical health conditions could impact a childs regular utilization of health care and could reflect the stability of a childs living


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
99
circumstances. The second domain, Contextual Variables, contains ICD-9-CM codes and variables that relate to the childs likely access to and quality of preventive care and the social stability of a childs family. These covariates also relate to the quality of health care to which the pediatric patients had access. Table 2 shows that the only continuous, time-varying variables in my dataset are those based on patients ages, or the number of encounters patients had with health providers such as emergency department or office visits.
Table 2 Pre-Exposure Covariates For Propensity Score Matching
PATIENT VARIABLES
Childs Demographic Characteristics Childs Health Status
Female Sex Dichotomous Age Group at First Visit Date Ordinal
Race Categorical Avg Temp Per Year Ordinal
Ethnicity Categorical Avg Heart Rate Per Y ear Ordinal
Child in Foster Care/Welfare Custody Dichotomous Avg Weight Per Year (kg) and (lbs.) Ordinal
Childs Visit History at CHCO Avg Height Per Year (cm) and (in) Ordinal
No visits Pre MLP Dichotomous Avg BMI Percentile Per Year Ordinal
Age Entered CHCO Clinic Continuous Childs Acute Complaint
Number Days CHCO Patient Number Days In CHCO Clinic= (Age At Exit Age At Entry) Continuous Strep Throat Dichotomous
Number Days in MLP Program Number Days In MLP Program = (Age At MLP -Age At Exit) Continuous Viral Infection Dichotomous
Childs Overall Health Condition Acute Bronchitis Dichotomous
Death by Age 7 Dichotomous Sinusitis Dichotomous
Death by Age 18 Dichotomous Acute Upper Respiratory Infection Dichotomous
Apgar Score at 1 minute Ordinal Allergic Rhinitis Dichotomous
Mental Illness Diagnosis Dichotomous Acute Pharyngitis Dichotomous


MEDICAL-LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES
100
Childs Specific Health Condition Chronic Illness
Surgical History Dichotomous Diabetes Diagnosis Dichotomous
Low BMI (<5th Percentile) Dichotomous Asthma / Asthma Diagnosis Dichotomous
Hi BMI (>85th Percentile) Dichotomous Obesity Diagnosis Dichotomous
Anemia Iron Deficiency/Anemia Unspecified Dichotomous Overweight Diagnosis Dichotomous
Child w/ Ambulatory Care Sensitive Condition (ACSC) Dichotomous Mental Illness
Specific ACSC Diagnosis Depression/ Depressed Adjustment Disorder Dichotomous
Dehydration/V olume Depletion Dichotomous Anxiety Disorder/ Anxiety Dichotomous
Low Birth Weight Dichotomous Sleep Disorder Dichotomous
Vaccine Preventable Disease Dichotomous Eating Disorder Dichotomous
Pneumonia Dichotomous Reaction Disorder Dichotomous
Traumatic Injury Personality Disorder Dichotomous
Fracture or Sprain Dichotomous Hyperkinetic Conduct Disorder Dichotomous
Burn Dichotomous Mental Retardation Dichotomous
Child Accidents (traumatic accident, all, fire, environment, suffocation, Poison, other poison or other) Dichotomous ADHD Dichotomous

Child V ehicle Accident (Railway, MV, Traffic, Road, Water, Air, Other) Dichotomous Autism Dichotomous
Head Injury Dichotomous Developmental Disorder/ Delay Dichotomous
Crushing Injury, Lower Limb Dichotomous Language Development/Dev Expr Language Dichotomous
Behavioral and Cognitive Issues Developmental Delays Dichotomous
Disturbance of Conduct Dichotomous Mild Cognitive Impairment Dichotomous
Disturbance of Emotions Dichotomous
Cognitive Delays Dichotomous
CONTEXTUAL VARIABLES
Family History Accessing Health Care Family Access to Diagnostic Care
Regularly Scheduled Office V isits Continuous Lead Screening Dichotomous
Missed Office Appointments Continuous Special Exams Dichotomous
ER Visits / 365 Days Continuous Special Screen Pulmonary TB Dichotomous
Hospitalizations / 365 Days Continuous Special Screen V iral Dichotomous


Full Text

PAGE 1

Medical Legal Partnerships: Reducing Health Disparities By Disrupting Cumulative Disadvantages Faced By Vulnerable Colorado Families by Dayna Bowen Matthew A.B., Harvard Radcliffe, 1981 J.D., University of Virginia, 1987 A Dissertation Submitted to the Faculty of the Graduate School of the University of Colorado i n partial fulfillment of the requirement s for the degree of Doctor of Philosophy Health and Behavioral Sciences 2 018

PAGE 2

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES ii 2018 Dayna Bowen Matthew All Rights Reserve d

PAGE 3

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES iii This dissertation for the Doctor of Philosophy degree by Dayna Bowen Matthew h as been approved for the Health and Behavioral Sciences Program by Ronica Rooks Sharon Devine Patrick Krueger Yvonne Ke ller Guenthar Tillman Farley Date: May 12, 2018

PAGE 4

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES iv Matthew, Dayna Bowen (Ph.D., Health and Behavioral Sciences Program) Medical Legal Partnerships: Reducing Health Disparities by Disrupting Cumulative Disadvantages Faced by Vulnerable Colorado Families Di ssertation directed by Research Assistant Professor Sharon Devine ABSTRACT In Colorado, low income and minority children live shorter and less healthy lives than wealthier and white children. Data on childhood asthma, obesity, and infant mortality dispari ties are disheartening. For example, although Colorado has the 5th lowest infant mortality rate nationally, Colorado's black infants die nearly three times more often than the state's white infants; Latino infants are 70% more likely to die during their fi rst year than whites. Data from 2015 not only show the immunization rate for Colorado toddlers continued a troubling downward trend so that Colorado children rank 30th in the nation, but a research collaborative between the University of Colorado School of Medicine and Children's Hospital Colorado reported that "[s]ignificant disparities exist in vaccination delivery by age group race/ethnicity, and socioeconomic status." This dissertation presents a study premised on the understanding that the inequita ble distribution of access to important social determinants of health is an underlying cause of these health disparities among Colorado children. The study examines the impact of medical legal partnerships ( MLP s), an intervention designed to equalize acces s to preventive medical care, decent housing, and other social determinants in order to disrupt the fundamental association between poverty, minority race and ethnicity status, and poor health. MLP s are an innovative

PAGE 5

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES v health delivery model that integrate la wyers into clinical delivery teams to help equalize vulnerable patients' access to the social determinants of health that are guaranteed by law. The MLP studied here operated between 2009 and 2013 between Children's Hospital of Colorado, Colorado Legal Se rvices, and a law school program called the Colorado Health Equity Project. This study retrospectively analyzed quantitative medical and qualitative legal data to examine whether MLP services improved children's health and social outcomes by reducing upstr eam social risk factors that could decrease health disparities over the life course. Specifically, this study analyzed the MLP s impact on young children's vaccination compliance and access to better neighborhood resources due to upward residential mobility Although this MLP did not have a statistically significant impact on either outcome of interest, the methodology developed here suggested a relationship between MLP s and other health outcomes that warrants further study. The form and content of this abst ract are approved. I recommend its publication. Approved: Sharon Devine

PAGE 6

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES vi DEDICATION This dissertation is dedicated to my family: My beloved husband Thomas L. Matthew, M.D., my forever brother, Vincent E. Bowen, III, and my sons and daughters Sara h Griffin, Mark Benjamin, Thomas William, Marion Lewis, Erin Joy, and Amelia Cizwala each of whom has chosen to live a life of consequence that profoundly inspires me, and is destined to help many to persevere and triumph over adversity.

PAGE 7

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES vii A CKNOWLEDGMENTS I thank the University of Colorado COMIRB for approval of my study Protocol Number 14 1475. I also thank to my committee members for generously sharing their intellectual guidance and encouragement. I owe a special debt of thanks to Patri ck Krueger for tirelessly reviewing prior drafts and research designs. However, my deepest gratitude is reserved for Sharon Devine, who exemplified excellence in legal scholarship, research integrity, and commitment to social justice; without these bright lights I would never have found my way.

PAGE 8

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES viii TABLE OF CONTENTS I. INTRODUCTION ................................ ................................ ................................ .......................... 1 5 Health Disparities: Background and Significance ................................ ................................ ....... 16 Health Disparities In Colorado ................................ ................................ ................................ ..... 19 Colorado's Poor And Minority Children ................................ ................................ ..................... 21 Hypothesizing Medical Legal Partnerships As A Solution To He alth Disparities In Colorado ................................ ................................ ................................ ................................ 26 The Role Of Medical Legal Partnerships ................................ ................................ ............ 32 Research Question And Specific Aims ................................ ................................ ................ 41 II. PUBLIC HEALTH LAW LITERATURE REVIEW ................................ ............................... 46 6 Public Health Law Literature Review ................................ ................................ ...................... 46 6 Medical Legal Partnership Literature Review ................................ ................................ ......... 50 0 Observational Studies In MLP Literature ................................ ................................ ................ 54 4 III THEORETICAL FRAMEWORK ................................ ................................ ............................ 64 4 An Ecosocial Theory Of Health Inequality ................................ ................................ ............ 64 4 Fundamental Social Causes Of Health Inequity ................................ ................................ ....... 68 The Life Course Cumulative Disadvantage Theory Of Health Inequity .............................. 74 4 Conceptual Framework For This Study: The MLP As A Disruptive Protective Resource ................................ ................................ ................................ ................................ ...... 77 IV Research Design ................................ ................................ ................................ .......................... 88 Overview ................................ ................................ ................................ ................................ ..... 88 Research Design ................................ ................................ ................................ ........................ 89 Data ................................ ................................ ................................ ................................ ............ 9 0 Methods ................................ ................................ ................................ ................................ ..... 91 2

PAGE 9

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES ix Treatment Group ................................ ................................ ................................ ....................... 92 2 Control Group ................................ ................................ ................................ ........................... 93 4 Pre Treatment Variables ................................ ................................ ................................ ............ 97 Outcome Variables ................................ ................................ ................................ .............. 109 09 Quantitative Data Analysis ................................ ................................ ................................ ...... 1 19 Pre / Post Analysis Within Treatment Group ................................ ................................ ........ 1 22 Qualitative and Mixed Methods Outcome Analysis ................................ ......................... 123 23 V RESULTS ................................ ................................ ................................ ................................ .... 127 The Outcome Variables ................................ ................................ ................................ ....... 127 27 Quantitative Results ................................ ................................ ................................ ............ 128 28 Outcome Means Comparison ................................ ................................ .............................. 129 29 Estimated Average Treatment Effect of MLP ................................ ................................ .... 132 32 Linear and Logistic Regression Analysis ................................ ................................ ........... 136 36 Pre /Post MLP Analysis Results ................................ ................................ ........................ 144 44 Qualitative And Mixed Methods Results ................................ ................................ .............. 150 0 VI DISCUSSION ................................ ................................ ................................ ....................... 157 57 Limitations and Implications For Future Study ................................ ................................ 1 64 62

PAGE 10

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES x LIST OF TABLES TABLE 1. Frequency And Mean Compari sons Of Treatment Group And MLP Negative Patient Groups ................................ ................................ ................................ ............................ 9 5 2. Pre Exposure Covariates For Propensity Score Matching ................................ ....................... 99 3. Comparison Of Matched And Un Matched Group Of Treated And Control Patients ........ 106 4. Standardized Mean Differences: M atched And Un Matched Groups ................................ 1 08 5 Recommended Age Ranges for Vaccine Dose Administration ................................ ............. 1 12 6 Median Family Income Tiers for Patient Census Tracts ................................ ........................ 1 17 7 Upward Residential Mobility Outcome Variable By Patient Cohort ................................ .... 1 18 8 Immunization Dosage Compliance Rate Variable ................................ ................................ 1 27 9 Upward Residential Mobility Outcome Variable ................................ ................................ ... 1 28 10 M ean Comparison, Vaccine Dose Compliance ................................ ................................ ...... 13 0 1 1 Mean Comparison, Upward Residential Mobility ................................ ................................ .. 1 3 1 1 2 Means of Pre and Post MLP Compliance Rates by Age Groups ................................ ......... 1 46 1 3 Frequency of Pre MLP Residential Mobility for Treated Patients by Age Group ............... 1 47 1 4 Frequency of Post MLP Residential Mobility for Treated patients by Age Group ............. 1 47 1 5 Life Table: Pre and Post MLP Immunization Compliance Hazard Rate for Treated Patients ................................ ................................ ................................ ................................ ...... 1 49 1 6 Life Table: Pre and Post MLP Upward Residential Mobility Hazard Rate for Treated Patients ................................ ................................ ................................ ................................ ...... 1 5 0 1 7 Statistically Significant Estimated Average Treatment Effects of MLP On Inpatient Hospitalizations and To tal Missed Appointments ................................ ................................ ... 1 59

PAGE 11

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xi 1 8 Racial Composition of Patients Under Age 3 at First Visit to CHCO MLP Clinics during Study Period ................................ ................................ ................................ ..... 193 1 9. Ethnicity of Patients Under Age 3 at First Visit to CHCO MLP Clinics During Study P e riod ................................ ................................ ................................ ................................ ........ 193 20 D istribution of Children Under Age 3 Included in Study Population ................................ ... 194 2 1 Age Distribution of Treated and Un Treated Patients ................................ ........................... 194 2 2 Age Distribution of Treatment Group Children, by Cohort ................................ .................. 195 2 3 Removing Variables with High Missing Values Increased Resulting Matches ................... 198 2 4 ATT of MLP on Immunization Compliance Rate After Removing Variables With High Percentage of Missing Data ................................ ................................ ............................ 199 2 5 ATT of MLP on Upward Residential Mobility After Removing Variables with High Percentage of Missing Data ................................ ................................ ................................ ..... 199

PAGE 12

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xii LIST OF FIGURES 1 Self Reported Health Disparities In Colorado ................................ ................................ ....... 19 19 2. Colorado Health By Income, 2013 ................................ ................................ ........................... 20 0 3. Colorado Poverty And Deaths Due To Diabetes And Heart Disease ................................ .... 20 0 4. Asthma Racial And Ethnic Health Disparities For Colorado Children ................................ 22 2 5. Colorado Infant Deaths Due To Asthma By Race ................................ ................................ .... 23 6. Colorado Infant Mortality By Race ................................ ................................ .......................... 23 3 7. Colorado Infant Mortality By Race And Income, 2014 ................................ .......................... 24 4 8. Childhood Obesity Among Colorado Children By Income, 2008 ................................ ......... 25 5 9. Limited And Inequitable Access To The Social Determinants Of Health Mediate The Social Gradient Relationship Between Poverty/Race And Inferior Outcomes ................ 3 0 10 Children's Hospital O f Colorado MLP A Disruptive Moderator ................................ ........ 31 1 11 Frequency And Type Of Problems Identified By Vulnerable Minnesotans, 2011 .............. 34 4 12. Inadequate Housing Among Households With Children By Rac e and Annual Income, 2011 ................................ ................................ ................................ ................. 3 7 13 S tudy Specific Aims Summarized ................................ ......... Error! Bookmark not defined. 1 14 Specific Aims And Re search Questions For Each Phase Of Study ................................ ...... 44 4 15 Summary Of Major Observational Studies Of MLP Impact ................................ ............... 54 5 4 16. Racism As A Fundame ntal Cause Of Health Inequality, Phelan & Link (2015) .................... 7 2 17 Ben Shlomo & Kuh Biological Psycho Social Pathways To Disease Disparities ................. 78 18. Conceptual Framework: MLP As Protective Reso urce To Disrupt Pathways To Health ................................ ................................ ................................ ................................ .......... 8 0 19. Three Phases Of This Study ................................ ................................ ................................ ..... 90 0

PAGE 13

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xiii 20 Inclusion Criteria For Children's Hospital Colorado (CHOC) MLP Treatmen t Group Patient Population, January 2009 December 2013 ................................ ................................ 9 4 21 Colorado And USA Immunization Rates Rising But Remain Below Healthy People 2020 Objective ................................ ................................ ................................ ............. 11 1 22 Testing Th e MLP As Moderator Of The Social Gradient Relationship Between Race, Low SES, And Inferior Outcomes ................................ ................................ ........................... 12 0 23 Linear and Logistic Regression Variables ................................ ................................ ......... 122 22 24 Qualitative Data Analysis Approach Based On Miles, Huberman, Saldana (2014) ....... 124 24 25 Low To High Intensity Legal Services Within The CLS Case Cohort Of Treatment Group ................................ ................................ ................................ ................................ ......... 1 25 26 Summary of Average Treatment Effect Results (ATT and ATE) on Outcome Variables ................................ ................................ ................................ ................................ ... 1 36 2 7 Odds Ratio for High Upward Residential Mobility For MLP Patients ............................ 142 42 2 8 Logistic Regression Results For High Residential Mobility Outcome Variable ............. 143 43 29. Linear Regression Results For Association Between Compliance Rate And High Intensity M LP Services ................................ ................................ ................................ ............ 1 43 3 0 Logistic Regression Results For Association Between High Upward Residential Mobility And High Intensity MLP Services ................................ ................................ ........... 1 44 3 1 Frequency Of Pre And Post MLP Immu nization Compliance For Treated Patients ...... 145 45 3 2 Comparison Of Pre And Post MLP Treatment Frequency Of Upward Residential Mobility Moves ................................ ................................ ................................ ........................ 1 48 3 3 Race Ethnicity, And Gender Of 92 CLS Case Patients (B=Black, H=Hispanic, W/H=White/Hispanic, O=Other, O/H=Other/Hispanic, W=White, A=Asian ..................... 1 5 1

PAGE 14

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES xiv 3 4 Types Of Legal Problems, 92 CLS Case Patients (MED=Medical Insurance, INC+Income/Public Benefits, FAM=Family Law, HSE=Housing Law .............................. 1 52 3 5 Types Of Legal Services And Legal Outcomes For 92 CLS Patients (C/E= Counseling And Education; LTR=Letter On Client's Behalf; APP=Appeal Of Adverse Decision; PER=Personal Court/Administrative Appearance ................................ ................ 1 53 3 6 Intensity of Legal Services and Outcomes Among 32 CLS Case Patients By Race And Ethnicity ................................ ................................ ................................ ............................ 1 54 3 7 Leg al Outcomes by Race/Ethnicity for 92 CLS Case Patients ................................ .......... 155 55 38. Immunization Compliance (Red) And Upward Residential Mobility (Blue) By MLP Service Intensity For 92 CLS Case Patients ................................ ................................ ........... 1 56

PAGE 15

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 15 CHAPTER I INTRODUCTION El iminating racial, ethnic, and socioeconomic health disparities remains one of the nation's top medical, scientific, and political priorities. Yet in Colorado, as in the nation overall health e qui ty for all regardless of income, race, or ethnicity remains a tragically elusive goal notwithstanding commitment from the highest levels. Existing efforts to combat health disparities are falling short. The United States Department of Health and Human Services' Agency for Healthcare Research and Quality (AHRQ) rele ased a report in 2012 showing that the fight against racial health disparities has stalled. Between 2002 and 2010, over 80% of all measures comparing the quality of medical care that whites and minorities receive in this nation have remained unchanged. Th e data show even less progress in narrowing the gap for measures of disparate access to care. Between 2002 and 2009, some access gaps between white and Asian Americans narrowed, but for all other groups white patients continued to enjoy consistently better access to health care than minorities (United States Department of Health and Human Services 2012 ) Similarly, research ers have shown only limited progress in reducing socioeconomic disparities. For example, Paula Braveman presented comprehensive data r elated to socioeconomic d isparities in the United States in 2010; Braveman reported that for 11 adult and child health indicators, the least educated and lowest income members of American population suffered the worst health outcomes, and improvements in h ealth were generally seen at higher income levels (Braveman Cubbin, Egerter, Williams, & Pamu k 2010). This study examines the impact of an intervention designed to disrupt this fundamental association described by the "social gradient."

PAGE 16

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 16 Medical Legal Pa rtnerships ( MLP s ) are an innovative health delivery model that integrates lawyers as part of the health delivery team to help equalize access to the social determinants of health. MLP lawyers and health care providers collaboratively deliver medical and l egal services in the clinical setting in order to mitigate the legal barriers to the fundamental determinants of healthy living that poor and minority patients face every day. MLP attorneys combat discriminatory practices in housing and employment argue to reverse erroneous public benefit denials enforce disability laws that grant access to quality education, and assist patients in a variety of legal actions that enforce existing legal rights to access the social determinants. Because law plays a central role in organizing, regulating, and distributing both the medical and social determinants of health, this study examines whether MLP s can help reduce disparities and increase health equity. Health Disparities: Background a nd Significance Two main reason s underlie the persistence of health inequity in America. First, the United States dedicates few resources to ensuring that all populations have a healthy social environment in which to live, work, and play. Populations especially vulnerable ones need much more than access to good health care to be healthy. Indeed w hile social scientists recognize five factors that may contribute to health outcomes ( M c G overn Miller, & Hughes Cromwick, 2014) the social factors are the most important (Tarlov 1999). These social factors include the social environment (e.g. i ncome gender, disability, and race discrimination transportation and food access ), the physical environment (neighborhood and housing conditions), (Heiman & Artiga, 2015) and access to health care ; th ey have at least as strong an association with poor health outcomes as genetics, biology, or effects of individual behaviors related to diet, smoking, exercise, or alcohol consumption (Forchuk, Dickins, & Corring, 2016)

PAGE 17

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 17 Yet the United States spends far le ss on ensuring adequate access to the social needs of populations than most other developed nations. A recent study published by the RAND Corporation importantly confirmed that countries with greater social expendit ure have better health outcomes than cou ntries that spend relatively more on medical care (Rubin, et al., 2016) Moreover, the association between higher social spending and better health outcomes holds within the United States as well. Elizabeth Bradley and colleagues found that states with a higher ratio of social and public health spending ( for housing, food and income benefits, etc.) t o healthcare spending ( for Medicare and Medicaid reimbursement ) had significantly better health outcomes for adult obesity, asthma, mentally unhealthy days, da ys with activity limitations, and mortality rates for lung cancer, heart attack, and Type 2 diabetes (Bradley, et al., 2016) My study proceeds from the understanding that eradicating health disparities depends principally upon ensur ing access to the socia l determinants of health. The second major contributor to the prevalence and persistence of health disparities is inequality. Specifically, social inequality that distorts access to the social determinants of health may in fact be the most powerful pred ictor of disparate mortality and morbidity than any other single or combined set of known factors (Marmot & Brunner 2005) Since the Whitehall Studies (Marmot & Brunner 2005) convincingly documented evidence of the near uniformly inverse relationship bet ween health, health behaviors, and social class in England knowledge about the "social gradient" has shed considerable light on the crucial role that an inequitable distribution of the social determinants of health plays in producing health disparities. The social gradient describes the relationship between the circumstances in which people live, play, and work and their poor health outcomes (Marmot & Brunner, 2005) These social determinants of health are essentially the "causes of the causes" (Institute of Medicine 2011) of health disparities. As such,

PAGE 18

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 18 they are an indispensable component to improving population health outcomes and reducing health disparities (Davey Smith 1998) To the extent that the social determinants are inequitably organized and d istributed, poor or minority communities suffer decreased opportunities to be healthy. The Whitehall S tudies (Marmot & Brunner 2005) revealed not only the fact that people with lower socioeconomic status (SES) bear at least twice the risk of shorter lif e expectancy and disease morbidity than people with higher SES, but also that universal access to health care such as the British have enjoyed since the end of World War II did not disrupt the social gradient. Instead, the Whitehall studies confirmed that conditions at work, home, and in communities primarily account for the social gradient (Marmot & Brunner 2005) Considerable evidence has further demonstrated that low income and minority populations suffer inferior health outcomes because they disproport ionately lack access to decent housing, safe recreational spaces, quality education, fair employment, income supports, and food security ( Woolf and Braveman, 2011) Inequality constructs the environments in which poor and minority people live, work, and pl ay. Moreover, these social inequalities also adversely influence health behaviors so that individuals living in poverty and racial segregation are more likely than others to smoke, eat poorly, consume alcohol excessively, develop drug dependence, and lead sedentary lives that result in inferior health outcomes (Pampel & Krueger, 2010). Based on the association between inequitable distribution of the social determinants and poor health outcomes, this study posits that improved distribution of the positive s ocial determinants wi ll also reduce health disparities. It tests this theory using data from a patient population in Colorado, where there is substantial evidence of health inequity.

PAGE 19

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 19 Health Disparities In Colorado Low income and minority Coloradans do n ot have the same access to the social determinants of health as wealthier and white Coloradans. Therefore, the disadvantaged experience inferior health outcomes. The Colorado Commission on Affordable Health Care released a report in March 2016 that ident ified self reported health differences among Colorado adults by income, race, and sexual orientation. Figure 1 summarizes the data from that report, which confirms that white, wealthier patients report better health than low income and minority patients. Figure 1 Self Reported Health Disparities In Colorado Income disparities perpetuate health disparities by ensuring an unequal distribution of the social determinants of health. Figure 2 shows that household income is closely ass ociated with the quality of physical, mental, and oral health that Coloradans report.

PAGE 20

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 20 Figure 2 Colorado Health By Income, 2013 Figure 3 shows that place matters. Average death rates due to diabetes on the left and heart diseas e on the right are shown for Colorado's highest (red) and lowest (blue) poverty counties in Colorado. Poor people die at a higher rate than their well off counterparts who have the same disease These data show that beyond subjective self reports, objecti ve data also reveal the inequities that affect all social determinants of health. Lowest Poverty Counties Highest Poverty Counties Diabetes Coronary Disease Figure 3 Colorado Poverty And Deaths Due To Diabetes And Heart Disease

PAGE 21

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 21 Thus socioeconomic, racial, and ethnic health disparities in Colorado remain a pervasive and persistent public health problem. Colorado's Poor And Minority Children Pover ty is a key determinant of poor health in children, and Colorado has one of the fastest growing child poverty rates in the nation (Office of Health Disparities 2009) Between 2000 and 2011, the poverty rate in Colorado increased from 10% to 18%, leaving m ore than 1 in 6 Colorado children living in poverty (Colorado Children's Campaign, 2013) Denver County has the second highest poverty level in the state with 26.2% of all children living in poverty. In 2016, KIDS COUNT Colorado (Co lorado Children's Campaign, 2016 ) reported that despite the good news that Colorado's overall child poverty rate of 15% had declined for two consecutive years, sign ificant racial, ethnic, and economic disparities persist. In Colorado, only 8% of white children live in poverty, while 27% of Latino children and 31% of black children are poor, according to the Colorado Department of Public Health and Environment. Pred ictably, low income and minority children in Colorado are less healthy than wealthier and white children. B lack and Latino families in Colorado earn substantially less than white and Asian families. As a result, Colorado social indicators confirm that Co lorado children from low income, black, and Hispanic families suffer disparate access to the social determinants of health. In Colorado and in Denver County, the regional home to the majority of patients in this study, food insecurity and nutrition data f urther demonstrate health disparities among Coloradans. According to the Colorado Department of Public Health and Environment (CDPHE), 20.4% of Colorado's white children ages 1 14 sometimes relied on low cost foods during the past year, while 57.6% of Ame rican Indian/Native Alaskan children, 49.0% of Latino children, and 34.4% of African

PAGE 22

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 22 American children consume low cost foods These inequities are directly related to vulnerable children's disparate health outcomes. Colorado's health disparities data on asthma, obesity, and infant mortality provide evidence of a fundamental association between poverty, race, and poor health for Colorado's children. During the period from 2006 2007, Figure 4 shows the asthma prevalence was 12.5% for black children, 7.7 % for white children, and 7% for Hispanic children. During the same period, death rates among persons with current asthma were higher for blacks (3.4 deaths per 100,000 persons) than whites (1.9 deaths per 100,000) ( Jacobellis et al., 2008). Figure 4 Asthma Racial And Ethnic Health Disparities For Colorado Children The incidence of chronic diseases such as asthma disproportionately impacts the health of Coloradans of color as compared to whites in the state, as shown in Figur e 5

PAGE 23

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 23 Figure 5 Colorado Infant Deaths Due To Asthma By Race Colorado's overall infant mortality data are particularly disturbing. Although Colorado has the 5th lowest infant mortality rate in the nation, Colorado's black infants die three times more often than the state's white infants. As shown in Figure 6 twelve in every 1,000 live births to African American women end in death. Latino infants are 70% more likely to die than whites. (Wilcox, 2016) Figure 6 Colorado Infant Mortality By Race Importantly, socioeconomic status does not fully explain these disparate outcomes as demonstrated by the differences in birth outcomes that black and white mothers at different

PAGE 24

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 24 income levels experience in Colorado Figure 7 shows that unlike other health disparities, African American infant mortality gaps do not narrow with improved socioeconomic status. Black mothers who earn $50,000 $75,000 per year suffer the same infant mortality rate as black families earni ng less than $15,000 annually. Middle income blacks have infant mortality rates that are twice the rate of white families below the poverty line and over 4 times as high as white middle income families. Figure 7 Colorado Infan t Mortality By Race And Income, 2014 Obesity data similarly confirm that health disparities remain prevalent in Colorado; t he s e data also portend continued health disparities over the life course for racial and ethnic minorities. As seen in Figure 8 the percent of Colorado children who are obese and therefore exposed to higher rates of disease follows the familiar pattern of the social gradient in which children of low

PAGE 25

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 25 income parents suffer greater rates of obesity than children of wealthier, better educa ted families ( Graham 2008) Source: Graham 2008 Figure 8 Childhood Obesity Among Colorado Children By Income, 2008 Morbidity and mortality among poor and minority children outpace wealthier, non minority children at ala rming rates in virtually every leading disease category providing further evidence of the social gradient Sadly, race and economic disparities plague Coloradans' access to health care as well. In 2015, the immunization rate for Colorado toddlers cont inued a troubling downward trend, dropping significantly so that Colorado children rank 30th in the nation (Colorado Health Foundation, 2015). In 2011, 75.8% of parents got recommended immunizations for children between 19 and 35 months old, but in 2013 on ly 69.2% of toddlers completed recommended immunizations at 35 months of age, down from 80.3% in 2007. This decline does not bode well for the long term health of Coloradans generally or low income and minority Coloradans specifically. Indeed a research collaborative between the University of Colorado School of Medicine and Children's Hospital Colorado reports that "[s]ignificant disparities exist in vaccination delivery by age group (e.g. Child vs. Adolescent vs. Adult ), race/ethnicity and

PAGE 26

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 26 socioeconom ic status (ACCORDS, 2016) On a national level, evidence of racial and socioeconomic disparities in vaccine coverage led the CDC to implement the Vaccines for Children Program (VFC) in 1994. Colorado's VFC program leverages federal funding to provide low or no cost vaccines for low income children. Colorado law requires documented immunization compliance for all children attending kindergarten. Nevertheless, the best evidence is that although racial and ethnic disparities in immunization compliance have na rrowed, gaps persist in Colorado despite the success of the VCF program (Colorado Health Institute, 2005) ; these gaps predictably contribute to health disparities in Colorado as in the nation. Hypothesizing Medical Legal Partnerships As A Solution To Healt h Disparities In Colorado The MLP approach is founded on the premise that addressing inequities in poor and minority patients' incomes, housing, educational attainment, personal safety, and other social determinants will improve health outcomes. Today near ly 300 MLP s operate in 155 hospitals, 139 health centers, and 34 health schools across the country. (National Medical Legal Partnership Center, 2017) Often the inequitable distribution of social determinants is due to legal problems or problems such as hou sing code violations or improper termination of benefits In these cases, it is fair to say that unmet legal needs are a social determinant of health. MLP s are designed to bridge the gap between the need to address unmet legal issues that exacerbate patie nts' health problems and assist the medical provider who is treating patient health problems but does not have the time or resources to include solving legal problems in the care model MLP s train clinicians to regularly screen patients for health harmin g legal needs and work through in clinic attorneys to address those needs preventively before they become serious In

