Citation
Longitudinal associations among disease severity, employment and quality of life in adults with cystic fibrosis

Material Information

Title:
Longitudinal associations among disease severity, employment and quality of life in adults with cystic fibrosis
Creator:
O'Donnell, Holly ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (49 pages) : ;

Thesis/Dissertation Information

Degree:
Master's ( Master of arts)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Psychology, CU Denver
Degree Disciplines:
Clinical health psychology

Subjects

Subjects / Keywords:
Cystic fibrosis ( lcsh )
Cystic fibrosis ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Currently, little is known about health-related quality of life in adults with Cystic Fibrosis (CF). Only recently a milestone was achieved such that adults (18+ years) now represent more than half of the population with CF in the United States. Previous research has shown that CF affects an individual's ability to work and that both cystic fibrosis and employment affect health-related quality of life. The sample for this study consisted of 303 adults with CF who received their care at one of ten medical centers across the United States. Using structural equation modeling (SEM), a latent growth curve (LGC) analysis allowed us to better understand the longitudinal relationships among disease severity, employment and health-related quality of life. Specifically, this study examined how employment and health-related quality of life change as a result of declining health in adults with CF. Analyses used data collected at four time points over the course of 41 months. Possible implications include a greater understanding of factors that influence health-related quality of life in adults with CF and more specifically, risk factors that negatively impact health-related quality of life. These findings are important to share with CF care team members in order to provide high quality psychosocial care.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Holly O'Donnell.

Record Information

Source Institution:
University of Colorado Denver Collections
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
987242093 ( OCLC )
ocn987242093
Classification:
LD1193.L645 2016m O46 ( lcc )

Downloads

This item has the following downloads:


Full Text
LONGITUDINAL ASSOCIATIONS AMONG DISEASE SEVERITY, EMPLOYMENT
AND QUALITY OF LIFE IN ADULTS WITH CYSTIC FIBROSIS
by
HOLLY ODONNELL B.A., Middlebury College, 2012
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Arts
Clinical Health Psychology Program
2016


This thesis for the Master of Arts degree by Holly ODonnell has been approved for the Clinical Health Psychology program by
Edward Dill, Chair Barbara Walker Walter Robinson
Date: December 17, 2016


Ill
ODonnell, Holly, M.A., Clinical Health Psychology
Longitudinal Associations Among Disease Severity, Employment, and Quality of Life in
Adults with Cystic Fibrosis
Thesis directed by Assistant Professor Edward Dill
ABSTRACT
Currently, little is known about health-related quality of life in adults with Cystic Fibrosis (CF). Only recently a milestone was achieved such that adults (18+ years) now represent more than half of the population with CF in the United States. Previous research has shown that CF affects an individuals ability to work and that both cystic fibrosis and employment affect health-related quality of life. The sample for this study consisted of 303 adults with CF who received their care at one of ten medical centers across the United States. Using structural equation modeling (SEM), a latent growth curve (LGC) analysis allowed us to better understand the longitudinal relationships among disease severity, employment and health-related quality of life. Specifically, this study examined how employment and health-related quality of life change as a result of declining health in adults with CF. Analyses used data collected at four time points over the course of 41 months. Possible implications include a greater understanding of factors that influence health-related quality of life in adults with CF and more specifically, risk factors that negatively impact health-related quality of life. These findings are important to share with CF care team members in order to provide high quality psychosocial care.
The form and content of this abstract are approved. I recommend its publication.
Approved: Edward Dill


IV
TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION...................................................1
Hypothesisl....................................................7
Hypothesis 2...................................................8
Hypothesis 3...................................................8
Hypothesis 4...................................................9
II. METHOD........................................................10
Sample........................................................10
Procedure.....................................................11
Instruments...................................................11
Data Analysis.................................................14
III. RESULTS.......................................................20
Descriptive Statistics........................................20
Hypothesis 1..................................................23
Hypothesis 2..................................................25
Hypothesis 3..................................................26
Hypothesis 4..................................................29


V
IV. DISCUSSION.....................................................32
Implications..................................................36
Limitations...................................................38
Future Directions.............................................39
REFERENCES...........................................................41


1
CHAPTERI INTRODUCTION
Cystic fibrosis (CF) is a life-threatening genetic disease that affects approximately 30,000 individuals in the United States and 70,000 individuals worldwide (CF Foundation, 2014). It is caused by mutations in both copies of the cystic fibrosis transmembrane conductance regulator (CFTR) gene, which interferes with production of the CFTR protein. In patients with clinical CF, either little to no CFTR is produced or a defective protein is produced. The most common clinical pattern in patients with CF is the development of chronic obstructive lung disease accompanied by pancreatic insufficiency, with some patients also developing diabetes, joint and bone disease, and severe sinus disease. There has also been documentation of a host of other symptoms and conditions, such as diabetes and sleep disturbances.
Cystic fibrosis was previously considered a childhood disease, because adults represented a relatively small proportion of the population with CF. However, in 2014, a milestone was reached such that over half of those with CF in the United States are now adults (CF Foundation Report, 2014), which is a drastic increase from 1986 when only 29.2% of people with CF were 18 years or older (CF Foundation Report, 2013). As of 2014, the median predicted survival age was 39.3 years, which means that 50% of patients with CF are expected to live to 39.3 years or beyond (CF Foundation Report, 2014).
The increased lifespan of adults with CF has led to an increasing number of patients with CF pursuing developmentally appropriate adult activities such as higher education and employment (Havermans, Colpaert, Vanharen, and Dupont, 2009). However, potentially because predicted survival age shifts are so recent, there is little research to date on how the disease influences patients daily tasks of adult living and health-related quality of life.


2
Impact of CF on Functioning and Quality of Life
There is conflicting evidence as to whether having CF is associated with increased rates of psychological difficulties in adolescence and adulthood (Pfeffer, Pfeffer, Hodson, 2003).
Some investigators found elevated rates of psychiatric symptoms in patients with CF (Pearson, Pumariega & Seilheimer, 1991; Lawler, Nakielny & Wright, 1966), whereas other studies (Blair, Cull, & Freeman, 1994; Sheperd et al., 1990 & de Jong et al., 1997) have found no differences between the psychological functioning of young adults with and without CF. These discrepancies may be partially due to methodology; the instruments used to measure psychosocial functioning and rates of psychiatric illness varied widely across the studies and the samples were heterogeneous. For example, one of the studies (Anderson, Flume & Hardy, 2001) used a sample that was significantly less sick than those who chose not to participate, which is important considering there is some evidence suggesting psychological and psychosocial functioning of adults with CF is on par with their peers until the disease becomes severe (Pfeffer et al., 2003; Pearson et al, 1991). Additionally, some of these studies are from Europe and therefore the results may not generalize to a U.S. population, especially because social welfare systems in Europe are very different from those in the U.S. Regardless, careful examination of these studies can provide some insight into the functioning of adults with CF.
In contrast with the conflicting evidence regarding links between CF and psychological disorders, there is consistent evidence suggesting that CF affects health-related quality of life, especially the physical functioning domain. Although the term health-related quality of life lacks a consistent definition across the literature, it is generally understood to be a measure of how ones medical illness affects ones psychiatric, psychosocial and physical functioning (Pfeffer et al., 2003). Related to this, there is consistent evidence that patients with CF have more


3
problems in daily living than their healthy peers (Anderson et al., 2001; Congleton, Hodson & Duncan-Skingle, 1996).
Disease severity was found to correlate with Health Related Quality of Life (HRQoL), but correlations were weak (Abbott et al., 2003; Dill et al., 2013), suggesting that variables other than disease severity also affect health-related quality of life. Work or the ability to maintain employment may be one of these variables. After controlling for medical variables such as disease severity, adults with CF who were working reported a higher health-related quality of life on physical functioning, role functioning and social functioning than those who were not working (Havermans et al., 2009). This suggests that employment plays an important role in determining health-related quality of life in adults with CF. Additionally, patients who were not working were found to have significantly higher levels of depression than those who were working (Burker et al., 2004). Although depression is not a direct measure of health-related quality of life, patients with anxiety and depression have been found to have lower scores on certain health-related quality of life domains (Havermans, Colpaert, & Dupont, 2008), and health-related quality of life encompasses psychological functioning.
Overall, adult CF patients experience difficulties in many areas of daily living including, but not limited to, sleep, home management, mobility, physical recreation, and employment (de Jong et al., 1997). Employment appears to be a consistent challenge for individuals with CF; patients with CF are less likely to be employed than their healthy peers (Blair et al., 1994) and report that the disease affects their ability to work (Laborde-Casterot et al., 2012). Some investigators have even found employment to be the area of daily living in which adult patients with CF experience the greatest difficulty (de Jong et al., 1997).


4
Relationship between Employment and Quality of Life
Similar themes have emerged in studies examining how employment affects individuals with illnesses and disabilities across a wide range of diagnoses (Freedman & Fesko, 1996; Johnson et al., 2004; Miller & Dishon, 2006). Financial benefits of employment are important as they allow individuals not only to support themselves but to receive quality health care and insurance through employer health insurance plans. Work has a therapeutic aspect in that it distracts individuals from their illness (Johnson et al., 2004) and makes individuals feel like productive members of society by assuming responsibility, completing tasks, and helping others (Freedman & Fesko, 1996; Johnson et al., 2004). Finally, social connections from work are important and valuable (Freedman & Fesko, 1996; Johnson et al., 2004). Employment can also increase ones selfesteem, in part because it becomes part of ones identity (Miller & Dishon, 2006).
These findings are consistent with Eriksons theory of psychosocial development and the importance of identity formation versus role confusion. Erikson emphasized the importance of a meaningful career in the development of ones identity. According to his theory, it is essential to achieve identity development in order to proceed to further developmental tasks. Those who struggle with identity formation are said to suffer from role confusion, which can lead to feelings of confusion and questions regarding their purpose (Sokol, 2009). Thus, according to this theory, individuals who have difficulty obtaining and maintaining employment are at risk for a wide range of psychosocial and developmental issues.
Impact of Cystic Fibrosis on Employment
Given the importance of work to psychosocial functioning, it is unfortunate that CF can have such a profound effect on ones ability to work. A majority (55% to 68%) of CF patients report that having CF impacts their ability to work (Hogg et al., 2007; Laborde-Casterot et al.,


5
2012). Patients have reported having to quit work due to CF (Hogg et al., 2007), and more CF patients work part-time (compared to full-time) than the national average in France (Larbode-Casterot, 2012). Despite these trends, half the patients in a sample of CF patients with highly compromised lung function were working, which suggests that employment is valuable and important to these patients (Burker et al., 2004).
Several studies have examined risk factors for difficulties maintaining employment among patients with CF. Educational attainment, socioeconomic status (SES), disease severity and treatment burden are the four most commonly reported risk factors in assessing for unemployment, employment difficulties, or disability in adults with cystic fibrosis (Laborde-Casterot, 2012; Havermans et al., 2009; Gillen et al., 1995; Taylor-Robinson, 2013; Goldberg, 1985; Burker et al., 2004 & Hogg et al., 2007). Lower levels of educational attainment and SES have both been associated with lower levels of employment in CF patients (Burker et al., 2004; Taylor-Robinson, 2013). Additionally, low SES has been found to amplify the effect that disease severity has on employment. For example, Taylor-Robinson et al. (2013) found that those with poor lung function in the lowest SES group had a 7.7% chance of reduction in employment while those with low lung function in the highest SES group had only a 3% chance of reduction in employment.
A few studies have investigated potential reasons that SES and education are positively correlated with employment by examining the types of jobs patients with CF were able to maintain. In one study, all but 6% of employed CF patients were working in skilled or semi-skilled jobs (Burker et al., 2005). Participants in the study who were not working at the time of data collection, but had worked in the past, reported working in a significantly higher percentage of unskilled jobs than those who worked at the time of data collection, suggesting that CF patients


6
who are able to obtain skilled jobs have an easier time maintaining these jobs. Other studies have similarly found greater rates of white-collar jobs (in contrast to blue-collar jobs) in CF patient populations compared to the general population (Gillen et al., 1995; Laborde-Casterot et al., 2012). This is likely because skilled jobs require less physical exertion than unskilled jobs, and patients with CF are unable to maintain jobs that require heavy physical exertion, especially as the disease progresses. Another factor that may contribute to these findings is that skilled jobs often offer more flexibility and are, therefore, easier to maintain as the disease progresses.
The most studied and perhaps most interesting risk factor for employment difficulties is disease severity. Several investigators have found that CF patients with more advanced disease and more impaired lung function work less and are at higher risk for being unemployed (Laborde-Casterot, 2012; Havermans et al.,2009; Gillen et al., 1995; Taylor-Robinson et al., 2013). However, other studies have found inconsistent results. For example, Burker et al., 2004, did not find a significant difference in lung function between those who were employed and those who were not employed. Similarly, Hogg et al., 2007, did not find number of hours worked to be dependent on lung function. However, number of hospital admissions was a significant predictor of hours worked per week in this study. Therefore, although a significant number of studies have examined the relationship between disease severity and employment, the relationship remains unclear due to inconsistent results. This may be partially a result of the heterogeneity of the studies that have examined these variables thus far. For example, studies have used several different measures of disease severity and have controlled for various clinical covariates. Finally, the studies are from several different countries, which may influence the results due to diverse economies and social welfare systems.


7
The Present Study
Given the evidence that employment affects health-related quality of life in individuals with CF and maintaining employment can be difficult for individuals with CF, it is important to further explore the relationships among disease severity, employment, and health-related quality of life in adults with CF. Almost all of the research to date regarding disease severity, employment, and health-related quality of life in the CF population has been cross-sectional. Cross-sectional studies do not allow examination of the relationship of these variables over time. These are important relationships to understand because disease severity increases over time and we currently know very little regarding how that change in disease severity may be associated with changes in employment and/or health-related quality of life. The current study seeks to further our understanding of the relationships among disease severity, employment, and health-related quality of life at a single time point and to provide novel information on the relationships among these variables over time.
Aim 1. To examine the relationship between employment and health-related quality of life in adults with cystic fibrosis at a single time point.
Hypothesis 1. Employment will be correlated with higher levels of health-related quality of life at the baseline time point. Specifically, those who are employed will score significantly higher on the emotional, social, role and physical health-related health-related quality of life (HRQoL) domains compared to those who are unemployed at the baseline time point.
Aim 2. To examine the relationship between disease severity and employment in adults with cystic fibrosis at a single time point.


8
Hypothesis 2. Disease severity will be negatively correlated with employment at the baseline time point. Higher forced expiratory volume in one-second (FEVi) will be positively correlated with employment and number of pulmonary exacerbations (defined as exacerbations requiring home IV treatment or hospitalization) will be negatively correlated with employment.
Aim 3. To examine the relationship between disease severity, employment and health-related quality of life over time in adults with Cystic Fibrosis.
Hypothesis 3a. Change in disease severity (FEVi and exacerbations) over time, measured by its slope, will be correlated with change in employment status over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEVi), employment status will decrease. The relationship between exacerbations and employment status will be negative (i.e., the more exacerbations, the less likely one is to be employed). The relationship between FEVi and employment will be positive (i.e., the higher the FEVi level, the more likely an individual is to be employed).
Hypothesis 3b. Change in employment level over time, measured by its slope, will be correlated with change in health-related quality of life (emotional, social, role functioning, and physical functioning domains). Specifically, the slopes of employment and health-related quality of life will be positively correlated.
Hypothesis 3c. Change in disease severity (FEVi and exacerbations) over time, measured by its slope, will be correlated with change in health-related quality of life over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEVi), health-related quality of life will decrease. The relationship between exacerbations and health-related quality of life will be negative (i.e., the more exacerbations, the lower


9
ones health-related quality of life). The relationship between FEVi and health-related quality of life will be positive (i.e., higher FEVi levels will be correlated with higher health-related quality of life scores).
Aim 4. To better understand the mechanism by which disease severity affects HRQoL in adults with CF.
Hypothesis 4. Employment level will mediate the relationship between disease severity and health-related quality of life. In other words, the relationship between disease severity and health-related quality of life will be at least partially explained by the relationship between disease severity and employment, as well as the relationship between employment and health-related quality of life. Therefore, increased disease severity will predict unemployment, which in turn will predict lower health-related quality of life.
Thus, the overall goal of all four aims is to examine and evaluate the cross-sectional and longitudinal relationships among disease severity, employment, and health-related quality of life in adults with cystic fibrosis.


