Citation
Treatment seeking behavior for symptoms of acute myocardial infarction among Latina and non-Latina women

Material Information

Title:
Treatment seeking behavior for symptoms of acute myocardial infarction among Latina and non-Latina women
Creator:
Leeman-Castillo, Bonnie A
Publication Date:
Language:
English
Physical Description:
xiv, 485 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Myocardial infarction -- Patients ( lcsh )
Hispanic American women -- Health and hygiene ( lcsh )
Women -- Health and hygiene -- United States ( lcsh )
Heart -- Diseases -- Prevention -- United States ( lcsh )
Myocardial infartion -- Treatment ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 475-485).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Bonnie A. Leeman-Castillo.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
71813801 ( OCLC )
ocm71813801
Classification:
LD1193.L566 2005d L43 ( lcc )

Full Text
TREATMENT SEEKING BEHAVIOR FOR SYMPTOMS OF ACUTE
MYOCARDIAL INFARCTION AMONG LATINA AND NON-LATINA
WOMEN
Bonnie A. Leeman-Castillo
B.S., University of Delaware, 1992
M.S., East Stroudsburg University, 1993
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences
2005
by


This thesis for the Doctor of Philosophy
degree by
Bonnie Ann Leeman-Castillo
has been approved
by
Susan L. Dreisbach
Mori J. Krantz

Date


Leeman-Castillo, Bonnie Ann (Ph.D., Health and Behavioral Sciences)
Treatment Seeking Behavior for Symptoms of Acute Myocardial Infarction among
Latina and Non-Latina Women
Thesis directed by Professor Kitty K. Corbett
ABSTRACT
Background: Many women suffering an acute myocardial infarction (AMI) delay
seeking care, or experience delays when they do seek care. It appears that many
women may not assign their symptoms importance or fail to recognize their
symptoms as heart related.
Objective: To determine the specific symptoms Latinas and non-Latina women
experience related to an AMI and to describe the processes used to seek care for
these symptoms. The specific aims were to:
Aim 1: Describe and compare the initial symptoms that Latina and non-
Latina women report for an AMI.
Aim 2: Identify the subjective meanings that the symptoms have for the
Latina and non-Latina women, including the words they use and explanatory
models they draw upon to talk about those symptoms with family, friends,
and healthcare personnel, and how those subjective meanings, and other
contextual and personal factors in their lives, influence them in the treatment
seeking process.
Aim 3: Create care seeking trajectories for these populations of women that
illustrate the relationship of these subjective perceptions and the process of
seeking and receiving care for an AMI.
Methods: A multi-method research design emphasized qualitative, semi-structured
interviews with 17 Latinas and 26 non-Latina women, along with a review of their
medical records. Individual interviews during the hospital admission and a follow up
2 weeks after admission were conducted with each respondent. Outcomes: Four
typologies illustrating symptom perception, communication, and treatment-seeking
processes were created. Conclusion: Sub-populations of women (e.g., Latinas, non-
Latinas, and Spanish-speaking women) vary in their symptomatology and
assessment of those symptoms. Latinas overall had less knowledge and awareness of
heart attack symptoms. Most women had difficulty recognizing symptoms as heart
related. Intermittent symptoms (identified by 76% Latinas, 58% non-Latinas) are a
hallmark attribute for Latina and non-Latina women. Primary care providers have a
key role in the early period of womens care seeking trajectories, particularly in risk
communication and symptom assessments. The role of primary care providers and
m


the occurrence of prodromal symptoms warrant further study to better understand
their role in decreasing patient delay.
This abstract accurately represents the content of the candidates dissertation. I
recommend its publication.
Signed
Kitty K. Corbett
IV


DEDICATION
I dedicate this thesis to my husband, Alberto, and to my family for their patience
and support throughout this process.


ACKNOWLEDGEMENTS
This research was supported through a pre-doctoral fellowship award from
the American Heart Association and from the Center for Research in the Health and
Behavioral Sciences Program at the University of Colorado at Denver and Health
Sciences Center.
I would like to express my gratitude to my dissertation committee, Dr. Kitty
Corbett, whose belief in my success, patience, and experience doing qualitative
research with Latinas opened my eyes to understand illness through an
anthropological and cultural lens; Dr. Susan Dreisbach, who listened and guided me
to through behavioral theory and shared her experiences with qualitative research;
Dr. Jack Westfall, who offered access to wonderful healthcare systems in rural and
northeast Colorado, and to his supportive staff in the High Plains Research
Network; Dr. Mori Krantz for his expertise in cardiology, advisement that allowed
me to understand the pathophysiology of acute myocardial infarction; and, to all my
committee for their support and enthusiasm throughout the dissertation process.
Last, but certainly not least, I wish to acknowledge the research subjects and their
families who willingly and enthusiastically shared their experiences during the acute
phase of their illness to be advocates to advance our knowledge of women and heart
disease.