PAGE 27

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 27 contrast, providers in most traditional health care settings that do not screen for unmet legal needs discover a patient's health harming, social problems only once the patient is in crisis ( e.g., evicted, without heat, suffering elevated blood lead levels), if at all. Even when the astute physician practici ng in a traditional setting suggest s that a patient consult an a ttorney outside the c linic, vulnerable patients are unlikely to get the help they need. This is because without any prior relationship, a low income patient seeking help from an attorney who is unaware of the health impacts caused by the legal crisis, and quite likely under re sourced, could be slow to resolve the legal issue and the associated adverse impacts on patient health may be prolonged. MLP attorneys provide three core services First, MLP lawyers provide legal representation to address adverse social conditions for which there are legal remedies and which have the potential to improve patient health. Examples include requiring landlords to remove lead paint, toxins or mold, appealing wrongful public benefit terminations, and enforcing educational accommodations for disabled children. Second, MLP s reform health and legal institutional practices by training clinical providers to screen for and identify patients' social and legal needs during office visits. The goal is to identify these needs while they may be addressed preventively in the same way that physicians seek to provide preventive rather than crisis medical care. Third, MLP s advocate for structural policy changes at an institutional, local, state, and federal level. MLP attorneys bring a "patient to policy" per spective, identifying needs in the communities they serve and then working to improve policies and laws that impact those communities and ultimately the social determinants of individual and population health. Colorado's first MLP began operating in 2009 at the Children's Hospital of Colorado (CHCO). This MLP operated in two CHCO clinics (the "CHCO MLP "). A total of 31,844 pediatric patients visited the two CHCO clinics from April 2009 to December 2013. The CHCO

PAGE 28

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 28 MLP attorneys used a brief survey to screen approximately 7,000 children of those 31,844 children from all age groups to identify those who "screened positive" because they identified an MLP need and requested contact from an MLP representative. The CHCO MLP served all patients who screened positiv e for need and of this cohort that screened positive, Colorado Legal Services (CLS) opened legal case files on 150 patients. The CHCO MLP provided legal services to patients at Children's Hospital for five years before it closed in December 2013. I worked with the Colorado Legal Services attorneys and CHCO MLP staff during its last year before it closed. In the following year, with the help of CLS attorneys, I co founded the Colorado Health Equity Project (CHEP) to continue building MLP s in Colorado. CHEP s mission is to remove legal barriers to better health outcomes for low income Coloradans by forming medical legal partnerships. In 2013 CHEP opened two new MLP s in order to expand this collaborative work and fulfill the three core components of the MLP mo del and for one year worked with CLS on CHCO patient cases. CHEP introduced students from the Colorado School of Public Health and the University of Colorado Law School to work with pro bono private attorneys as well as the CLS lawyers in serving vulnerabl e patients. CHEP opened MLP s at the Colorado Center for Refugee Wellness and the Salud Family Health Center in Commerce City. During its first year of operation, CHEP MLP s provided legal representation to six families, introduced training to frontline prov iders in two safety net health clinics, and provided public policy advocacy on several legislative initiatives during 2014. A CHEP student again worked on health equity advocacy in the 2016 Colorado General Assembly session. Moreover, MLP s initiated by CHE P served an additional 8 patients in 2015 and screened approximately 200 patients in 2016. In 2017, CHEP represented patients from the Globeville and Elyria Swansea neighborhoods in anti pollution litigation

PAGE 29

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 29 related to the Interstate 70 highway expansion. CHEP has been approached by at least two other major health providers to form new MLP s in collaboration with public and community hospital systems. The point of recounting CHEP's success is twofold. First, CHEP's good work addresses but a fraction of the need for health related legal services in Colorado's poor and minority patient populations. Second, despite my first hand experience with the CHCO and CHEP MLP s neither the real impact nor the optimal mi x of MLP services is well understood in Colorado or in the nation. Therefore, this study addresses an academic gap in the knowledge of how well MLP s work and also explores an incomplete pragmatic understanding of what type of legal services have the greatest impact on helping vulnerable populations. This st udy has the potential to have tangible impact on the health and social outcomes that poor and minority Coloradans experience by improving the knowledge base required to expand and better structure medical legal partnerships that provide services that demon strably improve patient outcome s throughout the state Figure 9 provides a graphic summary of my conceptual framework for understanding the relationship between poverty, race, and poor outcomes In this figure, limited and inequitable distribution of the social determinants of health explain (or mediate) the social gradient relationship between poverty and race on the left, and inferior health outcomes on the right (Baron & Kenny, 1986) Poverty and Race Inferior Health and Social Outcomes Limited and Inequitable Access to the Social Determinants of Health

PAGE 30

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 30 Figure 9 Limited And Inequitable Access To The Social Determinants Of Health Mediate The Social Gradient Relationship Between Poverty/Race And Inferior Outcomes Theoretically, as MLP s improve access to the social determinants, they improve access to healthy food, clean and decent housing, full educational opportunities, violence free home s and other social determinants (Regenstein et al., 2018). Thus, MLP s may also improve the opportunity to experience good health by disrupting the social gradient. Figure 10 illustrates the MLP role this study a nalyzes; I ask whether MLP s can weaken the relationship between poverty and race (independent variables) and poor health or social outcomes (dependent variables). Put another way, discussed further in Chapter 3, this study examines the moderator effect MLP s may have on the social gradient.

PAGE 31

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 31 Figure 10 Children's Hospital Of Colorado MLP A Disruptive Moderator Some researchers have attempted to identify the precise causal mechanisms that link specific soc ial determinants to poor health outcomes. Several studies have connect ed poverty and mortality (Brooks Gunn & Duncan, 1997) Others identify a strong relationship between income inequality and mental as well as physical health disparities (Marmot 2005) A number of studies have identified poor housing conditions (Krieger & Higgins 2002) stressful work environment s (Siegrist, 1996) and educational disparities (Zimmerman, Woolf, & Haley, 2015) as social mechanisms that have a deleterious impact on health Importantly, poor social conditions such as low income (Sorensen, Barbeau, Hunt, & Emmons, 2004; Barbeau, Krieger, & Soobader, 2004) and low educational attainment have been tied to resulting poor health behaviors such as Predictors: Poverty, Race and Ethnicity Moderator: CHCO Medical Legal Partners hip (Disrupting Cumulative disadvantages that limit access to health care, housing, income,)and Family Stability) Interactive Terms: Race x MLP Low SES x MLP Inferior Health and Social Outcomes (Immunization Compliance Rate and Upward

PAGE 32

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 32 smoking, alcohol consumption an d dr ug use The exact extent of the causal relationship between these social risks is a matter of debate In this study, I posit that in order to disrupt the association between higher mortality and morbidity on the one hand, and poverty race and ethnic ity on the other hand, effective health interventions must address inequality that affects both upstream and downstream social causes of health disparities. The conceptual framework for this study assumes that a comprehensive approach to eradicating dispa rities addresses the fundamental inequality that characterizes downstream social determinants such as access to preventive medical care and also the inequality that affects upstream social determinants of health like housing and education. Because law is a particularly useful tool for addressing social inequality broadly this study posits that low income populations require legal representation to address the inequality that characterizes all social determinants of health, which stand as a barrier to good health itself. In other words, this study identifies un met legal needs as a social determinant of health that, once addressed, can improve health outcomes for poor and minority families. My mixed method approach examines the health related impacts of a legal intervention aimed at equalizing access to two social determinants of health regular preventive medical care a nd improved neighborhood resources for vulnerable Colorado families. The Role Of Medical Legal Partnerships Many existing laws guarante e patients' access to social determinants. However often the laws are unenforced or enforced inequitably. Therefore, vulnerable patients can face an impossible situation in which they require legal assistance to gain access to social supports essential fo r healthy living but cannot find or afford an attorney to represent their cause. The evidence is that millions of A mericans are in this position (Legal Services Corporation, 2017).

PAGE 33

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 33 Families need legal help fighting unlawful eviction and compelling landlo rds to fix building code violations that produce unhealthy substandard living conditions. Victims of domestic violence need representation in court to obtain protective orders and to preserve safety for their homes and children. Workers and indeed entire c ommunities may be exposed to excessive toxins or other environmental health hazards unless an attorney intervenes. In these and many other circumstances, limited access to legal services disproportionately harms the health of vulnerable populations that ha ve the greatest social needs and the fewest resources to address them. For these families, t he MLP model integrates legal assistance to address their unmet legal needs as a vital component of patient health care The need for MLP representation is partic ularly acute among the poor. Studies consistently show that low income people have significantly more unresolved civil legal problems than higher income people and that low income people are less likely to obtain legal assistance for their problems (Greene 2016) As a result, a high proportion of the legal needs related to housing, family, and consumer issues that low income families face go es unaddressed (Scherer, 2016) While evidence of the relationship between unmet legal needs and health problems migh t be found in every state, Minnesota's state bar association has gathered exemplary and detailed data. A qualitative study of Minnesota's low income population shows the intersection between physical and mental disability and these unresolved legal issues in data that are representative of populations around the country. These frequency data ha ve been uniquely collected in Minnesota, though the data are shown in F igure 1 1 below as representative of what might be seen in Colorado and nationally based on the literature an d experience I have had representing MLP clients. Housing and healthcare problems were the leading problems that poor people with physical or mental disabilities suffered, and the vast majority of all problems

PAGE 34

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 34 they identified would likely req uire legal services to resolve (Minnesota State Bar Association, 2011) Figure 11 Frequency And Type Of Problems Identified By Vulnerable Minnesotans, 2011 Similarly, a survey of 600 Colorado patients conducted to determine the need for a medical legal partnership in Denver, confirmed similar problems are prevalent in this part of the state. A majority of low income Colorado patients surveyed (66%) indicated that they had a legal problem over the past 12 months that adversel y affected their health (Miller and Ayers, 2013) These patient responses reflect the fact that unmet legal needs are a social determinant of poor health (Tyler 2012) and potentially a mechanism that influen ces the cumulative disadvantages that lead to di sease risk s over the life course in Colorado's low income and minority children. Un represented (or pro se) litigants in civil matters suffer much worse outcomes than those with legal representation (Engler 2009) and are therefore most vulnerable to th e health effects of unmet legal needs. For these people, the fact that a law exists to protect their rights is meaningless without an attorney to help enforce the law on their behalf. Colloquially, MLP s are often said to improve vulnerable patents healt h by making the law "on the streets" conform to the law "on the books." More formally, the difference between the civil legal needs of low

PAGE 35

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 35 income people in the state and the availability of attorneys to address those needs is called a "justice gap." The ju stice gap in Colorado is considerable. Recent statistics show that in Colorado domestic relations cases approximately 75% of litigants are unrepresented. In Colorado county courts which handle cases common to low income families such as housing evictions and financial collection cases, approximately 98% of people are unrepresented. Colorado Legal Services, the federally funded pro bono legal services organization for the entire state, has only 51 lawyers. Thus, given that approximately 11% of Colorado's 5.5 million people live in poverty, the ratio of available Colorado licensed lawyers to all Coloradans is more than 1:200, while the ratio of available lawyers to low income Coloradans is a dismal 1:11,800 (Colorado Supreme Court Library 2016) These Colo radans, like most low income Americans have a broad range of civil legal problems (Legal Services Corporation 2017 ) that have a deleterious effect on health. L imited access to attorneys who can help address them (National Center for Access to Justice 20 16) contributes to their inferior health status. Former HHS Secre tary Sylvia Mathews Burwell explained, "c ivil legal aid ensures that more Americans have access to good nutrition, safe housing and other basic human necessities that are essential to overall health and well being" (White House Legal Aid Interagency Roundtable 2016) Scholars also have begun to recognize that the inequitable enforcement of laws that control s social determinants is a mechanism through which social structures contribute to the health harming disadvantages experienced by low income and minority populations (Burris, Kawachi, & Sarat, 2002) Therefore, when an MLP gives patients from vulnerable communities legal representation, the MLP itself can serve as mechanism to equalize ineq uitable social structures that are then translated into more equitable levels and distributions of health (Burris, 2001) Housing provides an illustrative example

PAGE 36

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 36 Numerous studies associate poor housing conditions with a broad range of health problems including asthma, lead poisoning, developmental delays, heart disease, and neurological disorders (Desmond & Bell, 2015) More than one researcher has called the state of inadequate housing in poor American communities a "public health crisis" (Bashir, 200 2) Two million people in the U.S. occupy homes with severe physical problems while 4.8 million occupy homes with moderate physical problems (Krieger & Higgins 2002 ) Annually, 13.5 million non fatal injuries occur in and around U.S. homes; 2,900 people d ie in house fires, and two million people visit the emergency room for asthma related illness. One million children in the U.S. have blood lead levels high enough to adversely affect their intelligence, development, and behavior. In Colorado, nearly one in three children live s in families who cannot afford decent housing (Colorado Children's Campaign, 2016) Statewide, the number of Colorado children who were homeless in 2014 increased to over 24,600 (Robles, 2016) Figure 1 2 demonstrates the fact that low income and minority children are the populations most vulnerable to the health hazards that result from substandard, unaffordable, or overcrowded housing.

PAGE 37

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 37 Figure 12 Inadequate Housing Among Households With Children By Race And Annual Income, 2011 An abundant body of local, state and federal laws organize the rights vulnerable families have to clean and safe housing, free from mold, pest infestation, toxins such as lead paint or radon, and adequate heat in cold weather. These l aws include the Fair Housing Act which prevents discrimination, state building codes which require safe and sanitary housing, and landlord tenant laws, which equalize the power relationships between poor renters and their wealthier landlords. However, l aws that regulate housing safety and sanitation issues tend to be under enforced in low income communities (Ross, 1993) S ubstantial evidence shows that legal representation improves housing code enforcement (Ortiz, 2014) and further that code enforcement can improve health. The majority of tenants in housing courts nationally do not have legal representation, while most landlords have attorneys (Engler 2009 ) There is a shortage of a ttorneys to help a poor asthmatic child's family with housing problems b y enforcing building sanitation codes, landlord tenant laws anti discrimination regulations, and improving fair access to public benefits that reduce homelessness As a result, an asthmatic child's health outcomes may remain poor despite access to medical care simply because the family lacks access to legal

PAGE 38

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 38 help that could improve the family home environment one of the most important social determinants of health. Still, h ealth care providers seldom regard legal services as an important part of the heal th delivery model. Moreover, health policy experts rarely regard legal representation as a tool to fight health disparities. MLP s seek to transform fragmented and siloed approaches to health disparities, which separate access to legal services from acces s to health care delivery. The existing approach works to the detriment of low income and minority patients who disproportionately suffer inequitable access to health care legal services, and the social determinants of health. MLP s seek to deliver integra ted medical care and legal representation to vulnerable patient populations as a way to break the chain of disadvantage and cumulative inequality that leads to health inequity Other researchers have posited an association between the MLP intervention's a bility to solve patient's legal problems and improved patient health. However, generally, these studies have involved small numbers of patients, a single disease or condition, or one type of MLP service. Moreover, previous estimated associations fail to c ontrol for the possibility that commonly shared personal, family, and structural circumstances may also confound the relationship between MLP services and improved health and social outcomes. Therefore, the extent to which statistical associations have ide ntified a cause and effect relationship between MLP s and better health outcomes remains unclear. Admittedly, the randomized clinical experiment is could be done prospectively, but the CH CO MLP did not design an experiment to test causation. Therefore, usin g the observational data collected for treatment purposes, I conducted a retrospective analysis to analyze the MLP 's impact. I have chosen a statistical method that reduces confounding bias, compares treated to comparison group populations, and in

PAGE 39

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 39 this stu dy involves over 2,000 participants who presented with a broad range of medical conditions and received a spectrum of MLP services. I have use d the propensity scoring method to estimate the causal relationship between MLP services as a treatment variable, and two outcomes children's immunization compliance and upward residential mobility Both of these outcome variables were refined during the project. Originally, I considered a dichotomous variable called immunization completion, but on the recommendat ion of my committee, I changed this to immunization compliance to examine a continuous variable over time. Also, I originally considered residential instability, counting all moves as a measure of instability but changed to upward residential mobility to minimize confounding bias, conform to the theoretical literature and better reflect an outcome relevant to the very young children in this study. I elected not to analyze r esidential instability primarily because it is a measure that affects older children whose social networks, school relationships, and established routines are disrupted by frequent moves. These are factors less relevant to children under age three. However, upward residential mobility has the potential to expose younger children to impro ved resources that could affect their health and social outcomes over time. This project takes a sequential, Qual QUANT Q ual Quant approach First, I interviewed practitioners and patients who participated in the CHCO MLP in order to understand the struct ure and scope of the MLP intervention. Next, I collected and retrospectively reviewed data from CHCO medical records to quantitatively analyze the medical legal partnership's treatment and moderator effects on the social determinants of health and social o utcome variables of interest in children between birth and three years of age. Next, I explored the documentary data contained in the legal files for those CHCO patients who received legal representation from

PAGE 40

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 40 Colorado Legal Services (CLS) attorneys as a re sult of the MLP intervention. Finally, I integrate d the quantitative health and social outcomes data with the qualitative data for patients who received MLP services to explore whether the intensity of MLP services is associated with the quality or quantit y of either the legal, social, or health outcomes of patients in this study. I selected the two dependent variables of interest in this study because their association with health outcomes is well established. Each has been shown to contribute to cumulat ive adversity or advantage over the life course. Additionally, MLP s have the potential to impact these outcomes positively. Existing evidence shows that improving vaccination coverage ( Aungst, 2011 ) and residential stability (Bures 2003) during childhoo d can improve physical and mental health into adulthood. At the same time, upward social mobility can constrain social inequality over the life course A s residential changes improve parental socioeconomic position children may move to neighborhoods where they might experience the cumulative advantages of improved access to resources available to higher socioeconomic groups (Blane, 2006) I hypothesize that MLP s can improve vaccination compliance by solving legal problems that stand in the way of famili es taking advantage of available preventive medical care. For example, t he CHCO MLP addressed health coverage issues such as wrongful Medicaid or Social Security Disability Insurance (SSDI) termination financial issues such as wrongful denials of Suppleme ntal Nutriti on Assistance Program (SNAP), Temporary Assistance for Needy Families (TANF), or Supplemental Security Income (SSI) benefits and immigration or asylum issues that limited patients' access to public benefits and regular preventive medical care. I further hypothesize that MLP s can improve upward residential mobility by reducing the number of moves that result from legal problems with housing such as eviction, foreclosure, sanitation violations, racial or ethnic discrimination, and overcrowding w hile increasing the moves to areas

PAGE 41

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 41 of greater economic opportunity by stabilizing employment, improving income, and reducing the stress of unsolved legal issue Thus, my study assumes that s howing improvement in immunization com pliance or upward residentia l mobility variables would suggest medical legal partnerships could have beneficial long term health impacts and may reduce the cumulative adversity that results in health disparities. In summary, my study tests the theory that medical legal partnerships a ct as a protective factor to improve access to the social determinants of health. Research Question And Specific Aims The overarching research question for this study is: Do MLP services improve the health and social outcomes that could reduce health dispa rities for low income and minority children? The study has five specific aims, each shown in Figure 13 below with the analytical approach I took to meet the specific aim Specific Aim Analytical Approach 1 To determine whether access to MLP services is a ssociated with better health and social outcomes than those experienced by children who do not receive MLP services Means comparison between treated and untreated cohorts 2 To estimate the average treatment effect of medical legal partnerships on immuniza tion compliance and residential mobility Average treatment effect on treated and untreated cohorts 3 To determine whether introducing the MLP intervention improves the post intervention measures of outcomes as compared to pre Intervention measures Pre pos t intervention survival analysis within treated cohort 4 To examine whether the type and intensity of MLP legal services is related to the quality of legal outcomes that patients who also became MLP law firm clients experienced Mixed method analysis 5 To determine whether the intensity of legal services that MLP s provide have an impact on either patients' housing mobility or immunization compliance Mixed method analysis Figure 13 Study Specific Aims Summarized In Colorado, as i n the nation, socioeconomic and racial inequality limits access to the resources that families need to live healthy lives. MLP s across the nation are built on the belief that lawyers can reduce this inequality. But how and by how much specific legal servic es impact

PAGE 42

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 42 health is not well understood. My study contributes to understanding whether the 290 MLP s that now operate nationally are just another well intentioned but ineffective health reform effort or are, in fact, measurably contributing to improved acce ss to the social determinants that allow vulnerable patient populations to experience improved outcomes that can reduce health disparities. This study proceeded in three phases; the specific aims and research questions associated with each are summarized b elow in Figure 14:

PAGE 43

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 43 Specific Aim Study Phase Phase I Aim 1: To determine whether access to MLP services is associated with better health and social outcomes than those experienced by children who do not receive MLP services RQ1(a) Are pediatric patients who receive MLP services more compliant with the Advisory Committee on Immunization Practices (ACIP)/CDC recommended immunization schedules than patients who do not receive MLP services? RQ1(b) Are pediatric patients who receive the MLP intervention mo re likely to experience upward residential mobility than children who do not receive the MLP intervention? RQ1(c) Does the MLP intervention moderate the social gradient relationship between poverty, race, and poor health outcomes? Aim 2: To estimat e the average treatment effect of medical legal partnerships on immunization compliance and residential mobility RQ2(a) What impact does the receipt of MLP services have on the odds of children's immunization compliance before age 3? RQ2(b) What impact d oes the receipt of MLP services have on the ability of a young child's family to move to a neighborhood with higher median family income? Aim 3: To determine whether introducing the MLP intervention improves the post intervention measures of outcomes as compared to pre Intervention measures among treated patients. RQ3(a): Do pediatric patients who receive MLP services show greater immunization compliance after the MLP intervention than they showed before the MLP intervention? RQ3(b): Do pediatric patien ts who receive MLP services have greater upward residential mobility after the MLP intervention than before the MLP intervention? Phase II Aim 4: To explore whether the type and intensity of MLP legal services is related to the quality of legal outcomes that patients who also became MLP law firm clients experienced RQ4(a): What types of legal problems do MLP s address among children whose families receive MLP services as clients of CLS? RQ4(b): What types of legal services do MLP s provide to children whose families receive MLP services through legal representation as CLS clients? RQ4(c): What types of legal outcomes do children whose families receive MLP services through legal representation as CLS clients obtain? RQ4(d): Do the types of legal pr oblems, MLP legal services, or outcome that patients represented by CLS receive vary by race or ethnicity? Phase III

PAGE 44

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 44 Figure 14 Specific Aims And Research Questions For Each Phase Of Study A recent literature review conducted by the National Center of Medical Legal Partnership (NC MLP ) concluded, "no systematic assessment of the impact of MLP integration has been measured to date" (Beeson et al. 2013) This study begins to fill that void. Moreover, this study directly addresses the absence of any qualitative research in the MLP literature. Filling this methodological gap in the literature (Robert Wood Johnson Foundation 2011) is essential in or der "to understand the extent to which, and the ways in which law is implemented and enforced" so that MLP s around the country can more effectively address the inequality that characterizes low income and minority patients' access to the medical and social determinants of health (Burris 2001) Thus, this study is designed to make a significant contribution in three ways: First, this study contribute s to equipping providers, physicians, and clinicians with informa tion they seek to better treat poor and minority patients. In 2011, the Robert Wood Johnson Foundation surveyed a randomly selected, representative sample of 1,000 U.S. primary care physicians. The results showed that 85% of physicians believed that patien t social needs are as important to address as their medical conditions. The percent of providers holding this belief climbed to 95% of physicians surveyed when asked about low income patients. Put another way, physician respondents reported that if they had the power to write prescriptions to address social needs, these would represent approximately 1 out of every 7 prescriptions they write (Robert Wood Johnson Foundation, 2011) Medical legal partnerships could represent a Aim 5: To determine whether the intensity of legal services that MLP s provide has an impact on either patients' housing mobility or imm unization compliance RQ5: Does immunization compliance or residential mobility vary among children whose families receive CLS representation depending upon the intensity of the MLP legal services provided?

PAGE 45

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 45 transformative model to equip p roviders to deliver health care that connects patients to the social resources physicians want to provide and that patients need to thrive. Second, this study contribute s to the long term sustainability of MLP s in Colorado. Nationally, the empirical recor d must be strengthened if MLP s are to create lasting, systemic change. But the need for evidence of the MLP model's impact in Colorado is particularly acute. Because of the lack of evidence for the MLP performance in Colorado, the first MLP established in Colorado has ended. Children's Hospital patients will no longer have an integrated partnership with legal aid lawyers addressing their social needs. This MLP closed for lack of funding in December 2013 and was replaced by a small, ad hoc effort sponsore d by a private law firm with limited time to dedicate to the documented breadth of children's need. My experience raising grant funding for CHEP confirms that the long term financial future of scalable MLP s that can serve all Colorado's low income and min ority patients depends upon generating the kind of empirical information that can show the return on investment that MLP s provide in medical, social, and financial terms. This study will contribute information about medical and social impacts that may hel p sustain the MLP model in Colorado financially. Finally, this s tudy contribute s to the MLP and public health law literature by introducing a new theoretical framework for understanding whether MLP s disrupt patients' cumulative adversity (Hatch, 2005) By adding a combined qualitative and quantitative understanding of the way that law generally and MLP s specifically affect low income patients in Colorado, this study will contribute to understanding and improving the role that MLP s play in addressing health disparities in Colorado and the nation.

PAGE 46

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 46 CHAPTER II LITERATURE REVIEW This study seeks to contribute to two related bodies of literature. Public health law research combines empirical legal scholarship, health services research, and public health law sch olarship to advance public health goals. A goal of this study is to add to the theory based, quantitative and qualitative approaches that inform new public health solutions. More precisely, this study adds to the emerging body of transdisciplinary public h ealth law research that has been called "legal epidemiology" (Burris et al., 2016) Next, this study adds to a sparse but growing body of medical legal partnership scholarship. These studies report evidence of the way that legal services change health care delivery, health outcomes, and health care costs. This chapter summarizes the existing literature in both these areas and concludes with the latest research pertaining specifically to the two outcome variables studied here. Public Health Law Literature Re view Elizabeth Tobin Tyler explains the intersectionality between law and public health by highlighting an increasing need for a "fuller understanding of the role of social justice in health Particularly as mounting evidence points to the role of soc ial conditions in health outcomes" (Tyler, 2012 ) Tyler explains that the law plays two essential roles in affecting the social determinants of health. First, laws organize and perpetuate the structures that we describe as social determinants. In the cas e of housing, for example, laws define the rights of home ownership or occupancy, enforce financing terms, and establish construction standards. Construction laws organize laborers to build and meet safety standards tax laws govern affordability zoning l aws control density and the neighborhood's physical built environment and demographic composition criminal laws seek to ensure neighborhood safety and public benefit

PAGE 47

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 47 laws affect accessibility and affordability. The second role for law is corrective. In t his role, law serves as a tool to transform inequitable social structures. "Law has been and will remain critical for creating the infrastructure that supports directed and accountable action, as well as for limiting some actions that diminish health or re quire actions that enhance it" (Tyler, 20 1 2 ) For example, public health laws can be used to penalize unsafe or unsanitary housing conditions. Environmental laws may mandate removal of hazardous toxins from a neighborhood. A landlord can be compelled to r emove lead paint or exterminate rodents when housing habitability laws are enforced. Positively, a city can increase recreational spaces or food security by enacting tax and zoning provisions to reward health enhancing neighborhood design. Professor Sco tt Burris' s model for integrating law and social epidemiology describes a dual role for law that is similar to Tyler's description but goes further. Burris introduces a textured and normative understanding of the way a legal system may effectuate better o r poorer health of populations by defining social position. He explains that law operates as a pathway through which lower social economic position and poor social cohesion translate into stress and resultant poor health. According to Burris, integrating law, social science, and social epidemiology "identifies two measurable relationships between law and health: law may operate as a pathway along which broader social determinants of health have an effect; and laws or legal practices may contribute to the development, and influence the stability, of social conditions that have been associated with population health outcomes" (Burris, Kawachi, & Sarat, 2002) For example, negative experiences with law say with law enforcement officers have far reaching, health harming psychosocial effects on individuals and neighborhoods. Disparate enforcement of laws may affect self esteem, mutual respect, and social cohesion. "Similarly, much of the effect of race or class difference on health seems to be mediated by th e accumulation of small stressors

PAGE 48

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 48 over a life course many of which are communicated through law. Burris explains how law mediates job insecurity, fear of neighborhood crime, and experiences with racial prejudice and can result in "wear and tear on the ca rdiovascular, endocrine, immunologic, and metabolic systems, eventually leading to a host of maladies ranging from hypertension, obesity, diabetes, as well as depression, asthma and susceptibility to infection" (Burris, Kawachi, & Sarat, 2002) Professor L awrence Gostin describes the intersectionality between law and health with a focus on the branch of law that governs public health particularly, and more generally on the role of the state in ensuring public health. Gostin explains that public health law s pecifically defines a set of legal powers and duties belonging to the state and its partners "to ensure the conditions for people to be healthy (to identify, prevent, and ameliorate risks to health in the population), and [define] limitations on the power of the state to constrain for the common good the autonomy, privacy, liberty, proprietary, and other legally protected interests of individuals. The prime objective of public health law is to pursue the highest possible level of physical and mental health in the population, consistent with the values of social justice" (Gostin, 2008) These theories of law and public health are applied practically when law is described as an integral part of the social determinants "strategy." Scott Burris translates th e large theoretical paradigm that essentially says "the law is all over" and presses the law into service to conform the lofty and theoretical "laws on the books" with the systems and institutions and practices that touch people's daily lives through the laws on the streets" (Burris, 2001) In a nutshell, the role of the medical legal partnership is to change the status of low income patients from having to accept inferior distribution of the protections, property, opportunity, goods and services to whic h the law entitles them. MLP s aim to improve the trajectory of those social conditions that most directly influence a patient's life chances and overall health and social outcomes. "If our

PAGE 49

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 49 theories are correct, [access to legal services will] improve both the level and distribution of health, because they address fundamental causes that find expression in a wide range of ultimate health states reached in a wide range of ultimate health states reached via a plethora of pathways across the life course" (Burr is, 2001) The most apt description of the work undertaken in this study only recently appeared in the literature. Scott Burris, Ma u rice Ashe, and others coined the term "legal epidemiology" to describe the work using social science methods to conceptuali ze, implement, and evaluate both laws and legal practices that change unhealthy behaviors and environments. (Burris et al. 2016 ) This study includes components of legal prevention and control as it explores MLP s as a way to intervene preventively in the l ives of vulnerable patient populations. Moreover, my study adds to legal etiology literature the study of the structural u nin tended consequences of laws that are inequitably enforced. In both cases, this study fills gaps in the literature identified by Burris and Ashe (2016) : Efforts to theorize social determinants and to prescribe reforms have been impressive and important, but our health research investments remain biased toward the individual risk factors and proximate causes of death. Because law pl ays such an important and pervasive role in structuring environments and behaviors and beliefs, research under the heading of legal etiology is crucial to charting a course of practical reform. It encompasses law's structural role in shaping the level and distribution of health in a community, law's contribution to cultural beliefs about how health is produced, protected, and distributed, and how we can use legal interventions, such as enforcing healthy housing codes, to improve health and health equity. I n this way, my study builds on and extends these analyses by Tyler, Gostin, Burris, and Ashe which describe the reason and methodology that inform my examination of whether the Children's Hospital of

PAGE 50

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 50 Colorado Medical Legal Partnership moderates immunicatio n compliance and upward residential mobility outcomes. Medical Legal Partnership Literature Review In 2013, the National Center for Medical Legal Partnership published a white paper collecting the evidence to support the efficacy of MLP s (Beeson et al., 20 13) This paper highlighted the limitations of the empirical record. As a result, the NC MLP has focused considerable attention on developing a series of common metrics for the more than 290 MLP s nationwide to use in evaluating their impact (Regenstein, 20 14) However, this process has been slow and incomplete. So far, the development process has yielded proposals on six metrics focused primarily on process and MLP structure evaluation but largely ignoring the type of outcome measures necessary to determine whether MLP s are contributing to reducing disparities or improving population health outcome (NC MLP Performance Measures Handbook 2016) The absence of such outcome measures not only affects the assessment of the quality and effectiveness of MLP s but th is void is also hindering the long term financial sustainability of MLP s The NC MLP reported in 2012 that 23 published articles descriptively present the components and function of the MLP model. These articles generally explain what medical legal partne rships are and why they are needed to address the social determinants of health. A few contain some empirical evidence that patients identify unmet legal needs as harmful to their health (Sandel, Suther, Brown, Wise, & Hansen, 2014) "Practice reports" or case studies comprise another large category of MLP literature; they outline the activities and structure of medical legal partnerships in a single setting (Newman, 2012) Observational studies make up the smallest category of published MLP research; the NC MLP identified 13 articles reporting

PAGE 51

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 51 observational studies of MLP impacts on patients, providers, or communities. The NC MLP white paper remains the most comprehensive compilation of MLP research literature. However, since its publication several notable studies have been published. Figure 1 5 describes the NC MLP reported observational studies and those that have been published since the white paper. Author Title Methods and/or Measures Used or Suggested Results Atkins, et al. 2014 Medical Legal Partner ship and Healthy Start Practice Report Number of cases opened (200), number legal consultations (150), resolved (70) and legal advocacy training hours (48) over 3 years Beck et al. 2014 Housing Code Violation Density Associated with Emergency Department and Hospital Use by Children with Asthma Correlated number of code violations in census tract with population level asthma morbidity Found density of housing code violations associated with and predicted asthma morbidity and hospitalized patients risk of subsequent morbidity Beck et al., 2012 Identifying and Treating a Substandard Housing Cluster Using a Medical Legal Partnership Identified number and type of cases, counted housing repairs completed 16 apartments in a 19 building complex treated. Housing and health outcomes such as asthma, developmental delays, and elevated blood levels more likely for 16 of 45 pediatric patients who were in poor quality housing Cohen et al 2010 Medical Legal Partnership: Collaborating with Lawyers to Identify and Add ress Health Disparities Practice Report Evaluated four MLP s Found established MLP s improve provider knowledge; reduce provider concerns over making patients nervous with legal questions; increase resident referrals to legal services after training; and one MLP found 67% medical residents uncomfortable addressing social issues

PAGE 52

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 52 Conover et al. 2002 Impact of Ancillary Services on Primary Care Use and Outcomes for HIV/AIDS Patients Using primary data from 1997 survey of low income patients in North Caroli na, multivariate logit analysis to estimate effect of receiving housing, legal services, and substance abuse treatment on use of primary care, CD 4 counts, viral load, and self rated health status Legal services one of several ancillary services analyzed. Legal services positively associated w/increased access to primary care but negatively associated with viral load. Hernandez et al. 2016 Extra Oomph: Addressing Housing Disparities through Medical Legal Partnership Interventions Qualitative interviews with 72 families about affordability, substandard conditions, and stability. 83% of MLP families improved their housing conditions. Legal services helped reinstate or prevent utility shut off, retain or regain housing subsidies, relocate to better resident ial environments, and appeal a rent hike. Patients in control group less likely to resolve need for safe, affordable housing; 64% of control group patients and 17% of MLP patients did not resolve housing problem Fleishman et al. 2006 The Attorney as the Newest Member of the Cancer Treatment Team Practice Report Observed patients' stress, financial situations, and reviewed appointments and clients served. At legalhealth New York, a fully staffed free legal services program for cancer patients, patients had reduced stress and maintained treatment regimen, families had better financials, and MLP tracked number of clients. O'Sullivan et al. 2012 Environmental Improvements Brought by the Legal Interventions in the Homes of Poorly Controlled City Adult Asth matic Patients: A Proof of Concept Retrospective study of 12 patient charts for adult patients with poorly controlled asthma; patients self reported allergen exposures; pre post intervention analysis of peak expiratory flow rate (PEFR, asthma severity cla ss, medications, ED Effect of MLP intervention to compel landlords to provide better living conditions for asthma patients improved mean PEFR, reduced ED visits from 22 to 2; reduced hospital admissions fro m 11 to 1 admission, and all patients had reduced need for medication and improved asthma severity.