10
CHAPTER II METHOD
Sample
The participants of this study were part of a longitudinal study of adults with cystic fibrosis called the Project on Adult Care in Cystic Fibrosis (PAC-CF). The participants were 18 years or older and received CF care at one of ten centers across the United States participating in the study. A stratified sample was used to maximize inclusion of adults with relatively lower probability of five-year survival. Adults with a very high probability of five-year survival are the least informative as to how health-related quality of life changes as the disease progresses, and a major goal of this study was to examine how health-related quality of life changes as the disease progresses and disease severity increases. Study coordinators were provided potential participants probability of five-year survival using equations by Liou et al., 2001. All adults with a
five-year predicted probability of survival less than 0.975 (=515) and a randomly selected 25% of adults with a five-year predicted probability of survival equal to or greater than 0.975 (n=60) were selected to be approached to participate in the study. Out of the 575 potential participants, 333 enrolled in the study and 242 either refused or could not be contacted. Compared to those who did not enroll, the participants were more likely to be white, female, and older. The participants also had better standardized weight for age score (i.e., higher weight) and a greater number of pulmonary exacerbations than those who chose not to participate. There were 303 participants who completed the first wave in April of 2005. Mean age of participants at the beginning of the study was 32.98 years (579=10.71), 42.0% identified as male, 3.3% identified as Hispanic/La-tino, and 97.0% identified as White, which is consistent with the racial distribution of CF patients. There were 185 participants who participated in the final wave in February of 2009. Of


11
those available at the beginning of each wave, between 70% and 93% completed and returned the surveys for each wave.
Procedure
The study was approved by the Institutional Review Boards at Education Development Center, Inc. in Waltham, Massachusetts, as well as the ten medical centers across the country with the participating CF centers. Participants were mailed surveys 11 times over the course of 46 months and returned them in an envelope by mail. The data presented in this study are from waves 1, 5, 8, and 10, which took place in April 2005, May 2006, June 2007 and August 2008. These waves were chosen because they are the most equally spaced apart. Having participants respond to the surveys outside of their clinical visits to their CF center was thought to promote anonymity and honesty, and reduce the possibility of response bias. Clinical data were collected at the time of the participants routine visits, which typically occurred on a quarterly basis, and were extracted from the CF Foundation patient registry. Therefore, the collection time points of survey data did not necessarily coincide with clinic visits. However, because most adults visited their CF care center on a quarterly basis, the clinical data were usually collected within three months of each survey.
Instruments
Demographic information. Demographic information such as age, gender, educational attainment, marital status, and income was collected via self-report surveys at several waves during the study. Demographic information used in the present study was collected at the initial time point.
Disease severity. This study used two measures of disease severity: FEVi and number of pulmonary exacerbations. FEVi is forced expiratory volume in one second and is an objective


12
measure of lung function used as a proxy for disease severity in patients with CF. Higher FEVi scores indicate better lung function and a lower disease severity. FEVi was collected when the participants visited their CF care center for routine clinical visits. Number of pulmonary exacerbations was defined as the number of hospitalizations or home IV treatments between each wave. For the first wave, number of exacerbations in the four months preceding the first data collection time point was used.
Employment. Employment status was measured via a self-report survey that participants completed at home outside of clinic visits. Employment was measured both continuously and di-chotomously. In the survey, patients were asked how many hours they worked in the previous week and if this was typical for them. They were also asked to respond yes/no to Are you currently employed or self-employed?
Quality of life. Quality of Life was measured using the Cystic Fibrosis Questionnaire-Revised (CFQ-R), which is a disease-specific measure of health related health-related quality of life for patients with cystic fibrosis. The CFQ-R is a revised version of the Cystic Fibrosis Questionnaire (CFQ), which was the first disease specific patient reported outcome measure for CF patients and caregivers (Quittner et al., 2012). There are currently four forms available: one for preschool age children, one for school age children, one for teens and adults, and one for parents. This study utilized the teen/adult version, because all participants were 18 years or older, and the teen/adult version is designed for patients 14 years and older. The teen/adult version consists of 50 items measured on 4-point Likert scales that include frequency (always, often, sometimes, never), intensity (a great deal, somewhat, a little, not at all), and true-false scales (very true, somewhat true, somewhat false, very false). Responses to the questions are converted to stand-


13
ardized scores ranging from 0 to 100 in the following domains: physical functioning, body image, eating disturbances, digestive symptoms, respiratory symptoms, weight, health perceptions, vitality, treatment burden, emotional functioning, social functioning, and role. Higher scores on the CFQ-R indicate higher health-related quality of life.
The psychometric properties of the CFQ-R were studied in patients enrolled in the Epidemiologic Study of Cystic Fibrosis (ESCF). From this sample, 13,437 CFQ-Rs from 7,330 patients were completed by patients and parents. According to the CF Foundation Patient Registry, the sample for this study was similar to the U.S. population in terms of demographic and health related variables, which suggests that the results of this study are generalizable to CF patients in the United States (Quittner et al., 2012).
The CFQ-R was found to successfully differentiate between patients who were being seen for sick versus well visits. With the exception of the Digestion scale, all comparisons were statistically significant, with lower scores reported during sick visits (Quittner et al., 2012). Moreover, all scales differed significantly as hypothesized between disease stage/severity except for Eating Problems, Role Functioning and Emotional Functioning, which differed significantly across all disease state groups except between the normal and mild disease groups. In order to further evaluate the construct validity of the CFQ-R, correlations were calculated between CFQ-R scales and health measures including FEVi% predicted, number of pulmonary exacerbations treated with IV antibiotics, and BMI (Quittner et al., 2012). Strong correlations were found between FEVi% predicted and number of exacerbations with the Physical Functioning, Role Functioning, Vitality, Health Perceptions, and Respiratory Symptoms scales. Significant correlations were also found between FEVi% predicted and the Social Functioning, Body Image, and Weight scales. Moderate correlations were found between BMI and the Body Image and Weight


14
scales (Quittner et al., 2012). Additional studies have shown the CFQ-R to correlate with measures of anxiety and depression, which is further evidence for the validity of this measure (Havermans et al., 2008).
In terms of reliability, strong internal consistency suggested the measure demonstrated good reliability. Cronbach alphas for the CFQ-R teen/adult version were as follows: Physical Functioning (0.94), Respiratory Symptoms (0.87), Role Functioning (0.83), Eating Problems (0.82), Vitality (0.80), Health Perceptions (0.79), Emotional Functioning (0.77), Digestive Symptoms (0.68), Body Image (0.65), Social Functioning (0.58), and Treatment Burden (0.51).
A Cronbach alpha was not calculated for the Weight subscale, as it is a single item scale. Therefore, all except two subscales of the teen/adult version of the CFQ-R demonstrated adequate internal consistency using the standard of 0.60 (Quittner et al., 2012).
Data Analysis
Hypothesis 1. Employment will be associated with higher levels of health-related quality of life at the baseline time point. Specifically, those who are employed will score significantly higher on the emotional, social, role and physical health-related health-related quality of life (HRQoL) domains compared to those who are unemployed at the baseline time point. I addressed this research question using an Analysis of Variance (ANOVA). The grouping variable for this analysis was employment status, and the dependent variable was the score on the CFQ-R subscales. I used SPSS software for this analysis.
Hypothesis 2. Greater disease severity, as measured by lower FEVi and a greater number of pulmonary exacerbations, will be negatively correlated with employment at the baseline time point. This question was analyzed using binary logistic regression. The independent variable (IV)


15
for this analysis was disease severity (measured by FEVi and number of pulmonary exacerbations), and the dependent variable (DV) was employment status (measured dichotomously). I used SPSS software to conduct this analysis.
Hypothesis 3a. Change in disease severity (FEVi and exacerbations) overtime, measured by its slope, will be correlated with change in employment status over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEVi), employment status will decrease. The relationship between exacerbations and employment status will be negative (i.e., the more exacerbations, the less likely one is to be employed). The relationship between FEVi and employment will be positive (i.e., the higher the FEVi level, the more likely an individual is to be employed). This was analyzed using Latent Growth Curve Analysis.
Hypothesis 3b. Change in employment level over time, measured by its slope, will be correlated with change in health-related quality of life (emotional, social, role functioning, and physical functioning domains). Specifically, the slopes of employment and health-related quality of life will be positively correlated. This hypothesis was analyzed using Latent Growth Curve Analysis.
Hypothesis 3c. Change in disease severity (FEVi and exacerbations) overtime, measured by its slope, will be correlated with change in health-related quality of life over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEVi), health-related quality of life will decrease. The relationship between exacerbations and health-related quality of life will be negative (i.e., the more exacerbations, the lower ones health-related quality of life). The relationship between FEVi and health-related


16
quality of life will be positive (i.e., higher FEVi levels will be correlated with higher health-related quality of life scores). This hypothesis was analyzed using Latent Growth Curve Analysis.
Structural equation modeling (SEM), in general, allows the investigator to test hypothesized relationships among observed and/or latent variables. A basic SEM model consists of several observed and/or latent variables and the regression paths between these variables. Indicator variables may be included if latent constructs are present. Each of the indicator variables has a factor loading (Preacher et al., 2008).
Latent growth curve (LGC) modeling (see Figure 1) is a specific type of SEM that allows a researcher to better understand change. The latent variables in LGC represent patterns of change of a variable that is measured a minimum of three times. The values of the variable at several time points allow the researcher to better understand the variables pattern of change, which is reflected in the latent construct (Preacher et al., 2008).
There are two main latent constructs in LGC, which are intercept and slope. The intercept is a measure of the variable of interest at the baseline time point. The second main latent construct, slope, is the rate at which the variable changes.
Conducting LGC analysis within a SEM framework allows two LGC models to be combined in order to see how they relate to each other. For example, a question of interest that requires two LGC models is: how does the rate of change of variable X and the rate of change of variable Y relate? Do they both grow in a positive direction together? Does one change in a negative direction while the other changes in a positive direction?
In order to address hypotheses 3a-3c, I examined the relationship of two variables intercepts and slopes. For example, in order to understand how change in employment over time relates to change in emotional well-being, I created a LGC model for each variable separately, and


17
then examined the correlations between the intercepts and slopes. The relationship between the two intercepts (a path in Figure 1) provides information on how the variables are related at the initial time point. The relationship between the two slopes (b path in Figure 1) provides information on how the variables rates of change are related.
LGC modeling within a SEM framework has a few additional advantages. For example, LGC modeling allowed me to examine both variables, each measured four times, within a single model. Analyzing these variables within a single model reduces the chance for type 1 error. Finally, using LGC modeling within a SEM framework allows for missing data within the longitudinal nature of the study to be addressed using Full Information Maximum Likelihood (FIML), which is a state-of-the-art method for handling missing data. For all Hypothesis 3 analyses, I used Mplus statistical software (Version 7.31; Muthen & Muthen, 2014).


18
Figure 1. Latent growth curve model
Hypothesis 4. Employment level will mediate the relationship between disease severity and health-related quality of life. This was analyzed using mediation analyses. The IV for this analysis was disease severity (FEVi or exacerbations), the mediator variable was employment status and the DV was score on the CFQ-R subscales. Eight separate mediation analyses were conducted.
To test the mediation hypothesis, I followed the procedure described in Baron & Kenny (1986). A variable is considered to function as a mediator if the following criteria are met: (1) the independent variable (IV) is a significant predictor of the hypothesized mediator variable (a path


19
in Figure 2), (2) the mediator significantly predicts the dependent variable (DV) (b path in Figure 2), (3) the IV predicts the DV (c path in Figure 2), and (4) when the correlations between the IV and mediator variable and between the mediator variable and DV are controlled, the IV is no longer a significant predictor of the DV (c path in Figure 2). The strongest mediation relationship occurs when the relationship between the IV and DV becomes zero after controlling for the regression paths between the IV and mediator and mediator and DV. More commonly, however, controlling the regression paths leads to a significant decrease in the relationship between the IV and DV (Baron & Kenny, 1986). I tested this mediation hypothesis using a series of regressions. To test the significance of the decrease in the relationship between the IV and the DV, I used the Sobel Test (1982). I used SPSS software to conduct the mediation analyses.
Figure 2. Mediation analyses diagram


20
CHAPTER III RESULTS
Descriptive Statistics
Survey response rates for the time points included in this study (1, 5, 8, and 10) were 93%, 77%, 75%, and 70%, respectively. The mean age of participants at the beginning of the study was 33.0 (SD= 10.7) years, and 42.0% were male. Almost half (46.4%) of the participants were married at the beginning of the study, and 81.1% had some education beyond a high school degree. The mean FEVi at baseline was 54.39 0579=20,43), mean exacerbations over four months was 0.76 (SD= 1.05), and mean clinical weight percentile was 36.13 (SD=21.66). Table 1 displays disease severity descriptive statistics over time.
Table 1. Correlations, Means, and Standard Deviations for Disease Severity
Variable Exacerbations Exacerbations Exacerbations Exacerbations FEV! FEV! FEVi FEVi
T1 T2 T3 T4 T1 T2 T3 T4
Exacerbations T1 -
Exacerbations T2 41** -
Exacerbations T3 24** 29** -
Exacerbations T4 .18* .46** .31** -
FEVlT1 -0.12 -0.10 -0.16* -0.23** -
FEV !_T2 -0.14* -0.10 -0.19* -0.19* £9** -
FEV !_T3 -0.24** -0.13 -0.22** -0.24** .86** 93** -
FEVlT4 -0.26** -0.14 -0.22** -0.28** 82** 90** 94** -
M .76 .68 .80 .78 54.4 52.9 51.3 51.2
SD 1.0 1.2 1.3 1.3 20.4 20.0 19.2 19.7
**p<.01, *p<.05


21
The percent of participants employed part-time or full-time ranged from 51.1% at the initial time point (wave 1) to 30.0% at the final time point included in this study (wave 10). Table 2 shows descriptive statistics for employment status over time in more detail.
Table 2. Percent Employed across Time
Variable Employed T1 n= 307 Employed T2 n= 307 Employed T3 n= 307 Employed T4 n= 307
Employed 51.1% 44.0% 37.1% 30.0%
Not Employed 41.7% 31.9% 30.6% 25.7%
Missing 7.2% 24.1% 32.2% 44.3%
Responses to the questions on the CFQ-R aggregated to form standardized scores from 0 to 100 on each of the domains. At the initial time point, mean scores on the emotion subscale ranged from 66.5 (SD=20.1) to 71.4 (SD= 18.3). For the social subscale, mean scores ranged from 58.1 (SD= 18.1) to 62.7 (Sl)= 18,8), Mean scores for the physical subscale ranged from 52.7 (SD=28.3) to 55.7 (SD=27.6). Finally, mean scores for the role subscale ranged from 70.7 (SD=23.0) to 74.0 (SD=23.3). Table 3 illustrates CFQ-R scores over time in more detail.