CONTENTS
Figures..............................................................xiii
Tables...............................................................xiv
CHAPTER
1. INTRODUCTION........................................................1
The Problem of Delayed Treatment Seeking Behavior for Myocardial
Infarction Symptoms............................................6
The Purpose of the Study.......................................8
Contribution of This Study: Furthering the Understanding of the Problem ...9
The Theoretical Framework................................... 10
The Organization of the Dissertation..........................11
2. REVIEW OF THE LITERATURE...........................................13
Prodromal Symptoms for AMI....................................14
Atypical Symptoms/Less Typical Symptoms Experienced by Women.18
Women and Treatment Seeking Behavior..........................20
Treatment Seeking Behavior in Non-White Populations...........22
Latino/as and AMI.............................................25
Perceptions of AMI Risk, Meaning of Symptoms, Discussions about
Symptoms with Others and Self-Treatment Strategies............31
Competing Demands.............................................35
3. THE THEORETICAL FRAMEWORK..........................................38
The Health Belief Model.......................................40
The Self Regulation Model of Illness Behavior.................42
Control Theory................................................43
Self-Efficacy.................................................44
4. RESEARCH DESIGN AND METHODOLOGY....................................46
The Case for In-Depth Interviews..............................46
vii


The Sites............................................................51
Gaining Entree...................................................52
Conducting Research Under the Health Insurance Portability and
Accountability Act (HIPAA)...........................................52
Selection of the Study Population....................................56
Inclusion and Exclusion Criteria.................................56
Recruiting Methods and Enrollment Procedures.........................57
The Impact of HIPAA on the Patient Enrollment Period.............58
HIPAA and Consent Procedures.....................................59
Protecting Patient Confidentiality...............................61
Hiring Spanish Interviewers..........................................62
Description of the Study Design......................................64
Study Instruments....................................................65
The Interview Guide..............................................65
The Contact Summary Form.........................................66
The Medical Record Review........................................66
Structure of the Data Collection Process.............................67
Initial Interview................................................67
Follow-up Interview..............................................69
The Medical Record Review........................................69
Analytic Methods for Quantitative Data...............................70
Data Reduction Processes.........................................71
Measures......................................................71
Analysis.........................................................73
Analytic Methods for Interview Data..................................77
Data Reduction Processes.........................................81
Crystallizing and Corroborating/Legitimating Connections......81
Vignettes.....................................................81
Data Management and Analysis..................................82
Reliability and Validity Checks...............................83
5. ANALYSIS OF DATA FROM MEDICAL CHART REVIEWS...............................85
The Sample...........................................................85
The Symptoms.........................................................91
Findings from Troponin Data.........................................107
Evaluating Prehospital Delay Time...................................108
The Chief Complaint.................................................115
viii


6. FINDINGS FROM INTERVIEW DATA
134
Patient Career Timelines...........................................135
Results............................................................136
Emerging Themes....................................................137
Communicating With Family and Healthcare Professionals..........137
Recent Encounters Impact Explanatory Models.................137
Providing Vague vs. Detailed Descriptions...................142
Treating the Symptoms...........................................144
Actively Obtaining Care.....................................144
Knowing the Treatment: Self Management vs.
Care Seeking vs. Waiting....................................148
Self-Treatment Strategies...................................149
Appraising the Symptoms.........................................150
Acknowledging Hunches and Gut Feelings......................151
Expecting vs. Experiencing MI Symptoms......................151
The Not Qualities......................................152
Being a Patient.................................................152
Crying Wolf.................................................155
Learning the System.........................................157
Additional Themes............................................. 160
Using Voice Inflection, Gestures, or Sounds to
Communicate Symptoms........................................160
Role of Pills...............................................161
Assessing Symptoms in the Night.............................164
Better to be Safe than Sorry.............................164
Not Wanting to Bother Others.............................165
Confident but Not Sure...................................165
Motivated by a Public Service Announcement...............166
Common Themes among Latinas.....................................167
Unanticipated Findings.............................................170
Lacking Awareness about Cardiac Rehabilitation..................170
Referencing Pregnancy and Childbirth Experiences................170
Involving the Family............................................171
Being Unaware of MI Diagnosis...................................172
7. TYPOLOGIES OF TREATMENT SEEKING BEHAVIOR................................173
IX