PAGE 53

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 53 visits, hospitalizations and steroid requirements O'Toole et al. 2012 Resident Confidence Addressing Social History Survey of 40 residents Survey of 40 residents show those with more legal and social reso urces have greater confidence, knowledge; screen more; and take longer social histories Pettignano et al., 2012 The Health Law Partnership: Adding a Lawyer to the Health Care Team Reduces System Costs and Improves Provider Satisfaction Practice Report Patients served, Medicaid cost savings/claims recovered by coverage cohort, provider satisfaction and CE credits rendered 4 year study increased Medicaid reimbursement payments, saved hospital $10,000 in continuing education costs, and increased MD satisfa ction Pettignano et al. 2011 Medical Legal Partnership: Impact on Patients with Sickle Cell Disease Practice report number of families referred, number of legal issues, cases closed, cases won Retrospective cohort of 81 SCD parents with 76 children ope ned 106 cases, 99 closed with 21 measuring legal gains Rodabaugh et al ., 2010 A Medical Legal Partnership as a Component of a Palliative Care Model 3 yr review of palliative care MLP for cancer patients 297 legal referrals for custody over 3.5 years, 17 benefits denial cases overturned, and MLP demonstrated financial sustainability as institutions recovered $923,188 for 17 benefits advocacy cases. Staffed by fulltime social worker and 0.5 FTE attorney doing advocacy in guardianship, estate, and advance c are planning, housing and legal services cases. Ryan et al 2012 Pilot Study of Impact of Medical legal Partnership Services on Patients' Perceived Stress and Wellbeing Pre post study of stress levels using questionnaires Changes in perceived stress st rongly related to participants' concern over legal issues addressed by MLP MLP reduced stress and improved overall wellbeing. Sege et al., 2015 Medical Legal Strategies to Improve Infant 330 Medicaid families r andomly assigned to MLP or safety intervention. MLP associated with improved access to income supports and better quality

PAGE 54

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 54 Health Care: A Randomized Clinical Trial Pre and Post intervention interviews care. At 6 and 12 month surveys more MLP group had obtained income, utility, housing, and fo od assistance than control grp Teufel et al., 2012 Rural Medical Legal Partnership and Advocacy: A Three Year Follow up Study Five years review of baseline data and three years follow up data showed benefit relative to cost of MLP increased from 2002 200 6 and 2007 2009. Number of people served and cases won increased. Health care recovery dollars showed 319% ROI between 2007 2009, and $4 million in health care debt relieved for patients. Weintraub et al. 2010 Pilot Study of Medical legal Partnership to Address Social and Legal Needs of Patients 54 low income families completed baseline and 6 month follow up assessments to test hypothesis that integrating legal services into pediatric settings would increase families' awareness of and access to legal a nd social services, decrease barriers to children's health care, and improve child health. 2/3 reported improved child health and well being, increased food and income support utilization, and decreased avoidance of health care due to lack of insurance. Pe rcent of patients accessing legal services, and patients with improved satisfaction increased. Figure 15 Summary Of Major Observational Studies Of MLP Impact Observational Studies In MLP Literature The observational MLP studies f rom the NC MLP white paper contain a limited body of evidence that MLP s affect financial, professional, and health outcomes. Financial impact studies typically analyze the return on investment that MLP s provide when the cost of volunteered services is meas ured against the savings to institutions or patients. These studies attempt to prove the financial viability of MLP s by focusing on a traditional cost benefit analysis. For example, a longitudinal study of an MLP in rural Illinois compared the financial benefit as compared to the initial investment to calculate a financial measure useful for analyzing stock and bond values. This article calculated the MLP 's ROI return on investment for the initial period

PAGE 55

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 55 from 2002 2006 and the relative benefit of the MLP during 2007 2009. Researchers reported the program produced a 149% return on investment (ROI) of $115,438 for the hospital partner during the first period, and the hospital's Medicaid ROI increased to 319% based on an investment of $116,250. Another approach in financial studies is to evaluate the impact the programs have on patients who have been denied health benefits or who have fallen into medical debt. For example, one researcher analyzed the return on investment hospitals received from MLP prog rams that resolve previously denied benefit claims. Rodabaugh and colleagues reported on an MLP that worked with cancer patients and generated nearly $1 million in benefits by resolving denied benefit claims (Rodabaugh, et al., 2010) In a retrospective s tudy of 71 parents or guardians with 76 children diagnosed with sickle cell disease, researchers measured patients' gains in benefits due to the MLP intervention (Pettignano, Caley, & Bliss, 2011) This MLP formed between a Children's Hospital and Georgi a Law School, measured the monthly amount of public benefits obtained or retained including housing, education, and health insurance benefits, as well as the number of patients who received end of life care following MLP l egal counsel (Rodabaugh, et al., 2 010). In still another approach, researchers at Johns Hopkins conducted a longitudinal study to determine that an Illinois MLP relieved patients of over $4 million in health care debt and obtained a total of $1 million in social security benefits (Teufel et al., 2012). The second group of observational studies report MLP impact on attorney or physician professional practices. Most of these studies are descriptive or anecdotal. Fleishman, for example, observed that attorneys benefitted from the collaborat ion by improving their preventive clinical skills, while health institutions experienced fewer missed appointments and treatment

PAGE 56

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 56 interruptions when cancer patients' social needs were addressed (Fleishman, Retkin, Brandfield, & Braun, 2006) Another study highlighted the benefit to law students of developing interdisciplinary communication and collaborative practice techniques (Wettach, 2008) A survey of 40 medical residents found the MLP training gave physicians greater confidence and knowledge, leading them to screen for patients' social and legal needs more frequently and spend more time taking social histories (Wettach, 2008) In addition to recovering previously unreimbursed Medicaid payments and saving continuing education costs, a Georgia MLP survey ed physicians, social workers, and nurses to find that the interdisciplinary MLP collaboration increased provider satisfaction. These respondents perceived that patient emergency department visits and readmissions declined, while the providers' ability to reallocate their time to other cases improved (O'Toole et al., 2012) Another exemplar from this group of observational studies presents descriptive data from an MLP in Chester, Pennsylvania. After reviewing the number of legal cases opened during one year the average number of legal issues addressed for each client, and the number of legal and advocacy training hours the MLP contributed to public health issues, Atkins concluded, "[a] current need exists for longitudinal data showing the impact of MLP serv ices over time on factors such as health birth outcomes, mortality, emergency room utilization treatment compliance, and absenteeism to provide even more compelling evidence on MLP efficacy" ( Atkins, 2014 ) While the growing evidence confirms that both patients and providers perceive the need for integrating legal services into the health care delivery model, the studies that evince the impact of integrating attorneys into the health care delivery model seldom quantify the impact MLP s have on patient hea lth outcomes and generally do not take a longitudinal perspective. Six studies

PAGE 57

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 57 currently comprise the third and smallest group of articles from the NC MLP report that present empirical evidence of MLP impact on patient health and well being. Interesting beh avioral health outcomes such as stress have received a large share of the attention in the MLP literature. In one study, researchers administered 10 item Perceived Stress Scale (PSS10) and Measure Yourself Concerns and Wellbeing i nstruments to find that 6 7 patient participants showed a slight decrease in their mean stress level and an even more slight improvement in their self rated well being ( Ryan 201 2 ) In the other study, a survey of 20 clients of a New York City MLP showed that 75% of patients report ed reduced stress as a result of the legal help they received; 30% of these patients said the MLP services helped them to maintain their treatment regimen and 25% said the MLP helped them to keep their medical appointments ( Fleishman, 2006 ) This finding t hat MLP participation positively impacts access to health care is consistent with an earlier study of HIV/AIDS patients. In 2002, researchers measured the impact of a variety of ancillary services on HIV/AIDS patients. In this study, legal services were on e of several ancillary services that were positively associated with increased patient access to primary care services. However, counter intuitively the availability of legal services was also negatively associated with patient viral loads. In other words the health outcomes of patients worsened with increased access to legal services. In this study, improved housing was the only ancillary service associated with improved primary health care access and health outcomes (Conover, 2002) A Stanford based me dical legal partnership conducted a 36 month, prospective study of the impact their MLP had on patients' access to legal and social services as well as access to health care (Weintraub, 2010). A cohort of 54 Children's Hospital and Health Center patient fa milies, who received free MLP legal services, completed baseline and six month follow up

PAGE 58

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 58 assessments. This pilot study found that integrating legal services into health care delivery not only decreased barriers to health care access significantly fewer participants reported they avoided going to the doctor for their children after the MLP intervention but the MLP was also associated with increased awareness of other social supports such as Supplemental Nutrition Program for Women, Infants, and Children ; Food Stamps ; and Social Security Income Benefits. Several MLP studies have examined the MLP 's impact on housing conditions. In 2012, Beck et al r eported a study involving an urban, 19 building complex owned by a single housing firm ( Beck et al 2012 ) This group studied the health of children receiving outpatient primary care in the neighborhood where housing was affected by known health harming risks such as pest infestation, water damage, and poor ventilation. Physicians identified patients through social risk screening and referred these families to an MLP advocate who identified a cluster of substandard housing conditions, formed a tenant association and advocated in court, before city council, city planners, and housing inspectors to cause the b uilding owner to repair the patient housing to meet code standards. This study did not measure children's health outcomes after the intervention but found an association between the substandard home environments and poor health outcomes. Beck and colleague s studied 45 children living within the identified units; 36% of these children had asthma, 33% had developmental delays and 9% showed elevated lead content in their systems, leading these researchers to conclude that if attorneys could improve housing co nditions, they could also improve health. Beck and colleagues published a second study in which he found that an MLP in Cincinnati, Ohio successfully intervened to enforce building ordinances requiring a landlord to make improvements in a large cluster of pest infested apartments ( Beck et al 201 4 )

PAGE 59

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 59 A 2012 study that measured MLP impacts on children's asthma is particularly instructive for this project. O'Sullivan and colleagues (2012) looked at the effect an MLP had on the health outcomes of 12 asthma p atients with poorly controlled disease. These researchers evaluated patient's pre and post intervention outcomes by measuring Peak Expiratory Flow Rates (PERF), asthma severity class, medications, ED visits, hospitalizations, and the requirement for stero ids before MLP intervention and 6 12 months after the intervention. The results of that study showed that patients all had increased PERF, decreased number of ED visits and hospitalizations, fewer required systemic steroids, and all had reductions in the d ose and/or number of medications used. Despite the limitation of this study which include its small number of patients, the lack of a control population, and the limited time period considered, the researchers concluded the medical legal collaboration is h ighly effective in improving the control of inner city asthmatics by effecting improvements in the domestic housing environment. O'Sullivan measured a 91% reduction in ED visits and hospital admissions for inner city New York asthmatic adults after an MLP intervention (O'Sullivan et al, 2012) More recently, Diana Hernandez published an influential qualitative study on an MLP 's impact on 72 families with housing issues. Hernandez examined the mechanisms through which MLP s work to resolve housing problem s that fell into three categories: affordability, substandard conditions, and stability. Hernandez' study found that the interaction of physicians and lawyers in a patient centered medical home allowed MLP participants to resolve housing affordability (e.g ., averting utility shut off), adequacy (e.g., substandard conditions) and stability (e.g., eviction) problems more successfully than families without MLP assistance ( Hernandez, 2016). This research team conducted interviews with low income families who pa rticipated in MLP s and with a control group of families. Before the MLP intervention, 53% of these families

PAGE 60

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 60 reported living in inadequate housing, 33% were struggling to afford rent or utilities, and 14% reported housing instability because they were at ri sk of eviction and homelessness. An overwhelming majority 83% of families in this study who received the MLP services improved their housing conditions. Legal services helped reinstate or prevent utility shut off, retain or regain housing subsidies, reloca te to better residential environments, and appeal a rent hike. In contrast, the patients in the control group were less likely to resolve their need for safe, affordable housing; 64% of the control group patients and 17% of MLP patients did not resolve the ir housing problem during the study period ( Hern‡ndez, 2016) In May 2015, researchers at the Boston Medical Center reported a randomized clinical trial testing the effect of an intervention that incorporates the MLP model in care delivered to 330 families of healthy newborns in Pediatrics (National Low Income Housing Coalition 2016) There, families were randomly assigned a specialist who provided support until the newborn's 6 month routine health care visit. The specialist conducted home visits and teleph one check ins, provided age related information on child development, and when necessary combined elements of an MLP program at Boston Medical Center. At the conclusion of the study, infants who received the support intervention were 14% more likely to hav e completed their 6 month immunization schedule by 6 months of age. They were also more likely to have five or more routine preventive care visits by age 1 and less likely to have visited the emergency department by 6 months. While it is difficult from thi s study to isolate the MLP 's impact on these improved outcomes, and while the study period followed children for only a short time after birth, this article importantly represents the first randomized clinical trial reported to test the MLP intervention. A t the outset, 73% of families reported economic hardships. More than half (61%) reported food insecurity, 28% reported they were unable to pay rent or mortgage during the

PAGE 61

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 61 previous 12 months, 42% reported missing a payment for gas, electricity or water uti lities, 12% reported a utility shut off for failure to pay, and almost half (44%) reported their telephone service had been disconnected for failure to pay sometime during the previous year. The MLP intervention accelerated access to concrete income suppor ts for newborn babies and their families. At six and 12 month post intervention surveys, significantly more patients in the MLP group had obtained income, utility, housing, and food assistance than patients in the control group. For example, 12 months aft er the birth of their children, 21.1% of families had obtained income assistance compared to 18.8% of control families (p value = .029). Because prior research confirms that concrete support in early months of a child's life may protect against child negle ct and abuse by reducing parental stress, and may improve primary health care, the results from this study point out the impact that MLP income supports can have on child health outcomes (Shonkoff, 2012). My study builds upon this work to isolate the treat ment effect of the MLP intervention itself on children's immunization completion in order to infer patients' participation in regular preventive care. In 2015, a study reported in the Journal of Health Care for the Poor and Underserved showed how an MLP im proved home heating security for poor children and their families (Sege et al., 2015) In this pre post study, researchers compared the number of families reporting energy insecurity on a waiting room screening questionnaire before and after MLP interventi on. The certification of medical need approvals for local utility companies increased by 65%, preventing utility cut offs for 396 more families with vulnerable children between the first and second year that the MLP operated. In another study using popula tion level asthma morbidity data, Cincinnati Children's Hospital researchers found that the density of housing code violations in a census track could predict a hospitalized child's risk of subsequent emergency department

PAGE 62

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 62 visits and hospitalizations. Child ren who had been hospitalized for asthma had 1.84 greater odds of a subsequent hospital visit due to asthma morbidity within a year if they lived in the areas with the highest number of housing code violations. This study supported the inference that MLP s could reduce asthma related morbidity in children by reducing housing code violations (Taylor et al 2015) My work in this project extends the Taylor (2015) and Beck (2014) studies of MLP impacts on internal housing conditions as a social determinant of children's health. My study uses median family income by census tracts to infer the quality of both internal housing and external neighborhood conditions that may impact child health. By measuring the housing variable based on median family income of the c ensus tract, I have aimed to capture a broader range of housing conditions relevant to health. Second, the housing mobility variable allowed me to connect the legal issues that accompany poor housing conditions and affect children's health on a population level. Finally, while the Beck study considers the effect of poor housing on children with asthma, my study includes a broad range of childhood diseases as co variates, and looks not only at housing related outcome variables but also at a variable related to immunization completion as a proxy for overall child health rather than singling out a single disease condition such as asthma. The current body of MLP literature is characterized by inquiries that have yet to examine the broader implications of the p otential impact of the MLP model. On one hand, inquiries about investment returns, financial benefits, and professional practices ignore the long term impact that an MLP may have on patient's biomedical outcomes. On the other hand, existing studies of ML P health impacts narrowly focus on one disease or health outcome and miss the opportunity to evaluate the broader social impact that MLP s are intended to have. My study begins to fill

PAGE 63

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 63 these gaps in the MLP literature and does so using data from thousands o f pediatric patients, over a five year period, and employing a methodology designed to identify the treatment effect of the MLP service specifically. The next chapter elaborates the theoretical framework that I have employed.

PAGE 64

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 64 CHAPTER III THEORETICAL FRAM EWORK First, i t is well accepted that the most fundamental causes of health disparities are inequities in socioeconomic status (SES) within the United States (Beck, Huang, Chandur, & Kahn, 2014) Moreover, a growing body of research supports the claim th at racism is also a fundamental cause of health inequity (Adler & Newman, 2002) Racism, an organized system of categorization and ranking of societal groups into hierarchical races that devalues, disempowers, and differentially allocates desirable opportu nities and resources, causes disparate health outcomes for minority patients (Phelan & Link, 2015) My theoretical starting point for this study is an ecosocial understanding of the impact that economic and social inequality have on patterns of health ineq uality in Colorado children. Second, I rely upon fundamental cause theory to inform my hypotheses about the vulnerable populations studied. Third, the dependent variables chosen to reflect the life course theoretical approach to health interventions. I tak e each concept in turn and conclude this chapter with a description of the theories that support my selection of outcome variables. An Ecosocial Theory Of Health Inequality Ecosocial theory explains that populations biologically embody adverse exposures fr om ecological and societal influences. The result of economically and socially skewed influences is a disparate distribution of health (Williams & Mohammed, 2013 ; Krieger 20 01 ) Disparate influences begin imposing harmful health effects early in life. Re searchers have described a process called "biological programming," to explain how socially mediated factors such as poverty and racism can adversely affect human growth even before birth and throughout infancy (Barker 19 98 ) J.L. Aber and others suggest that poverty is associated with higher neonatal and

PAGE 65

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 65 post natal mortality rates, increased risks of accidental injuries, asthma, and abuse as well as lower cognitive developmen t scores (Aber et al., 1997 ) Importantly, ecosocial theory describes harmful he alth impacts that begin in utero, persist during childhood, and reach into adulthood. Jack Shonkoff for example, explains that the stress of early childhood adversity leaves a "lasting signature" on genetic predispositions, brain architecture, and long term health so that the adverse health effects of inequa lity in childhood are permanent ( Shonkoff, 2011 ) Moreover, these health effects on children are both direct and indirect. For example, poor mothers cope with stress through higher rates of smoking, excess alcohol intake, drug misuse and dietary deficits. These factors affect not only the mother's health but also the child's health beginning in the developmental stages of pregnancy, leaving newborns vulnerable to later health insults In 1999, Her tzman used the term "biological embedding" to describe this interaction between physical development and the environment in early childhood that reduces weight and height growth and impacts mental functioning for life (Hertzman, 1999) The vulnerability of a child's physical development before birth continues during the postnatal period, as children and their families interact with their social environment. Children are particularly vulnerable to the psychosocial causes of poor health ( Haas, Krueger, and Rohlfsen, 2012 ) The evidence shows that children whose early lives are characterized by a poor socio economic environment are at increased risk of low birth weight, smoking, alcohol abuse, and mental health problems. The Adverse Childhood Experience (ACE) Study also convincingly demonstrates a strong link between adverse childhood experiences and higher rates of heart disease, diabetes, lung disease, hepatitis, d epression, obesity, and suicide (Felitti & Anda, 1997) The ACE Study linked childhood trauma t o increased incidence of risky health behaviors during childhood and adolescence (Shonkoff et al., 2011) Moreover, adverse effects of the interaction

PAGE 66

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 66 between poor social circumstances and poor health and health behaviors have been shown to accumulate. Fo r example, depressed socio economic circumstances experienced early in life are associated with an increased likelihood of poor educational attainment, which is associated with poor health behaviors that produce overweight adults, who are ultimately suscep tible to disease and early death later in life. In addition to explaining socioeconomic disparities, the ecosocial theory also provides a framework for my understanding racial and ethnic differences in health outcomes. I sought to capture environmental an d psychosocial variables that may change children's health outcomes over time. I examined immunization compliance during early childhood as a proxy for the access a child may have had to preventive health care which could improve long term health outcomes as well as for the direct benefits accruing from immunization compliance in avoiding infectious di s eases Also, I analyzed upward residential mobility to examine whether the MLP intervention could change neighborhood factors such as access to resources, ma terial goods, and social influences that might improve long term health outcomes. My resear ch asked whether a disruption that could improve those variables might improve long term health outcomes for participant patients. My interest in influences on th e relationship between race and poor outcomes arises from the understanding that racial inequality simultaneously benefits groups who claim superiority while harming groups deemed inferior. The impact of historic and contemporary racial and ethnic discrimi nation is to produce inequitable living, working, and environmental conditions. Exposures to toxins, trauma, and violence then are expressed in disparate biological patterns along racial and ethnic lines. Arline Geronimus' s (2006) groundbreaking work on th e weathering framework links racism and stress to explain health disparities. Geronimus found evidence that racial inequalities in health that manifest in biological systems cannot be explained solely by

PAGE 67

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 67 racial differences in poverty. Racial and ethnic min orities navigate the results of individual discrimination and structural racism on a daily basis. These vulnerable populations experience overt and covert racism, concentrated poverty, high crime, and chronic environmental pollution, and over the course of a lifetime the physiological and psychological responses lead to health problems at rates not experienced by majority populations who are free from such life stressors. By measuring racial differences in the allostatic load scores of her study particip ants, Geronimus (2006) was able to capture the biological responses to the stress of discrimination and its impact on cardiovascular, metabolic, and immune systems and attribute these health differences to "the weathering" effects of living in a race consc ious society (Geronimus, 2006) Race, for example, independently predicts low birth weight and infant mortality levels among babies such that babies born to black mothers with high income and education suffer poorer morbidity and mortality than babies born to white mothers with low income and educational attainment ( A delman, 2007 ) Colorado data are consistent with Geronimus' weathering hypothesis. Although Colorado has the fifth lowest infant mortality rate in the country, black babies have nearly three t imes the infant mortality rate of white babies and Latino babies are now 70% more likely to die before their first birthday than white infants Finally, ecosocial theory provides a theoretical basis for questions of accountability and agency that I raise i n this study. I ask whether MLP attorneys can disrupt the predicted association between race, poverty, and inferior health outcomes based on a theoretical understanding that societal actors both individual and institutional can be held responsible for changing the stressors that produce unequal population health outcomes. I assume that most patients who are victims of inequality do not, in their own right, have the agency to require the necessary changes. Thus, embodiment theories in social epidemiology counsel the understanding

PAGE 68

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 68 that children biologically incorporate the inequities present in the world around them and inform the core concept driving this study: inequitable population patterns of health, disease, and well being can be improved by compelli ng those responsible for unjust distribution of environmental and economic resources to rectify these inequities (Office of Health Equity 2013) Fundamental Social Causes Of Health Inequity The second theoretical construct important to this study is that social determinants are a fundamental cause of disproportionately poor health outcomes for vulnerable populations ( Krieger, 2001) More precisely, this study is based on the understanding that inequities in distribution of the social determinants of health including the conditions under which people work, live, eat, and the environment in which they live, more directly explain health disparities than any biological explanation alone (Marmot, 2004) An economic historian is credited with the first systema tic assertion that improvements in population health are associated with social, economic, and industrial changes rather than exclusively relying upon improvements in medical interventions ( M c K inlay, 1977) Later, in the Whitehall Studies" Sir Michael Ma rmot reported a pair of longitudinal studie s that analyzed UK health data in which he i dentified an inverse relationship between the social status of British civil servants and their relative risk of morbidity and mo rtality (Marmot et al., 1991). The White hall studies prospectively compiled over 15 years of longitudinal data for a cohort of over 10,000 study participants. The data showed that shorter life expectancy and most diseases occur more commonly further down the social ladder, less commonly in the middle class, and least among upper class populations ( Marmot, Rose, Shipley, & Hamilton, 1978) Access to health care could not explain this inverse relationship called the "social gradient" that the Whitehall studies revealed since all the study particip ants obtained health care through the National Health Service

PAGE 69

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 69 (Wilkinson & Marmot, eds., 200 6 ) In stead, differences in the social conditions faced by participants were shown as determinants of health outcomes. In 1998, the World Health Organization (WHO) published comprehensive evidence of the association between social determinants and population health globally, based on clear evidence of the inverse relationship between socioeconomic status and poor healt h (Wilkinson & Marmot, 2006). In 1995, Link and Phelan introduced evidence to explain the social gradient further. Their research explained that variable access to resources that affect social conditions, rather than individual and biological risk factors, results in unequal health outcomes and inequi table ability to address deficiencies in the social determinants. In their parlance, a "fundamental cause" of disease is one that influences multiple disease outcomes, through multiple risk factors, by limiting access to the resources that could avoid risk factors and that is replicated over time through various mechanisms (Link and Phelan, 1995) This paper asserts that a lack of legal services is a fundamental cause of poor health. Law organizes access to the social determinants of health in American soci ety. Therefore, I apply fundamental cause theory to postulate that a lack of access to justice through legal services satisfies the three constituent claims of fundamental causality posited by Lutfey and Freese (Luftey & Freese, 2005) First, lack of acc ess to legal services is multiply realized to affect diffuse proximate causes of poor health. Examples include lead poisoning suffered due to unenforced housing codes, lack of health care access suffered due to wrongly terminated benefits, and mental healt h impacts of police brutality and community violence. Second, lack of access to legal services satisfies the holographic claim of a fundamental cause. The relationship between poor health outcomes and the lack of legal services is replicated even when poor health is separated into a wide variety of health outcomes. Examples include inferior treatment of

PAGE 70

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 70 chronic diseases, recurring insults to personal safety, acute injury from trauma, environmental health issues, medical disability and any subclass of poor health outcome imaginable. Finally, lack of access to legal services satisfies the predictive claim of fundamental causality. Changing the structure of how poor health outcomes are realized will have only modest effects on the observed relationship between legal powerlessness and poor health. The best example of this relationship is the limited impact of health policy reforms that substantially increase access to health care for the poor. SCHIP for example, has broadened access to medical care for poor chi ldren but has had limited impact on the population wide health outcomes of unrepresented children, since they still lack the legal resources to enforce their SCHIP rights to receive health treatment absent advocacy by legal representative. In 2014, the Col orado Health Institute reported that 45,227 children were eligible for Medicaid but not enrolled and 36,380 children were eligible for CHP but not enrolled. While not all 81,607 of these uninsured children require lawyers to enforce their rights, the evide nce of pervasive health disparities in Colorado suggests that a significant number of them may. Poverty and minority status are both characterized by a general lack of resources. In this study I theorize that the inability to address legally enforceable r ights is one of the most important resources that low income and minority children in Colorado lack. I suggest that limited access to legal representation broadly influences myriad resulting resource limitations including a lack of money, knowledge, power prestige, and the kinds of interpersonal resources embodied in the concepts of social capital, social support and social networks. MLP s may increase families' social capital by giving them a way to increase and take control of their income and allocate funds and time needed to access pediatric preventive health care more easily. MLP representation may also increase available income by enforcing laws that control

PAGE 71

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 71 access to social security, Medicaid, TANF, SNAP, and other income benefits. I further theori ze that MLP s may improve residential conditions for children, by reducing destabilizing economic and legal stressors such as eviction and foreclosure, while increasing upward residential mobility so that patients' families can move to neighborhoods with gr eater social opportunity in the form of food security, educational opportunity, and transportation. Improved residential stability and upward mobility can lead to improved social networks for children and their families, in communities with improved built environments and access to populations that support healthy behaviors. In future studies, researchers may also find that MLP s increase self efficacy and mastery, decrease stress, and improve health behaviors that could allow low income and minority famil ies to better provide children's health care, food, and shelter. The lack of legal representation influences multiple risk factors that contribute to a persistent association between limited legal access and poor health. Existing evidence shows that even progressive improvements in the underlying laws that affect low income and minority families whether large sweeping pronouncements such as the federal Civil Rights Act of 1964 which prohibits all forms of discrimination or minor building code regulation s that require habitable housing conditions low income and minority families who lack legal resources will be largely unable to take advantage of their rights guaranteed by the law because they will be unable to enforce them as they pertain to their dail y lives and interactions (Seron et al., 2001) Low income families have more frequent and more urgent legal needs than other families (Rhode, 2004) Thus the fundamental relationship between their lack of legal services and poor health outcomes will remai n undisturbed. I posit, therefore, that the lack of legal services suffered by minority and low income Coloradans represents a fundamental social cause of poor

PAGE 72

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 72 health that limits access to resources that vulnerable populations require to avoid or correct t he disproportionate risks for morbidity and mortality to which they are exposed. I have considered MLP s' impacts on racial and ethnic minority patients because Link and Phelan (1995) have argued convincingly that race discrimination is a systemic, fundam ental cause of health inequality, independent of SES. Race is associated with a variety of flexible, race related resources such as occupational prestige and power, beneficial social connections, freedom, and neighborhood segregation that produce disparat e health outcomes (Phelan & Link, 2015) Moreover, Phelan and Link (2015) assert they find evidence of differences in multiple, flexible resource s related to health disparities, which can only be explained by racial disparities, not attributable to diffe rences in SES. Specifically, Phelan and Link (2015) identify neighborhoods as an important mechanism to understand racial differences in health. This, they explain, is because neighborhoods, independent of individual level SES, account for differences in t oxic environmental exposures, nutrition, poorer police and fire protection, inferior recreational spaces, higher levels of tobacco and alcohol advertising, and a technologically inferior quality of hospitals and other health providers. Figure 1 6 below des cribes the theory of race as a fundamental cause of health inequality. Figure 16 Racism As A Fundamental Cause Of Health Inequality, Phelan & Link (2015)

PAGE 73

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 73 The evidence that race, through the mechanism of racism, impacts housing i s particularly persuasive. Racial segregation in housing is a fundamental cause of racial disparities in health (Williams & Collins, 2001) Segregation is defined as the geographical separation of people, pr imarily unrelated to personal preferences, based on ethnicity or race. Residential segregation is detrimental to health outcomes for minority populations (Sudano, et al., 2013) This is because when black and Latino populations live in segregated neighbor hoods, they are isolated from the resources white populations can access to protect and improve their health. Residential segregation relegates black and Latino populations to areas of lower quality housing, schools, food, employment, and recreational spac es as well as increased exposure to violence, environmental hazards, and disparate law enforcement practices (Smith & Petrocelli, 2001 ; Briggs & Keimig, 2016) Residential segregation is also associated with inferior access to health care providers and acc ess to lower quality providers such as pharmacies with lower quality inventories, clinicians with inferior training and experience, hospitals with worse outcomes, older physical plants, and less medical equipment (Skinner et al., 2005 ; Bach et al., 2004) Higher rates of racial segregation are associated with higher rates of adult and infant mortality ( L aveist 1993 ; L aveist 2003) coronary heart disease (Diez Roux et al., 2001 ; Fang et al., 1998) infectious disease such as tuberculosis (Acevedo Garcia, 20 01 ; Acevedo Garcia, 2000) and with poorer mental health (Aneshensel & Sucoff 1996 ) even when researchers control for poverty rates. Residential segregation is also associated with higher homicide rates, one of the key drivers of the gap in black white li fe expectancy. While this study does not directly evaluate the impact of racial segregation on health outcomes, the study does evaluate the relationship between race and health outcomes as a way to infer whether moving from more to less segregated

PAGE 74

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 74 neighbor hoods, which are highly correlated with improved median family incomes, is a mechanism the MLP intervention might impact. The objective of this study is to evaluate medical legal partnerships as a disruption of the fundamental cause relationship between p overty, race, and poor health outcomes, due to the lack of legal access that vulnerable Colorado families suffer. I theorize that medical legal partnerships change not only the diffuse health consequences of poverty and discrimination, but also change the way in which health care delivery is fundamentally realized. I have chosen a methodology to infer a cause and effect relationship between MLP s and outcomes that may improve the outcomes that fundamental causation predicts. Whereas poverty and race predict poor access to health, preventive health care, and residential stability, this study will examine whether MLP s disrupt this relationship to improve access to primary care as evinced by immunization compliance and improved residential stability. The Life C ourse Cumulative Disadvantage Theory Of Health Inequity The final theoretical construct informing this study is life course theory. I hypothesize that the impact of medical legal partnerships may redirect life course trajectories for pediatric patients and their families who receive the intervention. The life course theoretical framework is appropriate to study the MLP influences, first because it is rooted in a contextualist perspective ( Elder & Johnson, 2003) and second because it examines changes in indi viduals and groups as large social forces alter environmental pathways. The life course theory examines changes by noting first how life trajectories are formed in a historical and social context and then how a substantial change at the individual level th rough legal representation and more broadly by policy advocacy may represent a turning point in that trajectory.