22
Table 3. Correlations, Means, and Standard Deviations for CFQ-R
Variable Emot Soc Role Phys Emot Soc Role Phys Emot Soc Role Phys Emot Social Role Phys
T1 T1 T1 T1 T2 T2 T2 T2 T3 T3 T3 T3 T4 T4 T4 T4
n = 291 n = 290 = 291 n = 292 n = 254 n = 250 n = 250 n = 252 n = 233 n = 233 n = 233 n = 232 n = 220 n = 220 n = 220 n = 220
Emot T1 -
SocTl .56** -
RoleTl .56** .53** -
PhysTl .43** .52** .64** -
Emot T2 .58** 39** 44** 24** -
Soc_T2 49** *70** .38** 45** 4g** -
Role_T2 39** 40** .56** .46** 47** 48** -
Phys_T2 .28** 28** 4^** 20** 29** 47** 59** -
Emot T3 .64** 45** .46** 27** .67** .43** 44** .25** -
Soc_T3 39** .64** .34** 40** 37** 73** .35** .35** 51** -
Role_T3 .38** 39** 54** .38** .38** .43** .56** 37** 59** .52** -
Phys_T3 .25** .30** 39** .66** 2i** .36** .43** 74** .36** 48** .58** -
Emot T4 .52** 42** 40** 29** 57** 39** 40** 28** .63** .50** .46** .35** -
Soc_T4 45** .63** 39** .43** 44** .69** 42** 41** .50** 74** 45** 40** .61** -
Role_T4 40** .43** .53** 45** .35** 39** .56** 40** 47** 44** .62** .50** 57** .60** -
Phys_T4 .34** .38** 42** .69** 22** .35** .50** .69** .34** .41* 41** 74** 49** .56** .65** -
M 66.49 62.66 72.07 53.53 71.40 61.3 73.4 54.5 70.4 59.9 74.0 55.7 67.4 58.1 70.7 52.7
SD 20.1 18.8 24.0 26.6 18.3 18.7 22.4 28.0 19.5 19.4 23.3 27.6 20.4 18.8 23.0 28.3
Note. T1 = Baseline; T2 = Time Point 2; T3 = Time Point 3; T4 = Time Point 4; Emot = CFQ-R Emotion scale; Soc = CFQ-R Social scale; Role = CFQ-R Role scale; Phy = CFQ-R Physical scale. **p < .001.


23
Hypothesis 1
My first hypothesis was that those who were employed at the baseline time point would score significantly higher on the emotional, social, role, and physical health-related quality of life domains compared to those who were unemployed at the baseline time point. Four one-way analyses of variance (ANOVAs) were conducted to examine the relationship between employment status and health-related quality of life. Employment status (yes/no) was the independent (grouping) variable for each of the ANVOAs. Four domains of health-related quality of life (emotional, social, role, physical) were the outcome variables of interest. Demographic and clinical covariates were included in the analysis. More specifically, FEVi and exacerbations were included as clinical covariates; age, gender, marital status, and educational attainment were included as demographic covariates. Results of the analyses are presented in Tables 4-7.
Table 4, Employment Status and Emotional HRQoL ANOVA at baseline___________________
Variable df F p value
Emotion 1,282 23.0 <001
FEVi 1,242 2.8 .096
Exacerbations 1,271 16.3 <001
Age 1,283 0.27 .60
Male 1,282 5.1 .024
Married 1,282 0.26 .61
Post High School Education 1,282 8.3 .004
Table 5. Employment Status and Social HRQoL ANOVA at baseline
Variable df F p value


24
Social 1,281 17.1 <001
FEVi 1,242 2.8 .096
Exacerbations 1,271 16.3 <001
Age 1,283 0.27 .60
Male 1,282 5.13 .024
Married 1,282 0.26 .61
Post High School Education 1,282 8.3 .004
Table 6. Employment Status and Role HRQoL ANOVA at baseline
Variable df F p value
Role 1,282 45.18 <001
FEVi 1,242 2.8 .10
Exacerbations 1,271 16.3 <001
Age 1,283 0.27 .60
Male 1,282 5.13 .024
Married 1,282 0.26 .61
Post High School Education 1,282 8.3 .004


25
Table 7. Employment Status and Physical HRQoL ANOVA at baseline
Variable df F p value
Physical 1,283 32.43 <001
FEVi 1,242 2.8 .096
Exacerbations 1,271 16.3 <001
Age 1,283 0.27 .604
Male 1,282 5.13 .024
Married 1,282 0.26 .61
Post High School Education 1,282 8.3 .004
The four one-way ANOVAs confirmed that individuals who were employed did indeed score higher on all four HRQoL domains (emotion, social, role and physical) included in the analysis (all /K.001) even when controlling for clinical and demographic covariates, which suggests that employment positively impacts several domains of self-reported health-related quality of life in individuals with CF. In addition, there were significant differences between the two groups in number of pulmonary exacerbations, gender, and educational attainment. Specifically, those in the employed group had less exacerbations, more males, and more individuals with some education beyond high school.
Hypothesis 2
My second hypothesis was that greater disease severity (lower FEVi and more pulmonary exacerbations) would be negatively correlated with employment. A single binary logistic regression analysis was conducted to examine the relationship between the two measures of disease severity (FEVi and exacerbations) and employment status. Both


26
measures of disease severity, as well as demographic covariates, were included in the analysis. Results are summarized in the Table 8.
Table 8, Logistic Regression at baseline with Dependent Variable= Employed (Yes/No)
Variable B Wald p-v alue Odds Ratio
FEVi 0.007 0.874 0.350 1.007
Exacerbations -0.421 8.957 0.003 0.657
Age 0.001 0.006 0.936 1.001
Male 0.710 5.946 0.015 2.034
Married -0.139 0.205 0.650 0.870
Post HS Education 0.941 5.906 0.015 2.563
Results of the binary logistic regression showed that number of exacerbations was indeed a significant predictor of employment status (p=.003); FEVi, however, was not a significant predictor of employment status (p=.35). Being male and having any education beyond high school were also significant predictors of employment status (p=.02 for both). Odds ratios provided further information on the strength of the relationship of these predictors of employment. For every additional pulmonary exacerbation, likelihood of employment decreased (OR=0.66). Males were approximately two times more likely to be employed than females (OR=2.03), and those with some education beyond high school were 2.5 times more likely to be employed than those without one (OR=2.56).
Hypothesis 3
My third hypothesis had several components. First, I predicted that a change in disease severity over time would be correlated with a change in employment status over time.


27
Specifically, a positive slope for exacerbations or a negative slope for FEVi would be correlated with a negative slope for employment status. I hypothesized that the sicker patients became, the less likely they would be employed. Second, I hypothesized that change in employment level over time would be correlated with change in health-related quality of life. Specifically, a declining employment status slope would be correlated with declining slopes for each of the four health-related quality of life domains (emotion, social, role, physical). Finally, I hypothesized that disease severity and health-related quality of life would change together over time. Specifically, I hypothesized that increasing disease severity, as measured by a positive slope for exacerbations or a negative slope for FEVi, would be associated with declining health-related quality of life. A significant correlation between the intercepts of two latent growth curve analyses (the a path in Figure 1) indicates that the variables are correlated at the initial time point. A significant correlation of the slopes signifies that the variables change together over time (the b path in Figure 1).
Fourteen separate latent growth curve analyses were conducted in order to examine the relationships of both measures of disease severity with each health-related quality of life domain (8 in total) and employment status (2 in total) as well as to examine the relationship of employment status with each health-related quality of life domain (4 in total). Results are presented in Table 9 below.


28
Table 9. Latent growth curve analysis standardized correlations
Variable a path (Intercept) b path (Slope)
Employment Status
Emotion .44 (p< 001) .80 (p=. 10)
Social .35 (p<001) .23 (p=34)
Role .58 (p<001) .48 (p=. 11)
Physical .46 (p<001) .32 (p=.24)
FEVi .17 (p=.03) .22 (p=.33)
Exacerbations -.38 (p<001) -.53 (p=.02)
Exacerbations
Emotion -.32 (p<001) -1.31 (p=. 10)
Social -.18 (p=.05) -.49 (p=.01)
Role -.52 (p<001) -.98 (p=.02)
Physical -.31 (p<001) -.56 (p=.06)
FEVi
Emotion .04 (p=.59) .75 (p=.29)
Social .15 (p=. 04) .15 (p=51)
Role .22 (p=.005) .58 (p=. 13)
Physical .51 (p<001) .89 (p=001)
Results of the intercept correlations (a paths) indicated that initial level of employment was significantly positively correlated with each health-related quality of life domain (all p<.001), and FEVi (p=.03). Initial level of employment was negatively correlated with


29
exacerbations (/;< 001) in the negative direction, indicating that those with more exacerbations were less likely to be employed at the baseline time point. Initial number of exacerbations was also negatively correlated with initial levels of all four health-related quality of life domains (all p<.001), except social (p=.05). Finally, initial level of FEVi was significantly positively correlated with three of the four health-related quality of life domains: social (p=.04), role (p=.005), and physical (/;< 001), Initial FEVi levels were not correlated with emotional health-related quality of life.
Results of the slope correlations (b paths) were less consistent. The slope of exacerbations and employment status was significantly negatively correlated (p=.02) such that as exacerbations increased, employment decreased. The slope of employment status only approached a positive significant correlation with the slope of emotional health-related quality of life (p=. 10). The correlations between slopes of exacerbations and health-related quality of life, however, were more consistent. The correlation between exacerbations and role functioning was significant (p=.02) in the negative direction, which indicates that as exacerbations increase, self-reported role health-related quality of life declines. The correlations between the exacerbations slope and emotion, social, and physical functioning slopes all approached significance (p=. 10, .07, and .06, respectively) and were all in the negative direction. Finally, the slope of FEVi was significantly positively correlated with the slope of physical health-related quality of life (p<. 001), indicating that as FEVi declines over time, so does self-reported physical health-related quality of life.
Hypothesis 4
I hypothesized that employment status would mediate the relationship between disease severity and health-related quality of life. Because this study examined two measures of


30
disease severity and four aspects of HRQoL, I proposed conducting eight separate mediation analyses according to the steps outlined in Baron & Kenny and using the Sobel Test (1982) to determine if the mediation effect was significant. Two markers of disease severity (FEVi and exacerbations) each served as the IV for four mediation analyses. The four domains of health-related quality of life (emotion, social, role, physical) each served as the DV for two mediation analyses. Employment status, measured dichotomously, served as the mediator variable for all eight analyses (see Figure 2).
The IVs were significant predictors of the DVs in all but one analysis (FEVi and emotion). For the analysis in which FEVi served as the IV and emotional health-related quality of life served as the DV, the relationship only approached significance (p=.08). In all other analyses, the relationship between the IV and DV (path c) was significant (all p<.05). The relationship between the mediator and DV (path b) was significant (p<.05) in all eight analyses. The relationship between the IV and mediator (path a), however, was only significant (p<.05) when using exacerbations as the IV. Path a only approached significance (p=. 10) when FEVi served as the IV, meaning that the prerequisites for mediation were not met in the analyses that used FEVi as the IV. Therefore, only the results for the four analyses in which exacerbations served as the IV are presented in Table 10.
Table 10. Results of mediation analysis with exacerbations as independent variable__________
Emotion Social Role Physical
a -0.47** -0.47** -0.47** -0.47**
b 9.58** 8.26** 14.54** 14.88**
c -4.09** -2.44* -8.29** -6.15**
c -2.99* -1.5 -6.64** -4.48*
Sobel test statistic -2.75** -2.62** -3.11** -2.96**
Percent mediated 17.0% 31.5% 7.8% 12.6%
Note: *p<.05. **/K.01.


31
Results of the mediation analyses using exacerbations as the independent variable were consistent with my hypothesis. Results of the Sobel test indicate that employment status is indeed a mediator of the relationship between exacerbations and the four domains of health-related quality of life I used in this study (emotion, social, role, and physical). In all four analyses, the strength of the relationship between the IV and DV was significantly reduced (all p<.01) when the mediator (employment status) was added to the model. When examining social health-related quality of life as the DV, the relationship between the IV and DV became non-significant with the inclusion of the mediator. However, the /i-value for the c path was largest for social health-related quality of life, so this finding may not be meaningful.


32
CHAPTER IV DISCUSSION
Cystic fibrosis was previously considered a childhood disease, because patients did not commonly survive long enough to reach adulthood. However, medical advances have greatly improved the prognosis for patients with CF. Adults with CF (i.e., ages 18 years and older) now make up almost 50% of the CF population, which is a drastic increase from 1986 when only 29.2% of people with CF were 18 years or older (CF Foundation Report, 2013).
The increased lifespan of adults with CF has led to an increasing number of patients with CF pursuing developmentally appropriate adult activities such as higher education and work (Havermans, Colpaert, Vanharen, and Dupont, 2009). However, because the adult survival rate shifts are so recent, there is little research to date on how the disease influences patients daily tasks of adult living and health-related quality of life. This study sought to expand our knowledge of employment in adults with CF, and explore how employment relates to disease severity and health-related quality of life.
Based on previous work, my first hypothesis was that those who were employed at the baseline time point would score higher on the four health-related quality of life scales included in this study (emotional well-being, social well-being, role functioning, and physical functioning). Results of the four ANVOAs confirmed this hypothesis. Emotion, social, role and physical health-related quality of life scores of those who were employed and those who were not employed were significantly different (all /K.001). These health-related quality of life domains were all significant even when controlling for demographic and clinical covariates, which suggests that work may provide a unique, positive contribution to an individuals health-related quality of life.


33
Additionally, gender, education, and pulmonary exacerbations were significantly different between those who were employed and those who were not employed. The employed group included more males, more individuals with some education beyond high school, and individuals with fewer exacerbations. Interestingly, although FEVi and exacerbations are both measures of disease severity, FEVi scores did not differ between the employed and unemployed groups. This suggests that perhaps exacerbations are more disruptive to maintaining a job than a chronically low FEVi score. Exacerbations require either hospitalization or home IV treatment, which may make it difficult to fulfill ones occupational responsibilities. This issue was further examined by hypothesis two (see below). Also of note, education level was significantly different between groups, suggesting that educational attainment may in part predict ones ability to maintain employment as a CF patient. This finding suggests it is important to pay attention to educational goals in the face of childhood illness.
To address my second hypothesis, I used logistic regression to further understand factors that predict employment status in adults with CF. Interestingly, exacerbations was a significant predictor with an odds ratio of 0.66, even when controlling for FEVi and demographic covariates. FEVi, however, was not a significant predictor when controlling for exacerbations and demographic covariates. In fact, FEVi had an odds ratio of nearly 1.0(OR= 1.007), suggesting that it did not have much influence on the likelihood of whether one would be employed. Again, since exacerbations require hospitalization or home IV treatment, and patients who suffer from more exacerbations may find themselves unable to fulfill their duties at work due the disruptive treatment required for these episodes. This suggests that exacerbations, rather than FEVi, may be a superior measure for identifying individuals with CF who are at risk of being unable to maintain their job. Given that FEVi is the most


34
commonly used marker of disease severity in individuals with CF and most studies examining HRQoL, employment and disease severity have used FEVi as a marker of disease severity, this finding may have important implications for future work in this field.
Being male and having some education beyond high school were also associated with an increased likelihood of being employed. Specifically, males were approximately two times more likely to be employed than females (OR=2.03), even when controlling for other demographic and clinical covariates. Individuals with a high school diploma were 2.5 times more likely to be employed than those without one even when controlling for other demographic and clinical covariates. These findings, like those from the first hypothesis, suggest that having some education beyond high school is a protective factor in maintaining employment for individuals with CF.
One reason that educational attainment may be such a strong predictor of employment status is that CF patients may have a particularly difficult time maintaining jobs that require physical labor (Burker et al., 2004; Burker et al., 2005; Taylor-Robinson, 2013). In one study that compared individuals with CF who were employed to those who were not employed (Burker, 2005), all but 6% of patients currently employed were working in skilled or semiskilled labor jobs. Further, those who were not working at the time of data collection, but had worked in the past, reported working unskilled jobs at significantly higher rates than those who were employed at the time of data collection (Burker, 2005).
Results from the latent growth curve (LGC) analysis conducted to address hypothesis 3 provided information on the longitudinal relationship of the variables of interest, which has not yet been studied. Results of LGC analysis showed that the correlations between intercepts were significant between employment and all four health-related quality of life domains, as


35
well as between exacerbations and all four health-related quality of life domains. These results are consistent with those found in by analyses that addressed hypotheses 1 and 2. Interestingly, the intercept variable for FEVi was significantly correlated with three domains of health-related quality of life (social, role, and physical). This finding may initially seem to contradict the results from the analyses that addressed hypotheses 1 and 2. The results from analyses 1 and 2 suggested that FEVli is not associated with being employed or is not a good predictor of employment. The reason for what may initially seem like contradictory results is most likely due to the fact that exacerbations were not controlled for in the latent growth curve analysis. They were, however, controlled for in the analyses that addressed hypotheses 1 and 2. Exacerbations and FEVi are highly correlated, and therefore when they are both entered in a model, FEVi is no longer a significant predictor of health-related quality of life. However, without exacerbations in the model, FEVi was indeed correlated with several health-related quality of life domains at the initial time point.
The correlations between the slopes were less consistent. Only the paths between the slopes of exacerbations and each of the four domains of health-related quality of life were consistently significant, and several of them only approached significance (p values greater than .05, but equal to or less than .10). This suggests that perhaps self-reported health-related quality of life decreases as number of exacerbations changes over time. The path between the slopes of FEVi and physical health-related quality of life was significant (p<. 001), indicating that as FEVi changes over time, self-reported physical health-related quality of life changes as well. The path between the slopes of employment and emotional health-related quality of life approached significance (p=. 10), suggesting that as employment status changes over


36
time, so does emotional well-being. Because many of the correlations approached significance, further studies with more participants and more time points are needed to confirm this relationship.
Results of the mediation analysis for hypothesis 4 provide additional knowledge on how these variables relate to each other. Results of the Sobel test indicate that employment status is indeed a mediator of the relationship between exacerbations and the four domains of health-related quality of life used in this study (emotion, social, role, and physical). In all four analyses, the strength of the relationship between the IV and DV was significantly reduced (all p<.01) when the mediator (employment status) was added to the model. Between 17.8% and 31.5% of the effect of exacerbations on the health-related quality of life domains was mediated by employment status. When examining social health-related quality of life as the DV, the relationship between the IV and DV became non-significant with the inclusion of the mediator. These results suggest that at least part of the relationship between exacerbations and self-reported health-related quality of life is accounted for by the fact that exacerbations disrupt ones ability to work.
Implications
Despite how often many people complain about work, it plays an important role in our lives much beyond a source of income and health care. Work can make individuals feel like productive members of society by assuming responsibility, completing tasks, and helping others (Freedman & Fesko, 1996; Johnson et al., 2004). Employment can also become part of ones identity and thus serve to increase ones self-esteem because (Miller & Dishon, 2006). According to Eriksons theory of psychosocial development, work also plays a key role in psychosocial development.