Typologies.........................................................173
Fast Actors: Short Delayers Who Recognized Symptoms
As Heart Related................................................174
Got Luckies: Short Delayers Who Did Not Recognize Symptoms
as Heart Related................................................174
The Derailed: Long Delayers Who Recognized Their Symptoms
as Serious or Heart Related.....................................175
Wonder a Lots: Long Delayers Who Misrecognized, or
Had No Recognition of, Their Symptoms...........................176
TYPOLOGY A: Short Delayers Who Recognized Symptoms
as Heart Related....................................................176
Vignette Example: Dottie Wilson, The Champion Billiards Clerk:
Im a womans advocate........................................177
Patient Career Timeline Example.................................182
TYPOLOGY B: Short Delayers Who Did Not Recognize Symptoms
as Heart Related....................................................184
Vignette Example: Taylor Robbins, NOT ONE SMIDGE,
said the lady from Laramie, Help others since you cant help me..... 184
Patient Career Timeline Example.................................189
TYPOLOGY C: Long Delayers Who Recognized Their Symptoms
as Serious or Heart Related.........................................191
Vignette Example: Ofelia Canela, Three months and three
clinic visits later, it ended with Mother Cabrini..............191
Patient Career Timeline Example.................................196
TYPOLOGY D: Long Delayers Who Misrecognized,
or Had No Recognition of, Their Symptoms............................198
Vignette Example: Stella Perez, The Caregiver
with an Unfilled Prescription...................................199
Patient Career Timeline Example.................................205
Care Seeking Trajectories...........................................207
8. CONCLUSION...............................................................209
System Factors that Influence Behavior Summary......................211
Medical Models that Emphasize Self-Treatment....................211
The Patients Role in Healthcare: Factors that Influence Behavior...212
Self-Efficacy...................................................213
What Can Be Done When Patients Hear, Nothing is Wrong?...........214
Perceived Susceptibility and Severity...........................214
x


Sense of Self-Care Boundary...............................215
The Influence of Co-Morbidities.............................217
Emotions....................................................219
The Providers Contribution to the Problem of Patient Delay....220
Use and Evaluation of the Prodrome Paradigm.................220
Summary: Determination of Symptom Onset.....................221
Chief Complaint................................................223
Provider Communication.........................................224
I Had a Heart Attack?.....................................225
Differences and Similarities among Latinas and Non-Latinas.....226
Differences Regarding the Seven Attributes of Symptoms......226
Location.................................................227
Quality of Symptoms......................................228
Associated (i.e. Atypical) Symptoms......................229
Severity.................................................231
Missed Cues: Alleviating and Exacerbating Factors and the Setting....232
Symptom Constellation.......................................233
Prodromal Symptoms......................................... 233
What Can Providers Do?.........................................235
Improve Communication with Patients....................... 235
Continue Educating Ancillary Staff..........................238
Focus on the Primary Care Provider..........................238
Consider the Troponin Measurement and Delay Time............239
The Next Steps for Patient Education...........................240
Using the Seven Attributes of a Symptom to Teach Women......240
Targeting and Tailoring Public Awareness Campaigns..........240
Challenges and Limitations.....................................241
Research Enrollment and Participation.......................241
Challenges..................................................241
Limitations.................................................242
Concluding Remarks.............................................244
APPENDIX
A. INTERVIEW GUIDE.............................................246
B. COMIRB INFORMED CONSENT.....................................249
C. COMIRB AUTHORIZATION A: RESEARCH RECRUITMENT................255
D. COMIRB AUTHORIZATION B: ENROLLMENT INTO
RESEARCH....................................................257
xi


E. PHYSICIAN NOTIFICATION LETTER.................260
F. PARTICIPANT INFORMATION FLYER.................262
G. MEDICAL CHART REVIEW FORM.....................264
H. CONTACT SUMMARY FORM..........................268
I. COLORADO DISCHARGE DATABASE, 1995-1997........270
J. COLORADO DISCHARGE DATABASE BY GENDER,
1995-1997.....................................272
K. ZIP CODES BY GEOGRAPHICAL LOCATION............274
L. TROPONIN-I REFERENCE RANGES...................276
M. PATIENT CAREER TIMELINES......................278
N. VIGNETTES.....................................328
O. SUBJECT CODING FORM...........................466
BIBLIOGRAPHY..........................................475
xii


FIGURES
Figure
3.0 Multidimensional Phenomenon: the Decision to Seek Care................39
3.1 The Health Belief Model...............................................41
3.2 The Self-Regulatory Model of Illness Behavior.........................43
3.3 Control Theory........................................................45
4.0 Enrollment Process Diagram............................................60
4.1 Peak Troponin Data Before and After Log Transformation................75
4.2 Total Delay Time Before and After Log Transformation..................76
4.3 Immersion/Crystallization Organizing Style............................78
4.4 Template Analysis Technique...........................................79
5.0 Association Between Peak Troponin and Total Delay Time...............Ill
7.0 Patient Career Timeline for Dottie Wilson............................183
7.1 Patient Career Timeline for Taylor Robbins...........................190
7.2 Patient Career Timeline for Ofelia Canela............................197
7.3 Patient Career Timeline for Stella Perez.............................206
7.4 Summary of Four Typologies...........................................207
7.5 Care Seeking Trajectories for the Fast Actors and Got Luckies........208
7.6 Care Seeking Trajectories for the Derailed and Wonder a Lots.........208
xiii