PAGE 75

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 75 The life course perspective has its origins in longitudinal studies examining individual lives over changing times. For example, Leonard D. Ca in, Jr.'s essay, "Life Course and Social Structure introduced the concept that aging represents a systematic series of stages or statuses that individuals occupy over time ( Cain, 1964; Marshall & Mueller, 2003) Later, the theory was both contextualized and collectivized as Matilda White Riley and others developed four central premises of life course theory which included the notions that a cohort of aging individuals could both be affected by social and environmental changes and also could produce social change as a group of individuals in the same c ohort shared common experiences (Riley, 1973). Organizing family, work, and other societal roles into "trajectories" shaped and influenced by changes he called "transitions," Glen Elder, Jr. i s credited with a rticulating the five principles of the life course, which have emerged as the dominant conceptualization of life course theory and that are widely used to study human development and aging. The principles are 1) human development and aging are lifelong pro cesses; 2) individuals construct their life course through human agency; 3) the life course is shaped by historical times and places; 4) events and transitions vary according to their timing over a person's life; and 5) individual lives are linked through a network of shared relationships (Elder & Johnson, 2003 ) According to Elder, the life course paradigm represented a major change in the way to think about and study human lives by focusing on the social forces that shape and produce developmental conseq uences throughout life (Elder, 1994) My study theorizes that all five of Elder's principles of development are affected by the level of legal power individuals may exercise to enforce their economic and social rights. Specifically, low income and minorit y children are thwarted in their ability to experience good health early in life, because the historic inequities, the environment, and the social relationships that surround them are shaped by inequitable distribution of opportunity and resources.

PAGE 76

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 76 Importa ntly for this study, life course theory has provided a method to study changes in health experienced by cohorts of individuals as they age. As a result, the perspective has highlighted the possibility that the roots of health inequalities begin in the biol ogical and social experiences that individuals and groups experience early in life, and further, that the relationship between socio economic inequality and health increases over time. This conceptualization, called d ivergence theory" or the "cumulative a dvantage/disadvantage hypothesis" predicts that socioeconomic differences in health increase with age. Differences in social resources between socioeconomic groups produce inequalities early in life, which increase over time and consequently produce increa singly greater health disadvantages. Taking a life course approach to inequality, in 1968, Merton argued that inequality results from the unequal distribution of resources that supports productivity with recognition leading to further productivity and incr easingly working to the advantage of the few, while also working to disadvantage the most (Merton, 1968) Lisa F. Berkman and Ichiro Kawachi identify three hypotheses that have been proposed by social epidemiologists as life course theories to explain the early social and biological influences in life that affect the onset of disease in middle and later life. I apply their second hypothesis of "cumulative disadvantage" in my study. Cumulative disadvantage theory posits that "disadvantage in early life sets in motion a series of subsequent experiences that accumulate over time to produce disease after y ears of disadvantage" (Berkman & Kawachi, 2000) Catherine Ross and Chia ling Wu (1996) convincingly applied this cumulative disadvantage theory to healt h inequality by demonstrating that the health advantages of high income and educational attainment increases with age (Ross & Wu, 1996)

PAGE 77

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 77 In 2005, Stephanie Hatch further elucidated the cumulative advantage hypothesis (Hatch, 2005) Hatch predicted that het erogeneity in trajectories through exposure to adversity and advantage are organized around key transitions that increase relative inequality over the life course. According to Hatch, the cumulative adversity processes can take many forms. A single hardsh ip or loss of a single protective factor might trigger a cumulative disadvantage. Alternatively, a chain of contingencies might occur such that each hardship sequentially surpasses its predecessor. In a third scenario, a cascading sequence of layering har dships may build increasingly intensifying effects that begin small but grow into larger consequences. Regardless of scenario, intervening adversity mediates the relationship between SES and poor health outcomes. Importantly, Hatch (2005) also posits th at protective resources that increase social mobility may affect the causal relationship between low SES and poor health outcomes by making the sources of adversity avoidable and by making it possible for individuals to achieve in and attach to social inst itutions over their life course. I suggest that MLP s may be such a protective resource. I further suggest that by measuring MLP s impact on early childhood experiences with the outcome variables chosen here immunization compliance and residential mo bilit y my study can shed light on the health and social advantages poor and minority children may enjoy over the life course given equal access to the social determinants of health. Conceptual Framework For This Study: The MLP As A Disruptive Protective Reso urce In an article that is particularly important for the conceptual model that informs this study, Yoav Ben Shlomo and Diana Ku g (2002) further elaborated the integrated biological and psychosocial pathways that a life course approach to disease suggests. Their diagram in Figure 1 7 below illustrates that exposures to adversity might take a predomina nt ly biological pathway

PAGE 78

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 78 (Path a), a predominantly social pathway (Path b), a socio biological pathway (Path c) or a bio social pathway (Path d) (Ben Shlomo & Kug, 2002) Figure 17 Ben Shlomo & Ku g (2002) Biological Psycho Social Pathways To Disease Disparities Ben Shlomo & Kuh (2002) capture the complexity of multiple pathways that combine to produce health disparities. My study as serts that regardless of which of these pathways social adversity travels whether that adversity has taken a biological, social, or mixed pathway to produce poor health outcomes an MLP intervention that re distributes access to the social determinants of health can disruptively weaken and redirect the association between social adversity and poor health to ultimately reduce disparities. I theorize that access to legal services can increase and equalize social determinants of health that affect all pathw ays between social adversity and inferior health outcomes. In the Figure 16 below, I condense the multiple pathways suggested by Ben Shlomo and Ku g (2002) My conceptual diagram summarizes the various exposures that affect a child's health over the life c ourse into two cumulative and interactive categories of disadvantage. One category contains social disadvantage (e.g. air pollution, poor educational attainment and occupational hazards from Ben Shlomo and Kuh's

PAGE 79

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 79 diagram), and the other contains biomedica l disadvantages such as poor growth in utero, nutrition, asthma, or childhood chest illness from Figure 1 7 The diagram of my theoretical conceptualization for this project focuses on the limited and inequitable access to the social determinants of healt h as the overarching mechanism that operates along all the various pathways to disease disparities that originate during early childhood. The grey arrows in Figure 18 connect limited and inequitable resources to poverty and race, and then show that each of these two independent, explanatory variables lead to inferior health. I intend to suggest, by this diagram, that poverty and race may overlap but are each independently associated with inferior health and are not individually but systemically generated. M oreover, I intend to illustrate in Figure 18 that I adopt Stephanie Hatch's (2005) view, that the adversities, which result from a misdistribution of the social determinants of health, are interconnected and cumulative. For example, housing and education inequity combine to produce employment and income deficiencies, which exacerbate the inability of the poor and of minority families to equitably access health care and healthy food for their children. Figure 18 depicts the role I theorize MLP s play as a p rotective resource to disrupt the pathways of cumulative adversity. The blue arrows point to the role that MLP s can play addressing structural inequality through legal and policy advocacy, inequities at the individual and family level through legal service s and representation, and health care delivery disparities through training and education. Moreover, these blue arrows illustrate my conceptual hypothesis that MLP s may disrupt all pathways that connect social adversity to poor health outcomes, whether soc ial, biological, or mixed. MLP s may intervene to improve limited and inequitable access to the social determinants of health that prevent poor and minority populations from coping with the effects of biological as well as social adversity. Figure 1 8 offers a detailed,

PAGE 80

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 80 graphic look at my overarching theory that limited and inequitable access to the social determinants of health explains why poverty and race are associated with poor health through a variety of pathways over the life course and how MLP s may di srupt this association. Figure 18 Conceptual Framework: MLP As Protective Resource To Disrupt Pathways To Health Disparities I theorize that medical legal partnerships may provide legal representation as a protective resource and thus potentially weaken the effect of inequities that characterize the social determinants of health. As with other protective resources, MLP s provide services that act on medi ator s (e.g. s ocial and biological disadvantages) and that can imp rove causal pathways ; MLP s can intervene in any one of the socio biological pathways that lead to health disparities, thus creating disruptions in the established relationship between race, poverty and poor health. In sum, I have applied ecosocial theor y, fundamental cause theory, and a theory of cumulative advantage to achieve the specific aims of this study. I s ought to determine whether medical legal partnerships are an intervention that can redirect inequitable access to the social !"#"$%&'()&'*)%+,"$(-.%' /00%11'$2'$3%'420"(.' 5%$%6#")()$1'27'8%(.$3' 9,#,.($":%';"2#%&"0(.' 5"1(&:()$(<%1 *)7%6"26'8%(.$3' =,$02#%1 >2:%6$?'()&'@(0% 9,#,.($":%'420"(.' 5"1(&:()$(<%1 MLP Improves Access to Social Determinan ts of Health MLP Improves Access to Social Determinan ts of Health

PAGE 81

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 81 determinants of he alth for poor and minority populations in Colorado, thereby moderating the intervening adversity that connects race and low socio economic status to disparately adverse health outcomes. Although I could not accurately measure changes over the life course w ithout more extensive longitudinal data, I expected my analysis to show improvement in immunization compl ia n ce and residential mo bility that is associ ated with the CHCO MLP services. I predicted based on the life course/cumulative advantage theory, that t he likely impact of these improvements in children's lives will last longer than the five year period of this study. Theoretical Support For The Choice Of Study Measures: Immunization Compliance And Residential Mobility First, this study considers whet her children whose families have legal representation through MLP s are more likely to adhere to immunization schedules recommended by the Advisory Committee on Immunization Practices (ACIP). Children's immunization compliance is theorized as a proxy for ev idence the child has regular access to preventive health care required to adhere to the recommended schedule and objectively better protection against infectious diseases Of course, immunization compliance only suggests but does not confirm that a patient had regular preventive care, because a number of non clinical settings in Colorado can offer vaccinations needed to remain immunization compliant, without providing any accompanying preventive care. Nevertheless, it is not unreasonable to use immunizatio n as a proxy for preventive care access because compliance represents some level of clinical encounter. The ACIP recommends that children receive routine childhood vaccinations by age 2 years, yet the evidence suggests that fewer than 25% of children rec eive vaccine doses at the appropriate time. 1 Children who do complete the recommended series of vaccinations between birth and age 2 are protected against 14 childhood diseases. Pediatric immunizations have been

PAGE 82

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 82 linked to improved health outcomes and healt h cost savings over the life course. The CDC reports that among U.S. children born during 1994 2013, vaccination will prevent an estimated 322 million illnesses, 21 million hospitalizations, and 732,000 deaths during their lifetimes (CDC, 2014) Economist s have estimated that between 2005 2009, compliance with the ACIP immunization program could have prevented 42,000 early deaths and 20 million cases of disease over a lifetime and could have saved approximately $13.5 billion in direct costs and $68.8 billi on in total societal costs in the United States (Kurosky, Davis, & Krishnarajah, 2016) Estimated vaccine coverage among poor children is lower than among children who live above the poverty level. Median vaccination coverage rates are near 95% of nationa l Healthy People 2020 targets for U.S. children generally, however, low vaccination coverage can cluster within poor communities leaving low income and minority children more vulnerable to vaccine preventable diseases. In Colorado, however, vaccination cov erage patterns are more complex. Overall, vaccine coverage rates are declining among all children in Colorado. In this state, vaccine exemption rates are higher and completion rates are among the lowest in the nation, yet it is wealthier and typically whi te families who avail themselves of vaccine exemption in Colorado (CDC 2014) Moreover, although black and Hispanic children are disproportionately represented among Colorado's poor, a group that has lower vaccination coverage than wealthier families nati onwide, and although Hispanic children are three times more likely to be uninsured than whites in Colorado, Latino vaccination coverage is slightly higher than for whites in this state. This reversal is possibly because the state's Vaccines for Children (V CF) program is an effective outreach to low income and Hispanic families, while Colorado's wealthier, white families are more often the demographic group that uses Colorado's exemption from

PAGE 83

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 83 vaccination. This study will, in any event, provide additional inf ormation about the influences that affect the complex picture of vaccine coverage in Colorado. This study hypothesizes that MLP s may improve ACIP compliance rates for low income and minority children in Colorado in four ways that could reduce health dispa rities. First, the MLP intervention may provide a child's household with improved access to health insurance to reduce financial barriers that impede access to regular preventive health care for children. When MLP services improve access to Medicaid or CHI P benefits, children may experience better immunization compliance because their caregivers have insurance coverage to pay for preventive health services. Second, when MLP representation improves a family's access to other income supports such as SNAP, TAN F, disability, or unemployment benefits, a child's caregivers may redirect funds from providing immediate needs such as food or shelter to cover the cost of routine health care. Moreover, MLP s may also give caregivers the financial freedom to avail themsel ves of transportation, childcare, or time off work making health care for children not only affordable but also accessible. Third, my interviews with former MLP clients suggest that MLP s may reduce the stress caregivers experience as a result of unsolved legal problems thereby reducing underlying psychosocial factors that may stand in the way of immunization compliance (Zhou et al., 2014) A primary example is resolving immigration status concerns. CHEP has successfully adjusted legal status for several c lients who then explained they had one less barrier to obtaining regular health care for their children (Matthew, 2010) Fourth, CHEP faculty and student attorneys have seen families respond to receiving MLP services by engaging in better health behaviors such as missing fewer clinic appointments, suggesting that resolving legal problems may improve self mastery and control in a way that increases access of preventive health care services such as childhood immunization.

PAGE 84

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 84 The second outcome variable that thi s study examines is upward residential mobility. This variable is derived from the absolute number of residential moves participants made during the study period. The frequency of address changes suggests the level of a child's residential stability. Child ren who enjoy stable living conditions are more likely to belong to households that have food security, well developed social networks, and social cohesion that significantly influence children's mental and physical health (Drukker et al., 2003). In contra st, the evidence shows that housing instability is associated with food insecurity and limited health care access, affecting routine and acute health care utilization in children. Such residential disruptions in a child's life can produce increased suscept ibility to health and health disparities throughout a lifetime (Ma, Gee, & Kushel, 2008) Studies demonstrate that housing instability is independently associated with not having a usual source of care, postpone d medical care and medications, and increased hospitalizations and ED use. Moreover, researchers have concluded that policies that decrease housing insecurity can promote the health of young children, thereby affecting their health in later life (Cutts et al., 2011) However, the outcome measure here is a refinement on a simple count of the number of residential moves a child may have experienced because f requent residential relocations may not always signal instability. Research based on the Moving to Opportunity experiment showed that children whose families move to lower poverty areas experience significant improvement in mental and physical health, as well as in family safety (Kushel, 2006) In addition, r ecent research offers compelling evidence that for younger children, moving to lower poverty neighborhoods has positive long term economic impacts as well. Moreover, the best evidence is that family moves from high poverty to lower poverty neighborhoods ar e associated with reduced dependence on public benefits, improved health status, and improved school and

PAGE 85

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 85 behavior outcomes for children (Ludwig et al., 2013) Chetty Hendren and Katz showed that improving neighborhoods has a causal exposure effect on chi ldren's educational achievement and earnings outcomes ( Chetty, Hendren, & Katz, 2016). Importantly, the length of time that a child spends in improved neighborhood conditions matters to their long term economic prospects. Specifically, they found that "e very year spent in a better area during childhood increase [s] college attendance rates and earnings in adulthood. In other words, the gains from moving to a lower poverty neighborhood are greater for children who are younger when they move. Indeed, Chetty describes the treatment effects at younger ages as "substantial": children who move to better neighborhoods when they are less than 13 years old earn annual incomes that are 31% higher on average than children who do not move, while children whose families move to lower poverty neighborhoods when they are over age 13 may experience a negative impact on their earnings when compared to children who remain in higher poverty neighborhoods (Chetty & Hendren 2015) Therefore, I consider the MLP 's impact on the neighborhood economics of housing changes that pediatric patients experienced during the study period. I hypothesize that the MLP intervention will encourage upward housing mobility. I expect that treated families will show more upward mobility tha n untreated families, and families will show more upward residential mobility after the MLP intervention than before. My study considers housing mobility at an important time in Colorado's economic history. Colorado is experiencing a decline in affordable housing, especially in the Denver metropolitan region where many of this study's patients live. Colorado wages have not risen as fast as housing costs throughout the state. Therefore, the lack of affordable housing can spell danger for children whose fam ilies spend more than 30% of their income on housing. They may find housing affordable only when conditions are overcrowded or substandard. Frequent moves

PAGE 86

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 86 and even homelessness may result. Moreover, children may more frequently live in conditions that lead to illness and school absences. Residential security has been linked not only to disruption in school attendance but also to the quality of school instruction. High mobility students tend to receive slower, repetitive curricular content. They also may los e the benefit of stable social networks. Kidscount Colorado concluded that "[f]orty one percent of highly mobile students are low achievers, compared with 26 percent of stable students. Mobile students are half as likely to graduate" ( Chetty, Hendren, & Ka tz, 2016) These are the long term consequences that I theorize could be changed over the life course if MLP interventions prove effective. I theorize that MLP s may improve families' financial resources to allow them to provide housing for children in bett er neighborhoods with access to better social capital and networks MLP s may help low income and minority families enforce legal rights that protect against eviction, foreclosure, housing discrimination, and other law based affronts to residential stabilit y. Social mobility during childhood may narrow social inequality during adulthood as a move may also expose children to better schools, safer streets, and increased food security. Where health outcomes are concerned, social mobility may serve as a "gradien t constraint" so that the size of health inequalities among populations that are socially stable are greater than the size of inequalities experienced by the socially mobile (Colorado Children's Campaign 2013) In other words, the effect of social mobilit y can be to constrain long term inequality (Blane et al., 1999 ). To the extent that MLP s improve not only financial resources but also the social networks and resources that children are exposed to, the intervention may affect positive social mobility for children that could last a lifetime. In short, this study conceptualizes medical legal partnerships as a protective resource that mitigates the adversity low income and minority children experience

PAGE 87

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 87 because of inequality. The next chapter presents the rese arch design and methodology for this study.

PAGE 88

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 88 CHAPTER IV RESEARCH DESIGN The goal of this study was to examine whether MLP services can improve health and social outcomes of disadvantaged children. Overview As more fully described below, I analyzed data f rom a group of 1,077 ped iatric patients who received MLP services and were all unde r age 3 during the study period The two outcomes of interest in this study are immunization compliance rates and upward residential mobility. The treated patients we re divi ded into two groups; 985 patients were treated with MLP services without formally becoming legal aid clients, and 92 patients were treated with MLP services as legal aid clients. I obtained retrospective data from both treated and un treated patients' elec tronic health records and compared the two groups immunization compliance and upward residential mobility. I also compared the MLP 's effect on these outcomes for a subset of 272 patients from the T reatment Group, selected using propensity score matching to form a control group for analysis. Next, I examined how the MLP af fected t he predicted association between patients' race and SES and the outcome variables In order to compare the outcome variables pre and post MLP intervention, I calculated the hazard rate as the conditional probability that each child would experience immunizations or upward residential mobility at a given age group Following the methodology outlined by Singer and Willett, I defined the beginning of time in my study as 2009, when the MLP program began (Garg, Marino, Vikant, & Solomon, 2012). The metric for clocking time was the age of patient participants in years Lastly, I conducted a qualitative and mixed method analysis on the 92 patients who received MLP services from legal aid. I explored relationships among those patients' legal problems, services, and outcomes by

PAGE 89

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 89 race and gender, and then explored whether the intensity of legal aid services they received affected the two outcome variables of interest. Research Design I employe d an explanatory, sequential, mixed methods, retrospective, cohort design to gain several advantages (Leech, Nancy L. And Anthony J. Onwuegbuzie, 2009) First, the mixed methods design fits the theoretical framework that informs this study. The quantitative analysis of associations between the MLP intervention and the dependent variables that reflect health and social conditions in early childhood sheds light on the ecosocial theory of health disparities that is the theoretical foundation of this study. The mixed methods design tests not only whether the MLP intervention changes the patients' outcomes after receiving services but also examines how those outcomes correlate with specific legal services and outcomes reflected in the legal records (Bartley & Plewis, 2007) A second benefit of this design is that it afforded flexibility in addressing research variables that have been under explored (e.g. immunization compliance ) or wholly i gnored (e.g. type of MLP service type and residential mobility ) in the MLP literature to date. Third, this study design could help to promote collaboration among health providers, legal professionals, policy makers, and social scientists by using a mixed method approach that relates data from medical and legal sources. Finally, the study design accounted for the respective weaknesses in quantitative and qualitative analysis by combining both methodologies to study the impact of MLP interventions. For exam ple, while a quantitative study of the CHCO data may show the relative extent to which children received health related benefits from the MLP intervention, the qualitative study added an enriched understanding of whether some MLP services are more benefici al than others. Together, the qualitative and quantitative aspects of this study provided a more contextualized picture of the MLP 's impact.

PAGE 90

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 90 Once approved by the COMIRB in February 2015, (Protocol Number 14 1475, Attachment A) this study proceed ed in three phases as shown in Figure 1 9 below: Figure 19 Three Phases Of This Study I selected the research design based on my MLP experience described in Chapter 1, the literature reviewed in Chapter 2, the theoretical framework presente d in Chapter 3, and my interviews with the pediatricians, attorneys, and paralegals who staffed the CHCO MLP from 2009 2013. These interviews helped me to identify the patient population for this study and to understand the MLP treatment model they employ ed. I then chose an analytical approach to answer the questions that arose based on these experiences and existing data. DATA I compiled the quantitative datasets for my study from the de identified, EPIC electronic medical records maintained by CHCO an d from data on immunizations from the Colorado Department of Public Health and the Environment (CDPHE). I collected immunization records and all other health indicators in two person period datasets in which each patient had a separate !"#$%&'& ()#*+,+#+,-%&#*#./$,$&01& 23!4$&%11%5+$ !"#$%&'' ()#.,+#+,-%&#*#./$,$&01& 6%.#+,0*$",7&8%+9%%*& .%:#.&#,;&$%6-,5%$&#*;& 0)+50<%$&$"09*&,*& ;05)<%*+#6/&.%:#.& 6%506;$ !"#$%&''' 2,=%;&<%+"0;$&#*#./$,$& 01&6%.#+,0*$",7&8%+9%%*& .%:#.&,$$)%$>&23!& $%6-,5%$>&.%:#.&0)+50<%$& #*;&"%#.+"&0)+50<%$

PAGE 91

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 91 row for each visit t hey made to the CHCO clinic Most covariates in these datasets are static however each patient's age in days from birth i s time varying I also collected explanatory, demographic information for each patient in a person level dataset and then merged the t hree files into a single person period dataset I gathered and coded the documentary data for the qualitative phase of this study from legal fi les obtained from Colorado Legal Services (CLS) the state's legal aid law firm that provides legal assistance a nd advocacy for lo w income Coloradans. CLS provided de identified, redacted copies of the legal files for the CHCO patients for whom Colorado Legal Services entered a formal retainer agreement ; from this group, I analyzed data from the patients who were u nder age 3 during the study period. For these patients I collected data from the following legal documents : Legal Services Agreement with Colorado Legal Services, Inc. (CLS) This document states the legal problem CLS is retained to address. CLS Intak e/Eligibility Information Sheet Checklist to establish client meets CLS low income and citizenship criteria. Case notes and correspondence Containing attorney's notes, letters written and received in the case. Case Closing Checklist and Certification or Closing Memo Which includes dates case opened and closed, code description of case outcome, and code or description of services provided. Based on these documents, I extracted data from the legal records and then coded fields on an Excel spreadsheet to identify which legal problem, services, and outcome each participant had. Next, I imported the excel spreadsheet into Dedoose, a qualitative research software program developed by sociocultural Research Consultants, a software design group based at UC LA. Methods

PAGE 92

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 92 I analyzed observational and longitudinal data for all 11,553 pediatric patients who visited the CHCO cl inics where the MLP operated and who were under age 3 during the five year study period. Inclusion criteria w ere based on the information I received in interviews about s creening procedures All patients who visited the clinics were eligible to be screened and therefore eligible to be included in the study. However, patients who visited the clinic were not uniformly screened during the five y ear period. During some years, patients were screened one day per week, but it is not clear they were screened on the same day each week. During other years, patients were screened multiple days per week. On still other occasions patients may have been re ferred to the MLP for screening based on the ad hoc judgment of an individual physician or social worker. Different staff members took responsibility for screening and each followed record keeping protocols to differing extents. Therefore, I could not rule out selection bias based on the screening procedures. I divided the 11,553 patients into two groups: 4,898 patients who visited the clinic during the study period and were screened comprised one group. The remaining 6,655 patients who visited the clinic b ut were not screened comprised the un screened group Treatment Group Next, I identified which of the screened patients received the MLP treatment and which patients remained untreated All staff interviewed agreed that 1) MLP representatives, or in some c ases a hospital representative screened for MLP patients by presenting the instrument shown in Attachment B to the caregiver accompanying the pediatric patient to the clinic visit; 2) any patient whose caregiver responded to the questions on Attachment B w ere considered "screened" and therefore entered into a the study database; 3) patients whose caregiver answered "Yes" to any of the first four screening questions and also answered "Yes" to the fifth screening question to indicate they wished to receive a contact from the MLP were considered to have s creened

PAGE 93

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 93 positive for MLP need" (" MLP Positive") ; 4) all patients who screened MLP Positive patients received some sort of MLP services; 5) patients who did not answer with two Yes" responses were considered ne gative for MLP need (" MLP Negative") and did not receive MLP services. This interview process resulted in a sample group of 1,077 treated patients. S taff interviews and legal records further revealed that 92 of the 1,077 treated patients were referred to C LS and became legal aid clients ("CLS Case" patients) The remaining 985 treated patients received other MLP services. Together, these 1,077 patients comprised the "Treatment Group." Figure 20 provides a flowchart to graphically describe the inclusion crit eria for this study, which I have confirmed with the CHCO MLP staff. Attachment C contains demographic information about each patient group including distribution by race, ethnicity, and age group. Figure 20 Inclusion Criteria F or Children's Hospital Colorado ( CHOC ) MLP Treatment Group Patient Population January 2009 December 2013 Control Group A6%($#%)$'B62,C D)EFGHIIJ K!>'>21"$":%':L'K!>' M%<($":%'>($"%)$1 406%%)%&':L'N)106%%%)%&' >($"%)$1 O."<"-.%'>($"%)$1'N)&%6' /<%'P'Q32'R"1"$%&'9.")"01' Q3%6%'K!>'406%%)%& FFGSSP'>($"%)$1' /<%1'H T ULVV''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' ?%(61 WGXVX'>($"%)$1'/<%' H T ULVV'?%(61' 406%%)%& FGHII'>($"%)$1'/<%' H T ULVV'K!>'>21"$":%' 726'K!>'M%%& VXS'>($"%)$1'/<%'H T ULVV'>21"$":%'726' K!>'M%%& VU''K!>'>21"$":%' >($"%)$1'/<%'H T ULVV' Q"$3'9!4'!%<(.''/"&' 9(1%'Y".% PGXUF'>($"%)$1'/<%' H T ULVV'M%<($":%'726' K!>'M%%& ZGZSS'>($"%)$1'/<%' H T ULVV'?%(61'M2$' 40%%)%&

PAGE 94

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 94 I experimented with several approaches to constructing the control group First I considered comparing the 1,077 Treatment Group pa tients to a control group comprised of the 3,821 MLP N egative patients However, when I compared these groups, t he data suggested that patients who screened positive for MLP need were generally from lower socio economic populations than patients who screen ed negative for MLP need. Indeed, s everal studies confirmed the family and household income levels among Treatment Group patients would be unlike the control group if selected from patients who screened negative for MLP need. For example, in Development of a Brief Questionnaire to Identify Families in Need of Legal Advocacy to Improve Child Health, Dr. David Keller found, "families that identified themselves as Hispanic, with lower incomes, and families seen at community health centers were independently mo re likely than families without those characteristics to be screened positive" and referred for legal counsel (Keller et al., 2008). Gottieb et al f ound similar results among patients who screened positive in a randomized trial testing electronic versus f ace to face social screening tools (Gottlieb et al 2014) Similarly, a group at Johns Hopkins Children's Center reported the most prevalent need among 1,059 families studied was for employment (Brookhart, et. a l ., 2006) And medical homes have regularly confirmed that income related concerns such as the need for employment are the most frequently identified patient concerns in social needs assessments if low income children (Garg et al., 2008) The literature correctly predicted the income differences am ong patients in the Treatment and MLP Negative groups empirically as well. I found a significant difference between the two groups with respect to insurance sources, which I interpreted as a proxy for income. MLP Negative patients were nearly twice as like ly as patients in the Treatment Group to have private health insurance, the type of coverage that is more often associated w ith steady employment.