37
Erikson divides psychosocial development into eight stages, each characterized with a developmental conflict or crisis that must be resolved. He theorizes that failure to resolve one conflict may lead to difficulties with future conflicts and/or developmental tasks. For example, failure to make occupation commitments (Munley, 1977, p.262) in adolescence may lead to poor ego formation and social connections in young adulthood. Munley (1977) discusses Eriksons theory within the context of vocational development and highlights that career development is integrated with several of Eriksons stages, including developing a sense of initiative, industry, identity, generativity, and integrity. According to Erikson, work is an integral piece of several developmental stages and failure to successfully resolve these developmental crises will negatively impact the individual in the future.
The findings from this study are consistent with Eriksons theory of psychosocial development. Results of this study showed that those who were employed reported higher health-related quality of life than those who were not employed even when controlling for clinical and demographic covariates. Eriksons theory highlights the important role work plays in an individuals psychosocial development and this study highlights the important role work plays in the related concept of well-being or health-related quality of life. It is important for CF clinicians to be made aware of these findings, especially when discussing issues such as disability with patients. Results of this study suggest that the relationship between disease severity, work and health-related quality of life may not be that simple, and work may have a positive impact on ones life.
Previous research suggests that CF affects many patients ability to work (Hogg et al., 2007; Laborde-Casterot et al., 2012). This study further examined which factors may impact an adult with CFs ability to work. Perhaps the most important was education. Those with


38
any education beyond high school were more likely to be employed than those without. This is perhaps due to the fact that jobs that require higher levels of education tend to (1) require less physical exertion, and (2) be more flexible with regards to working from home or adjusting ones work schedule. These findings will hopefully provide a basis for future interventions, such as educational and vocational counseling, aimed at increasing CF patients ability to work.
This study provides important information on regarding CF patients who may be at higher risk for unemployment. Those with more exacerbations appear to have more difficulty maintain employment than those who experience fewer exacerbations. Lung functioning, as measured by FEVi, does not seem to impact ones ability to work in the same way as exacerbations. This has important implications for future work with this population, as FEVi is the most commonly used marker of disease severity in CF patients. Results from this study show that perhaps number of pulmonary exacerbations is a superior marker of disease severity when trying to identify patients at risk for unemployment who may benefit from vocational counseling.
Limitations
One limitation in this study is that some of the variables of interest are not manifest variables, but rather hypothetical constructs, and therefore there is inherent error in measurement. Two of our manifest variables are physiological measures (FEVi and number of pulmonary exacerbations), and therefore likely to be accurate with the exception of systematic error due to the measurement tool or random error. Self-report measures, however, may lead to under or over reporting. The most complicated variable of interest is health-related quality of life because it is a hypothetical construct that was measured using a self-report survey.


39
Studies conducted on the psychometric properties of the CFQ-R have demonstrated strong reliability and validity (Quittner et al., 2012). However, hypothetical constructs are more difficult to measure than manifest variables, and this is important to consider when interpreting the results of this study.
There is also a potential issue of external validity in this study. The study was conducted at ten CF centers across the country. Although this provides a certain degree of geographic diversity, our results may not generalize to patients who live in rural communities and do not receive their care at a large CF center, which are usually located at academic medical centers in large cities. Additionally, this project began almost ten years ago and due to medical advances, an individuals experience with CF today could be different than an individuals experience ten years ago. However, the medical staff and lead investigators of this study are confident that the variables measured in this study remain relevant.
Future Directions
Several slope correlations in the LGC analysis approached significance. Thus, it is necessary to repeat this study, or a similar one, with a larger sample, more data points, and/or a sicker population in order to confirm the findings. A larger sample would increase power, more data points would provide better information on the shape of the slopes, and a sicker population should provide more variability in the measures.
If a study similar to this one is repeated, it would be interesting to consider a possible cohort effect. The adults enrolled in PAC-CF were between the ages of 18 and 65 when the study began in 2005. This means that the youngest subject was bom in 1987, and the majority of participants were born well before that, considering the average age at time of enrollment was 33 years. Given that children bom with CF in the 1980s were not even expected to


40
reach adulthood, there may be important differences between the PAC-CF cohort and a cohort of adults studied in the future. An individual growing up not expecting to live much past their 18th birthday may approach school, higher education, and vocational choices differently than someone who grows up expecting to live well into adulthood. Additionally, medical advances could have important implications educationally, as patients may now be able to miss less school and live a more normal life.
Education was found to be a strong predictor of employment in adults with CF. Additionally, previous studies have suggested that perhaps it is easier for educated, white-collar CF workers to maintain employment than those in blue-collar jobs (Burker et al., 2005). This is logical given the physical effects of the disease, but requires more research to understand. The present study found education to be an important predictor of employment, suggesting that education perhaps has a protective factor. In order to better understand this finding, it could be beneficial to compare the education odds ratio in our CF sample to the education odds ratio in the general public. It is well-known that education is associated with employment in the general population. However, the question of interest is if education is even more important in helping those with CF maintain employment due to their physical limitations and the physical exertion often required in an unskilled or blue-collar job.
In conclusion, this study contributed to our understanding of the relationships among disease severity, employment and health-related quality of life in individuals with CF and was the first study to examine the variables longitudinally. However, more research is still needed on this topic in order for comprehensive CF care teams to best serve the relatively new, increasingly large cohort of adults with CF.


41
REFERENCES
Abbot, J., Hart, A., Morton, A., Gee, L., & Conway, S. (2008). Health-related health-related quality of life in adults with cystic fibrosis: The role of coping. Journal of Psychosomatic Research, 64, 149-157.
Anderson, D.L., Flume, P.A., & Hardy, K.A. (2001). Psychological functioning in adults with cystic ffibrosis. Chest, 119. 1079-1084.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistic considerations. Journal of Personality and Social Psychology, 51, 1173-1182.
Blair, C., Cull, A., Freeman, C.P. (1994). Psychosocial functioning of young adults with cystic fibrosis and their families. Thorax, 49, 798-202.
Burker, E.J., Sedway, J., & Carone, S. (2004). Psychological and educational factors: Better predictors of work status than FEVi in adults with cystic fibrosis. Pediatric Pulmonology, 38, 413-418.
Burker, E.J., Sedway, J., Carone, S., Trombley, C., & Yeatts, B.P. (2005). Vocational attainment of adults with CF: Success in the face of adversity. Journal of Cystic Rehabilitation, 77(2), 22-27.
Congleton, J., Hodson, M.E., & Duncan-Skingle, F. (1996). Quality of life in adults with cystic fibrosis. Thorax, 51, 936-940.
Corrigan, J.D., Bogner, J.A., Mysiw, J.W., Clinchot, D., Fugate, L. (2001). Life satisfaction after traumatic brain injury. Journal of Head Trauma Rehabilitation, 16, 543-555.
Cystic Fibrosis Foundation. (2013). Cystic Fibrosis Foundation Annual Report. Retrieved from https://www.cff.org/About-Us/Assets/2013-Annual-Report/
Cystic Fibrosis Foundation. (2014). Cystic Fibrosis Foundation Annual Report. Retrieved from https://www.cff.org/About-Us/Assets/2014-Annual-Report/
de Jong, W., Kaptein, A.A., van der Schans, C.P., Mannes, G.P.M., van Aalderen, W.M.C., Grevink, R.G., Koeter, G. H. (1997). Quality of life in patients with cystic fibrosis. Pediatric Pulmonology, 23, 95-100.
Dill, E.J., Dawson, R., Sellers, D.E., Robinson, W.M., Sawicki, G.S. (2013). Longitudinal trends in health-related health-related quality of life in adults with cystic fibrosis. Chest, 144(3), 981-989.
Freedman, R.I., Fesko, S.L. (1996). The meaning of work in the lives of people with significant disabilities: Consumer and family perspectives. Journal of Rehabilitation, 62(3), 49-55.


42
Gee, L., Abbott, J., Hart, A., Conway, S.P., Etherington, C., Webb, A.K. (2005). Associations between clinical variables and health-related quality of life in adults with cystic fibrosis. Journal of Cystic Fibrosis, 4, 59-66.
Gillen, M., Lallas, D., Brown, C., Yelin, E., & Blanc, P. (1995). Work disability in adults with cystic fibrosis. American Journal of Respiratory and Critical Care Medicine, 752,153-156.
Goldberg, R.T., Isralsky, M., & Shwachmann, H. (1985). Prediction of rehabilitation status of young adults with cystic fibrosis. Archives of Physical Medicine and Rehabilitation, 66, 492-495.
Havermans, T., Colpaert, K., Dupont, L.J. (2008). Quality of life in patients with cystic fibrosis: Associations with anxiety and depression. Journal of Cystic Fibrosis, 7, 581-584.
Havermans T., Colpaert, K., Vanharen, L., and Dupont, L. J. (2009). Health related health-related quality of life in cystic fibrosis: To work or not to work? Journal of Cystic Fibrosis, 8, 218-223.
Hogg, M., Braithwaite, M., Bailey, M., Kotsimbos, T., & Wilson, J.W. (2007). Work disability in adults with cystic fibrosis and its relationship to health-related quality of life. Journal of Cystic Fibrosis, 6, 223-227.
Johnson, K.L., Yorkston, K.M., Klasner, E.R., Kuehn, C.M., Johnson, E., Amtmann, D.
(2004). The Cost and benefits of employment: A qualitative study of experiences of persons with multiple sclerosis. Archives of Physical Medicine and Rehabilitation,
85, 201-209.
Laborde-Casterot, H., Donnay, C., Chapron, J., Burgel, P.R., Kanaan, R., Honore, I.,...Hubert, D. (2012). Employment and work disability in adults with cystic fibrosis. Journal of Cystic Fibrosis, 11, 137-143.
Lawler, R.H., Nakielny, W., & Wright, N.A. (1966). Psychological implications of Cystic Fibrosis. Canadian Medical Association Journal, 94, 1043-1046.
Little, T. D. (2013). Longitudinal structural equation modeling: Methodology in the social sciences. New York: Guilford Press.
Miller, A. & Dishon, S. (2006). Health-related health-related quality of life in Multiple Sclerosis: The impact of disability, gender and employment status. Quality of Life Research, 15, 259-271.
Munley, P.H. (1977). Eriksons theory of psychosocial development and career development. Journal of Vocational Behavior, 10, 261-269.


43
Muthen & Muthen. (2014). MPLUS Version 7.31. [Computer Software], Los Angeles, CA.
Orenstein, D.M., Nixon, P.A., Ross, E.A., & Kaplan, R.M. (1989). The quality of well-being in cystic fibrosis. Thorax, 95, 344-347.
Pearson, D.A., Pumariega, A. J., Seilheimer, D. K. (1991). The development of psychiatric symptomology in patients with Cystic Fibrosis. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 290-297.
Pfeffer, P.E., Pfeffer, J.M., Hodson, M.E. (2003). The psychosocial and psychiatric side of cystic fibrosis in adolescents and adults. Journal of Cystic Fibrosis, 2, 61-68.
Platten, M.J., Newman, E., Quayle, E. (2013). Self-Esteem and its relationship to mental
health and health-related quality of life in adults with cystic fibrosis. Journal of Clinical Psychology in Medical Settings, 20, 392-399.
Preacher, K.J., Wichman, A.L., MacCallum, R.C., and Briggs, N.E. (2008). Latent Growth Curve Modeling. Thousand Oaks, California: Sage Publications.
Quittner, A.L., Sawicki, G.S., McMullen, A., Rasouliyan, L., Pasta, D.J., Yegin, A., Constan, M.W. (2012). Psychometric evaluation of the Cystic Fibrosis Questionnaire-Revised in a national sample. Quality of Life Research, 21, 1267-1278.
Sawicki, G.S., Sellers, D.E., & Robinson, W.M. (2009). High treatment burden in adults with cystic fibrosis: Challenges to disease self-management. Journal of Cystic Fibrosis, 8, 91-96.
Shepherd, S.L., Hovell, M.F., Harwood, I.R, Granger, L.E., Hofstetter, C.R., Molgaard, C., & Kaplan, R.M. (1990). A comparative study of the psychosocial assets of adults with cystic fibrosis and their health peers. Chest, 97, 1310-1316.
Sobel, M.E. (1982). Asymptomatic confidence intervals for indirect effects in structural
equation models. In S. Leinhart (Ed.), Sociological methodology 1982 (pp.290-312). San Francisco: Jossey-Bass.
Sokol, J.T. (2009). Identify development throughout the lifetime: An examination of Erik-sonian theory. Graduate Journal of Counseling Psychology, 7(2), 139-148.
Staab, D., Wenninger, N.G., Rupprath, K., Bisson, S., Trettin, M., Paul, K.D., ... and Wahn, U. (1998). Quality of life in patients with cystic fibrosis and their parents: what is important besides disease severity? Thorax, 53, 727-731.
Taylor-Robinson, D.C., Smyth, R., Diggle, P.J., & Whitehead, M. (2013). A longitudinal
study of the impact of social deprivation and disease severity on employment status in the UK cystic fibrosis population. PLoS One, 5(8), e73322.


44
Walters, S., Britton, J., & Hodson, M.E. (1993). Demographic and social characteristics of adults with cystic fibrosis in the United Kingdom. British Medical Journal, 306. 549-552.
Whitley, B.E. & Kite, M.E. (2013). Principles of research in behavioral science. New York: Routledge.
Widerman, E. (2003). Knowledge, interests, and education needs of adults diagnosed with cystic fibrosis after age 18. Journal of Cystic Fibrosis, 2, 97-104.