PAGE 95

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 95 Table 1 shows that while only 7.7% of the MLP Positive cohort had private insurance, 14.7% of the MLP Negati ve cohort had private insurance. Tab le1 also shows the Treatment and MLP Negative groups were otherwise quite similar with respect to ethnicity and race; medical debt owed; frequency of in patient hospitalization; number of ER visits; and APGAR scores (a p roxy for the health of these patients at birth). Table 1 Frequency And Mean Comparisons Of Treatment Group And MLP Negative Patient Groups Characteristic MLP Positive Patient Group MLP Negative Patient Group P value Patients (n) 1,077 3,821 First Age (years) .44 .44 .00 Race Black (%) White (%) Other (%) 28.0 27.7 44.4 27.6 40.0 42.4 Ethnicity Hispanic (%) 49.0 44.7 Sex (% Female) 45.7 47.3 First Insurance % Public (income contingent) % Private % Uninsured 91.1 7.7 20.5 84.0 14.8 19.1 % With Medical Debt to CHCO 27.2 26.3 ER Visits, mean (SE) 4.7 (.16) 4.0 (.07) .00 Inpatient Hospitalizations, mean (SE) .6 (.04) .4 (.019) .24 Apgar at 1 Minute, mean (SE) 7.6 (.09) 7.5 ( .05) .0001 I concluded from this analysis that the differences between the MLP Positive (Treatment Group) and MLP Negative patient group were too large to use the latter as a source for creating a

PAGE 96

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 96 control group. Instead I used the propensity score matchi ng method to construct a control group from the unscreened group to compare with the Treatment Group outcomes. Propensity score matching I used propensity score matching as a method to approximate a randomized experiment from which I could infer causality by creating a control group of 1,231 patients from among the 6,655 unscreened patients who had no prior contact with MLP staff or services. The propensity score matching method uses proportional quota sampling to select control group members that have obs ervable pre treatment characteristics in approximately the same proportion as patients in the T reatment G roup In 1983, Rosenbaum and Rubin introduced the propensity score, a number to represent the conditional probability that a person might be assigned to receive a particular treatment, given a vector of pre treatment, observable covariates (Rosenbaum & Rubin, 1983) Propensity score matching is an appropriate sampling method for this study that analyzes observational data usin g retrospective analysis, w here I could not employ a randomized technique to sample patients for comparison. Yet the method allowed me to estimate a cause effect relationship between the MLP intervention and outcome variables of interest. According to the propensity matching theory, the bias between the MLP Tr eatment G roup and the control patient group was reduced to the extent that this sampling methodology compares individuals from the two groups who have the same value propensity scores. Thus, any differences that remain may reas onably be attributed to the MLP intervention. Bingenheimer, Brennan, and Earles used this method to estimate the causal impact of exposure to firearm violence and the subsequent perpetration of serious violence (Bingenheimer, Brennan, & Earles, 2005). In their study, these researchers claimed that "comparing individuals with identical propensity scores but different realized exposures is [sic.] a nalogous to conducting

PAGE 97

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 97 a randomized experiment, and therefore provides a valid basis for measuring a cause effe ct relationship between exposure and outcome" (Ri chards, 1998) My overall approach to propensity matching follows the analysis reported by Bingenheimer and colleagues in their estimate of the average treatment effect of exposure to firearm violence on ser ious violent behavior by adolescents. To apply the Bingenheimer methodology, I used Stata's propensity estimators described by Becker and Ichino to execute the propensity score matching method. A limitation of the propensity matching approach is that it ca nnot account for unobserved variables which may also confound outcomes. Another important caveat is that this method depends on having appropriate observed, pre treatment variables to match the patients in the Treatment Group with patients in the control group. The next section describes the pre treatment variables I selected for to calculate the propensity scores used to match patients in the Treatment Group with patients from the unscreened group in order to construct a control group for this study. Pre Treatment Variables I selected the pre treatment variables to create a control group for this study based on the propensity score literature, the theoretical framework underlying this study, and my understanding of the MLP policies and procedures from inte rviews with CHCO MLP staff. I collected data for 202 pre treatment variables from patient medical records in CHCO's Epic Systems Corporation, electronic health record software (Verona, Wisconsin). The original data for the pre treatment variables in th is study appeared in Epic either as text (such as descriptive information about race, gender, or age) or as International Classification of Diseases, Ninth Revision, Clinical Modification (ICD 9 CM) codes, a standard list of six character

PAGE 98

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 98 alphanumeric cod es to describe diagnoses. (See, excerpt from my codebook shown in Attachment D ). The variables I selected initially were guided in large part by a general, though not uniform, consensus in the literature that all available variables that are related to both the MLP exposure and the outcomes should be included in the propensity score model in order to increase the precision of my estimated exposure effect without increasing bias. The logic behind this approach is to reduce bias the distance of the estim ated treatment effect from the true effect, and also increase efficiency the precision of the estimated treatment effect (Garrido et al, 2014) Because I am using observational data, the patient characteristics in my datasets are likely to be associated with both the MLP treatment selection and the outcomes of interest. Without propensity scoring, there is likely to be a different distribution of these covariates in treatment and control groups. However, the overall goal of the propensity matching method is to balance covariates between patients who are treated and those who are untreated so that I can isolate the effect of the MLP D ue to high percentages of missing data, only the 87 pre treatment covariates listed in Table 2 below were available to use in calculating propensity scores I divided these 87 covariates into two domains to help identify pre treatment variables for the model ; the domains served as a way to select the covariates that related to the treatment selection and outcomes as discussed earlier T he first domain, "Patient Variables," contain s codes and descriptive variables about each patient individually This domain contains ICD 9 CM codes that are related to patients' general and specific health conditions as well as demographic des criptors These co variates include i ndicators of underlying mental and physical health conditions could impact a child's regular utilization of health care and could reflect the stability of a child's living

PAGE 99

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 99 circumstances. The second domain, "Contextual Variables," contains ICD 9 CM codes and variables that relate to the child's likely access to and quality of preventive care and the social stability of a child's family. These covariates also relate to the quality of health care to which the pediatric pat ients had access Table 2 shows that the only continuous, time varying variables in my dataset are those based on patients' ages, or the number of encounters patients had with health providers such as emergency department or office visits. Table 2 Pre Exposure Covariates For Propensity Score Matching PATIENT VARIABLES Child's Demographic Characteristics Child's Health Status Female Sex Dichotomous Age Group at First Visit Date Ordinal Race Categorical Avg Temp Per Year Ordinal Ethnicity Categorical Avg Heart Rate Per Year Ordinal Child in Foster Care/Welfare Custody Dichotomous Avg Weight Per Year (kg) and (lbs.) Ordinal Child's Visit History at CHCO Avg Height Per Year (cm) and (in) Ordinal No visits Pre MLP Dichotomous Avg BMI Percentile Per Year Ordinal Age Entered CHCO Clinic Continuous Child's Acute Complaint Number Days CHCO Patient Number Days In CHCO Clinic= (Age At Exit Age At Entry) Continuous Strep Throat Dichotomous Number Days in MLP Program Number Days In MLP Program = (Age At MLP Age At Exit) Continuous Viral Infection Dichotomous Child's Overall Health Condition Acute Bronchitis Dichotomous Death by Age 7 Dichotomous Sinusitis Dichotomous Death by Age 18 Dichotomous Acute Upper Respiratory Infection Dichotomous Apgar Score at 1 minute Ordinal Allergic Rhinitis Dichotomous Mental Illness Diagnosis Dichotomous Acute Pharyngitis Dichotomous

PAGE 100

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 100 Child's Specific Health Condition Chronic Illness Surgical History Dichotomous Diabetes Diagnosis Dichotomous Low BMI (<5th Percentile) Dichotomous Asthma / Asthma Diagnosis Dichotomous Hi BMI (>85th Percentile) Dichotomous Obesity Diagnosis Dichotomous Anemia Iron Deficiency/Anemia Unspecified Dichotomous Overweight Diagnosis Dichotomous Child w/ Ambulatory C are Sensitive Condition (ACSC) Dichotomous Mental Illness Specific ACSC Diagnosis Depression/ Depressed Adjustment Disorder Dichotomous Dehydration/Volume Depletion Dichotomous Anxiety Disorder/ Anxiety Dichotomous Low Birth Weight Dichotomous Sleep Di sorder Dichotomous Vaccine Preventable Disease Dichotomous Eating Disorder Dichotomous Pneumonia Dichotomous Reaction Disorder Dichotomous Traumatic Injury Personality Disorder Dichotomous Fracture or Sprain Dichotomous Hyperkinetic Conduct Disorder D ichotomous Burn Dichotomous Mental Retardation Dichotomous Child Accidents (traumatic accident, all, fire, environment, suffocation, Poison other poison or other) Dichotomous ADHD Dichotomous Child Vehicle Accident (Railway, MV, Traffic, Road, W ater, Air, Other) Dichotomous Autism Dichotomous Head Injury Dichotomous Developmental Disorder/ Delay Dichotomous Crushing Injury, Lower Limb Dichotomous Language Development/Dev Expr Language Dichotomous Behavioral and Cognitive Issues Developmental D elays Dichotomous Disturbance of Conduct Dichotomous Mild Cognitive Impairment Dichotomous Disturbance of Emotions Dichotomous Cognitive Delays Dichotomous CONTEXTUAL VARIABLES Family History Accessing Health Care Family Access to Diagnostic Care Regularly Scheduled Office Visits Continuous Lead Screening Dichotomous Missed Office Appointments Continuous Special Exams Dichotomous ER Visits / 365 Days Continuous Special Screen Pulmonary TB Dichotomous Hospitalizations / 365 Days Continuous Spec ial Screen Viral Dichotomous

PAGE 101

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 101 CHCO Outpatient Prescriptions / 365 Days Continuous Special Screen Bacterial Dichotomous Social Characteristics of Family Special Screen Blood Dichotomous First Insurance Type Categorical Special Screen Other (Eye, ear, urin ary, mental) Dichotomous Insurance Type at Visit Categorical Family Trauma History Medical Debt Written Off Dichotomous Family Substance Abuse Dichotomous Median Income Tier Categorical Family Disruption (Non Specified) Dichotomous Primary Language Cat egorical Homicide Dichotomous Parent Child Problems Dichotomous Family War Injury Dichotomous Family Residential Environment & History Sexual Abuse Dichotomous Frequent Movers (> Population Median) Dichotomous Physical Abuse/ Abuse Unspecified Dichotomo us Household Mold or Algae Dichotomous Family legal matter (Not MLP ) Dichotomous Ultimately, I sought to identify between 15 and 20 pre exposure covariates for matching purposes, locating my analysis between that of researchers who conducted studies to compare propensity scores based on as many as 250 variables (Peikes, Moreno and Orzol, 2015), and those who used approximately 15 variables to perform matching analysis (Schneider, Zaslavsky and Epstein, 2002). I focused on limiting the biases that may dis tort estimates of the effect the MLP exposure had on outcome variables in my analysis and on minimizing selection bias by using the available patient data to approximate randomization as closely as possible. I ran several iterations, combining various pre treatment covariates from the list of 87 available in Table 2 I evaluated the quality of the matches using several criteria: whether variable combinations resulted in collinearity, whether the balancing property was satisfied among the Treatment and con trol group patients, and whether the matches reduced the standard mean differences between matched and unmatched groups. I also sought to maximize the number of treated and control patients matched by Stata by using several of Stata's matching approaches. Based on these criteria, the resulting model included 10 co variates that 1) reflected patient use of available health care (ER visits, office visits, missed appointments, first age at CHCO ); 2)

PAGE 102

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 102 overall health status as exhibited by health at birth (Apgar at 1 min, low birthweight); and 3) general health condition over time (inpatient hospitalizations, average temperature at office visits, whether the patient has ever had an ambulatory care sensitive condition). Many of the contextual co variates that I c ollected had too many missing values to be useful to the matching exercise. Others proved unhelpful to the quality of my matches. In the final analysis, I used a combination of 10 pre treatment covariates which provided the best balance of factors related to my analysis, and for which my data were complete Attachment F shows different combinations of pre treatment covariates I tried and explains further my reasons for selecting the 10 covariates in this model. Calculating Propensity Score Using the 10 c ovariate model, I calculated a propensity score for each patient. I used Stata to balance the treatment and control cohorts on all measured pre exposure covariates and compute the mean difference between the MLP treatment cohort and the control sub jects on two outcome measures. After experimenting with Stata's radius matching, kernel matching, and stratification methods, I used Stata's "Nearest Neighbor Method" as de scribed by Becker and Ichino because this approach maximized the number of control and Trea tment Group matches. The resulting MLP frequency of the MLP treatment in my sample was: I used Stata to estimate the propensity score through multiple iterations using a logit regression to calculate the probability of treatment given ten selected cova riates The nearest neighbor

PAGE 103

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 103 determination searches backwards and forwards repeatedly for the closest match for Treatment Group patients, considering all available control group observations from among the 6,655 unscreened patient possibilities This is wh y the number of observations (1,577) exceeds the number of patients in the Treatment Group and the number of ultimate matches (Becker and Ichino, 2002) :

PAGE 104

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 104 Stata's pscore command stratified my data across successively created blocks, based on the propen sity score, and then tested the balancing propensity for each covariate, matching a total of 1,231 patients in the control group with 272 of the 1,077 Treatment Group patients. This is a relatively sm all number of matches because two variables the Apgar score and missed office visits had high levels of missing data I later repeated this analysis, removing these two covariates that had significant missing values While this adjustment significantly increase d the number of Treatment Group patients matched by propensity scores to include almost all 1,077 the quality of the matches was not significantly improved by increasing the number of matched patients. ( See, Attachment F) and I was unable to satisfy the balancing property as described below. Therefore, I opted to use the combination of 10 pre treatment covariates that satisfied Stata's balancing property though the number of matches was small

PAGE 105

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 105 The pscore command provided a single score for each patient to match the MLP treatment and comparison groups. The mean is highlighted, showing that of the control patient group, on average, the likelihood that any given individual in it would be in the treatment group was 17.24%, given the set of measured characteristics I used; this probability is similar to the estimated propensity scores found by Bingenheimer et al., whose study of firearm violence exposure and serious violent behavior informed my methodology. This graph summarizes the balance property achieved by this estimated propensity score, shown in the graph below where the y axis is proportional by group: The 10 covariates in the resulting model ensured that the propensity score balanced across the T reatment and control groups The balancing property means that the same propensity scores appear in th e matched control and treatment populations in similar proportions Stata d oes this by automatically splitting blocks and performing t tests until the program calculated the smallest number of blocks where the propensity score was equivalent across the tr eated and controls in each block:

PAGE 106

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 106 To interrogate the quality of the matched and unmatched populations, I compared the characteristics of treated and un treated groups after identifying the control group throough propensity scores. This analysis had seve ral limitations. First, because Stata does not identify the exact patients who have been matched as controls, some demographic information about this group was not available (N/A). Secon d, as noted earlier, some variables had substantial missing data. In T able 3 below, the extent to which missing values reduced the number of patients (n) available for comparison are noted parenthetically. The un matched Treatment and Control groups are demographically similar except importantly, the Treatment group has con siderably more low socioeconomic patients than the control group. Generally, the differences among the matched and unmatched participants in the Treatment Group were statistically signifi cant, while the differences between the matched and un matched contr ol group participants were typically neither substantial nor statistically significant.

PAGE 107

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 107 Table 3 Comparison Of Matched And Un Matched Group Of Treated And Control Patients Treatment Group Control Group Matched Unmatched p valu e Matched Unmatched p value Patients (n) 272 1,077 -1,231 6,655 -Race Black (%) White (%) Other (%) N/A 22.13% 27.67 % 44.29 % -N/A 28.04% 35.78 % 42.09 % -Ethnicity Hispanic (%) N/A 49.03 % -N/A 42.43 % -Low SES N/A 92.0 8 % (n=2,974) -N/A 80.61 % (n=568) -Pre_ip_total .17 .2 8 0.0 8 .24 .2 3 0.3 4 Pre_er_total .96 1.6 4 0.00 .92 .7 6 0.86 Pre_missed_total 2.05 2.4 0 (n=1 038) 0.00 2.30 1.10 0.46 Pre_office_total 6.01 5.97 (n=1,036) 0.00 6.81 7.3 3 (n=2,523) 0.22 Pre_rx_tot al 4.7 9 8.3 8 (n=1,038) 0.00 5.4 9 5.64 0. 50 First_age .1 4 .44 0.00 .11 .8 5 0.42 Apgar_1_min 6.8 3 7.60 (n=283) 0.00 7.35 6.82 (n=1,363) 0.0 3 Avg_temp 98 00 98.17 (n=1,038) 0.00 98.01 98.03 (n=6,618) 0. 80 Acsc_ever .1 2 .064 0.00 .064 .1 1 0. 10 Low_birthwe ight .09 .02 7 0.00 .0 6 .073 0.29 Standardized Mean Differences Analysis To Test The Qual ity of Propensity Score Matches Next, I tested the quality of the propensity score matches in the model by using a Standardized Mean Difference (SMD) analysis. For both the matched and un matched sample groups, I calculated the difference between (the mean of the treatment group minus the mean of all patients/ standard deviation of all patients) and (the mean of the non treatment group minus the mean of all patients / standard deviation of all patients). I did this for both the whole sample of 11,553 patients then for the matched samples in order to see how the propensity score technique reduces the Standardized Mean Differences bringing them closer to zero. This provided some evidence that the matched groups are more similar and thus statistically

PAGE 108

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 108 comparable than the unmatched groups Ideally, when comparing the covariates of standardized mean difference, the absolute magnitude should be no larger than .25 and l ess than .1 (Stuart, 2007) When I compared the Standardized Mean Differences shown in Table 4 below though not uniformly, on balance the propensity score technique reduced the standardized mean difference s showing that the matched group is more similar to the Treatment Group and provides a better statistical comparison than the unmatched group Table 4 Standardized Mean Differences: Matched And Un Matched Groups *Missing data treated as missing for 39 patients Sample Size A nd Power Estimate The literature concerning power estimates for propensity matching is not uniform or well developed. Therefore, I have used power estimates based on control and experimental group means estimated from my preliminary pilot data. My sampl e size and power estimates are based on = 0.05 and 2 sided t tests on between group differences. This study is powered at > .80 to detect an effect size of delta = 12.08% with 1,077 participants each in the control and comparison groups, which is suffic ient to reject a false null hypothesis. I treated missing data as missing, not imputed, and all statistical analyses were performed using Stata 14. The two outcome study variables had no missing values. However, as discussed further below, several of Unmatched Group Matched Group

PAGE 109

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 109 the pre treatment variables had a significant number of missing values. I discarded any variables for which there were more than 10% of the variables missing with the exception of two variables: number of missed office visits ( 36.3 % missing) and APGAR at 1 mi nute ( 78 % missing). Despite the high number of missing values, I retained this variable in the moel because counterintuitively, the matching algorithm worked better wi th this variable than without. Outcome Variables I tested my hypothesis by analyzing th e MLP 's impact on two outcome variables that are determinants of long term health for children over their life course. The first outcome variable is immunization compliance (compliance_rate) The second outcome variable is upward residential mobility (resi dential_mobility) The data for both measures were obtained from two sources. First, I collected immunization and residential data for each patient from the CHCO electronic health records. Next, in order to minimize missing data on the two study measure s for this project, I entered into Data Use Agreements with the Colorado Department of Public Health and Environment (CDPHE) to collect missing data on each patient for these two variables. The CDPHE returned immunization records to Children's Hospital fo r all patients who visited the CHCO clinics and in this manner, I obtained the most complete record available of immunizations recei ved by patients in this study. The same Agreement allowed CHCO to provide patient address data to CDPHE. In turn, CDPHE appl ied its geocoding software to standardize patient addresses and return these datasets to Children's Hospital for de identification and disclosure to me. I then received data on each patient that combined information from the CHCO ehrs and state's immunizat ion records from the Colorado Immunization Information System (CIIS). The next two sections describe each outcome variable in detail.

PAGE 110

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 110 Immunization Compliance Rate A young child's compliance with the recommended Immunization schedule is a well accepted he alth measure that also suggests the quality of patients' access to preventive care between ages 0 and 3. In order to complete the immunization doses in accordance with the ACIP recommended schedule by the third birthday, a child must regularly visit a hea lth care provider at ages 0, 1, 2, 4, 6, 9, 12, 15, and 18 months old and between 19 23 months. The immunization compliance outcome is measured by comparing the number of recommended immunization doses to the number of doses received during age appropriat e windows, before age 3. Adherence to this regular immunization schedule is a proxy in this study for regular access to preventive care. The ACIP recommends children between birth and age 3 receive vaccinations at specific ages and intervals in order to provide the greatest protection against disease and cost savings. Yet, the CDPHE reports that immunization rate for children between 19 35 months of age in Colorado, as well as in the United States continues to be considerably lower than the Healthy Peop le 2020 target. Figure 2 0 shows the immunization completion rates regardless of timeliness, in Colorado and in the United States for vaccines recommended for children before age 3 by the Advisory Committee on Immunization Practices (ACIP) of the American Academy of Pediatrics. The recommendation protects against 14 diseases and includes four doses of diphtheria, tetanus and acellular pertussis vaccine ( dtap ); three doses of inactivated poliovirus vaccine (IPV), one does of measles, mumps and rubella va ccine (MMR) ; three doses of Haemophilus influenza type b vaccine (Hib); three doses of hepatitis B vaccine ( hepb ); one dose of varicella vaccine; and four doses of pneumococcal conjugate vaccine (PCV) (i.e.

PAGE 111

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 111 4:3:1:3:3:1:4). In addition, as shown below, my study also considers the ACIP recommended receipt of two doses of rotavirus vaccine (RV) and two doses of hepatitis A vaccine ( hepa ) by three years of age. However, Figure 2 1 illustrates the significant gap between the target vaccination coverage and actu al coverage in Colorado and the nation. Figure 21 Colorado And USA Immunization Rates Rising But Remain Below Healthy People 2020 Objective I generated the outcome variable, compliance_rate to operationalize the first outcome me asure. The basic methodological approach to creating this outcome variable calculated the number of each patient's re ceived vaccination doses as a proportion of the recommended immunization doses according to the schedule published by the ACIP of the Ameri can Academy of Pediatrics (See Attachment E ). I reviewed the ACIP recommended schedule with a CHCO MLP pediatrician and identified a series of nine vaccinations recommended for all children, ages birth to 2.99 years old, from this schedule. I have exclude d vaccinations recommended only for certain high risk populations, such as extra doses for mumps, measles and rubella, as well as meningococcal and annual influenza doses. The immunization data in this study is not time varying. Originally I conducted t his analysis using a dichotomous outcome variable that

PAGE 112

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 112 measured immunization completion at age 3. However, after discussing my preliminary results with my committee, I changed my immunization outcome variable from a dichotomous variable to follow the appro ach taken by Kurosky, Davis, and Krishnarajah I generated three dichotomous variables for each patient for each recommended vaccination dose. I created a dichotomous variable for each immunization dose where the variable = 1 if the child had received the i mmunization dose within the recommended age window and equaled 0 otherwise. I created separate dichotomous variables to denote whether the patient received the immunization dose within the recommended age interval, before (pre ) or after (post ) enrollmen t into the MLP Thus, for the first dtap dose would be as follows: Dtap1 _compliant = 1 if (age at visit = (between 56 89 days)) AND pre_dtap_1 or post_dtap_1 == "Y" Dtap1 _pre_compliant = 1 if (ag e at visit = (between 56 and 89) And pre dtap _1 == "Y" oth erwise = 0 Dtap1 _pos t _compliant = 1 if (age at visit = (between 56 89) And post_dtap_1 =="Y" otherwise = 0 Table 5 shows age windows used to create each dichotomous variable, based on the age ranges for vaccine doses recommended for each immunization. I f a vaccine dose was administered during the recommended age interval, or within the 4 day grace period before the child reached the recommended age, I included that dose as received included as timely or "compliant However, vaccine doses administered be fore this grace period, or after the AICP recommended endpoint, were considered non compliant.

PAGE 113

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 113 Table 5 Recommended Age Ranges For Vaccine Dose Administration Vaccine Dose Recommended Age in Months Minimum Acceptable Age in Days fr om DOB Age in days for Age Appropriate" Vaccination Compliance Dtap 2 1 2 months 56 Days 56 89 Days 2 4 months 116 Days 116 149 Days 3 6 months 176 Days 176 209 Days 4 15 18 months 446 Days 446 539 Days IPV 1 2 months 56 Days 56 89 Days 2 4 months 116 Days 116 149 Days 3 6 18 months 176 Days 176 569 Days MMR 12 15 months 356 Days 356 479 Days Hib 1 2 months 56 Days 56 89 Days 2 4 months 116 Days 116 149 Days 3 12 18 months 361 Days 361 479 Days Hepb 1 Birth (0 3 days) 0 Days 0 3 Days 2 1 2 months 28 Days 28 89 Days 3 6 18 months 176 Days 176 569 Days Varicella 12 15 months 361 Days 361 479 Days Rotavirus 1 2 months 56 Days 56 89 Days 2 4 months 116 Days 116 149 Days 3 6 months 176 Days 176 209 Days PCV13 1 2 months 56 Days 56 89 Days 2 4 months 116 Days 116 149 Days 3 6 months 176 Days 176 209 Days 4 12 15 months 361 Days 361 479 Days Hepa 1 12 23 months separated by 6 months 361 Days 361 719 Days 2 536 Da ys 536 719 Days I then calculated a compliance rate outcome variable for each patient at each age. To accomplish this, I created a pre MLP post MLP and overall compliance_rate outcome variable by

PAGE 114

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 114 aggregating the positive compliant dichotomous variab les in the numerator, and aggregating the number of recommended doses at each age in the denominator I added the number of dichotomous dose variables equal to 1 at each visit and calculated the proportion of th at total when divided by the total number o f recommended age appropriate doses for the child's age at the time of each visit Thus, while I do not have a time varying, continuous compliance rate outcome variable, I do have a n ordinal pre post and overall compliance rate for each patient at eac h age interval, that equals the ratio of timely completed vaccine doses, divided by the number of ACIP recommended doses. Upward Residential Mobility The second outcome variable counts the number of residential address changes recorded for each child, w here the initial census tract where the child lived had a lower median family income than the median income in the new census tract to which the child moved. In short, the upward residential mobility variable was derived by counting the number of times pa tients moved to neighborhoods with higher median household incomes. This outcome variable reflects the literature that shows while frequent family moves overall can adversely influence a child's ability to enjoy the strong social networks that are a social determinant of good health, upwardly mobile moves to neighborhoods with more resources can positively affect a very young child's health by giving access to better schools, food, personal safety, and environment, all of which are important social determin ants of health (Berkman & Krishna, 2014). I generated the variable, residential_mobility to operationalize the upward residential mobility outcome variable I based this variable on the underlying theory and methodology used in Regina M. Bures' s study of Childhood Residential Stability and Health at Midlife. Dr. Bures defined residential stability as a dichotomous variable with a value of one whenever participants

PAGE 115

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 115 in her study reported changing neighborhoods not more than 2 times as a child. The variable value was zero if a participant experienced more than two moves in a lifetime and this was counted as a lack of residential stability. However, Bures' measure did not take into account that a move to better neighborhood could improve stability and socia l capital rather than signal instability. My residential variable addresses this oversight. First, it is ordinal rather than dichotomous. Second, it considers the median family income of neighbo rhoods to which patients move. My preliminary analysis did al so consider the impact an MLP had on the absolute number of residential moves a patient experienced during the study period, but I ultimately opted to measure upward mobility rather than residential instability for two reasons. First, the theoretical liter ature, such as Dr. Bures' study, describes the advantages of residential stability for children older than those in my study. For example Bures theorizes that w hen a child is old enough to attend school unlike the children in my study, frequent moves dis rupt established friendship and educational networks ; these factors are not yet relevant in the life of children under age 3. Second, residential instability can be affected by many confounding factors the medical legal partnership cannot influence. Genera l economic conditions, neighborhood violence, and gentrification are three examples. In contrast, upward residential mobility can be the result of improved income and increased social support, as the information in my dataset that allowed me to associate m edian household income levels with neighborhood changes allowed a rich use of my data that is consistent with the life course theory. The long term impact of improved neighborhood resources that might come from upward residential mobility has been shown to affect children when the moves occur even from birth and therefore could be important to the very young children in my study. Bures hypothesized the residential stability measure as a reflection of the opportunity to build positive relationships that are associated with improved

PAGE 116

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 116 physical and mental health. In contrast, because the children in my study are too young for moves to directly impact their social networks, I focus on the quality of residential changes rather than the frequency of moves a child's family makes by counting moves from a neighborhood where families have less income to a neighborhood where families generally have more income. I hypothesized that the MLP intervention might improve upward residential mobility by providing legal representa tion that improved family income stability through employment or benefit advocacy, decreased financial pressures, and reduced overall stress such as by resolving a custody or child support matter. I was unable to collect accurate data at the level of each individual home address change. Therefore, I collected data on each address change that resulted in a move to a different census tract, defined by the United States Census Bureau to be a geograph ic area roughly equivalent to a neighborhood and established by the Bureau of Census for analyzing populations. Census tracts generally encompass a population between 2,500 to 8,000 people and are described as "relatively permanent though they do change over time. Based on the 2010 census, the Census Bureau esta blished 1,249 census tracts in Colorado. I selected census tracts for analysis because they are small geographic units. Each census tract population ideally equals approximately 4,000 people and represents the most granular residential data available to me for the de identified group of patients in my study. Moreover, although census tracts vary widely, they were designed to be relatively homogeneous with respect to a population's economic status, which is a characteristic of interest for this study. Ea ch patient's address records have been standardized by the Colorado Department of Public Health and Environment (CDPHE) using geocoded software to minimize errors due to data entry or reporting variability such as street name spelling differences.

PAGE 117

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 117 Each pa tient's address was identified with a census tract for which I recorded the 2013 median family income (MFI). I selected the census tract variable, median family income ($MFI) to analyze the financial characteristics of patients' neighborhoods. I chose t his measure because I am interested in the economic status of young children in this study. The Census Bureau states a family "consists of two or more people Related by birth, marriage, or adoption residing in the same housing unit." Since by definit ion, there are no 1 person families (in contrast to $MHI which includes 1 person households), I determined the median family income measure best aligns with my expectation there are no children in this study living alone. Using the Census Bureau definition of $MFI best fits the expected living situation of my study participants. I then stratified all Colorado census tracts into six levels based on median family income. I divided Colorado census tracts into five quintiles using the following income catego ries: Low, Moderate, Middle, High, and Very High. I added a sixth category for homeless patients to capture the population for which CHCO enters the hospital's address for a residence. Table 6 below describes the $MFI tiers that I use d to determine wheth er a child's move is to a higher median income tier and therefore evinces upward social mobility. Table 6 Median Family Income Tiers For Patient Census Tracts Tier Median Family Income Range Very low Income (CHCO Address = Home less) $0 Low Income $1 to $35,500 Moderate Income $35,501 to $56,800 Middle Income $56,801 to $85,200 High Income $85,201 to $ 249,999 Very High Income $250,000 and above

PAGE 118

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 118 To generate the upward residential mobility outcome variable, I used a dichoto mous variable (v_neighbchange). The v_neighbchange variable equals "Y" each time the address a patient's caregiver reports at a CHCO clinic visit reveals a move to a different address from the one recorded from the immediately prior visit, and the new add ress is located in a different census tract than was reported at the patient's prior CHCO visit. I next used the string variable (v_changetype) which the CDPHE had set to equal "Increase," "Decrease," or "No Change" based on the comparison between the medi an family incomes in the prior and new address census tracts. I counted the number of neighborhood changes that were also increases and set this total equal to a new variable called residential_mobility. The resulting dependent outcome variable, residentia l_mobility is an ordinal variable, ranging as shown in Table 7 below from 0 to 12 upward moves for the children in my study. As this variable increases, it reflect a patient's improved potential access to improved neighborhood level conditions and resource s.

PAGE 119

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 119 Table 7 Upward Residential Mobility Outcome Variable By Patient Cohort Table 7 suggests that a higher percentage of patients who do not have legal needs ( MLP Negative) experience upward mobility than patients who d o have legal needs ( MLP Positive). Also, it appears that patients who enjoy a high number of upward moves are more frequently patients without legal needs. Quantitative Data Analysis The goal of my quantitative analysis was to analyze the data to determin e whether they allowed an inference as to the causal impact the MLP s may have on patient health. I used three methods. First, to determine whether access to MLP services is associated with better access to preventive health care and upward social mobilit y (Aim 1) I compared post treatment means between the Treatment Group of 1,077 patients and all 10,475 untreated patients Next, I estimated the average treatment effect of medical legal partnerships on the two outcome variables (Aim 2) comparing the tr eated and untreated groups matched by propensity scores I also used maximum likelihood linear and logistic regression models to analyze the association

PAGE 120

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 120 between the MLP and race, ethnicity, and socioeconomic status. Third I conducted a pre post analysi s, using survival rate models to determine whether introducing the MLP intervention improves outcomes among the treated patient group (Aim 3). This section summarizes each of these three quantitative methods. Comparisons between treatment and control grou ps I began my analysis with a basic comparison that quantified the size of the differences between treated and untreated patient groups. This is a crude measure of the MLP 's effect because it provides no way to ensure that the compared groups are reasonab ly similar so that differences can be attributed to the MLP 3 Still, this starting point allowed me to make basic observations about the impact that MLP s may have. I then refined this analysis by estimating the average treatment effect of the MLP comparin g the matched members of the Treatment Group with their counterparts selected from the unscreened patient group by propensity score matching Next, I used linear and regression analysis to analyze the association between the MLP and outcomes of interest. The purpose the regression analysis was to build a model to determine whether the relationship between race, ethnicity, SES and poor outcomes is moderated by the MLP intervention. I also analyzed whether the intensity of the MLP services has any impact o n the moderator effect of the MLP overall. The hypothesis I tested in this analysis was whether the MLP could disrupt the predicted relationship between poverty, minority status, and poor health and social outcomes. I theorized that poor health and social outcomes are mediated by inferior access to preventive care and the social determinants of health. (Figure 9 above). The MLP intervention might moderate this relationship by weakening the association between poverty, race, and unequal access to preventiv e care or improved neighborhood conditions. Figure 22 below shows the interactive terms I used to test the MLP 's impact.

PAGE 121

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 121 Figure 22 Testing The MLP As Moderator Of The Social Gradient Relationship Between Race Low SES And Infe rior Outcomes I theorized that being black, Hispanic, or poor is inversely related to immunization compliance rates This is because disadvantaged families have inferior access to health care due in part to the lack of insurance, the quality of health c are located in minority neighborhoods, and limited access to discretionary income and time needed to access health care I also hypothesized that race, ethnicity, and low SES is inversely related to upward residential mobility because disadvantaged famili es face multiple obstacle s to moving into better neighborhoods including a lack of financial capital, under or un employment, lower educational attainment and discrimination However, I predicted that the MLP c ould moderate these relationships, reducing or reversing the inverse relationships between race, ethnicity and low SES and the outcome variables by providing legal services that improved access to better housing, income supports, and lower stress Lastly, I expected to find that as the intensity o f the MLP services provided increased, the quality of legal outcomes might also improve. My outcome variables for the regression analysis were the ordinal variable, compliance rate, and a dichotomous residential mobility variable I created I generated a

PAGE 122

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 122 dichotomous variable called "hi_mobility" which equals 1 for patients who experienced a number of upward moves above the mean number of upward residential moves (residential_mobility) for the entire population and 0 for patients equal to or below th at va riable's mean. The residential mobility outcome is a count variable and not susceptible to linear analysis I also created interaction terms to measure the moderating impact of the MLP and thus test the increase or decrease in predicted probability of im munization compliance and residential mobility associated with an MLP intervention. Figure 23 summarizes the variables used in the regression model. D ependent Variables Compliance_rate Ordinal response variable for proportion of timely number of rec eived to recommended vaccination doses Hi_mobility Dichotomous response variable = for patients whose number of moves to higher $MFI neighbourhoods exceeds median of all upward residential_mobility moves MLP Dummy independent variable for screened posi tive for MLP treatment Race Dummy variable (1=white) Ethnicity Ethnic variable (1 = Hispanic, 0 = All other) Low _ses Dummy variable created combining public insurance, income dependent + uninsured = 1, otherwise 0 to serve as proxy for low SES Coh ort = MLP Treatment Group Indicator for the sample the individual is in: CLS = Legal File at CLS = "High intensity" MLP Service + MLS Pos = "Low Intensity MLP Service" Figure 23 Linear and Logistic Regression Variables Pre / Pos t Analysis Within Treatment Group I next used survival analysis to calculate hazard ratios that measured the risk of immunization compliance over three age group intervals for patients who received the MLP intervention during the study period. Survival an alysis is frequently used by medical clinicians to test the efficacy of a treatment regime until a final event -death -occurs. This analytical

PAGE 123

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 123 procedure allows the researcher to measure the rate of the event (i.e. The hazard rate) at any moment in tim e, and then compare s the cumulative probability of events over time among cohorts (i.e. the survival function) (Singe and Mukhopadhyay, 2011) Because my immunization compliance data from CHCO did not have sufficient time variability to calculate survival ratios for each patient individually, I calculate d survival ratios over three, 1 year intervals for patient age groups using patient ages as the time metric in this study. The study beginning date in 2009 was the "beginning time" in my analysis. I was a ble to determine the probability of immunization compliance in the treated (Treatment_Group) patient population and thereby measured the hazard rate for each treated patient's timely immunization the discrete event occurrence -and the corresponding sur vival function for all patients cumulatively before and after receiving the MLP intervention. Qualitative and Mixed Methods Outcome Analysis The qualitative analysis for this project began with interviews that informed my quantitative and qualitative resea rch design. These initial interviews with the physician, attorney, and paralegal who ran the CH CO medical legal partnership helped me to understand how the data for this project were generated and therefore defined my general research design. Therefore, d uring the first phase of study, I used the information from those interviews to frame questions about the measurable impact that MLP s may have on patients and their families. However, my data for one group of patients allowed me to go beyond these quantita tive measures to pursue a more in depth analysis of the work that MLP lawyers provided. For the 92 patients who became CLS clients, I used the attorneys' notes and documents generated by legal staff to further explore the impact MLP s might have.

PAGE 124

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 124 Patients in the CLS Case cohort were each assigned an individual lawyer to represent them on an ongoing basis to resolve a legal problem that affected their health D uring Phase II of this study I was able to examine the relationship of these patients' cases to on e another. That is, within the group of 92 CLS Case patients, I analyzed whether the type and intensity of MLP legal services is related to the quality of legal outcome of patients who got high intensity CLS services. (Aim 4). To accomplish this, I read an d coded the qualitative data from the CLS legal files. I condensed the qualitative data by "quantitizing" (Miles, Huberman & Saldana, 2014) the information revealed in the legal files. These data helped me better understand the legal problems, services, an d outcomes that the 92 patients who became CLS clients received. Moreover, I was able to explore whether the MLP legal services, problems, or outcomes that patients experienced w ere affected by their race or ethnicity. Figure 24 summarizes my approach to t he data from the CLS legal files. Methodological Step Process Objective Data Condensation Select data from legal documents to focus and simplify the full body of legal files Select the data that is analytically related to study theoretical framework and research questions Coding Assign codes to legal documents to elucidate themes, relationships, and patterns Develop basis for elaborating set of assertions, propositions, and generalizations discerned in data Display Data Organize a compressed assembly of information in the form of matrices and charts Provide understanding to allow and support drawing conclusions from data Draw and Verify Conclusions Interpret the patterns, causal flows, and relationships revealed in the data Identify responses to resear ch questions identified for the study Figure 24 Qualitative Data Analysis Approach Based On Miles Huberman Saldana (2014)

PAGE 125

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 125 In Phase III, I concluded the mixed method analysis of my study. I linked the qualitative and quantitat ive data together to determine whether the intensity of legal services that MLP s provide has an impact on either patients' housing mobility or immunization compliance (Aim 5). My objective was to take a closer look at the impact of the intensity and type of legal services that MLP s provide. In the quantitative analysis, I made a macro level distinction between hi intensity and low intensity MLP services based on a general assumption that becoming a CLS client meant closer attention to client problems than receiving less formal legal services from MLP staff. There, I made the assumption that CLS Case patients received more intense legal services than patients whose legal problems were addressed without opening a legal file to resolve their matters. In contr ast, during Phase III, I completed a mixed methods analysis that included a closer look at the specific type of legal service the hi intensity, CLS Case clients received. These patients from the hi intensity group received a range of services and provided a more granular look at the impact a lawyer might have on social and health outcomes. During this final phase, I divided the hi intensity legal services into four sub categories. When a patient received legal advice and recommendations on a course of acti on, they had the benefit of the lowest intensity services. If a lawyer wrote a letter such as a request to an employer or landlord to change a course of action, this represented the next level of intensity. When CLS filed a formal appeal, seeking to rever se a decision made that adversely affected a patient these services required legal research, writing, and persuasive efforts that represented the second highest level of intensity that I coded. Finally, CLS patients who were accompanied by an attorney in person to make a court or administrative appearance received the highest intensity services of this cohort. I based these categorizations on my interviews with the CHCO MLP attorneys and on my own experience as an attorney representing MLP clients in a w ide variety of

PAGE 126

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 126 cases. This final category of services requires factual investigation, legal research, preparation for oral arguments and presentations, witness preparation, and in person confrontation with adverse parties. Figure 25 summarizes these assu mptions about the intensity of CLS services received by the hi intensity cohort of patients in this study. Figure 25 Low To High Intensity Legal Services Within The CLS Case Cohort Of Treatment Group My research during this ph ase focused on the relationship between the intensity of CLS legal services and 1) race and ethnicity of patients, 2) legal outcomes, and 3) outcome variable s of interest (immunization compliance and upward residential mobility) for this study. To examine these relationships, I combined the quantitative outcomes for each patient cohort related to immunization compliance and residential mobility with the qualitative data related to the type of legal problems CLS addressed for those patients. I then compared the quantitative outcomes for patients who received CLS services with those who received other MLP services, dividing the former group by the legal problem they presented, the method CLS used to address that problem, and the legal outcome the patient's fam ily experienced. 92,)1%.")<'()&' O&,0($"2) !%$$%6'A2' /&:%61%'>(6$? /CC%(.'27' /&:%61%' 5%0"1"2) >%612)(.'92,6$' [A6"-,)(.' /CC%(6()0%'

PAGE 127

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 127 CHAPTER V RESULTS The Outcome Variables T he maximum compliance rate (compliance_rate) for patients in my study was 37.5 % as shown in Table 8 below This low compliance rate is likely due to a combination of the lower vaccination rates experienced in Colorado generally and among the vulnerable populations I studied, as well as the way I constructed this variable. I adhered precisely to the recommended age intervals in order to define compliance rates that excluded patient vaccinations that may have been achieved by getting immunization dose s close to, though not exactly within the recommended time or as part of a catch up schedule. Table 8 Immunization Dosage Compliance Rate Variable

PAGE 128

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 128 The number of upward re sidential moves for the categorical variable, residential_mobility, ranged from 0 to 12 moves Table 9 shows there were no missing variables and the overwhelming majority of patients in this study did not experience upward residential mobility. Earlier in the study, I also created and analyzed a residential_instability variable aggregating the total number of neutral and negative residential moves a family made but I opted not to use this as an outcome measure because it provided little improvement as a co ntinuous variable and did not fit my conceptual theory of disrupting cumulative disadvantage as well as the upward residential mobility outcome variable 4 Table 9 Upward Residential Mobility Outcome Variable Quantitative Result s My quantitative analysis included four steps. First, I used a simple means comparison to compare outcomes for the treated MLP group to the unscreened patients under age 3. Next, I used propensity scoring to create a matched control group and then usin g Stata treatment effects

PAGE 129

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 129 software, estimate d the causal relationship that may be inferred between receipt of MLP services and improved patient health and social outcomes. Using this matched comparison group, I estimated the average treatment effect of the MLP treatment within the treatment group (ATE) and the average treatment effect that would have been observed had the entire patient population been treated (ATT) Third, using linear and logistic regression, I analyz e d the moderator effect of the MLP in tervention on the association between race, low SES and poor health outcomes. Finally, I performed a pre and post analysis to evaluate the MLP impact on two outcome variables among treated patients. The next section presents the results from each of thes e four steps. Outcome Means Comparison The simplest estimate of the MLP 's effect on outcomes is shown in Table 10 which compares the mean outcome s in the control group with the average outcome in the MLP treated group.

PAGE 130

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 130 Table 10 Mean Comparison, Vaccine Dose Compliance Vaccine Dose Recommended Age in Months Minimum Acceptable Age in Days from DOB[i] Age in days for Age Appropriate" Vaccination Compliance[ii] Mean Patients Compliant MLP Control Differenc e P value Dtap [iii] n =1,038 n =10,475 1 2 months 56 Days 56 89 Days 0.0019268 0.0191885 0.0172617 0.0001 2 4 months 116 Days 116 149 Days 0 0.011074 0.011074 0.0007 3 6 months 176 Days 176 209 Days 0 0.0042959 0.0042959 0.0344 4 15 18 months 446 Days 446 539 Days 0.0028902 0.0085919 0.0057017 0.0503 IPV n =1,038 n =10,475 1 2 months 56 Days 56 89 Days 0.2957611 0.2900239 0.0057372 0.70 2 4 months 116 Days 116 149 Days 0.3131021 0.2888783 0.0242238 0.101 3 6 18 mo nths 176 Days 176 569 Days 0.6213873 0.6062053 0.015182 0.34 MMR n =1,038 n =10,475 12 15 months 356 Days 356 479 Days 0.0028902 0.0168974 0.0140072 0.0005 Hib n =1,038 n =10,475 1 2 months 56 Days 56 89 Days 0.0019268 0.0189021 0.0 169753 0.0001 2 4 months 116 Days 116 149 Days 0 0.0107876 0.0107876 0.0008 3 12 18 months 361 Days 361 479 Days 0.0028902 0.0168974 0.0140072 0.0005 Hepb n =1,038 n =10,475 1 Birth (0 3 days) 0 Days 0 3 Days 0.0028902 0 0 2 1 2 mo nths 28 Days 28 89 Days 0.0096339 0.0258711 0.0162372 0.0012

PAGE 131

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 131 3 6 18 months 176 Days 176 569 Days 0.0279383 0.0832458 0.0553075 0.0000 Varicella n =1,038 n =10,475 12 15 months 361 Days 361 479 Days 0.0028902 0.0143198 0.0114296 0.0022 Rot avirus n =1,038 n =10,475 1 2 months 56 Days 56 89 Days 0.0019268 0.0170883 0.0151615 0.0002 2 4 months 116 Days 116 149 Days 0 0.0074463 0.0074463 0.0053 3 6 months 176 Days 176 209 Days 0.0120706 0.0154639 0.0033933 0.3853 PCV13 n =1,077 n =10,476 1 2 months 56 Days 56 89 Days 0.0019268 0.0189021 0.0169753 0.0001 2 4 months 116 Days 116 149 Days 0 0.0105967 0.0105967 0.0009 3 6 months 176 Days 176 209 Days 0 0.0041046 0.0041046 0.0352 4 12 15 months 361 Days 361 479 Days 0.0028902 0.0124105 0.0095203 0.0062 Hepa n =1,077 n =10,476 1 12 23 months separated by 6 months 361 Days 361 719 Days 0 0.014223 0.014223 0.0001 2 536 Days 536 719 Days Th is crude means comparison showed h o w similar c ompliance rates are in the MLP treatment and control groups with very small difference between the two groups Nevertheless, w ith very few exceptions, the differences between the two groups were statistically significant a s shown by the small p values T he MLP group generally had slightly lower compliance rates than the control group, contradicting my hypothesis. Missing data for the 39 patients was not imputed Table 11 Mean Comparison, Upward Residential Mobility MLP n =1 ,077 Control n =10,476 Difference P value Residential _mobility .3500464 .3513746 .0013281 0. 9698

PAGE 132

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 132 Residential mobility appears slightly higher in the control group than in the MLP treatment group ; however the means comparison for this variable is not sta tistically significant. Estimated Average Treatment Effect of MLP I used two different approaches to estimate the ave rage treatment effect. First, I used Stata's "teffects" command to calculate the Average Treatment Effect (ATE) defined as the effect of t he treatment on the entire population observed using Stata 14's propensity matching command This command produced the output shown below for the outcome variable compliance_rate, showing that the average treatment effect (ATE) of the MLP was to slightly i mprove immunization compliance rates by .27% This outcome is different from the crude means comparison shown in Table 10 above because the ATE was calculated using the propensity score matched populations. These were more similar comparisons shown in the Stata output below, and therefore produced a more sensitive result. In contrast, the ATE of the MLP on the residential_mobility variable was disappointing The average treatment effect of the MLP was to slight ly decrease the number of upward residential mobility moves by .1262 moves :

PAGE 133

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 133 Second, I used Stata's pscore command, to calculate the average treatment effect on the treated (ATT) in four analytical steps. T his calculation differs from the ATE calculation above in that t he ATE estimates the MLP s treatment effect had the entire population been observed under the MLP treatment versus had the entire population been observed without the MLP treatment. In contrast, the ATT calculation estimates the treatment effect of the MLP using only the treated p opulation participants who matched in the propensity scoring process rather than the entire population. ( Becker and Ichino, 2002 ) I estimated the average treatment effect (ATT) of the MLP intervention on compliance_rate using the Nearest Neighbor matchin g command. The final results are presented below. They show that the 272 matched patients from the Treatment Group who received the MLP intervention, experienced a very small (.004, t=0.842) improvement in their mean compliance rate over 2 26 matched pati ents Stata selected from the control group.

PAGE 134

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 134 I also estimated the average treatment effect (ATT) of the MLP intervention on residential_mobility using the Nearest Neighbor matching command. The final results are presented below. They show that the gr oup of 272 matched patients from the Treatment Group who received the MLP intervention experience .081 fewer upward mobility moves during the study period than the matching 226 control group patients that Stata selected using the propensity score

PAGE 135

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 135 Fig ure 2 6 below summarizes the results from both approaches to estimating the a verage treatment effect of the MLP intervention on the two outcome variables of interest :

PAGE 136

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 136 Number Treatment Group Patients Number Control Group Patients Treatment Effect Estima te Std. Error p value Vaccination Compliance Rate ATT 272 226 0.004 0.005 0.842 ATE 1,0 77 1,577 0.002695 .0059148 0.649 Residential Mobility ATT 272 226 0.081 0.091 0.888 ATE 1,0 77 1,577 .126189 .0841597 0.134 Figure 26 S ummary of Average Treatment Effect Results (ATT and ATE) on Outcome Variables Linear a nd Logistic Regression Analysis I built a regression model, beginning with covariates to analyze the association between the poverty, race, and the outcome variables I then successively added the MLP treatment variable, as well as interaction terms in order to analyze the effect of MLP intervention. I evaluated the results of this model by comparing the coefficients for each variable and interaction term to determine wh ether the MLP moderator variable alters the strength of the fundamental relationship between low income, race, ethnicity, and poor health and social outcomes. I conducted linear regression to analyze the MLP 's association with the continuous compliance_rat e variable, and logistic regression to analyze the MLP 's association with a dichotomous variable generated from the categorical residential_mobility variable. Immunization Compliance Rate Linear Regression Results Model # 1: First, I regressed compliance_ra te on race variables to show the relationship between outcome variable compliance_rate and race (using white as the reference group) and between the outcome and Hispanic ethnicity. I analyzed race and ethnicity separately in case the

PAGE 137

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 137 data did not capture i ndividuals who identified as both being of Hispanic ethnicity and a member of a specific racial category : These results confirm the social gradient I hypothesized, showing that being black d ecreases compliance rate by .04 percentage points. Similarly b eing Hispanic was associated with a small ( .023 percentage points) and statistically significant increase in immunization rate compliance as shown below: Model #2 This model adjusted for low socioeconomic status and demonstrated that low SES had a sta tistically significant association with the outcome variable. Adjusting for low SES reduce d the extent to which being a member of another racial group is associated with lower immunization compliance.

PAGE 138

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 138 Moreover, adjusting for low SES only slightly reduced the inverse association between Hispanic ethnicity and immunization compliance. Model #3 : I next added the MLP variable to the model and found that the MLP had no statistically significant, direct rate association with compliance rates and no impact o n the inverse relationship between being black and immunization compliance

PAGE 139

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 139 The results were similar when I added the MLP variable to the model for Hispanic patients. Model #4: The next model added interaction terms to analyze the effect that MLP s migh t have on associations between race, ethnicity, and the outcome variable. The interaction term created by the MLP and black predictors did not change the strength or direction of the association between being black and the outcome Neither did the interact ion term between the MLP and Hispanic ethnicity predictors change the strength or direction of the association between Hispanic ethnic ity and immunization compliance

PAGE 140

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 140 Model #5 : This model a dds an interaction term between the MLP and low SES to analyze whether MLP s moderate the relationship between the low socioeconomic status and poor immunization compliance The results below show that neither the relationships between race or ethnicity and immunization outcomes are affected significantly by adding the interaction terms between the low_ses predictor and the MLP variable.

PAGE 141

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 141 Residential Mobility Logistic Regression Results As discussed in the previous chapter, I analyzed the association of race, ethnicity, and socioeconomic predictors and residential m obility using a dichotomous outcome variable, hi _mobility Figure 2 7 below summarize s output from the logistic regression models run using this outcome variable. As expected, t his analysis showed that patients who had the MLP intervention had 1.0 7 times gr eater odds of experiencing high upward residential mobility However, this association was not statistically significant.

PAGE 142

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 142 Figure 27 Odds Ratio for High Upward Residential Mobility For MLP Patients I successively added race, His panic ethnicity, and low SES to the model, as shown in Figure 29 below, and found that the MLP intervention continued to have a small, statistically significant association with the high upward residential mobility variable. Variable Model 1 Model 2 Model 3 Model 4 Model 5 Z stat Z stat Z stat Z stat Z stat Constant .15 3 51.94 1.12 2.29 .9 2 0.87 0.92 0.85 0 91 0.85 MLP 1.0 7 0. 74 1. 66 7.25 2.30 10.49 2.31 5.54 2.37 3.09 Black Other_Race Hispanic 1.02 1.08 1. 18 0.33 1.42 3.15 0. 91 1.03 1. 05 0.89 0.32 0.55 .93 1.02 1.05 0.66 0.22 0.55 0.92 1.0 2 1.04 0.67 0. 22 0.55 Low SES 1. 10 1.00 1. 10 0. 99 1.12 0. 9 6 Black* MLP Other_Race MLP 0.96 1. 02 0.22 0.10 0.96 1.02 0.21 0.11 Low SES* MLP 0.97 0. 10 P value 0. 46 0.00 0.0 0 0.0 0 0. 0 0

PAGE 143

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 143 Figure 28 Logistic Regression Results For High Residential Mobility Outcome Variable Finally, I did a "sub analysis" to examine whether the difference in the intensity of MLP intervention services has any influence on the ass ociations between independent and dependent variables observed in the previous analysis. The high intensity variable described CLS Case patients (n=92) that received legal representation by becoming formal legal aid clients of Colorado Legal Services attor neys in contrast to MLP Positive patients (n=985) who received legal services through a more informal (i.e. Low intensity) MLP relationship Figure 29 shows t hat variable has a statistically significant but negative i mpact on immunization compliance rates Patients who screened positively for MLP services and then received higher intensity MLP services were less vaccine compliant than patients in the treatment group who received low intensity MLP services by a factor of .034. Figure 29 Linear Regression Results For Association Between Compliance Rate And High Intensity MLP Services Similarly, Figure 3 0 shows that the odds of upward residential mobility for patients in the treatment group who received high rather than low intensi ty MLP services were lower; they were only 81% as likely as low intensity patients to be upwardly mobile.

PAGE 144

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 144 Figure 30 Logistic Regression Results For Association Between High Upward Residential Mobility And High Intensity MLP Serv ices Pre /Post MLP Analysis Results I used the person period immunization level dataset to conduct a pre post study analyzing the immunization compliance rates before and after the MLP intervention. The variables for this analysis were reported by CHCO ; we constructed the visit level dataset so that f or each patient, for each of the 24 recommended vaccination doses for children up to age 3 CHCO reported a dichotomous pre MLP and post MLP variable where the value "Y" indicated the dose was given timely eith er before or after the MLP intervention, and "N" indicated the patient did not receive a timely dose. My dataset also includes each child's age in days from the date of birth. Therefore, for each patient, I separately aggregated the sum of pre MLP doses gi ven timely, and post MLP doses given timely, and then divided each sum by the total number of recommended doses appropriate for the child's age, in order to calculate overall pre MLP and post MLP compliance rate s for each patient. I then tabulated the age of patients and the aggregate compliance rate s to get the frequency for each age group completing their immunization before and after the intervention These frequencies are shown in Figure 31

PAGE 145

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 145 Figure 31 Frequency Of Pre And Post MLP Immunization Compliance For Treated Patients Notably, immunization compliance rates among the youngest children in the study (before age 2) are higher before MLP treatment than after MLP intervention. This was not an expected o u tcome. The re sults shown in Table 12 are the means of pre and post MLP compliance rates, and are based on the fact that treated children do not all enter the dataset at birth but rather enter at the first age that they became CHOC patients.

PAGE 146

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 146 Table 12 Means of Pre and Post MLP Compliance Rates by Age Groups I created a variable to aggregate the total upward residential moves before and after MLP interventions for each patient. Table 1 3 displays the probability of pre MLP residential mob ility for each age group in the study, while Table 1 4 shows the frequency with which children in each age group experienced upward residential mobility after the MLP intervention.

PAGE 147

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 147 Table 13 Frequency of Pre MLP Residential M obility for Treated Patients by Age Group Table 14 Frequency Of Post MLP Residential Mobility For Treated Patients By Age Group To examine these findings further, I combined all age groups together and compared the frequency of upward mobility moves within the treated population, before (pre) and after (post) MLP treatment. The results are shown in Figure 3 2 below and confirm that as overall instability increased, the MLP had less influence on residential mobility patterns. In deed, the least stable

PAGE 148

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 148 patients those who moved eight or more times during the study period were unaffected by the MLP intervention with respect to this outcome variable. Figure 32 Comparison Of Pre And Post MLP Treatment Fr equency Of Upward Residential Mobility Moves Finally, I also calculated the hazard rate (probability) of pre MLP and post MLP compliance and upward residential mobility for each of three age groups of treated patients using the Stata Life Table function The results from this analysis is summarized in the life tables below which show that the probability of immunization compliance increases significantly for the youngest children in the study after the MLP but this advantage disappeared quickly as childr en age. Table 1 5 shows that f or children in their first year of life, the probability of full immunization compliance is 44.5%. It declines to 39% and 33% in the second and third year of life respectively before the MLP After the MLP this table suggests a 62% chance of compliance in the first year, dropping precipitously to 3.5% in the second year and zero in the third year. H\ UH\ WH\ ZH\ XH\ FHH\ H F U P W X FU Y6%+,%)0?' M,#-%6'NC](6&'K2-".."$?'K2:%1 >6% T >21$'K!>'NC](6&'@%1"&%)$"(.'K2-"."$? DA6%($#%)$'B62,C')EFGHIIJ >6%^K!>'A6%($#%)$ >21$^K!>'A6%($#%)$

PAGE 149

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 149 Table 15 Life Table: Pre and Post MLP Immunization Compliance Hazard Rate for Treated Patients The p re and post MLP result for residential mobility are disappointing. Table 16 shows that while there is 9% probability that a child in this study could experience upward residential mobility in the first year of life before the MLP intervention, the probabi lity of experiencing upward residential mobility after the MLP intervention in the first year was less than 3% and declined each year thereafter.

PAGE 150

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 150 Table 16 Life Table : Pre And Post MLP Upward Residential Mobility Hazard R ate For Treated Patients Qualitative And Mixed Methods Results The qualitative analysis for this study was conducted by entering coded data into an excel spreadsheet and uploading it into a Dedoose analytical platform. I analyzed the gender, race, and ethnicity of CLS Case patients. In aggregating the data for this analysis, I found that one of the 92 patients who received CLS Legal services w as actually the same patient for whom more than one legal file was opened to resolve different legal matters. T hese results, as shown in Figure 3 3 below, demonstrate that the majority of patient families in this study who received legal services from Colorado Legal Services wer e Hispanic, Black, and female.