Full Text

PAGE 1

! LONGITUDINAL ASSOCIATIONS AMONG DISEASE SEVERITY, EMPLOYMENT AND QUALITY OF LIFE IN ADULTS WITH CYSTIC FIBROSIS by HOLLY O'DONNELL B.A., Middlebury College, 2012 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Arts Clinical Health Psychology Program 2016

PAGE 2

ii This thesis for the Master of Arts degree by Holly O'Donnell h as been approved for the Clinical Health Psychology program by Edward Dill, Chair Barbara Walker Walter Robinson Date: December 17 2016

PAGE 3

iii O'Donnell, Holly, M.A., Clinical Health Psychology Longitudinal Associations Among Disease Severity, Employment, and Quality of Life in Adults with Cystic Fibrosis Thesis directed by Assistant Professor Edward Dill ABSTRACT Currently, little is known about health related quality of life in adults with Cystic Fibrosis (CF) Only recently a milestone was achieved such that adults (18+ years) now represent more than half of the population with CF in the United States. Previous research has shown that CF affects an individual's ability to work and that both cystic fibrosis and employment affect health related quality of l ife Th e sample for this study consist ed of 303 adults with CF who received their care at one of ten medical centers a cross the United States. Using structural equation m odeling (SEM), a latent growth curve (LGC) analysis allow ed us to better understand th e longitudinal relationship s among disease severity, employment and health related quality of life Specifically, this study examined how employment and health related quality of life change as a result o f declining health in adults with CF Analyses used data collected at four time points over the course of 41 months. Possible implications include a greater understanding of factors that influence health related quality of life in adults with CF and more specifically, risk factors that negatively impact he alth related quality of life These findings are important to share with CF care team members in order to provide high quality psychosocial care. The form and content of this abstract are approved. I recommend its publication. Approved: Edward Dill

PAGE 4

iv T ABLE OF CONTENTS CHAPTER I. INTRODUC TION ................................ ................................ ............................. 1 Hypothesis1 ................................ ................................ ................................ ........ 7 Hypothesis 2 ................................ ................................ ................................ ....... 8 Hypothesis 3 ................................ ................................ ................................ ....... 8 Hypothesis 4 ................................ ................................ ................................ ....... 9 II. METHOD ................................ ................................ ................................ ........ 10 Sample ................................ ................................ ................................ .............. 10 Procedure ................................ ................................ ................................ ......... 11 Instruments ................................ ................................ ................................ ....... 11 Data Analysis ................................ ................................ ................................ ... 14 III. RESULTS ................................ ................................ ................................ ........ 20 Descriptive Statistics ................................ ................................ ........................ 20 Hypothesis 1 ................................ ................................ ................................ ..... 23 Hypothesis 2 ................................ ................................ ................................ ..... 25 Hypothesis 3 ................................ ................................ ................................ ..... 2 6 Hypothesis 4 ................................ ................................ ................................ ..... 29

PAGE 5

v IV. DISCUSSION ................................ ................................ ................................ .. 32 Implications ................................ ................................ ................................ ...... 36 Limitations ................................ ................................ ................................ ....... 38 Future Directions ................................ ................................ ............................. 39 REFERENCES ................................ ................................ ................................ ............ 41

PAGE 6

1 CHAPTER I INTRODUCTION Cystic f ibrosis (CF) is a life threatening genetic disease that affects approximately 30,000 individuals in the United States and 70,000 individuals worldwide (CF Foundation, 2014). It is caused by mutations in both copies of the cystic fibrosis transmembrane cond uctance regulator (CFTR) gene, which interferes with production of the CFTR protein. In patients with clinical CF, either little to no CFTR is produced or a defective protein is produced. The most common clinical pattern in patients with CF is the developm ent of chronic obstructive lung disease accompanied by pancreatic insufficiency, with some patients also developing diabetes, joint and bone disease, and severe sinus disease There has also been documentation of a host of other symptoms and conditions, su ch as diabetes and sleep disturbances. Cystic f ibrosis was previously considered a childhood disease because adults represented a relatively small proportion of the population with CF However, in 2014, a milestone was reached such that over half of tho se with CF in the United States are now adults (CF Foundation Report, 2014) which is a drastic increase from 1986 when only 29.2% of people with CF were 18 years or older (CF Foundation Report, 2013). As of 2014, the median predicted survival age was 39.3 years, which means that 50% of patients with CF are expected to live to 39.3 years or beyond (CF Foundation Report, 2014). The increased lifespan of adults with CF has led to an increasing number of patients with CF pursuing developmentally appropriate adult activities such as higher education and employment (Havermans, Colpaert, Vanharen, and Dupont, 2009). However, potentially because predicted survival age shifts are so recent, there is little research to date on how the disease influences patients' daily tasks of adult living and health related quality of life

PAGE 7

2 Impact of CF on Functioning and Quality of Life There is conflicting evidence as to whether having CF is associated with increased rates of psychological difficulties in adolescence and adulthood (Pfeffer, Pfeffer, Hodson, 2003). Some investigators found elevated rates of psychiatric symptoms in patients with CF (Pearson, Pumariega & Seilheimer, 1991; Law ler, Nakielny & Wright, 1966) whereas other studies (Blair, Cull, & Freeman, 1994; Sheperd et al., 1990 & de Jong et al., 1997) have found no differences between the psychological functioning of young adults with and without CF. These discrepancies may be partially due to methodology; the instruments used to measure psychosocial functioning and rates of psychiatric illness varied widely across the studies and the samples were heterogeneous. For example, one of the studies (Anderson, Flume & Hardy, 2001) us ed a sample that was significantly less sick than those who chose not to participate, which is important considering there is some evidence suggesting psychological and psychosocial functioning of adults with CF is on par with their peers until the disease becomes severe (Pfeffer et a l., 2003; Pearson et al, 1991). Additionally, some of these studies are from Europe and therefore the results may not generalize to a U.S. population, especially because social welfare systems in Europe are very different from those in the U.S. Regardless, careful examination of these studies can provide some insight into the functioning of adults with CF. In contrast with the conflicting evidence regarding links between CF and psychological disorders, there is consistent evid ence suggesting that CF affects health related quality of life especially the physical functioning domain. Although the term health related quality of life lacks a consistent definition across the literature, it is generally understood to be a measure o f how one's medical illness affects one's psychiatric, psychosocial and physical functioning (Pfeffer et al., 2003). Related to this, there is consistent evidence that patients with CF have more

PAGE 8

3 problems in daily living than their healthy peers (Anderson e t al., 2001; Congleton, Hodson & Duncan Skingle, 1996). Disease severity was found to correlate with Health Related Quality of Life (HRQ oL), but correlations were weak (Abbott et al., 2003 ; Dill et al., 2013), suggesting that variables other than disease severity also affect health related quality of life W ork or the ability to maintain employment may be one of these variables. After controlling for medical variables such as disease severity, adults with CF who were working reported a higher health relat ed quality of life on physical functioning, role functioning and social functioning than those who were not working (Havermans et al., 2009). This suggests that employment plays an important role in determining health related quality of life in adults with CF. Additionally, patients who were not working were found to have significantly higher levels of depression than those who were working (Burker et al., 2004). Although depression is not a direct measure of health related quality of life patients with an xiety and depression have been found to have lower scores on certain health related quality of life domains (Havermans, Colpaert, & Dupont, 2008) and health related quality of life encompasses psychological functioning. Overall, adult CF patients experie nce difficulties in many areas of daily living including, but not limited to sleep, home management, mobility, physical recreation, and employment (de Jong et al., 1997). Employment appears to be a consistent challenge for individuals with CF; patients wi th CF are less likely to be employed than their healthy peers (Blair et al., 1994) and report that the disease affects their ability to work (Laborde CastÂŽrot et al., 2012). Some investigators have even found employment to be the area of daily living in wh ich adult patients with CF experience the greatest difficulty (de Jong et al., 1997).

PAGE 9

4 Relationship between Employment and Quality of Life Similar themes have emerged in studies examining how employment affects individuals with illnesses and disabilities across a wide range of diagnoses (Freedman & Fesko, 1996; Johnson et al., 2004; Miller & Dishon, 2006). Financial benefits of employment are important as they allow individuals not only to support themselves but to receive quality health care and insurance through employer health insurance plans. Work has a therapeutic aspect in that it distracts individuals from their illness (Johnson et al., 2004) and makes individuals feel like productive members of society by assuming responsibility, completing tasks, a nd helping others (Freedman & Fesko, 1996; Johnson et al., 2004). Finally, social connections from work are important and valuable (Freedman & Fesko, 1996; Johnson et al., 2004). Employment can also increase one's self esteem in part because it becomes pa rt of one's identity (Miller & Dishon, 2006). These findings are consistent with Erikson's theory of psychosocial development and the importance of identity formation versus role confusion. Erikson emphasized the importance of a meaningful career in the d evelopment of one's identity. According to his theory, it is essential to achieve identity development in order to proceed to further developmental tasks. Those who struggle with identity formation are said to suffer from "role confusion," which can lead t o feelings of confusion and questions regarding their purpose (Sokol, 2009). Thus, according to this theory, individuals who have difficulty obtaining and maintaining employment are at risk for a wide range of psychosocial and developmental issues. Impact of Cystic Fibrosis on Employment Given the importance of work to psychosocial functioning, it is unfortunate that CF can have such a profound effect on one's ability to work. A majority (55% to 68%) of CF patients report that having CF impacts their abili ty to work (Hogg et al., 2007; Laborde CastÂŽrot et al.,

PAGE 10

5 2012). Patients have reported having to quit work due to CF (Hogg et al., 2007), and more CF patients work part time (compa red to full time) than the national average in France (Larbode CastÂŽrot, 2012 ). Despite these trends, half the patients in a sample of CF patients with highly compromised lung function were working, which suggests that employment is valuable and important to these patients (Burker et al., 2004). Several studies have examined risk f actors for difficulties maintaining employment among patients with CF. Educational attainment, socioeconomic status (SES), disease severity and treatment burden are the four most commonly reported risk factors in assessing for unemployment, employment diff iculties, or disability in adults with cystic fibrosis (Laborde CastÂŽrot, 2012; Havermans et al., 2009; Gillen et al., 1995; Taylor Robinson, 2013; Goldberg, 1985; Burker et al., 2004 & Hogg et al., 2007). Lower levels of educational attainment and SES hav e both been associated with lower levels of employment in CF patients (Burker et al., 2004; Taylor Robinson, 2013). Additionally, low SES has been found to amplify the effect that dise ase severity has on employment. F or example, Taylor Robinson et al. (201 3) found that those with poor lung function in the lowest SES group had a 7.7% chance of reduction in employment while those with low lung function in the highest SES group had only a 3% chance of reduction in employment. A few studies have investigated potential reasons that SES and education are positively correlated with employment by examining the types of jobs patients with CF were able to maintain. In one study, all but 6% of employed CF patients were working in skill ed or semi skilled jobs (Burker et al., 2005). Participants in the study who were not working at the time of data collection, but had worked in the past, reported working in a significantly higher percentage of unskilled jobs than those who worked at the t ime of data collection, suggesting that CF patients

PAGE 11

6 who are able to obtain skilled jobs have an easier time maintaining these jobs. Other studies have similarly found greater rates of "white collar jobs" (in contrast to "blue collar" jobs) in CF patient po pulations compared to the general population (Gillen et al., 1995; Laborde CastÂŽrot et al., 2012). This is likely because skilled jobs require less physical exertion than unskilled jobs, and patients with CF are unable to maintain jobs that require heavy p hysical exertion, especially as the disease progresses. Another factor that may contribute to these findings is that skilled jobs often offer more flexibility and are therefore easier to maintain as the disease progresses. The most studied and perhaps m ost interesting risk factor for employment difficulties is disease severity. Several investigators have found that CF patients with more advanced disease and more impaired lung function work less and are at higher risk for being unemployed (Laborde CastÂŽro t, 2012; Havermans et al.,2009; Gillen et al., 1995; Taylor Robinson et al., 2013). However, other studies have found inconsistent results. For example, Burker et al., 2004, did not find a significant difference in lung function between those who were empl oyed and those who were not employed Similarly, Hogg et al., 2007, did not find number of hours worked to be dependen t on lung function. However, number of hospital admissions was a significant predictor of hours worked per week in this study. Therefore, a lthough a significant number of studies have examined the relationship between disease severity and employment, the relationship remains unclear due to inconsistent results This may be partially a result of the heterogeneity of the studies that have exam ined these variables thus far. For example, studies have used several different measures of disease severity and have controlled for various clinical covariates. Finally, the studies are from several different countries, which may influence the results due to diverse economies and social welfare systems.

PAGE 12

7 The Present Study Given the evidence that employment affects health related quality of life in individuals with CF and maintaining employment can be difficult for individuals with CF, it is important to f urther explore the relationship s among disease severity, employment and health related quality of life in adults with CF. Almost all of the resear ch to date regarding disease severity, employment, and health related quality of life in the CF pop ulation ha s been cross sectional Cross sectional studies do not allow examination of the relationship of these variables over time. These are important relationships to understand because disease severity increases over time and we currently know very little regard ing how that change in disease severity may be associated with changes in employment and/or health related quality of life The current study seeks to further our understanding of the relationship s among disease severity, employment and health related qua lity of life at a single time point and to provide novel information on the relationship s among these variables over time. Aim 1 To examine the relationship between employment and health related quality of life in adults with c ystic f ibrosis at a single time point. Hypothesis 1. Employment will be correlated with higher levels of health related quality of life at the baseline time point. Specifically, those who are employed will score significantly higher on the emotional, social, role and physical health related health related quality of life (HRQoL) domains compared to those who are unemployed at the baseline time point. Aim 2. To examine the relationship between disease severity and employment in adults with c ystic f ibrosis at a single time point.

PAGE 13

8 Hypothesis 2 Disease severity will be negatively correlated with employment at the baseline time point. Higher forced expiratory volume in one second (FEV 1 ) will be positively correlated with employment and number of pulmonary exacerbations (defined as ex acerbations requiring home IV treatment or hospitalization) will be negatively correlated with employment. Aim 3 To examine the relationship between disease severity, employment and health related quality of life over time in adults with Cystic Fibrosis Hypothesis 3a Change in disease severity (FEV 1 and exacerbations) over time, measured by its slope, will be correlated with change in employment status over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEV 1 ), employment status will decrease. The relationship between exacerbations and employment status will be negative (i.e., the more exacerbations, the less likely one is to be employed). The relationship between FEV 1 and employment will be positive (i.e., the higher the FEV 1 le vel, the more likely an individual is to be employed). Hypothesis 3b Change in employment level over time, measured by its slope, will be correlated with change in health related quality of life (emotional, social, role functioning and physical function ing domains). Specifically, the slopes of employment and health related quality of life will be positively correlated. Hypothesis 3c Change in disease severity (FEV 1 and exacerbations) over time, meas ured by its slope, will be correlated with change in h ealth related quality of life over time. Specifically, as the disease severity increases over time (positive slope for exacerbations and negative slope for FEV 1 ), health related quality of life will decrease. The relationship between exacerbations and heal th related quality of life will be negative (i.e., the more exacerbations, the lower

PAGE 14

9 one's health related quality of life ). The relationship between FEV 1 and health related quality of life will be positive (i.e., higher FEV 1 levels will be correlated with higher health related quality of life scores). Aim 4 To better understand the mechanism by which disease severity affects HR QoL in adults with CF. Hypothesis 4 Employment level will mediate the relationship between disease severity and health related quality of life In other words, the relationship between disease severity and health related quality of life will be at least partially explained by the relationship between disease severity and employment as well as the relationship between employment a nd health related quality of life Therefore, increased disease severity will predict unemployment, which in turn will predict lower health related quality of life Thus, the overall goal of all four aims is to examine and evaluate the cross sectional an d longitudinal relationship s among disease severity employment, and health related quality of life in adults with cystic f ibrosis.