PAGE 151

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 151 Figure 33 Race Ethnicity And Gender Of 92 CLS Case Patients ( B = Black H = Hispanic W / H = White / Hispanic O = Other O / H = Other / Hispanic W = White A = Asian ) As the national literature reflects, Figure 3 4 shows the medical legal partnership patients studied here most often receive assi stance with legal problems related to public benefits, such as health insurance, that have adversely affected their income. The next largest category of legal problems this group experienced were housing problems, which is also consistent with the nationa l data. H S FH FS UH US PH PS WH ;.(0_ Q3"$% 8"1CL K"`%& =$3%6 /1"() @(0%'()&'O$3)"0"$?'27'9!4'9(1%'>($"%)$1 H FH UH PH WH SH ZH K Y B%)&%6'27'9!4'9(1%'>($"%)$1' B%)&%6'27'9!4'9(1%a Legend M Male F Female

PAGE 152

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 152 Figure 34 Types Of Legal Problems 92 CLS Case Patients ( MED = Medical Insurance INC = Income / Public Benefits FAM = Family Law HSE = Housing Law ) For the group of 92 CLS Case patients who became legal aid clients, the inte nsity of available legal services ranged from low intensity counseling and education, to high intensity personal court appearances as shown in Figure 3 4 above The legal services provided by the MLP in this study were largely lower intensity services, conc entrated in providing counseling, education, and letters written to potentially adverse parties on behalf of MLP clients. However, the data also show that the MLP attorneys won many more of the cases that they undertook for representation than they lost. S till, the category "O" that contains other outcomes that could not be categorized as either wins or losses is the largest category of legal outcomes. Of this group, 13 were lost to follow up and outcomes could not be determined; 13 had problems for which the under resourced CLS staff did not have an available attorney to address; 1 client refused legal services or terminated the representation after the file had been opened; and the remaining clients' cases H S FH FS UH US PH KO5'*M4N@/M9O *M9=KO Y/K*!b KO5'*M4N@[*M9 8=N4*MB KO5'*M4N@Ga KO5'*M4N@Ga *M9G'8=N4*MB Y/K*!bG'KO5a KO5'*M4N@Ga 9!4'9(1%'>($"%)$1c'A?C%1'27'!%<(.'>62-.%#1

PAGE 153

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 153 were closed after clients received legal consulta tion. Figure 3 5 summarizes the quantitative outcome showing the types of legal services and outcomes CLS Case patients experienced. Figure 35 Types Of Legal Services And Legal Outcomes For 92 CLS Patients ( C / E = Counseling And Education ; LTR = Letter On Client's Behalf ; APP = Appeal Of Adverse Decision ; PER = Personal Court / Administrative Appearance ) In order to combine this information from the qualitative data with the data from the qualitative analysis performed, I analyzed the rel ationship between race and the intensity of legal services provided, as well as legal outcomes as shown in the graphs that comprise Figure 3 6 below. Legen d C/E Counseling and Education LTR Letter to Adverse Party APP Appeal Adverse Decision PER Personal Appearance in Court

PAGE 154

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 154 Figure 36 Intensity of Legal Services and Outcomes Among 32 CLS Case Patients By Race and Ethnicity Interestingly, a larger sample might reveal whether it is possible that as the intensity of the legal service increased, the number of minority patients receiving the service decreased. In a sample this size, I could not determine an y statistically significant pattern, though it warrants further investigation to determine whether any biases might be revealed by the fact that among these patients, the majority of patients who received the lowest intensity counseling and education servi ces were black and Hispanic, while no patients in these minority groups received the highest intensity service Legend C/E Counseling and Education LTR Letter to Adverse Party APP Appeal Adverse Decision PER Personal Appearance i n Court

PAGE 155

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 155 Figure 37 Legal Outcomes by Race/Ethnicity for 92 CLS Case Patients The data did not provide sufficient number o f lost cases to determine whether there was a relationship between the intensity of legal services provided and case outcomes ; however, Figure 38 below presents a summary of the relationship between the two outcomes of interest for this study and the high est intensity of MLP legal services provided to clients in the CLS C ase patient group. The graph is designed to examine the relationship between the best outcomes (Hi Compliance and High Upward Residential Mobility), and the level of MLP service intensity. I sought to examine whether receiving higher intensity services such as personal legal representation in a court appearance (PER) might correlate with better outcomes. Unfortunately, the number of cases in this dataset was too small to find a statistica lly significant relationship. H U W Z X FH FU FW FZ FX UH Q*M =A8O@ !=44 ;!/9d 8*4>L K*eO5 =A8O@ =[8 Q8*AO /4*/M Legal Outcomes by Race and Ethnicity

PAGE 156

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 156 Figure 38 Immunization Compliance ( Red ) And Upward Residential Mobility ( Blue ) By MLP Service Intensity For 92 CLS Case Patients Although the distribution of these services is shown above, is inter esting, the fact that low intensity services produced the highest number of favorable outcomes, suggests further study is warranted, but does not provide evidence of any conclusive correlation between MLP service intensity and Hi immunization compliance or upward residential mobility rates. W hen I analyzed the Chi 2 for each outcome, given = .05, with 92 degrees of freedom, the critical 2 va i ue = CHIINV (0.95,92). Thus, I found that neither the relationship between MLP service intensity and Hi Resident ial Mobility (Chi 2 = 5.12), n or Hi Compliance (Chi 2 = 7.68) proved to be statistically insignificant. Yet, this graph shows that for the "Series 1" Hi Residential Mobility outcome and "Series 2" Hi Compliance outcome, most clients received the less inte nsive MLP services in order to achieve positive legal outcomes. H FH UH PH WH SH ZH IH XH VH FHH />> 9[O !A@ >O@ B6()&'A2$(. 8* T @%1"&%)$"(.'K2-"."$?'D;.,%J'()&'8" T 92#C."()0%' D@%&J';?'K!>'4%6:"0%'*)$%)1"$? 9!4'9(1%'>($"%)$1G')EVU LTR Letter to Adverse Party PER Personal Appearance in Court

PAGE 157

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 157 CHAPTER VI DISCUSSION This study sought to determine whether medical legal partnerships can improve two social determinants of health that contribute to health disparities. The theoretical starting point for my inquiry was the ecosocial model tha t posits that structural inequality expresses itself at an individual level through poor health a nd social outcomes. In many ways, this theoretical framework provides the conclusion for my study as we ll. The findings I report, confirm, perhaps more than I expected, that profound poverty and inequality, are tenaciously resistant to anything less than profoundly systemic change. T he intervention studied here seeks to hold le gally enabled institutions accountable for the inequality that produces health dispa rities. Howev e r, my stud y results reflect that the effort to use legal agency must be broad, if it is to reduce health disparities on a population level. The discussion that follows reports the results I obtained, using methods c ommonly employed to estimate causal effects when randomized experiments are not possible. However, the important message fr om this chapter goes beyond a discussion of the the results and outcomes I measure d. Indeed, the implications for future study point to the importance of developing further research methods that w ill not underestimate the persistent harms and harmfulness of health inequality Discussion Beginning with the comparison of post outcome means between the treated group of patients who received the MLP interventions, and a control group that did not, the data presented in Table s 10 and 11 above confirm neither my original hypothesis that an inverse relationship between the MLP intervention and immunization compliance exists, nor my hypothesis that the

PAGE 158

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 158 MLP might improve upward residential mobility. For most of the 24 vaccination doses recommended for children before age 3, immunization compliance among the MLP patients was slightly lower than the immunization compliance among unscreen ed, non treated patients. Only compliance means for the three recommended polio vaccine doses (IPV) were slightly greater among MLP patients than among control group patients, but those results were not statistically significant. The result of my means c omparison for upward residential mobility was similarly non confirming. I conclude that the general control group that I used for this outcome means comparison was not sufficiently similar to the treatment result to yield meaningful results. In contrast, when I constructed a matched control group of patients using the propensity score method, I saw some evidence that suggests that further research may reveal that the MLP intervention can slightly improve immunization compliance if not upward residential mo bility. The second of the two approaches to propensity score matches used here gave a clearer picture of the potential impact that the MLP model may have. However, both methods yielded similar results. The MLP treatment reduced the number of upward mobi lity moves that patients experienced by a factor of .123, p =0.134, 95% CI [ .291, .039]. Though this result is inconsistent with my hypothesis, it is also statistically insignificant. The average treatment effect analysis showed that the MLP improved i mmunization compliance ever so slightly by .2% over non treated groups, p =.649, 95% CI [ .009, .014]; this improvement appeared in the youngest age groups of children in the study and dropped off sharply as children grew older but again I am unable to re ject the null hypothesis. The average treatment effect of the intervention on upward residential mobility among treated participants, produced similar results, where the ATT showed a slightly less negative impact on residential mobility ( 0.081) and a smal ler positive influence on immunization compliance (0.004). These results, presented in Figure 26, do not

PAGE 159

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 159 support my hypothesis about these variables but do have relatively high p values in both instances and therefore more study is warranted before my hyp otheses can be accepted or rejected. The best explanation for the failure to find a statistically significant effect in this study may be that the outcome variables chosen for this study are not sensitive enough to the effects of the MLP intervention to reveal any association. In other words, the MLP intervention did not have a strong enough influence on the measures I studied to show improved outcomes. Other variables may be more directly impacted by the legal intervention. To ex amine this possibility I estimated the average treatment effects on outcome variables that were not the subject of this study and found encouraging results Using the ATE analysis approach, I found the MLP 's estimated treatment effect was statistically significant in reducing i npatient hospitalizations and missed office visits as shown in Table 17 below Table 17 Statistically Significant Estimated Average Treatment Effects Of MLP On Inpatient Hospitalizations And Total Missed Appointments

PAGE 160

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 160 These ou tcomes are promising because they suggest that further study using the methodological approach developed here is warranted to determine the impact that medical legal partnerships can have on health outcomes among disadvantaged patient populations. Moreover they confirm that the theoretical framework for this study is sound. The linear and logistic regression analys e s allowed me to further explore these results. In the case of both outcome variables, the initial regression model s showed the MLP improved th e odds of immunization compliance and upward residential mobility. Adding terms for race and ethnicity in the second model, as would be expected, only slightly reduced the impact that MLP had on the odds of improved immunization compliance ( # =1.36) but vir tually eliminated the impact the MLP had on upward residential mobility ( # =1.04). However, addition of the interaction terms that included race and socioeconomic status suggested the MLP is the only statistically significant variable that impacts the outco me variables. Moreover, there is evidence from the sub analysis to suggest that the intensity of the MLP intervention can have a statistically significant impact on some outcomes. In this study, the hi intensity MLP services had a counter intuitive negativ e effect on immunization compliance and upward residential mobility The best explanation for these results, and indeed for the failure to confirm my hypotheses overall, is that patients who need MLP services are very likely to have worse health indicators and more severe social deficiencies than patients who d o not need or need only low

PAGE 161

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 161 intensity MLP services. These patients do not compare favorably because the cir c umstances and conditions that lead them to require legal services have already negatively im pacted their health and social outcomes even before the MLP interventions. If this is true, then it is no wonder that their outcomes do not more favorably compare with patients who do not require substantial MLP intervention. I did not account for the sign ificant difference that may be signaled by the fact that MLP positive patients were much more likely to be the lowest SES families in the participant group. Perhaps the most informative results in Phase I of my study came from the pre /post analysis I did on the group of treated patients. However, these results failed again to confirm my initial hypotheses. In the case of immunization compliance, patients in the CLS Case subgroup of the study population exhibited better compliance rates before the MLP in tervention than they did after the treatment. These patients also showed higher rates of upward residential mobility before they enrolled in MLP s than after. However, in examining the post treatment residential mobility data further, I found that a plausi ble explanation for the post treatment decline in upward mobility was that after receiving MLP services, children and their families had less need to change their residential locations and instead were able to remain in relatively stable living conditions for longer. Said another way, while post MLP upward mobility did not improve, post MLP residential stability did improve. I saw evidence of this when I examined all neighborhood changes for the study population and found there were very few address and n eighborhood changes of any type after the MLP intervention. This could explain why upward residential mobility post MLP was flat. However, I cannot ignore the possibility that the results from this pre post study simply show that the MLP does not increase access improved upward mobility as hypothesized.

PAGE 162

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 162 A different explanation may apply to the comparison between pre and post MLP immunization compliance. First it is notable that immunization compliance is greatest between birth and two years of age, but t hen drops significantly during the child's third year of life. However, I interpret the overall inconsistency of my results with the original hypothesis of my study as evidence that the MLP population is generally a more at risk population than the genera l patient group served at Children's Hospital. The outcome means comparison results would certainly suggest this. The MLP group almost uniformly under performed the control group of all other unscreened patients with respect to immunization compliance an d upward residential mobility. When my analysis matched the treated patients with a more similarly situated control group, the differences between their outcomes was smaller. However, in the final analysis, when considering only the treated patients in t he pre / post studies, the quantitative data could not confirm that the MLP intervention was a significant factor in improving either of the outcomes of interest studied here. The qualitative data helped confirm that the patient population that receive d the most intensive MLP assistance were characterized by great need. For example, the average income of the CLS Case clients equaled $1,172 per month. This, of course, is due to the selection criteria CLS uses to serve the poorest patients in Colorado. Unsurprisingly, these patients are mostly black or Hispanic, and the care givers seeking MLP assistance are mostly women. A smaller proportion (11.96% and 44.5% respectively) of this patient group experienced high upward residential mobility and high compl iance as compared to the proportion of patients in the unscreened group (13.28% and 50.10% respectively). The overwhelming majority of legal problems these MLP patients had were related to their need for income supports.

PAGE 163

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 163 My results point to important theo retical strengths and weaknesses in my study and its design. First the lack of change noted in outcome variables despite the MLP intervention could be interpreted to confirm the fundamental cause theory that supports this study. As that theory predicts, th e extremely low socioeconomic status of many of the patients treated means depite that the MLP intervention, disparities between groups that have resources to avail themselves of the intervention's benefits will improve faster than those groups who lack re sources. This may explain why the very neediest patients who populated the MLP treatment group, did not experience significant improvement in the outcomes measured here. Said another way, Conditions of extreme poverty continued to overwhelm the strength of the MLP intervention, demonstrating that socio economic status remains a fundamental cause of poor health outcomes. Further, the continued poor outcomes for the very young children in this study are likely, in accordance with the life course perspective to continue to experience cumulative adversity over the lifecourse without a stronger intervention than the medical legal partnership. Similarly, consistent with ecosocial theory, the adversity faced by the patients in the MLP treatment group may have bi ologically expressed significant social adversity that was not addressed by the MLP representation that was studied here. For example, the interventions provided by lawyers studied here did not include criminal justice representation. Thefore clients who l ive in concentrated poverty, and who therefore are exposed to high levels of violence, did not receive MLP legal interventions that could have mitigated the physical harms that result from social risk factors that remained unaddressed despite the MLP inter vention. The third theory that I explored in this study the life course theory was not a good match for a study conducted over the five year length of m y study period. Five years is simply not sufficient time to uncover changes over the lifecourse th at would be revealed over a longer period of time, with more

PAGE 164

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 164 robust longitudinal data. Still the overarching social epidemiological framework that informs this study is suitable to consider the distribution of social determinants among vulnerable families ; the medical legal intervention remains a tool for increasing both community resources and people's demands for those resources. A final theoretical observation relates to the selection of outcome variables. The upward mobility measure proved too attenua ted as an outcome that could be influenced by legal representation. This reflects the fact that variables not measured here social capital, social networks, for example play a role in changing health outcomes that a targeted intervention cannot singula rly influence. Limitations and Implications For Future Study While it is not possible to infer that the MLP intervention improves access to preventive care or upward residential mobility from this study's results, future study of this hypothesis is warr anted using the study design presented here. Certainly, one of the primary limitations of these results is the extent to which missing data limited the robustness of the propensity score matching method of constructing a control group for comparison. Futur e studies can improve on these results by filling in the dataset used here from the CHCO electronic health records and other sources. A second major limitation of the study is the indirect impact that MLP services may have on the two outcome variables of i nterest. However, while immunization compliance and upward residential mobility may not be sensitive to the improvements that MLP s make in patients' lives, the evidence suggests that MLP s can impact more direct measures of access to health care such as inp atient hospitalizations and emergency department utilization. These outcome variables offer important and exciting opportunities for future study or whether and how MLP s might reduce health disparities among vulnerable families

PAGE 165

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 165 One of my most important f indings from this analysis may be related to the process information I gathered Importantly, I learned that law firms currently have no systematic approach to record keeping thus making data analysis quite challenging. For example, although the files for each patient contained a standard set of documents for all clients, the 92 files did not contain the same level of detail in the documents for each patient. Nor did they reflect the same approach to simple procedures such as opening and closing legal matte rs for each patient. This made the task of "cross walking" the legal files to match the identification numbers in the CHCO datasets particularly challenging and represents a limitation of my study. Moreover, t he legal files contain no information about the patients beyond the codes checked boxes, and sparse demographic information that a rushed and over worked poverty lawyer might quickly collect. The legal files do not contain statements about family relationships, details about financial burdens or obli gations faced, or even evidence of changed circumstances from one legal visit or conversation to the next. They are devoid of information about families' feelings, personal experiences or perceptions. In fact, the legal files which are now largely treated as record keeping for the management of law firm business matters (fund raising, billing, headcount), could be greatly improved if used as a tool to assess patients entire psycho social context. The inequitable social stratification among the attorneys a nd patients in this study is captured in this finding. Despite their good intentions, the power differential between lawyer and patient reinforced social stratifications among this population Lawyers could learn to treat the "whole client," in the way doct ors are urged to treat the "whole patient." In this way, the legal community could learn better record keeping techniques that are not time consuming. For example, the "SOAP" note taking system commonly used by physicians could work well to expand the info rmation (and therefore qualitative data) that legal

PAGE 166

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 166 files contain. In addition to the objective assessment of patients' legal needs, a legal file might also include a lawyer's subjective observations and plan for resolving patients' legal dilemma. Instead, the files I examined had terse references giving details only to account for the legal services rendered. Attorney notes were limited to "Reinstated Medicaid," "Terminated Food Stamps Restored or "filed request for reinstatement SSI benefits wi thheld (sic.) Incorrectly alleged incarceration." It is not difficult to imagine how relieved families were to receive CLS services to improve their precarious financial conditions. In several other cases, the notations about consultation included Advice re foreclosure" or "counselled regarding custody matter." These are the kind of details attorneys need for billing purposes and so the files tend to consistently contain this information. Another important insight from the qualitative study is confirmat ion that Denver, Colorado's low income patients currently suffer a significant lack of access to justice. T he CLS legal files studied here contain ample clues to the fact that the CLS staff felt stretched to the breaking. All too often files contained refe rences to legal problems that were within the scope of CLS' s practice parameters, but followed by a notation that says "no attorneys available to assist." I counted ten such cases (over 10% of the total), including one in which the client sought help gett ing a landlord to exterminate a bed bug infestation. These files show that even clients who passed all the requisite screens to get help from lawyers were still unable to access legal help to improve their social determinants of health because the institut ions that are available to help this population are under funded A final observation is evident from the qualitative data. A number of legal matters repeat often enough within this small sample of 92 cases to suggest a structural or systemic legal proble m may be present that is not being handled effectively by addressing as individual

PAGE 167

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 167 problems for a small sample of patients who happened to end up getting legal attention through the CHCO MLP Eleven of 92 cases involved agencies wrongly reducing or withdra wing food stamp benefits from families. In all of these cases, CLS was able to reinstate benefits. This seems an indication that there is a systemic problem that affects families with young children surrounding the policies and implementation of food stamp programs. Similarly, substandard housing conditions frequently mold, bugs, and utility problems dominated the housing issues patients presented. These clearly health harming violations of building code and warranties of habitability might better be ap proached as a policy problem that affects population health, rather than on an individual case by case basis. In short, these qualitative data suggest that the MLP model might be improved to give greater legal attention to structural interventions, rather than individual legal cases that are representative of a larger problem affecting public health outcomes. The data that compared MLP servi ce intensity with outcomes yielded surprising results. Most importantly, the sub analysis of hi intensity MLP services and the study outcome s were inversely related. This meant that the most intense level services were associated with lower immunization compliance rates than patients who received less intensive services. I attribute this result to the likelihood that the patient population that needed the most intensive legal interventions were also patients whose lives were least stable and therefore less likely to regularly access preventive health care. The MLP intervention does not appear robust enough to change the di rection of the relationship between housing stability overall and poverty. The patients who received MLP services in this study were those with the highest need, and the most serious problems; a single intervention could not change residential stability i n a statistically significant way.

PAGE 168

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 168 As explained earlier, the CLS cohort of patients, who are characterized in this study as having received "high intensity MLP services" by virtue of having a CLS lawyer assigned to them. The other MLP patients who did not become CLS clients also received legal services, but they were deemed "low intensity" for this study's purposes. Within the high intensity patient group, the majority of CLS clients received the lowest intensity legal services that CLS offers counseling and education. Most of the patients who received these services were black or Hispanic. But I could not tell from my data whether race and ethnicity made a difference with respect to the type of legal services CLS provided or the type of legal problems th ey presented. Indeed, the relationship between the intensity of legal services and overall outcomes was not statistically significant. The number of patients who received the most intensive personal representation and appeals was small. Yet, even though th e majority of CLS Case patients received only the lowest intensity MLP legal services, the majority of them experienced "win" outcomes, showing that a little legal help went a long way with this group. Not only did the CLS interventions more often yield wi nning than losing results, but the largest share of the CLS Case group's hi upward mobility and hi compliance patients were those who received the lowest intensity counseling and education services. Although my study hypotheses were largely unconfirmed by my analysis, the methodological a pproach developed in this study, and the evidence that other health outcome measures than those studied here may be significantly improved by the MLP intervention, does provide a promising foundation for further research an d has contributed to the understanding of the way that MLP services may influence patients' social risk factors and potentially their vulnerability to health disparities.

PAGE 169

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 169

PAGE 170

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 170 REFERENCES Aber, J. L., Bennet, N.G., Conley, D.C., & Li, J. (1997). "The Ef fects of Poverty on Child Health and Development." Annual Rev. Public Health 18: 463 484. Acevedo Garcia, D. (2000). Residential segregation and the epidemiology of infectious diseases. Social Science & Medicine, 51(8): 1143 1161. Acevedo Garcia, D. (200 1). Zip code level risk factors for tuberculosis: neighborhood environment and residential segregation in New Jersey, 1985 1992. American Journal of Public Health, 91(5): 734 741. Adelman, L. (2007). Unnatural causes: Is inequality making us sick?. Preven ting Chronic Disease 4 (4). Adler, N. E., & Newman, K. (2002). "Socioeconomic Disparities in Health: Pathways and Policies." Health Affairs 21(2): 60 76. Adult and Child Consortium for Health Outcomes Research and Delivery Science (ACCORDS), Vaccine Pre ventable Diseases, (2016) available here: http://www.ucdenver.edu/academics/colleges/medicalsc hool/programs/ACCORDS/childr ensoutcomesreserach/research/Pages/vaccinepreventablediseases.aspx Aneshensel, C. S. & Sucoff, C. A. (1996). The neighborhood context of adolescent mental health. Journal of Health and Social Behavior, 37(4): 293 310. Atkins, D (2014). "Medical legal Partnership and healthy start: Integrating civil legal aid services into public health advocacy." Journal of Legal Medicine 35(1): 37 413. Aungst, R. B. (2011). Healthy people 2020. Perspectives on Audiology 7 (1), 29 33. Bach, P. B., Pham, H. H., Schrag, D., Tate, R. C., & Hargraves, J. L. (2004). Primary care

PAGE 171

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 171 physicians who treat blacks and whites. New England Journal of Medicine 351 (6), 575 584. Barker, D. J. (1998). In utero programming of chronic disease. Clinical science 95 (2), 115 128. Baron, R.M. and Kenny, D.A. (1986). The Moderator Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations, Journal of Personality and Social Psychology, 51(6):1173 1182. Bartle y, M., & Plewis, I. (2007). "Increasing Social Mobility: An effective policy to reduce health inequalities." Journal of the Royal Statistical Society 170(2): 469 481. Bashir, S. A. (2002). Home Is Where the Harm Is: Inadequate Housing as a Public Health C risis. American Journal of Public Health, 92(5): 733 738. Beck, A. F., Huang, B., Chandur, R., & Kahn, R.S. (2014). "Housing Code Violations Density Associated with Emergency Department and Hospital Use by Children with Asthma." Health Affairs 33(11): 19 93 2002. Beck, A. F., Klein, M.D., Schaffzin, J.K., Tallent, V., Gillam, M., & Kahn, R.S. (2012). "Identifying and treating a substandard housing cluster using a medical legal partnership." Pediatrics 130(5): 831 838. Becker, S. O., & Ichino, A. (2002). Es timation of average treatment effects based on propensity scores. The States Journal, 2(4), 358 377. Barbeau, E. M., Krieger, N., & Soobader, M. J. (2004). Working class matters: socioeconomic disadvantage, race/ethnicity, gender, and smoking in NHIS 200 0. American journal of public health 94 (2), 269 278. Beeson, T., et al. (February 2013). Making the case for medical legal partnerships: A review of T he evidence, National Center for Medical Legal Partnership.

PAGE 172

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 172 Ben Shlomo, Y., & Kug, D. (2002). "A life c ourse approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives." International Journal of Epidemiology 31: 285 293. Berkman, L. F., & Kawachi, I. (2000). Chapter 3: A historical framework for soc ial epidemiology. Social Epidemiology. New York, Oxford University Press. Berkman, L.F. and Krishna, A., 2014. "Social Network Epidemiology," Chapter 7 in Social Epidemiology, (L.F. Berkman, I.Kawachi, and M. Maria Glymour, ed: 2nd edition, Oxford Univer sity Press: New York). Bingenheimer, J. B., Brennan, R.T., & Earles, F.J. (2005). Firearm violence exposure a nd serious violent behavior. Science, 308(1323 1326). Doi :10.1126/science.1109730 Blane, D. (2006). The life course, the social gradient, and heal th. In Social Determinants of Health. (M. Marmot and R. G. Wilkinson, eds.) Blane, D., Smith, G. D., & Hart, C. (1999). Some social and physical correlates of intergenerational social mobility: Evidence from the West of Scotland Collaborative Study. Soc iology 33 (1), 169 183. Boyle, P.J., Norman, P., Popham, F. (2009). Social Mobility: Evidence that it can Widen Health Inequalities. Social Science & Medicines 68:1835 1842. Braveman, P. A., Cubbin, C., Egerter, S., Williams, D. R., & Pamuk, E. (2010). So cioeconomic Disparities in Health in the United States: What the Patterns Tell Us. American Journal of Public Health,100(S1). Doi :10.2105/ajph.2009.166082 Bradley, E.H. et al. (2016). Variation in Health Outcomes: The Role of Spending on Social Services, Public Health, and Health Care, 2000 09. Health Affairs, 35(5): 760 768. Briggs, S. J. & Keimig, K. A. (2016). The Impact of Police Deployment on Racial Disparities in

PAGE 173

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 173 Discretionary Searches. Race and Justice, doi: 10. 1177/2153368716646163 Brookhart, M. A., et. Al (2006). Variable selection for propensity score models. American Journal of Epidemiology, 163(12), 1149 1156. Brooks Gunn, J. & Duncan, G.J. (1997). The Effects of Poverty on Children. Children and Poverty, 7(2): 55 71. Budrys, Grace," Soci al Inequality" in Unequal Health: How Inequality Contributes to Health or Illness (Rowan & Littlefield: Lanham, 3rd Ed., 2017). Bures, R. M. (2003). "Childhood residential stability and health at midlife." American Journal of Public Health 93(7): 1144 11 48 Burris, S. (2001). "Law in a social determinants strategy: A public health law research perspective." Public Health Reports 126 / Supplement 3: 22 27. Burris, S., Ashe, M., Levin, D. Et al., (2016). "A Transdisciplinary Approach to Public Health Law: Emerging Practice of Legal Epidemiology," Annual. Rev. Public Health 37:135 48. Burris, S., Kawachi, I., & Sarat, A. (2002). "Integrating Law and social epidemiology." The Journal of Law, Medicine & Ethics 30(4): 510 518. Cain, L. D. (1964). Life course and social structure. Handbook of modern sociology 272 309. Centers for Disease Control and Prevention (CDC) (2014). "Benefits from Immunization During the vaccines for Children Program Era United States, 1994 2013" (Morbidity and Mortality Weekly Rep ort, April 25, 2014). Retrieved from, https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6316a4.htm ). Chetty, R. And N. Hendren (May 2015). "The Impacts of Neighborhoods on Intergenerational

PAGE 174

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 174 Mobility: Childhood Exposure Effects and County Level Estimates," Harva rd University mimeo. (Retrieved from http://scholar.harvard.edu/files/hendren/files/nbhds_paper.pdf) Chetty, R., Hendren, N. And L.F. Katz, (2016). "The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment," American Economic Review 106(4): 855 902 (Retrieved from http://scholar.harvard .edu/files/lkatz/files/chk_aer_mto_0416.pdf) Cohen, E., Fullerton, D. F., Retkin, R., Weintraub, D., Tames, P., Brandfield, J., & Sandel, M. (2010). Medical legal partnership: collaborating with lawyers to identify and address health disparities. Journ al of general internal medicine 25 (2), 136 139. Colorado Commission on AffordableHealth Care, (2016). Retrieved from: https://www.colorado.gov/pacific/sites/ default/files/Cost%20Commission%20November %202017%20report%20v3%20draft_1.pdf Colorado Children's Campaign, "Colorado is 20th in Nation for Well Being of Kids," (June 21, 2016). Retrieved from http://www.aecf.org/m/databook/2016KC_newsrelease_CO.pdf Colorado Children's Campaign (2013). Kids Count in Colorado 2013! Denver, Colorado, Colorado Children's Campaign. Colorado Health Foundatio n (2015). The 2015 Colorado Health Report C ard. Avaialbe at: www.ColoradoHealth.org/ReportCard. Colorado Health Institute, (May 2005). Childhood Immunization Rates: Policy and Practice, CHI White Paper. Retrieved from https://www.coloradohealthinstitute.org/sites/default/files/file_attachments/Childhood_I mmunization_Full_Report.pdf Colorado Supreme Court Library, "Colorado Equal Access Center: Connecting Unreprese nted

PAGE 175

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 175 Litigants to Legal Resources through Technology," (2016). Retrieved from http://cscl.colibraries.org/colorad o equal access center connecting unrepresented litigants legal resources through technology ( These ratios are calculated using 2015 information. As of July 1, 2015, Colorado's population was approximately 5.5 million. The number of active Colorado law yers as of 2016 was 25,801 according to the Colorado Office of Attorney Regulation ) Conover, C. J. (2002). "The impact of ancillary services on primary care use and outcomes for HIV/AIDS patients with public insurance coverage." AIDS Care 14(sup001): 59 71. Creswell, J. W. (2014). Research Design qualitative, quantitative, and mixed method a pproaches (4th ed. Ed .). Thousand Oaks, California: Sage Publication. Cutts, D. B., Meyers, A. F., Black, M. M., Casey, P. H., Chilton, M., Cook, J. T., ... & Rose Jacobs, R. (2011). US housing insecurity and the health of very young children. American Journal of Public Health 101 (8), 1508 1514. Davey Smith, G. (1998). Adverse socioeconomic conditions in childhood and cause Specific adult mortality prospective ob servational study. Health Inequalities Lifecourse Approaches. G.Davey Smith. Desmond, M. & Bell, M. (2015). Housing, Poverty, and the Law. Annual Review of Law and Social Science, (11): 15 35. Drukker, M., Kaplan, C., Feron, F., & Van Os, J. (2003). Chi ldren's health related quality of life, neighbourhood socio economic deprivation and social capital. A contextual analysis. Social science & medicine 57 (5), 825 841. Diez Roux, A. V. Et al. (2001). Neighborhood of Residence and Incidence of Coronary Hea rt Disease. New England Journal of Medicine, 345: 99 106.

PAGE 176

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 176 Elder, J. G. H., & Johnson, M.K. (2003). The life course and aging: Challenges, lessons, and new directions. Invitation to the life course: Toward new understanding of later life: 49 81. Elder, J. G. H. (1994). "Time, human agency, and social change: perspective on the life course." Social Psychology Quarterly 57(1): 4 15. Engler, R. (2009). Connecting Self Representation to Civil Gideon: What Existing Data Reveal About When Counsel is Most Neede d. Fordham Urban Law Journal, (37). Retrieved from http://ir.lawnet.fordham.edu/ulj/vol37/iss1/2/ Fang, J. Et al. (1998). Residential segregation and mortality in New York City. Social Science & Medicine, 47(4): 469 476. Felitti, V. J., & Anda, R.F. (1997). The Adverse Childhood Experience (ACE) Study, Centers for Disease Control and Prevention. Fleishman, S. B., Retkin, R., Brandfield, J., & Braun, V. (2006). "The attorney as the newest mem ber of the cancer treatment team." Journal of Clinical Oncology 24(13): 2123 2126. Forchuk, C., Dickins, K. & Corring, D. J. (2016). Social Determinants of Health: Housing and Income. Healthcare Quarterly, (18)SE: 27 31. Garg, A., Marino, M., Vikant, A. & Solomon, B. (2012). Addressing Families' Unmet Social Needs within Pediatric Primary Care: the health leads model. Clinical Pediatrics. Garrido, M., Kelley, A., Parris, J., Roza, K., Meier, D., Morrison, R., and Aldridge, M. (2014). Methods for construc ting and assessing propensity scores. Health Services Research 49:5:1701 1720. Geronimus, A. T. (2006). ""Weathering" and Age Patterns of Allostastic Load Scores Among Blacks and Whites in the United States." American Journal of Public Health 96(5): 826 8 33.