PAGE 15

10 CHAPTER II METHOD Sample The participants of this study were part of a longitudinal study of adults with cystic fibrosis called the Project on Adult Care in Cystic Fibrosis (PAC CF). The participants were 18 years or older and received CF care at one of ten centers across the United States participating in the study. A stratified sample was used to maximize inclusion of adul ts with relat ively lower probability of five year survival. Adults with a very high probability of five year survival are the least informative as to how health related quality of life changes as the d isease progresses, and a major goal of this study was to examine how health related quality of life changes as the disease progresses and disease severity increases. Study coordinators were provided potential participants probability of five year survival using equations by Liou et al., 2001. All adults with a five year predicted probability of survival less than 0.975 ( n =515) and a randomly selected 25% of adults with a five year predicted probability of survival equal to or greater than 0.975 ( n= 60) were selected to be ap proached to participate in the study. Out of the 575 potential participants, 333 enrolled in the study and 242 either refused or could not be contacted. Compared to those who did not enroll, the participants were more likely to be white, female, and olde r. The participants also had better standardized weight for age s core (i.e., higher weight) and a greater number of pulmonary exacerbations than those who chose not to participate There were 303 participants who completed the first wave in April of 2005. Me an age of participants at the beginning of the study was 32.98 years ( SD= 10.71), 42.0% i dentified as m ale, 3.3% identified as Hispanic/Latino, and 97.0 % identified as White which is consistent with the racial distribution of CF patients There were 185 pa rticipants who participated in the final wave in February of 2009. Of

PAGE 16

11 those available at the beginning of each wave, between 70% and 93% completed and returned the surveys for each wave. Procedure The study was approved by the I nstitutional R eview B oard s at Education Development Center, Inc. in Waltham, Massachusetts as well as the ten medical centers across the country with the participating CF centers. Participants were mailed surveys 11 times over the course of 46 months and returned them in an envelo pe by mail. The data presented in this study are from waves 1, 5, 8, and 10, which took place in April 2005, May 2006, June 2007 and August 2008. These waves were chosen because they are the most equally spaced apart. Having participants respond to the sur veys outside of their clinical visits to their CF center was thought to promote anonymity and honesty, and reduce the possibility of response bias. Clinical data were collected at the time of the participant's routine visits, which typically occurred on a quarterly basis, and were extracted from the CF Foundation patient registry. Therefore, the collection time points of survey data did not necessarily coincide with clinic visits. However, because most adults visited their CF care center on a quarterly bas is, the clinical data were usually collected within three months of each survey. Instruments Demographic information Demographic information such as age, gender, educational attainment, marital status, and income was collected via self report surveys at several waves during the study. Demographic information used in the present study was collected at the initial time point. Disease severity. This study used two measures of disease severity: FEV 1 and number of pulmonary exacerbations. FEV 1 is forced ex piratory volume in one second and is an objective

PAGE 17

12 measure of lung function used as a proxy for disease severity in patients with CF. Higher FEV 1 scores indicate better lung function and a lower disease severity. FEV 1 was collected when the participants vis ited their CF care center for routine clinical visits. Number of pulmonary exacerbations was defined as the number of hospitalizations or home IV treatments between each wave. For the first wave, number of exacerbations in the four months preceding the fir st data collection time point was used. Employment. Employment status was measured via a self report survey that participants completed at home outside of clinic visits. Employment was measured both continuously and dichotomously. In the survey, patients were asked how many hours they worked in the previous week and if this was typical for them. They were also asked to respond yes/no to "Are you currently employed or self employed?" Quality of life Quality of Life was measured using the Cystic Fibrosis Q uestionnaire Revised (CFQ R), which is a disease specific measure of health related health related quality of life for patients with cystic fibrosis. The CFQ R is a revised version of the Cystic Fibrosis Questionnaire (CFQ), which was the first disease spe cific patient reported outcome measure for CF patients and caregivers (Quittner et al., 2012). There are currently four forms available: one for preschool age children, one for school age children, one for teens and adults, and one for parents. This study utilized the teen/adult version because all participants were 18 years or older, and the teen/adult version is designed for patients 14 years and older. The teen/adult version consists of 50 items measured on 4 point Likert scales that include frequency ( always, often, sometimes, never), intensity (a great deal, somewhat, a little, not at all), and true false scales (very true, somewhat true, somewhat false, very false). Responses to the questions are converted to stand-

PAGE 18

13 ardized scores ranging from 0 to 100 in the following domains: physical functioning, body image, eating disturbances, digestive symptoms, respiratory symptoms, weight, health perceptions, vitality, treatment burden, emotional functioning, social functioning, and role. Highe r scores on the CFQ R indicate higher health related quality of life The psychometric properties of the CFQ R were studied in patients enrolled in the Epidemiologic Study of Cystic Fibrosis (ESCF). From this sample, 13,437 CFQ Rs from 7,330 patients were completed by pati ents and parents. According to the CF Foundation Patient Registry t he sample for this study was similar to the U.S. population in terms of demographic and health related variables, which suggests that the results of this study are generalizable to CF pati ents in the United States (Quittner et al., 2012). The CFQ R was found to successfully differentiate between patients who were being seen for "sick" versus "well" visits. With the exception of the Digestion scale, all comparisons were statistically sign ificant, with lower scores reported during "sick" visits (Quittner et al., 2012). Moreover, all scales differed significantly as hypo thesized between disease stage/ severity except for Eating Problems, Role Functioning and Emotional Functioning, which diffe red significantly across all disease state groups except between the normal and mild disease groups. In order to further evaluate the construct validity of the CFQ R, correlations were calculated between CFQ R scales and health measures including FEV 1 % pre dicted, number of pulmonary exacerbations treated with IV antibiotics, and BMI (Quittner et al., 2012). Strong correlations were found between FEV 1 % predicted and number of exacerbations with the Physical Functioning, Role Functioning, Vitality, Health Per ceptions, and Respiratory Symptoms scales. Significant correlations were also found between FEV 1 % predicted and the Social Functioning, Body Image, and Weight scales. Moderate correlations were found between BMI and the Body Image and Weight

PAGE 19

14 scales (Quittn er et al., 2012). Additional studies have shown the CFQ R to correlate with measures of anxiety and depression, which is further evidence for the validity of this measure (Havermans et al., 2008). In terms of reliability, strong internal consistency sugge sted the measur e demonstrated good reliability. Cronbach alpha s for the CFQ R teen/adult version were as follows: Physical Functioning (0.94), Respiratory Symptoms (0.87), Role Functioning (0.83), Eating Problems (0.82), Vitality (0.80), Health Perceptions (0.79), Emotional Functioning (0.77), Digestive Symptoms (0.68), Body Image (0.65), Social Functioning (0.58), and Treatment Burden (0.51). A Cronbach alpha was not calculated for the Weight subscale as it is a single item scale. Therefore, a ll except two subscale s of the teen/adult version of the CFQ R demonstrated adequate internal consist ency using the standard of 0.60 (Quittner et al., 2012). Data Analysis Hypothesis 1 Employment will be associated with higher levels of health related quality of l ife at the baseline time point. Specifically, those who are employed will score significantly higher on the emotional, social, role and physical health related health related quality of life (HRQoL) domains compared to those who are unemployed at the base line time point I addressed this research question using an Analysis of Variance ( ANOVA ) The grouping variable for this analysis was employment status and the dependent variable was the score on the CFQ R subscales. I used SPSS software for this analysi s. Hypothesis 2 Greater disease severity, as measured by lower FEV 1 and a greater number of pulmonary exacerbations, will be negatively correlated with employment at the baseline time point This question was analyzed using binary logistic regres sion. The independent variable (IV)

PAGE 20

15 for this analysis wa s disease severity (measured by FEV 1 and number of p ulmonary exacerbations), and the dependent variable ( DV ) wa s employment status (measured dichotomously) I used SPSS software to conduct this analysis. H ypothesis 3a Change in disease severity (FEV 1 and exacerbations) over time, measured by its slope, will be correlated with change in employment status over time. Specifically, as the disease severity increases over time (positive slope for exacerbations a nd negative slope for FEV 1 ), employment status will decrease. The relationship between exacerbations and employment status will be negative (i.e., the more exacerbations, the less likely one is to be employed). The relationship between FEV 1 and employment will be positive (i.e., the higher the FEV 1 level, the more likely an individual is to be employed). This w as analyzed using Latent Growth Curve Analysis. Hypothesis 3b Change in employment level over time measured by its slope, will be correlated with change in health related quality of life ( emotional, social, role functioning and physical functioning domains ) Specifically, the slopes of employment and health related quality of life will be positively correlated This hypothesis was analyzed using Latent Growth Curve Analysis. Hypothesis 3c Change in disease severity (FEV 1 and exacerbations) over time, measured by its slope, will be correlated with change in health related quality of life over time. Specifically, as the disease se verity increases over time (positive slope for exacerbations and negative slope for FEV 1 ), health related quality of life will decrease. The relationship between exacerbations and health related quality of life will be negative (i.e., the more exacerbation s, the lower one's health related quality of life ). The relationship between FEV 1 and health related

PAGE 21

16 quality of life will be positive (i.e., higher FEV 1 levels will be correlated with higher health related quality of life scores). This hypothesis was analy zed using Latent Growth Curve Analysis. Structural equation m odeling (SEM), in general, allows the investigator to test hypothesized relationships among observed and/or latent variables. A basic SEM model consists of several observed and/or latent variab les and the regression paths between these variables. Indicator variables may be included if latent constructs are present. Each of the indicator variables has a factor loading (Preacher et al., 2008) Latent growth curve (L G C ) modeling (see Figure 1) is a specific type of SEM that allows a researcher to better understand change The latent variable s in L G C represent patterns of change of a variable that is measured a minimum of three times. The values of the variable at several time points allow the researcher to better understand the varia ble's pattern of change, which is reflected in the latent construct (Preacher et al., 2008) There are two main latent constructs in L G C which are intercept and slope. The intercept is a measure of the variable of interest at the baseline time point The second main latent construct, slope is the rate at which the variable changes. C onducting LGC analysis within a SEM framework a llows two LGC models to be combined in order to see how they relate to each other. For example, a question of interest that requires two LGC models is : how does the rate of change of variable X and the rate of change of variable Y relate? Do they both grow in a positive direction together? Does one change in a negative direction while the other change s in a positive direction? In order to address h ypotheses 3a 3c, I examined the relationship of two variables' intercepts and slopes. For example, in order to understand how change in employment over time relates to change in emotional well being, I created a LGC model for each variable separately, and

PAGE 22

17 then examined the correlations between the intercepts and slopes. The relationship between the two intercepts (a path in Figure 1) provides information on how the variables are related at the in i tial time point. The relationship between the two slopes (b p ath in Figure 1) provides information on how the variables rate s of change are related. L G C modeling within a SEM framework has a few additional advantages. For example, LGC modeling allowed me to examine both variables, each measured four times, withi n a single model. Analyzing these variables within a single model reduces the chance for type 1 error. Finally, using LGC modeling within a SEM framework allows for missing data within the longitudinal nature of the study to be addressed using Full Informa tion Maximum Likelihood (FIML), which is a state of the art method for handling missing data. For all H ypothesis 3 analyses, I us ed Mplus statistical software (Version 7.31; MuthÂŽn & MuthÂŽn, 2014 ).

PAGE 23

18 Figure 1. Latent growth curve model Hypothesis 4 Employment level will mediate the relationship between disease severity and health related quality of life This was analyzed using mediation analyses. The IV for this analysis was disease severity (FEV 1 or exacerbations), the mediator variable was employment status and the DV was score on the CFQ R subscales. Eight separate mediation analyses were conducted. To test the mediation hypothesis, I followed the procedure described in Baron & Kenny (1986). A variable is considered to function as a mediator if the following criteria are met: (1) the independent variable (IV) is a significant predictor of the hypothesized mediator variable (a path Var1_1 Var1_2 Var1_3 Var1_4 Var2_1 Var2_2 Var2_3 Var2_4 i1 Intercept Variable 1 i2 Intercept Variable 2 s1 Slope Variable 1 s2 Slope Variable 2 a path b path

PAGE 24

19 in Figure 2) (2) the mediator significantly predicts the depend ent variable (DV) (b path in Figure 2), (3) the IV predicts the DV (c path in Figure 2), and (4) when the correlations between the IV and mediator variable and between the mediator variable and DV are controlled, the IV is no longer a significant predictor of the DV (c' path in Figure 2) The strongest mediation relationship occurs when the relationship between the IV and DV becomes zero after controlling for the regression paths between the IV and mediator and mediator and DV. More commonly, however, contr olling the regression paths leads to a significant decrease in the relationship between the IV and DV (Baron & Kenny, 1986). I tested this mediation hypothesis using a series of regressions. To test the significance of the decrease in the relationship betw een the IV and the DV, I used the Sobel Test (1982 ). I used SPSS software to conduct the mediation analyses. Figure 2. Mediation analyses diagram IV: Disease Severity (1) FEV 1 and (2) exacerbations Mediator: employment status DV: Quality of Life (1) emotion, (2) social, (3) role, (4) physical a path b path c path (without mediator) c' path (with mediator)

PAGE 25

20 CHAPTER III RESULTS Descriptive Statistics Survey r esponse rates for the time points included in this study (1, 5, 8, and 10) were 93%, 77%, 75%, and 70%, respectively. The mean age of participants at the beginning of the study was 33.0 ( SD = 10 .7) years, and 42.0% were male Almost half (46.4%) of the pa rticipants were married at th e beginning of the study, and 8 1 .1 % had some education beyond a high school degree. The mean FEV 1 at baseline was 54.39 ( SD =20.43), mean exacerbations over four months was 0.76 ( SD =1.05), and mean clinical weight percentile was 36.13 ( SD =27.66). Table 1 displays disease severity descriptive statistics over time. Table 1 Correlations, Means, and Standard Deviations for Disease Severity ** p< .01, p <.05 Variable Exacerbations T1 Exacerbations T2 Exacerbations T3 Exacerbations T4 FEV 1 T1 FEV 1 T2 FEV 1 T3 FEV 1 T4 Exacerbations_T1 Exacerbations_T2 .41** Exacerbations_T3 .24** .29** Exacerbations_T4 .18* .46** .31** FEV 1 _T1 0.12 0.10 0.16* 0.23** FEV 1 _T2 0.14* 0.10 0.19* 0.19* .89** FEV 1 _T3 0.24** 0.13 0.22** 0.24** .86** .93** FEV 1 _T4 0.26** 0.14 0.22** 0.28** 82** .90** .94** M .76 .68 .80 .78 54.4 52.9 51.3 51.2 SD 1.0 1.2 1.3 1.3 20.4 20.0 19.2 19.7

PAGE 26

21 The p ercent of participants employed part time or full time ranged from 51.1% at the initial time point (wave 1) to 30.0% at the final time point included in this study (wave 10). Table 2 shows descriptive statistics for employment status over time in more detail. Table 2 Percent Employed across Time Responses to the questions on the CFQ R aggregated to form standardized scores from 0 to 100 on e ach of the domains. At the initial time point, mean scores on the emotion subscale ranged from 66.5 ( SD =20.1) to 71.4 ( SD =18.3). For the social subscale, mean scores ranged from 58.1 ( SD =18.1) to 62.7 ( SD =18.8). Mean scores for the physical subscale ranged from 52.7 ( SD =28.3) to 55.7 ( SD =27.6). Finally, mean scores for the role subscale ranged from 70.7 ( SD =23.0) to 74.0 ( SD =23.3). Table 3 illustrates CFQ R scores over time in more detail. Variable Employed T1 n = 307 Employed T2 n = 307 Employed T3 n = 307 Employed T4 n = 307 Employed 51.1% 44.0% 37.1% 30.0% Not Employed 41.7% 31.9% 30.6% 25.7% Missing 7.2% 24.1% 32.2% 44.3%

PAGE 27

22 Table 3 Correlations, Means, and Standard Deviations for CFQ R Note T1 = Baseline; T2 = Time Point 2 ; T3 = Time Point 3; T4 = Time Point 4; Emot = CFQ R Emotion scale; Soc = CFQ R Social scale; Role = CFQ R Role scale; P hy = CFQ R Physical scale. ** p < .0 0 1. Variable Emot T1 n = 291 Soc T1 n = 290 Role T1 n = 291 Phys T1 n = 292 Emo t T2 n = 254 Soc T2 n = 250 Role T2 n = 250 Phys T2 n = 252 Emot T3 n = 233 Soc T3 n = 233 Role T3 n = 233 Phys T3 n = 232 Emot T4 n = 220 Social T4 n = 220 Role T4 n = 220 Phys T4 n = 220 Emot _T1 Soc _T1 .56** Role_T1 .56** .53** Phys _T1 .43** .52** .64** Emot _T2 .58** .39** .44** .24** Soc _T2 .49** .70** .38** .45** .48** Role_T2 .39** .40** .56** .46** .47** .48** Phys _T2 .28** .28** .41** .70** .29** .47** .59** Emot _T3 .64** .45** .46** .27** .67** .43** .44** .25** Soc _T3 .39** .64** .34** .40** .37** .73** .35** .35** .51** Role_T3 .38** .39** .54** .38** .38** .43** .56** .37** .59** .52** Phys _T3 .25** .30** .39** .66** .21** .36** .43** .74** .36** .48** .58** Emot _T4 .52** .42** .40** .29** .57** .39** .40** .28** .63** .50** .46** .35** Soc _T4 .45** .63** .39** .43** .44** .69** .42** .41** .50** .74** .45** .40** .61** Role_T4 .40** .43** .53** .45** .35** .39** .56** .40** .47** .44** .62** .50** .57** .60** Phys _T4 .34** .38** .42** .69** .22** .35** .50** .69** .34** .41* .41** .74 ** .49 ** .56** .65** M 66.49 62.66 72.07 53.53 71.40 61.3 73.4 54.5 70.4 59.9 74.0 55.7 67.4 58.1 70.7 52.7 SD 20.1 18.8 24.0 26.6 18.3 18.7 22.4 28.0 19.5 19.4 23.3 27.6 20.4 18.8 23.0 28.3

PAGE 28

23 Hypothesis 1 My firs t hypothesis was that those who were employed at the baseline time point would score significantly higher on the emotional, social, role and physical health related quality of life domains compared to those who were unemployed at the baseline time point. Four one way analyses of variance (ANOVAs) were conducted to examine the relationship between employment status and health related quality of life Employment statu s (yes/no) was the independent (grouping) variable for each of the ANVOAs. Four domains of h ealth related quality of life (emotional, social, role, physical) were the outcome variable s of interest. Demographic and clinical covariates were included in the analysis More specifically, FEV 1 and exacerbations were included as clinical covaria tes; a ge gender, marital status, and educational attainment were included as demographic covariates. Results of the analyses are presented in Tables 4 7 Table 4 Employment Status and Emotional HR QoL ANOVA at baseline Variable df F p value Emotion 1 282 23.0 <.001 FEV 1 1 242 2.8 .096 Exacerbations 1, 271 16.3 < .001 Age 1, 283 0.27 .60 Male 1, 282 5.1 .024 Married 1, 282 0.2 6 .61 Post High School Education 1, 282 8.3 .00 4 Table 5. Employment Status and Social HR QoL ANOVA at baseline Variable df F p value

PAGE 29

24 Social 1, 281 17.1 < .001 FEV 1 1, 242 2.8 .096 Exacerbations 1, 271 16.3 < .001 Age 1, 283 0.27 .60 Male 1, 282 5.13 .024 Married 1, 282 0.26 .6 1 Post High School Education 1, 282 8.3 .00 4 Table 6. Employment Status and Role HR QoL ANOVA at baseline Variable df F p value Role 1, 282 45.18 < .001 FEV 1 1, 242 2.8 .10 Exacerbations 1, 271 16.3 < .001 Age 1, 283 0.27 .60 Male 1, 282 5.13 .024 Married 1, 282 0.26 .6 1 Post High School Education 1, 282 8.3 .00 4

PAGE 30

25 Table 7. Employment Status and Physical HR QoL ANOVA at baseline Variable df F p value Physical 1, 283 32.43 < .001 FEV 1 1, 242 2.8 .096 Exacerbations 1, 271 16.3 < .001 Age 1, 283 0.2 7 .604 Male 1, 282 5.13 .024 Married 1, 282 0.26 .6 1 Post High School Education 1, 2 82 8.3 .00 4 The four one way ANOVAs confirmed that individuals who were employed did indeed score higher on all four HRQoL domains (emotion, social, role and physical) included in the analysis (all p < .001) even when controlling for clinical and demographic covariates, which suggests that employment positively impacts several domains of self reported health related quality of life in individuals with CF. In addition, there were significant differences b etween the two groups in number of pulmonary exacerbations, gender, and educational attainment. Specifically, those in the employed group had less exacerbations, more males and more individuals with some education beyond high school. Hypothesis 2 My sec ond hypothesis was that greater disease severity (lower FEV 1 and more pulmonary exacerbations) would be negatively correlated with employment. A single bin ary logistic regression analysis was conducted to e xamine the relationship between the two measures of disease severity (FEV 1 and exacerbations) and employment status. Both

PAGE 31

26 measures of disease severity as well as demographic covariates were included in the analysis. Results are summarized in the Table 8 Table 8. Logistic Regression at baseline with Dependent Variable = Employed (Yes/No) Variable B Wald p value Odds Ratio FEV 1 0.00 7 0. 874 0. 350 1.00 7 Exacerbations 0.42 1 8.957 0. 00 3 0.65 7 Age 0.001 0.00 6 0.9 36 1.001 Male 0.7 10 5.946 0.01 5 2.0 34 Married 0. 139 0.205 0. 650 0. 870 Post HS Education 0.9 41 5. 906 0.01 5 2.5 63 Results of the binary logistic regression showed that number of exacerbations was indeed a significant predictor of employment status ( p = .003 ); FEV 1 however, was not a significant predictor of employment status ( p =.35 ). Being male and having any education beyond high school were also significant predictors of employment status ( p =.02 for both ). Odds ratios provided further information on the strength of the relationship of these predictors of employment. F or every additional pulmonary exacerbation, likelihood o f employment decreased (OR=0.66 ). Males were approximately two times more likely to be employed than females (OR=2.03 ) and those with some education beyond high school were 2.5 times more likely to b e employed than those without one (OR=2.5 6 ). Hypothesis 3 My third hypothesis had several components First, I predicted that a c hange in d isease severity over time would be correlated with a change in employment status over time

PAGE 32

27 Specifically, a positive slope for exacerbations or a negative slope for FEV 1 w ould be correlated with a negative slope for employment status. I hypothesized that the sicker patients became the less likely they would be employed. Second, I hypothesized that change in employment level over time would be correlated with change in health related quality of life Specifically, a declining employment status slope would be correlated with declining slopes for each of the four health related quality of life domains (emotion, social, rol e, physical). Finally, I hypothesized that disease severity and health related quality of life would change together over time. Specifically, I hypothesized that increasing disease severity as measured by a positive slope for exacerbations or a negative s lope for FEV 1 would be associated with declining health related quality of life A significant correlation between the intercepts of two latent growth curve analyses ( the a path in Figure 1 ) indicates that the variables are correlated at the initial time point. A significant correlation of the slopes signifies that the variables change together over time ( the b path in Figure 1 ). Fourteen separate latent growth curve analyses were conducted in order to examine the relationships of both measures of disease severity with each health related quality of life domain (8 in total) and employment status (2 in total) as well as to examine the relationship of employment status with each health related quality of life domain (4 in total). Results are presented in Tab le 9 below.

PAGE 33

28 Table 9. Latent growth curve analysis standardized correlations Variable a path (Intercept) b path (Slope) Employment Status Emotion .44 ( p <.001) .80 ( p =.10) Social .35 ( p <.001) .23 ( p =.34) Role .58 ( p <.001) .48 ( p =.11) Physical .46 ( p <.001) .32 ( p =.24) FEV 1 .17 ( p =.03) .22 ( p =.33) Exacerbations .38 ( p <.001) .53 ( p =.02) Exacerbations Emotion .32 ( p <.001) 1.31 ( p =.10) Social .18 ( p =.05) .49 ( p =.07) Role .52 ( p <.001) .98 ( p =.02) Physical .31 ( p <.001) .56 ( p =.06) FEV 1 Emotion .04 ( p =.59) .75 ( p =.29) Social .15 ( p =.04) .15 ( p =.51) Role .22 ( p =.005) .58 ( p =.13) Physical .51 ( p <.001) 89 ( p =.001) Results of the intercept correlations (a paths) indicated that initial level of employment was significantly positively correlated with each health related quality of life domain (all p <.001), and FEV 1 ( p =.03). I nitial level of employment was negatively correlated with

PAGE 34

29 exacerbations ( p <.001) in the negative direction, indicating that those with more exacerbations were less likely to be employed at the baseline time point Initial number of exacerbations was also negatively correlated with initial levels of all four health related quality of life domains (all p <.001), except social ( p =.05). Finally, initial level of FEV 1 was significantly positively correlated with three of the four health related quality of life domains : social ( p =.04), role ( p =.005), and physical ( p <.001). Initial FEV 1 levels were not correlated with emotional health related quality of life Results of the slope correlations (b paths) were less consistent. The slope of exacerbations and employment status w as significantly negatively correlated ( p = .02 ) such that as exacerbations increased, employment decreased. The slope of employment status only approached a positive significant correlation with the slope of emotional health related quality of life ( p =.10). The correlation s between slope s of exacerbations and health related quality of life however, were more consistent. The correlation between exacerbations and role functioning was significant ( p =.02) in the negative direction which indicates that as exacerbations increase, self reported role health related quality of life declines. The correlations between the exacerbations slope and emotion, social and physical functioning slopes all approached significance ( p =.10, .07, and .06, respectively) and were all in the negative direction Fin ally, the slope of FEV 1 was significantly positively correlated with the slope of physical health related quality of life ( p <.001), indicating that as FEV 1 declines over time, so does self reported physical health related quality of life Hypothesis 4 I hypothesized that employment status would mediate the relationship between disease severity and health related quality of life Because this study examined two measures of

PAGE 35

30 disease severity and four aspects of HR QoL, I proposed conducting eight separate med iation analyse s according to the steps ou tlined in Baron & Kenny and using the Sobel Test (1982) to determine if the mediation effect was significant Two markers of disease severity (FEV 1 and exacerbations) each served as the IV for four mediation analyse s. The four domains of health related quality of life (emotion, social, role, physical) each served as the DV for two mediation analyses. Employment status measured dichotomously, served as the mediator variable for all eight analyses (see Figure 2). Th e IVs were significant predictors of the DVs in all but one analysis (FEV 1 and emotion). For the analysis in which FEV 1 served as the IV and emotional health related quality of life served as the DV, the relationship only approached significance ( p = .0 8 ). In all other analyses, the relationship between the IV and DV (path c) was significant (all p < .05). The relationship between the mediator and DV (path b) was significant ( p <.05) in all eight analyses. The relationship between the IV and mediator (path a ), however, was only significant ( p <.05) when using exacerbations as the IV. Path a only approached significance ( p =.10 ) when FEV 1 served as the IV meaning that the prerequisites for mediation were not met in the analyses that used FEV 1 as the IV. Therefo re, only the results for the four analyses in which exacerbations served as the IV are prese nted in Table 10 Table 10. Results of mediation a nalysis with exacerbations as independent variable Emotion Social Role Physical a 0 .47 ** 0 .47 ** 0 .47 ** 0 .47 ** b 9.58 ** 8.26 ** 14.54 ** 14.88 ** c 4.0 9 ** 2.44 8. 29 ** 6.15 ** c' 2.99 1.5 6. 64 ** 4.4 8 Sobel test statistic 2.75 ** 2. 62 ** 3. 11 ** 2.96 ** Percent mediated 17.0% 31.5% 7.8% 12.6% Note: p <.05 ** p <. 01

PAGE 36

31 Results of the mediation analyses using exacerbations as the independent variable were consistent with my hypothesis Results of the Sob el test i ndicat e that employment status is indeed a mediator of the relationship between exacerbations and the four domains of health related qua lity of life I used in this study (emotion, social, role, and physical). In all four analyses, the strength of the relationship between the IV and DV was significantly reduced (all p <.0 1 ) when the mediator (employment status) was added to the model. When e xamining social health related quality of life as the DV, the relationship between the IV and DV became non significant with the inclusion of the mediator. However, the p value for the c path was largest for social health related quality of life so this f inding may not be meaningful.

PAGE 37

32 CHAPTER IV DISCUSSION Cystic f ibrosis was previously considered a childhood disease because patients did not commonly survive long enough to reach adulthood. However, medical advances have greatly improved the prognosis for patients with CF. Adults with CF (i.e., ages 18 years and older) now make up almost 50% of the CF population, which is a drastic increase from 1986 when only 29.2% of people with CF were 18 years or older (CF Foundation Report, 2013 ). The increased lifespan of adults with CF has led to an increasing number of patients with CF pursuing developmentally appropriate adult activitie s such as higher education and work (Havermans, Colpaert, Vanharen, and Dupont, 2009). However, because the adult survival ra te shifts are so recent, there is little research to date on how the disease influences patients' daily tasks of adult living and health related quality of life This study sought to expand our knowledge of employment in adults with CF, and explore how emp loyment relates to disease severity and health related quality of life Based on previous work, my first hypothesis was that those who were employed at the baseline time point would score higher on the four health related quality of life scales included in this study (emotional well being, social well being, role functioning, and physical func tioning). Results of the four ANVOAs confirmed this hypothesis. Emotion, social, role and physical health related quality of life scores of those who were employed a nd those who were not employed were signif icantly different (all p <.001). These health related quality of life domains were all significant even when controlling for demographic and clinical covariates, which suggests that work may provide a unique, positi ve contribution to an individual's health related quality of life

PAGE 38

33 Additionally, gender, education and pulmonary exacerbations were significantly different between those who were employed and those who were not employed. The employed group included more m ales, more individuals with some education beyond high school and individuals with fewer exacerbations Interestingly, although FEV 1 and exacerbations are both measures of disease severity, FEV 1 scores did not differ between the employed and unemployed gr oups. This suggests that perhaps exacerbations are more disruptive to maintaining a job than a chronically low FEV 1 score. Exacerbations require either hospitalization or home IV treatment, which may make it difficult to fulfill one's occupational responsibilities. This issue was further examined by hypothesis two (see below) Also of note, e ducation level was signific antly different between groups, suggesting that educational attainment may in part predict one's ability to maintain employment as a CF patient. This finding suggests it is important to pay attention to educational goals in the face of childhood illness. To address my second hypothesis, I used logistic regression to further understand factors that predict employment status in adults with CF. Interestingly, exacerbations was a significa nt predictor with an odds ratio of 0 .66 even when controlling for FEV 1 and demographic covariates. FEV 1 however, was not a significant predictor when c ontrolling for exacerbations and demographic covariates In fact, FEV 1 had an odds ratio of nearly 1 .0 ( OR= 1.00 7 ), suggesting that it did not have much influence on the likeli hood of whether one would be employed. Again, since exacerbations require hosp italization or home IV treatment, and patient s who suffer from more exacerbations may find themselves u nable to fulfill their duties at work due the disruptive treatment required for these episodes This suggests that exacerbations, rather than FEV 1 may be a superior measure for identifying individuals with CF who are at risk of being unable to maintain their job. Given that FEV 1 is the most

PAGE 39

34 commonly used marker of disease severi ty in individuals with CF and most studies examining HR QoL, employment and disease severity have use d FEV 1 as a marker of disease severity, this finding may have important implications for future work in this field. Being male and having some education beyond high school were also associated with an increased likelihood of being employed. Specifically, males were approximately two times more likely to b e employed than females (OR=2.03 ) even when controlling for other demographic and clinical covariates. Individuals with a high school diploma were 2.5 times more likely to be employed than those without one even when controlling for other demographic and clinical covariates. These findings like those from the first hypothesis suggest that having some edu cation beyond high school is a protective factor in maintaining employment for individuals with CF One reason that educational attainment may be such a strong predictor of employment status is that CF patients may have a particularly difficult time maint aining jobs that require physical labor (Burker et al., 2004; Burker et al., 2005; Taylor Robinson, 2013). In one study that compared individuals with CF who were employed to those who were not employed (Burker, 2005), all but 6% of patients currently empl oyed were working in ski lled or semi skilled labor jobs. Further, those who were not working at the time of data collection, but had worked in the past, reported working unskilled jobs at significantly higher rates than those who were employed at the time of data collection (Burker, 2005) Results from the latent growth curve (LGC) analysis conducted to address hypothesis 3 provided information on the longitudinal relationship of the variables of interest, which has not yet been studied Results of LGC ana lysis showed that the correlation s between intercept s were significant between employment and all four health related quality of life domains as

PAGE 40

35 well as between exacerbations and all four health related quality of life domains. These results are consisten t with those found in by analy ses that addressed hypotheses 1 and 2 Interestingly, the intercept variable for FEV 1 was significantly correlated with three domains of health related quality of life (social, role, and physical). This finding may initially seem to contradict the results from the analyses that addressed hypothes e s 1 and 2 The results from analyses 1 and 2 suggested that FEV1 1 is not associated with being employed or is not a good predictor of employment. The reason for what may initially seem like contradict ory results is most likely due to the fact that exacerbations were not controlled for in the latent growth curve analysis. They were, howeve r, controlled for in the analyse s that addressed hypotheses 1 and 2 Exacerbations an d FEV 1 are highly correlated, and therefore when the y are both entered in a model, FEV 1 is no longer a significant predictor of health related quality of life However, without exacerbations in the model, FEV 1 was indeed correlated with several health rela ted quality of life domains at the initial time point. The correlations between the slopes were less consistent. Only the paths between the slopes of exacerbations and each of the four domains of health related quality of life were consistently significan t, and several of them only approached significance ( p values greater than 0 5, but equal to or less than .10 ). This suggests that perh aps self reported health related quality of life decreases as number of exacerbations changes over time. The path between the slopes of FEV 1 and physical health related quality of life was significant ( p <.001), indicating that as FEV 1 changes over time, self reported physical health related quality of life changes as well. The path between the slopes of employment and emotional health related quality of life approached significance ( p =.10 ), suggesting that as employment status changes over

PAGE 41

36 time, so does emotional well b eing. Because many of the correlations approached significance, further studies with more participants and more time points are needed to confirm this relationship. Results of the mediation analysis for hypothesis 4 provide additional knowledge on how these variables relate to each other. Res ults of the Sobel test indicate that employment status is indeed a mediator of the relationship between exacerbations and the four domains of health related quality of life used in this study (emotion, social, role, and physical). In all four analyses, the strength of the relationship between the IV and DV was significantly reduced (all p <.0 1 ) when the mediator (employment status) was added to the model. Between 17.8% and 31.5% of the effect of exacerbations on the health related quality of life domains was mediated by employment status. When examining social health related quality of life as the DV, the relationship between the IV and DV became non significant with the inclusion of the mediator. These results suggest that at least part of the relationship b etween exacerbations and self reported health related quality of life is accounted for by the fact that exacerbations disrupt one's ability to work. Implications Despite how often many people complain about work it plays an important role in our lives mu ch beyond a source of income and health care. Work can make individuals feel like productive members of society by assuming responsibility, completing tasks, and helping others (Freedman & Fesko, 1996; Johnson et al., 2004). Employment can also become part of one's identity and thus serve to increase one's self esteem because (Miller & Dishon, 2006). Ac cording to Erikson's theory of psychosocial d evelopment, work also plays a key role in psychosocial development.