PAGE 177

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 177 Gostin, L. O. (2008). Conceptual foundations of public health law. Public health law: Power, duty, restraint, University of California Press: 3. Gottlieb, L., Hessler, D., Long, D., Amaya, A., & Adler, N. (2014). A randomized trial on screening for s ocial determinants of health: the iScreen study. Pediatrics 134 (6), e1611 e1618. Graham, S. (2008). Income, Education and Obesity: A Closer Look at Inequities In Colorado's Obesity Problem, Colorado Health Foundation Supplement to Colorado Health Repor t Card. Greene, S.S. (2016). Race, Class, and Access to Civil Justice. Iowa Law Review, (101):1263 1321. Haas, S. A., Krueger, P. M., & Rohlfsen, L. (2012). Race/ethnic and nativity disparities in later life physical performance: the role of health and s ocioeconomic status over the life course. Journals of Gerontology Series B: Psychological Sciences and Social Sciences 67 (2), 238 248. Hatch, S. I. (2005). "Conceptualizing and identifying cumulative adversity and protective resources: Implications for understanding health inequalities." Journal of Gerontology 60B (Special Issue): 130 134. Heiman, H. J. & Artiga, S. (2015). Beyond Health Care: The Role of Social Determinants in Promoting Health and Health Equity [Issue Brief]. Kaiser Family Foundation. Retrieved from http://kff.org/disparities policy/issue brief/beyond health care the role of social d eterminants in promoting health and health equity/ Hepburn, John .R. And Albonetti, C.A., (1004). Recidivism Among Drug Offenders: A Survival

PAGE 178

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 178 Analysis of the Effects of Offender Characteristics, Type of Offense, and Two Types of Intervention, Journal of Quantitative Criminology 10(2):159 179. Hern‡ndez, D. (2016). Extra oomph:' addressing housing disparities through Medical Legal Partnership interventions. Housing Studies, 31(7): 871 890. Higgins, G. E., Ricketts, M.L., Griffith, J.D., Jirard, S.A. (20 13). Race and Juvenile Incarceration: A propensity score matching examination. American Journal of Criminal Justice, 38, 1 12. Institute of Medicine Committee on Public Health Strategies to Improve Health (2011). For the Public's Health: Revitalizing l aw and policy to meet new challenges. Washington, D.C., Institute of Medicine. Jacobellis, J., Latimer, G., River, M.M., Gabella, B., Juhl, A., mcfarlane II, A.E., & Kaplan, G. (2008). The Colorado asthma surveillance report 2008. Denver, Colorado, Colora do Department of Public Health and Environment Prevention Services Division. Jakobsen, Tor G. (2012) "Survival Analysis The Forgotten Tool of the Social Scientist," Popular Social Science (August 10, 2012). Retrieved from: www.popularsocialscience.com/2012/08/10/survival analyis the forgotten tool of the social scientist 2/ Johnson, M.P., et al, (2002). The Benefits and Costs of Residential Mobility Programmes for the Poor, Housing Studies 17(1):125 138. Hertzman, C. (1999). The biological embedding of early experience and its effects on health in adulthood. Annals of the New York Academy of Sciences 896 (1), 85 95. Keller, D., Jones, N., J. Savageau, & S. Cashman. (2008). Development of a Brief Questionnaire to Identify Families in Need of Legal Advocacy to Improve Child Health.

PAGE 179

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 179 Ambulatory Pediatrics, 8(4), 266 269. Krieger, J. & Higgins, D. L. (2002). Housing and Health: Tim e Again for Public Health Action. American Journal of Public Health, 92(5): 758 768. Krieger, N. (2001). "Theories for Social Epidemiology in the 21st Century: An Ecosocial Perspective." International Journal of Epidemiology 30: 668 677. Leech, N. L., & Onwuegbuzie, A. J. (2009). A typology of mixed methods research designs. Quality & quantity 43 (2), 265 275. Legal Services Corporation. (2009). Documenting the Justice Gap in America: The Current Unmet Civil Legal Needs of Low Income Americans. Retriev ed from http://www.americanbar.org/content/dam/aba/migrated/marketresearch/publicdocuments/ justicegainamerica2009.aut hcheckdam.pdf Kurosky, S. K., Davis, K.L., & Krishnarajah, G., (2016). "Completion and Compliance of Childhood Vaccinations in the United States." Vaccine 34(3): 387 394. Kushel, M. B. (2006). "Housing instability and food insecurity as barriers to healt h care among low income Americans." Journal of General Internal Medicine 21(1): 71 77. Laveist T. (2003). Racial Segregation and Longevity among African Americans: An Individual Level Analysis. Health Services Research 38(6 Pt 2): 1719 1734. Laveist T. (1993). Segregation, Poverty, and Empowerment: Health Consequences for African Americans. The Milbank Quarterly, 71(1), 41 64. Legal Services Corporation, 2017 Justice Gap Report : Measuring the Unmet Civil Legal Needs of Low income Americans (June 2017) (Prepared by NORC at the University of Chicago for legal services Corporation, Washington, D.C.) Retrieved from https://www.lsc.gov/sites/default/files/imag es/thejusticegap executivesummary.pdf

PAGE 180

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 180 Link, B. G., & Phelan, J. (1995). "Social Conditions as fundamental causes of disease." Journal of Health and Social Behavior 35: 80 94. Ludwig, J., Duncan, G.J., Gennetian, L.A., and L.F. Katz, et al (2013). "Long Term Neighborhood Effects on Low Income Families: Evidence from Moving to Opportunity." American Economic Review Papers and Proceedings, 103(3): 226 31. Luftey, K., & Freese, J. (2005). "Toward some fundamentals of fundamental causality: Socioeconomic st atus and Health in the routine clinic visit for diabetes." American Journal of Sociology 110: 1326 1372. Ma, C. T., Gee, L., & Kushel, M. B. (2008). Associations between housing instability and food insecurity with health care access in low income childr en. Ambulatory Pediatrics 8 (1), 50 57. Marmot, M. (2005). "Social Determinants of Health Inequalities." The Lancet 365: 1099 1104. Marmot, M. (2004). The status syndrome: How social standing affects out health and longevity. New York, New York, Owl Book s. Marmot, M. G., Stansfeld, S., Patel, C., North, F., Head, J., White, I., ... & Smith, G. D. (1991). Health inequalities among British civil servants: the Whitehall II study. The Lancet 337 (8754), 1387 1393. Marmot, M. G. ; Rose, G.; Shipley, M.; Hamilton, P. J. (1978). "Employment Grade and Coronary Heart Disease in British Civil Servant. Journal of Epidemiology and Community Health. 32 (4): 244 249. Marmot, M. & Brunner, E. (2005). Cohort Profile: The Whitehall II Study. International Journal of Epidemiology, 34(2): 251 256. Marshall, V. W., & Mueller, M.M. (2003). Theoretical roots of the life course perspective.

PAGE 181

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 181 Social dynamics of the life course: Transitions, institu tions, interrelations. W. R. H. V. W. Marshall. New York, Walter de Gruyter, Inc.: 3 18. Matthew, D. B. (2010). The Social Psychology of Limiting Healthcare Benefits for Undocumented Immigrants Moving Beyond Race, Class, and Nativism. Hous. J. Health L. & Pol'y 10 201. Mccaffrey D.F., et al., (2013). A Tutorial on Propensity Score Estimation for Multiple Treatments Using Generalized Boosted Models, Statistical Medicine 32(19):3388 3414. Mcgough J. And Faraone, S.V., (2009). Estimating the Size of Treat ment Effects Moving Beyond P Values. Psychiatry 10:21 29l. Mcgovern L., Miller, G. & Hughes Cromwick, P. (2014). The Relative Contribution of Multiple Determinants to Health Outcomes [Policy Brief]. Health Affairs. Retrieved from http://healthaffairs.org/healthpolicybriefs/brief_pdfs/healthpolicybrief_123.pdf Mckinlay J. B. M., S.M. (1977). "The questionable contribution of medical measures to the declin e of mortality in the united states in the twentieth century." The Milbank Memorial Fund Quarterly Health and Society 55(3): 405 428. Merton, R. K. (1968). "The matthew effect in science." Science 159(3801): 56 63. Minnesota State Bar Association, Legal Assistance to the Disadvantaged Committee. "Overcoming Barriers that Prevent Low Income Persons from Resolving Civil Legal Problems," (September, 2011) (available here: http://www.mncourts.gov/Documents/0/Public/administration/Final_MN CABS_Study_September_2011.pdf ) MLP "I HELP." Retrieved from http://medical legalpartnership.org/ MLP response/how civil legal aid helps health care address sdoh/

PAGE 182

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 182 National Center for Access to Justice. (2016). The Justice Index, 2016. Fordham Law School. Retrieved from http://justiceindex.org/2016 findings/attorney access/ National Center for Medical Legal Partnership, Retrieved from http://medical legalpartnership.org/ MLP response/how civil legal aid helps health care address sdoh/ National Center for Medical Legal Partnership, "How MLP Works: A Utilities Case Study." Retrieved from http://medical legalpartnership.org/ MLP response/how medical legal partnership works/ National Center for Medical Legal Partnership, "The MLP Response." Retrieved from http://medical legalpartnership.org/ MLP response/ National Center for Medical Legal Partnership, "The Need for Medical Legal Partnership." Retrieved from: http://medical legalpartnership.org/need/ National Low Income Housing Coalition (2016). Medical Legal Partnership Interventions Improve Access to Safe, Decent, Affordable Housing. Retrieved from: http://nlihc.org/article/medical legal partnership interventions improve access safe decent affordable housing NC MLP Performance Measures Handbook (April 2016). Retrieved fro m http://medical legalpartnership.org/wp content/uploads/2015/08/ MLP Performance Measures Handbook Apr 2016.pdf Newman, J. (2012) "Miami's medical legal partnership: Preparing Lawyers and physicians for holistic practice." Indiana Health Law Review 9: 471. Office of Health Disparities. (2009). Racial and ethnic health disparities in Colorado 2009. Denver, Colorado, Colorado Depar tment of Public Health and Environment. Office of Health Equity. (2013). Health disparities the 2013 report: Exploring health equity in

PAGE 183

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 183 Colorado's 10 winnable battles. Colorado Department of Public Health and the Environment. Denver, Colorado. Onwuegbuz ie, A. J., & Leech, N.L. (2007). On becoming a pragmatic researcher: The importance of combining quantitative and qualitative research methodologies. International Journal of Social Research Methodology, 8(5), 357 387. Doi :10.1080/13645570500402447 Ortiz, P.C. (2014). How a Civil Right to Counsel Can Help Dismantle Concentrated Poverty in America's Inner Cities. Stanford Law & Policy Review, 25(1): 163 191. O'Sullivan, M. M., Brandfield, J., Hoskote, S.S., Segal, S.N., Chug, L., Modrykamien, A., & Eden, E. (2012). "Environmental improvements brought by the legal interventions in the homes of poorly controlled inner city adultashthmatic Patients: A proof of concept study." Journal of Asthma 49(9): 911 917. O'Toole, J. K. Et al (2012). "Resident Confidence addressing social history: Is it influenced by availability of social and legal resources." Clinical Pediatrics 51(7): 625 631. Pampel, F. C., Krueger, P. M., & Denney, J. T. (2010). Socioeconomic disparities in health behaviors. Annual review of sociolo gy 36 349 370. Page Reeves, J., Kauvman, W., Bleeker, M., et al. (2016). Addressing Social Determinants of Health in a Clinical Setting: The wellrx Pilot in Albequerque, New Mexico. Journal of the American Board of Family Medicine, 29(3), 414 418. Pe ttignano, R., Caley, S.B., & Bliss, L.R. (2011). "Medical legal partnership: Impact on patients with sickle cell disease." Pediatrics 128(6): e1482. Pettignano, R., Caley, S.B. & mclaren S. (2012). "The health law partnership: Adding a lawyer

PAGE 184

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 184 to the hea lth care team reduces system costs and improves provider satisfaction." Journal of Public Health Management Practice 18(4): E1 E3. Phelan, J. C., & Link, B.G. (2015). "Is Racism a Fundamental Cause of Inequalities in Health?" Annual Rev. Sociology 41: 311 330. Regenstein, M., Trott, J., Williamson, A., & Theiss, J. (2018). Addressing Social Determinants Of Health Through Medical Legal Partnerships. Health Affairs 37 (3), 378 385. Regenstein, M. (2014). In Medical Legal Partnership Programs. Measures Devel opment Background Memorandum. Washington, D.C., National Cener for Medical Legal Partnership. Rhode, D., (2004). Access to Justice: Connecting Principles to Practice, Georgetown Journal of Legal Ethics 17:359. Richards, E.G. Mapping Time, (1998, Oxford U niversity Press). Riley, M. W. (1973). Aging and cohort succession: Interpretations and misinterpretations. Public Opinion Quarterly 37 (1), 35 49. Robert Wood Johnson Foundation. (2011). Health care's blind side the overlooked connection between socia l needs and good health. Princeton, New Jersey, Robert Wood Johnson Foundation. Robles, Yesinia. "Fewer Colorado Kids Living in Poverty, But More Reported Homeless," The Denver Post (April 14, 2016). Retrieved from http://www.denverpost.com/2016/03/27/fewer colorado kids living in poverty but more identified as homeless/ Rodabaugh, J. Et al (2010). "A medical legal partnership as a component of a palliative care model." Journal of Palliative Medicine 13(1): 15 18. Rosenbaum, P. R., & Rubin, D.B. (1983). The central role of the propensity score in

PAGE 185

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 185 observational studies for causal effects. Biometrika, 70(1), 41 55. Ross, C. E., & Wu, C. (1996). "Education, age, and the cumulative advantage in health." Journal of Health and Social Behavior 37(1): 104 120. Ross, H. Laurence, (1993). Housing Code Enforcement and Urban Decline, Journal of Affordable Housing & Community Development Law, 6 (1):29 46. Rubin, J. Et al. (2016). Are better health outcomes related to social expenditure? A cross national empirical analysis of social expenditure and population health measures. Santa Monica, CA: RAND Corporation. Retrieved from : http://www.rand.org/pubs/research_reports/RR1252.html Ryan, A. M., Kutob, R.M., Suther, E., Hansen, M., & Standel, M. (2012). "Pilot study of impact of medical legal partnership services on patient's perc eived stress and wellbeing." Journal of Health Care for the Poor and Underserved 23(4): 1536 1546. Sandel, M., Suther, E., Brown, C., Wise, M., & Hansen, M. (2014). "The MLP vital sign: Assessing and managing legal needs in the healthcare setting." Journa l of Legal Medicine 35: 41. Scherer, Jasper. (October 14, 2016), "D.C. Council to Consider Free Legal Help for Poor Residents in Housing Cases." Sege, R. Et al. (2015). "Medical legal strategies to improve infant health care: a randomized trial." Pediatr ics 136(1): 2014 2955. Seron, C. Et al., (2001). The Impact of Legal Counsel on Outcomes for Poor Tenants in New York City's Housing Court: Results of a Randomized Experiment, 35 Law & Society Review 419 (2001); Abel, Laura K., Evidence Based Access to Ju stice, 13 Univ. Of Pennsylvania Journal of Law and Social Change 295 (2009).

PAGE 186

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 186 Shonkoff, J. P. e t al. (2011). "The Lifelong Effects of Early Childhood Adversity and Toxic Stress." Pediatrics 129(1): e232 e246. Siegrist, J. (1996). Adverse health Effects of High Effort/Low Reward Conditions. Journal of Occupational Health Psychology, 1(1): 27 41. See also, Robert Wood Johnson Foundation, (March 2013) "How Does Employment, or Unemployment Affect Health?" Singer, Judith D. And Willett, John B., Applied Longitu dinal Data Analysis: Modeling Change and Event Occurrence (Oxford University Press, 2003). Singh, Ritesh and Keshab Mukhopadhyay, (2011). Survival Analysis In Clinical Trials: Basics and Must Know Areas, Perspectives in Clinical Research 2(4): 145 148. Skinner, J. Et al. (2005). Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation, 112(17): 2634 2641. Smith, M. R. & Petrocelli, M. (2001). Racial Profiling? A Multivariate Analysis of Police T raffic Stop Data. Police Quarterly, 4(1): 4 27. Sorensen, G., Barbeau, E., Hunt, M. K., & Emmons, K. (2004). Reducing social disparities in tobacco use: a social contextual model for reducing tobacco use among blue collar workers. American journal of pu blic health 94 (2), 230 239. Starkes, H., & Garrido, M.M. (October 20, 2014). Observational & Quasi experimental research methods 8th Annual Kathleen Floey Palliative Care Retreat Methods Workshop. Steffen, J. (March 19, 2013). Colorado's childhood povert y rate almost doubles in 10 years. The Denver Post. Stuart, E.A., (2007). Estimating Causal Effects Using School Level Data Sets. Educational Researcher, 36(4):187 198. Sudano, J. J. Et al. (2013). Neighborhood racial residential segregation and changes in health or

PAGE 187

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 187 death among older adults. Health Place, (19): 80 88. Tarlov, Alvin R., (1999). Public Policy Frameworks for Improving Population Health, Annals of the New York Academy of Sciences, 896:281 293. Taylor, D. R. Et al (2015). "Keeping the Heat on for Children's Health: A successful medical L egal partnership initiative to prevent utility shutoffs in vulnerable children." Journal of Health Care for the Poor and Underserved 26(2015): 676 685. Teufel, J. A. Et al. (2012). "Rural medical legal partn ership and advocacy: A three year follow up study." Journal of Health Care for the Poor and Underserved 23(2): 705 714. Tyler, E. T. (2012). "Aligning public health, health care, law and policy: Medical legal partnerships as a multilevel response to the social determinants of health." Journal of Health and Biomedical Law 8: 211. United States Department of Health and Human Services. Agency for Health Research and Quality (March 2012). National Healthcare Disparities Report 2011.(AHRQ Publication No. 12 0 006). Weintraub, D., Rogers, M.A., Botcheva, L., Loeb, A., Knight, R., Ortega, K., Huffman, L. (2010). "Pilot study of medical legal partnerships to address social and legal needs of patients." Journal of Health Care for the Poor and Underserved 21( 2): 157 168. Wettach, J. (2008). "The law school clinic as a partner in a medical legal partnership." Tennessee Law Review 75: 305 315. White House Legal Aid Interagency Roundtable, "Expanding Access to Justice, Strengthening Federal Programs," First Ann ual Report of the White House Legal Aid Interagency Roundtable (November 2016) Wilcox, Katie. "Denver study seeks to unlock mystery of infant mortality disparities," The

PAGE 188

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 188 Gazette, (February 27, 2016). Available here: http://gazette.com/denver study sees to unlock mystery of infant mortality disparities/article/1571014 Wilkinson, R. & Marmot, M. (Eds.) 2003. Social Determinants of Health: The Solid Facts (2 nd ed.). World Health Organization: Denmark. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0005/98438/e81384.pdf Williams, D. R., & Coll ins, C. (2001). Racial residential segregation: a fundamental cause of racial disparities in health. Public health reports 116 (5), 404. Williams, D. R., & Mohammed, S.A. (2013). "Racism and health I: Pathways and Scientific Evidence." American Behaviora l Scientist 57(8): 1152 1173. Woolf, S. H., & Braveman, P. (2011). Where health disparities begin: the role of social and economic determinants and why current policies may make matters worse. Health affairs 30 (10), 1852 1859. Zhou, F., Shefer, A., Weng er, J., et al. (2014). "Economic Evaluation of the Routine Childhood Immunization program in the United States, 2009." Pediatrics 135(4): 1 9. Zimmerman, E. B., Woolf, S. H. & Haley, A. (2015). Understanding the Relationship Between Education and Health: A Review of the Evidence and an Examination of Community Perspectives. Agency for Healthcare Research and Quality. Retrieved from http://www.ah rq.gov/professionals/education/curriculum tools/population health/zimmerman.html

PAGE 189

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 189

PAGE 190

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 190 APPENDIX

PAGE 191

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 191 Ap pendix A

PAGE 192

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 192

PAGE 193

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 193

PAGE 194

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 194 Appendix B CHOC MLP Screening Questionnaire

PAGE 195

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 195 Appendix C Demographic Characteristics of Study Participants Table 17 shows the racial composi tion of this patient population, and Table 1 8 shows the ethnicity of the same patient population. Table 18 Racial Composition of Patients Under Age 3 at First Visit to CHCO MLP Clinics during Study Period Table 19 Ethnicity of Patients Under Age 3 at First Visit to CHCO MLP Clinics During Study Period Next, I categorized patients by age group, using the "first_age" variable to evaluate the number of patients in each group for immunization analysis. Then, using the Stata command, "egen_agecat = cut (first_age), at (0,1,2)," I categorized patients into age category groups of 1

PAGE 196

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 196 year increments. Patients between 0 and .99 years, I defined as 0 years old; patients between 1.00 1.99 years as 1 years old; pati ents between 2.00 2.99 years as 2 years old excluding any patients above 2.99 years from the study's patient population. Table 19 shows the ages of the 11,553 patients included in my study population and Table 20 shows the distribution of those ages am ong the treated and Un treated patient groups Table 20 Age Distribution of Children Under Age 3 Included in Study Population Table 21 Age Distribution of Treated and Un Treated Patients The treatment group includes two sub populations: the cohort of 92 patients who became CLS clients belonging to the CLS Case group, and the cohort of 985 MLP Positive patients who

PAGE 197

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 197 received other MLP services apart from CLS. Table 21 shows the age distribution for patie nts in each of the two cohorts within the treatment group of 1,077 patients. Table 22 Age Distribution of Treatment Group Children, by Cohort Cohort 0 to 0.99 Years 1 to 1.99 Years 2 to 2.99 Years Total CLS Case 66 13 13 92 M LP Positive 810 96 79 985 TOTAL 876 109 92 1,077

PAGE 198

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 198 Appendix D Excerpt from Final Data Dictionary (March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�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�+*/&+`#'/&4G */&+/0.71+%'+I+;0.-+,LM+$.+;%04$+@%4%$]+:#$&+:70%'/+$2&+$%-&;0#-&+.;+")")ST+$.+"<)H")"HN++O2&+@%4%$+$I1&4+%'3D7:&:+X,A+%'1#$%&'$+2.41%$#DA+.0+.7$1#$%&'$+%'+#'I+.;+$2&+10%-#0I+.0+41&3%#D$I+3#0&+3D%'%34+Q2&0&+>?!+430&&'%'/4+Q&0&+:.'&N !T !#$%&'$ =$7:I !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 [%04$J6'4 6'470#'3&+OI1&+,70%'/+=$7:I K2.D&+(7-C&0 O2&+10%-#0I+%'470#'3&+$I1&+;.0+$2&+1#$%&'$a4+;%04$+@%4%$+C&$Q&&'+")")ST+#':+"<)H")"HN++O2&+@%4%$+$I1&4+%'3D7:&:+X,A+%'1#$%&'$+2.41%$#DA+.0+.7$1#$%&'$+%'+#'I+.;+$2&+10%-#0I+.0+41&3%#D$I+3#0&+3D%'%34+Q2&0&+>?!+430&&'%'/4+Q&0&+:.'&N !"S !#$%&'$ X@&0 !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 */&J#$JX'$0I */&+#$+5U5L+X'$0I K2.D&+(7-C&0 */&+%'+:#I4+;0.-+,LM+$.+%'%$%#D+@%4%$]+:#$&+#$+5U5LN++\%4%$4+%'3D7:&+$2&+;.DD.Q%'/+$I1&4b++X,A+%'1#$%&'$+2.41%$#DA+.0+.7$1#$%&'$+%'+#'I+.;+$2&+10%-#0I+.0+41&3%#D$I+3#0&+3D%'%34+Q2&0&+>?!+430&&'%'/4+Q&0&+:.'&N !"" !#$%&'$ =$7:I+X':+57$.;;+E07'+$20.7/2+"9"R9"WG !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 */&J#$JXc%$ */&+#$+5U5L+Xc%$ K2.D&+(7-C&0 */&+%'+:#I4+;0.-+,LM+$.+D#$&4$+@%4%$]+:#$&+#$+5U5L+$20.7/2+"<)H")"RN++\%4%$4+%'3D7:&+$2&+;.DD.Q%'/+$I1&4b++X,A+%'1#$%&'$+2.41%$#DA+.0+.7$1#$%&'$+%'+#'I+.;+$2&+10%-#0I+.0+41&3%#D$I+3#0&+3D%'%34+Q2&0&+>?!+430&&'%'/4+Q&0&+:.'&N !"< !#$%&'$ !0&9>?! [#-%DId4+*33&44+$.+U&#D$2+ 5#0& 5U5LJV04J!0&>?! V+#$+5U5L+!0&9O0&#$-&'$ ,&3%-#D (7-C&0+.;+I+4&&'+#$+5U5L+10%.0+$.+>?!)$0&#$-&'$+:#$&+E&%$2&0+%'%$%#D+430&&'%'/+:#$&+;.0+430&&'&:+/0.714A+.0+R)")ST+;.0+3.'$0.D)d(.$+=30&&'&:d+/0.71G+;.0+X,A+%'1#$%&'$+2.41%$#DA+.0+.7$1#$%&'$+@%4%$4+#$+.'&+.;+$2&+3D%'%34+Q2&0&+>?!+430&&'%'/4+Q&0&+:.'&N++O2%4+%4+3.-17$&:+C#4&:+.'+$2&+$0&#$-&'$+:#$&+47C$0#3$+$2&+;%04$+@%4%$+:#$&+#$+5U5L+&@&0N++6;+#+1#$%&'$+'&@&0+2#:+#+@%4%$+10%.0+$.+$2&+$0&#$-&'$+:#$&A+$2%4+@#0%#CD&+%4+CD#'P)-%44%'/N !"H !#$%&'$ B':&0+Y =&D&3$&:+,%32.$.-.74+ \#0%#CD&4 ,&#$2JCIY !#$%&'$+,&#$2+MI+*/&+Y V)(+%':%3#$.0 6;+1#$%&'$+:%&:+C&;.0&+$2&+#/&+.;+Y+IA+$2&'+Ve+1#$%&'$4+Q%$2+7'P'.Q'+:&#$2+:#$&4+Q%DD+2#@&+CD#'P+@#D7&4A+#':+1#$%&'$4+Q2.+Q&0&+'.$+I&$+#/&+Y+#4+.;+07'+:#$&+E"<)H")"FG+Q%DD+2#@&+fBf+E7'P'.Q'G+@#D7&4 !"R !#$%&'$ B':&0+"^ =&D&3$&:+,%32.$.-.74+ \#0%#CD&4 ,&#$2JCI"^ !#$%&'$+,&#$2+MI+*/&+"^ V)(+%':%3#$.0 6;+1#$%&'$+:%&:+C&;.0&+$2&+#/&+.;+"^+IA+$2&'+Ve+1#$%&'$4+Q%$2+7'P'.Q'+:&#$2+:#$&4+Q%DD+2#@&+CD#'P+@#D7&4A+#':+1#$%&'$4+Q2.+Q&0&+'.$+I&$+#/&+"^+#4+.;+07'+:#$&+E"<)H")"FG+Q%DD+2#@&+fBf+E7'P'.Q'G+@#D7&4 !"F !#$%&'$ 5700&'$+E#4+.;+"<9H"9"FG =&D&3$&:+,%32.$.-.74+ \#0%#CD&4 ,&#$2J5700&'$,#$& !#$%&'$+,&#$2+CI+5700&'$+,#$& V)(+%':%3#$.0 6;+1#$%&'$+:%&:+C&;.0&+$2&+07'+:#$&+E"<)H")"FGA+$2&'+Ve+$2&0&+42.7D:+C&+'.+fBf+.0+CD#'P+@#D7&4+;.0+$2%4+@#0%#CD& !"W !#$%&'$ 5700&'$+E#4+.;+"<9H"9"FG !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 =&c =&c =$0%'/ `#Q+@#D7&4+;0.-+XU`b+>#D&A+[&-#D& !"Y !#$%&'$ 5700&'$+E#4+.;+"<9H"9"FG !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 `#3& `#3& =$0%'/ `&3.'3%D&:+0#3&+4.+$2#$+1#$%&'$4+Q%$2+-.0&+$2#'+"+:%4$%'3$+0#3&+@#D7&+E%'3D7:%'/+.$2&0G+.0+Q2.+2#@&+#+0#3&+.;+f>.0&+$2#'+.'&+0#3&f+#0&+D#C&D&:+f>7D$%1D&+`#3&4fN++!#$%&'$+Q%$2+'.+:&430%1$%@&+0#3&+@#D7&4+Q2#$4.&@&0+#0&+D#C&D&:+fB'P'.Q')(.$+`&1.0$&:fN++L$2&0Q%4&A+0#3&+%4+4&$+&g7#D+$.+$2&+4%'/D&+@#D7&+;.7':+%'+$2&+XU`+9+K2%$&A+MD#3P)*;0%3#'+*-&0%3#'A+*-&0%3#'+6':%#')*D#4P#+(#$%@&A+*4%#'A+(#$%@&+U#Q#%%#')L$2&0+!#3%;%3+64D#':&0A+L$2&0N !"^ !#$%&'$ 5700&'$+E#4+.;+"<9H"9"FG !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 X$2'%3%$I X$2'%3%$I =$0%'/ `#Q+@#D7&4+;0.-+XU`b+U%41#'%3+.0+?#$%'.A+(.$+U%41#'%3+.0+?#$%'.A+(.$+`&1.0$&:A+.0+B'P'.Q' !"T !#$%&'$ 5700&'$+E#4+.;+"<9H"9"FG !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 ?#'/7#/& !0%-#0I+?#'/7#/& =$0%'/ `#Q+@#D7&4+;0.-+XU`N !6( *!_*`J"J>6( =$0%'/ `#Q+@#D7&4+;0.-+XU`b+*!_*`+5#$&/.0I+#$+"+-%'7$& !<" !#$%&'$ *$+M%0$2 !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 *!_*`JFJ>6( *!_*`JFJ>6( =$0%'/ `#Q+@#D7&4+;0.-+XU`b+*!_*`+5#$&/.0I+#$+F+-%'7$&4 !<< !#$%&'$ *$+M%0$2 !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 *!_*`J"SJ>6( *!_*`J"SJ>6( =$0%'/ `#Q+@#D7&4+;0.-+XU`b+*!_*`+5#$&/.0I+#$+"S+-%'7$&4 !?! !#$%&'$+,&-./0#12%34+ 52#0#3$&0%4$%34 !7CD%3J65 !7CD%3+6'3.-&+5.'$%'/&'$ V)(+%':%3#$.0 6;+1#$%&'$+&@&0+2#:+>&:%3#%:A+565!A+%':%/&'$+3#0&A+.0+5U!h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

PAGE 199

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 199 199 Appendix E ACIP Recommended Immunization Schedule, United States, 2017 5

PAGE 200

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 200 200 Appendix F I estimated the MLP 's treatment eff ect using a variety of covariate combinations. I found several that increased the number of matches from the control population but none that improved my analysis. For example, I estimated a propensity score after removing the two covariates with the great est number of missing variables, leaving 8 variables with no missing observations. Table 22 shows the number of treated patients who matched increased to 1,038 but the balancing property was not satisfied using this combination of covariates. Table 23 Removing Variables with High Missing Values Increased Resulting Matches Nevertheless, I estimated the ATT with the propensity scores obtained from this analysis. Tables 23 and 24 show the output which estimated the average treatmen t effect of the MLP among the treated population. Although the number of matches improved significantly to over 1,000 patients, the ATT estimate was identical for vaccination compliance rate and only slightly smaller ( 0.057 versus 0.081) for residential mobility.

PAGE 201

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 201 201 Table 24 ATT of MLP on Immunization Compliance Rate After Removing Variables With High Percentage of Missing Data Table 25 ATT of MLP on Upward Residential Mobility After Removing Variables with High Percentage of Missing Data

PAGE 202

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 202 202 I concluded that simply increasing the number of matches had virtually no impact on the treatment effect I calculated. The reason for this is shown in Figure 39 below. Removing missing data significantly increased the numb er of matches but not the quality of matches as shown by the overlap and distribution between propensity scores in the treated and control populations with and without variables that have significant percentages of missing data. Figure 39 Quality and Number of Propensity Score Matches Before and After Removing Variables with High Percentage of Missing Data I concluded that my estimates were more reliable with fewer matches but a propensity score that satisfied the balancing pro perty.

PAGE 203

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 203 203 ENDNOTES 1 Kurosky, S. K., Davis, K.L., & Krishnarajah, G., (2016). "Completion and Compli ance of Childhood Vaccinations in the United States." Vaccine 34 (3): 387 394. 2 DTaP = Diphtheria, tetanus, pertussis (whooping cough); IPV=inactivated poliovirus; MMR= measles, mumps, and rubella (German measles); Hib=meningitis and other bacterial dise ases; HepB= Virus infection that attacks liver including cirrhosis, liver cancer, and liver failure; Varicella= chickenpox; Rotavirus=diarrhea and dehydration; PCV13=pneumococcal pneumonia and meningitis; HepA=viral liver disease. 3 See, McGough, J. and F araone, S.V., (2009). Estimating the Size of Treatment Effects Moving Beyond P Values. Psychiatry 10:21 29l. 4 I also calculated the absolute number of moves each patient had. That variable, residential_instability, was similarly clustered as the reside ntial_mobility variable I used for this study. 5 All relevant recommendations in this schedule are identical to recommendations made in 2013 found here: https://www.cdc.gov/mmwr/prev iew/mmwrhtml/su6201a2.htm

PAGE 204

MEDICAL LEGAL PARTNERSHIPS: REDUCING HEALTH DISPARITIES 204 204