PAGE 42

37 Erikson divides psychosocia l development int o eight stages, each characterized with a developmental conflict or "crisis" that must be resolved. He theorizes that failure to resolve one conflict m ay lead to difficulties with fu ture conflict s and/or developmental tasks. For example, failure to make "o ccupation commitments" ( Munley, 1977, p.262) in adolescence may lead to poor ego formation and social connections in young adulthood. Munley (1977) discusses Erikson's theory within the con text of vocational development and highlights that career development is integrated with several of Erikson's stages, including developing a sense of initiative, industry, identity, generativity, and integrity. According to Erikson, work is an integral pie ce of several developmental stages an d failure to successfully resolve these developmental "crises" will negatively impact the individual in the future. The findings from this study are consistent with Erikson's theory of psychoso c ial development. Result s of this study showed that those who were employed reported higher health related quality of life t han those who were not employed even when controlling for clinical and demographic covariates. Erikson's theory highlights the important role work plays in an individual's psychosocial development and this study highli g h ts the important role work plays in the related concept of well being or health related quality of life It is important for CF clinicians to be made aware of these findings especially when d iscussing issues such as disability with patients Results of this study suggest that the relationship between disease severity, work and health related quality of life may not be that simple and work may have a positive impact on one's life Previous res earch suggests that CF affects many patients' ability to work (Hogg et al., 2007; Laborde CastÂŽrot et al., 2012). This study further examined which factors may impact an adult with CF's ability to work. Perhaps the most important was education. Those with

PAGE 43

38 any education beyond high school were more likely to be employed than those without. This is perhaps due to the fact that jobs that require higher levels of education tend to (1) require less physical exertion, and (2) be more flexible with regards to work ing from home or adjusting one's work schedule These findings will hopefully provide a basis for future interventions, such as educational and vocational counseling, aimed at increasing CF patients' ability to work. This study provides important informa tion on regarding CF patients who may be at higher risk for unemployment. Those with more exacerbations appear to have more difficulty maintain employment than those who experience fewer exacerbations Lung functioning, as measured by FEV 1 does not seem to impact one's ability to work in the same way as exacerbations. This has important implications for future work with this population as FEV 1 is the most commonly used marker of disease severity in CF patients. Results from this study sh ow that perhaps n umber of pulmonary exacerbations is a superior marker of disease severity when trying to identify patients at risk for unemployment who may benefit from vocational counseling. Limitations One limitation in this study is that some of the variables of inte rest are not manifest variables, but rather hypothetical constructs, and therefore there is inherent error in measurement. Two of our manifest variables are physiological measures (FEV 1 and number of pulmonary exacerbations), and therefore likely to be acc urate with the exception of systematic error due to the measurement tool or random error. Self report measures however, may lead to under or over reporting. The most complicated variable of interest is health related quality of life because it is a hypoth etical construct that was measured using a self report survey.

PAGE 44

39 Studies conducted on the psychometric properties of the CFQ R have demonstrated strong reliability and validity (Quittner et al., 2012). However, hypothetical constructs are more difficult to m easure than manifest variables, and this is important to consider when interpre ting the results of this study. There is also a potential issue of external validity in this study. The study was conducted at ten CF centers across the country. Although this provides a certain degree of geographic diversity, our results may not generalize to patients who live in rural communities and do not receive their care at a large CF center, which are usually located at academic medical centers in large cities. Addition ally, this project began almost ten years ago and due to medical advances, an individual's experience with CF today could be different than an individual's experience ten years ago. However, the medical staff and lead investigators of this study are confid ent that the variables measured in this study remain relevant. Future Directions Several slope correlations in the LGC analysis approached significance. Thus, it is necessary to repeat this study, or a similar one, with a larger sample more data points, and/or a sicker population in order to confirm the findings. A l arger sample would increase power, more data points would provide better information on the shape of the slopes, and a sicker population should provide more variability in the measures. If a study similar to this one is repeated, it would be interesting to consider a possible cohort effect. The adults enrolled in PAC CF were between the ages of 18 and 65 when the study began in 2005. This means that the youngest subject was born in 1987 and the majority of participants were born well before that considering the average age at time of enrollment was 33 years. Given that children born with CF in the 1980s were not even expected to

PAGE 45

40 reach adulthood, there may be important differences between the PAC CF cohort and a cohort of adults studied in the future An individual growing up not expecting to live much past their 18 th birthday may approach school, higher education, and vocational choices differently than someone who grows up expecting to live well into adulthood. Additionally, medical advances could have important implications educationally as patients may now be able to miss less school and li ve a more "normal" life Education was found to be a strong predictor of employment in adults with CF. Additionally, previous studies have suggested that perhaps it is easier for educated, "white collar" CF workers to maintain employment than those in "blue collar" jobs (Burker et al., 2005). This is logical given the physical effects of the disease, bu t requires more research to understand. The present study found education to be an important predictor of employment suggesting that education perhaps has a protective factor. In order to better understand this finding it could be beneficial to compare the education odds ratio in our CF sampl e to the education odds ratio in the general public It is well known that education is associated with employment in the general population. However, the question of interest is i f education is even more important in helping those with CF maintain employment due to their physical limitations and the physical exertion often required in a n "unskilled" or "blue collar job." In conclusion, this study contributed to our understanding of the relationship s among disease severity, employment and health related quality of life in individuals with CF and was the first study to examine the variables longitudinally. However, more research is still needed on this topic in order for comprehensi ve CF care teams to best serve the relatively new, increasingly large cohort of adults with CF.

PAGE 46

41 REFERENCES Abbot, J., Hart, A., Morton, A., Gee, L., & Conway, S. (2008). Heal th related health related quality of life in adults with cystic fibrosis: The r ole of coping. Journal of Psychosomatic Research, 64, 149 157. Anderson, D.L., Flume, P.A., & Hardy, K.A. (2001). Psychologic al functioning in adults with cystic f fibrosis Chest, 119. 1079 1084. Baron, R .M., & Kenny, D.A. (1986). The m oderator mediator v ariable distinction in social psychological research: Conceptual, strategic and statistic considerations. Journal of Personality and Social Psychology, 51, 1173 1182. Blair, C., Cull, A., Freeman, C.P. (1994). Psychosocial functionin g of young adults with cystic fibrosis and their families. Thorax, 49, 798 202. Burker, E.J., Sedway, J., & Carone, S. (2004). Psychological an d educational factors: Better predictors of work status than FEV 1 in adults with cystic fibrosis. Pediatric Pulmonology, 38, 413 418. Burker, E.J., Sedway, J., Carone, S., Trombley, C., & Yeatts, B.P. (2005). Vocational attainment of adults with CF: Success in the face of adversity. Jour nal of Cystic Rehabilitation, 71 (2) 22 27. Congleton, J., Hodson, M.E., & Duncan Skingle, F. (1996) Quality of life in adults with cystic fibrosis. Thorax, 51, 936 940. Corrigan, J.D., Bogner, J.A., Mysiw, J.W., Clinchot, D., Fugate, L. ( 2001). Life satisfaction after traumatic brain injury. Journal of Head Trauma Rehabilitation, 16, 543 555. Cystic Fibrosis Foundation. (2013). Cystic Fibrosis Foundation Annual Report. Retrieved from https://www.cff.org/About Us/Assets/2013 Annual Report/ Cystic Fibrosis Foundation. (2014). Cystic Fibrosis Foundation Annual Report. Retrieved from https://www.cff.org/Abou t Us/Assets/2014 Annual Report/ de Jong, W., Kaptein, A.A., van der Schans, C.P., Mannes, G.P.M., van Aalderen, W.M.C., Grevink, R.G., Kšeter, G. H. (1997). Quality of life i n p atients with cystic fibrosis. Pediatric Pulmonology, 23, 95 100. Dill, E.J. Dawson, R., Sellers, D.E., Robinson, W.M., Sawicki, G.S. (2 013). Longitudinal trends in he a l th related health related quality of life in adults with cystic fibrosis. Chest 14 4 (3), 981 989. Freedman, R.I., Fesko, S.L. (1996). The meaning of work in the liv es of people with significant disabilities: Consumer and family perspectives. Journal of Rehabilitation, 62 (3), 49 55.

PAGE 47

42 Gee, L., Abbott, J., Hart, A., Conway, S.P., Etheringt on, C., We bb, A.K. (2005). Associations between clinical variables and health related quality of life in adults with cystic fibrosis. Journal of Cystic Fibrosis, 4 59 66. Gillen, M., Lallas, D., Brown, C., Yelin, E., & Blanc, P. (1995). W ork disability in adults with cystic fibrosis. American Journal of Respiratory and Critical Care Medicine, 152, 153 156. Goldberg, R.T., Isralsky, M., & Shwachmann, H. (1985). Predictio n of rehabilitation status of young adults with cystic fibrosis. Archiv es of Physical Medicine and Rehabilitation, 66, 492 495. Havermans, T., Colpaert, K., Dupont, L.J. (2008). Quality of life in pa tients with cystic fibrosis: Associations with anxiety and depression. Journal of Cystic Fibrosis, 7, 581 584. Havermans T., C olpaert, K., Vanharen, L., and Dupont, L. J. (2009). Health related health related quality of life in cystic fibrosis: To work or not to work? Journal of Cystic Fibrosis, 8, 218 223. Hogg, M., Braithwaite, M., Bailey, M., Kotsimbos, T., & Wilson, J.W. (2007). Work disability in adults with cystic fibrosis and its relationship to health related quality of life Journal of Cystic Fibrosis, 6, 223 227. Johnson, K.L., Yorkston, K.M., Klasner, E.R., Kuehn, C.M., Joh nson, E., Amtmann, D. (2004) The Cost and benefits of employment: A qualitative study o f experiences of persons with multiple sclerosis. Archives of Physical Medicine and Rehabilitation, 85, 201 209. Laborde CastÂŽrot, H., Donnay, C., Chapron, J., Burgel, P.R., Kanaan, R., HonorÂŽ, I.,Hubert, D. (2012). Employment and work disability in adults with cystic fibrosis. Journal of Cystic Fibrosis, 11, 137 143. Lawler, R.H., Nakielny, W., & Wright, N.A. (196 6). Psychological implications of Cystic Fibrosis. Canadian Medical Association Journal, 94, 1043 1046. Little, T. D. (2013). Longitudinal structural equation modeling: Methodology in the social sciences. New York: Guilford Press. Miller, A. & Dishon, S. (2006). Health related health related quality of life in Multiple Sclerosis: The impact of disability, gender and employment status. Quality of Life Research, 15 259 271. Munley, P.H. (1977). Erikson's theory of psychosocial development and career dev elopment. Journal of Vocational Behavior, 10, 261 269.

PAGE 48

43 MuthÂŽn & MuthÂŽn. (2014). MPLUS Version 7.31. [Computer Software]. Los Angeles, CA. Orenstein, D.M., Nixon, P.A., Ross, E.A., & Kaplan, R.M. (1989). The quality of well being in cystic fibrosis. Thor ax, 95, 344 347. Pearson, D.A., Pumariega, A. J., Seilheimer, D. K. (1991). The development of psychiatric symptomology in patients with Cystic Fibrosis. Journal of the American Academy of Child and Adolescent Psychiatry, 30 290 297. Pfeffer, P.E., Pfeffer, J.M., Hodson, M.E. (2003). The psychosocial and psychiatric side of cystic fibrosis in adolescents and adults. Journal of Cystic Fibrosis, 2, 61 68. Platten, M.J., Newman, E., Quayle, E. (2013). Self Esteem and its relationship to mental health and health related quality of life in adults with cystic fibrosis. Journal of Clinical Psychology in Med ical Settings, 20, 392 399. Preacher, K.J., Wichman, A.L., MacCallum, R.C., and Briggs, N.E. (2008). Latent Growth Curve Modeling. Thousan d Oaks, California: Sage Publications. Quittner, A.L., Sawicki, G.S., McMullen, A., Rasouliyan, L., Pas ta, D.J., Yegin, A., Constan, M.W. (2012). Psychometric evaluation of the Cystic Fibros is Questionnaire Revised in a national sample. Quality of Life Res earch, 21, 1267 1278. Sawicki, G.S., Sellers, D.E., & Robinson, W.M. (2009). High tr eatment burden in adults with cystic fibrosis: Challenges to disease self management. Journal of Cystic Fibrosis, 8, 91 96. Shepherd, S.L., Hovell, M.F., Harwood, I.R, Gr anger, L.E., Hofs tetter, C.R., Molgaard, C., & Kaplan, R.M. (1990). A comparative study of the psychosocial assets of adults with cystic fibrosis and their health peers. Chest, 97, 1310 1316. Sobel, M.E. (1982). Asymptomatic confidence intervals for indir ect effects in structural equation models. In S. Leinhart (Ed.), Sociological methodology 1982 (pp.290 312). San Francisco: Jossey Bass. Sokol, J.T. (2009). Identify development throughout the lifetime: An examination of Eriksonian theory. Graduate Journal of Counseling Psychology, 1 (2), 139 148. Staab, D., Wenninger, N.G., Rupprath, K., Bisson, S., Trettin, M., Paul, K.D., and Wahn, U. (1998). Quality of life in patients with cystic fibrosis and the ir parents: what is important besides disease severity? Thorax, 53, 727 731. Taylor Robinson, D.C., Smyth, R., Diggle, P.J., & Whitehead, M. (2013). A longitudinal study of the impact of social deprivation and disease severity on employment status in the UK cystic fibrosis population. PLoS On e, 8 (8), e73322.

PAGE 49

44 Walters, S., Britton, J., & Hodson, M.E. (1993). Demographic and so cial characteristics of adults with cystic fibrosis in the United Kingdom. British Medical Journal, 306. 549 552. Whitley, B.E. & Kite, M.E. (2013). Principles of researc h in behavioral science. New York: Routledge. Widerman, E. (2003). Knowledge, interests, and education needs o f adults diagnosed with cystic fibrosis after age 18. Journal of Cystic Fibrosis, 2, 97 104.