TABLES
Table
4.0 Multi-Method Approach.................................................47
4.1 A Priori Codes........................................................80
5.1 Sample Characteristics by Ethnicity-Language..........................88
5.1.1 Sample Characteristics by Geographical Location.......................90
5.2 Chest Characteristics of MI Symptoms by Ethnicity-Language............93
5.2.1 Back Characteristics of MI Symptoms by Ethnicity-Language.............94
5.2.2 Jaw Characteristics for MI Symptoms by Ethnicity-Language.............95
5.2.3 Throat/Neck Characteristics for MI Symptoms by Ethnicity-Language....95
5.2.4 Shoulder Characteristics for MI Symptoms by Ethnicity-Language........96
5.2.5 Arm Characteristics for MI Symptoms by Ethnicity-Language.............97
5.2.6 Stomach and Abdominal Characteristics for MI Symptoms
by Ethnicity-Language.................................................98
5.2.7 Leg Characteristics for MI Symptoms by Ethnicity-Language.............99
5.2.8 Head Characteristics for MI Symptoms by Ethnicity-Language...........100
5.2.9 Feet Characteristics of MI Symptoms by Ethnicity-Language............100
5.3 Intermittent/Stuttering (Timing) MI Symptoms by Ethnicity-Language...101
5.4 Associated MI Symptoms by Ethnicity-Language.........................102
5.5 Characteristics of MI Symptoms by Ethnicity-Language.................104
5.6 Setting Where MI Symptoms Occurred by Ethnicity-Language.............105
5.7 Behaviors Exacerbating or Alleviating MI Symptoms by Ethnicity-
Language.............................................................106
5.8 Chart Reported Troponin-I (ng/dl) by Ethnicity-Language..............107
5.8.1 Chart Reported Troponin-I (ng/dl) by Geographical Location...........108
5.9 Total Delay Time by Ethnicity-Language...............................109
5.9.1 Total Delay Time by Geographical Location............................110
5.10 Frequencies of Provider Visits Prior to the Index MI.................112
5.11 Comparison of Symptoms Described in Vignette to Symptoms
Reported to Others...................................................127
6.2 Relationship Between Having a Co-Morbidity and Delay Time
by Ethnicity.........................................................147
6.3 Examples of the Not Quality as Expressed by Patients...............153
6.4 Themes Mentioned by Latinas..........................................168
xiv


CHAPTER 1
INTRODUCTION
Ms. Kaufman is a 56 year old Caucasian with two grown children and
seven grandchildren. She was raised in a childrens home and learned that you
werent allowed to be sick... If you got sick it was like... murder if you got
sick, she laughed. She has been married to her third husband for 32 years. She
enjoys being a couch potato. She has been a diabetic for 12 years, has high
cholesterol, high blood pressure, osteoporosis, and suffers with COPD. Both
parents had a history of heart disease. She is on disability and has health
insurance. She completed the 10th grade. She agrees with her doctors
explanation and is aware that with all her illnesses, she is a walking time
bomb.
Ten days prior to her MI she was hospitalized with borderline
pneumonia which was the reason why her MI surprised her. Because Ms.
Kaufman had been feeling fine recently, she associated the heart attack
symptoms with the recent pneumonia. I thought that that was just a virus left
over or something else, she explained.
The day before the heart attack, she felt pretty tired. Although her stairs
are always a challenge, this time her chest hurt like when you breathe in cold
air and she couldnt catch her breath. The hurt in her right chest was
described as just a whole ache... You know when you get a brain freeze? ...
Well, this was a lung freeze, she laughed, ... on the right side. By the time I
walked from my front door to my kitchen, it was just like I was going to
collapse, Mrs. Kaufman said. Because of her diabetes and the fact that she had
1


not eaten anything on Friday, she thought those things caused her tiredness. This
tiredness, a heavy feeling, was different from when she is sleepy or experiences
her diabetic tiredness. It was as though she could not do another step or
pick up a cup. She sat down, watched TV for about an hour, and focused on
her trip to deliver family Christmas gifts.
Around 9pm, as she lay down in bed, her whole right arm was aching
and she just couldnt sleep. I just plain didnt feel very good... It just ached
from the shoulder down to the finger tips, Mrs. Kaufman explained. The ache
was constant for most of the time, when it was there... But in her mind, it
would be stupid for it to be heartburn because it did not bother before.
However, the ache changed and then it would be quiet for just a very short
period of time and then it would come back for a long time, she explained, I
was trying to get comfortable and I couldnt. She also got cold enough that she
turned on the heat. Mrs. Kaufman could not sleep that Friday night because the
pain was constant and the chest pain never went away.
By midnight, the bottom of her jaw, particularly near the joint, began to
ache and feel painful. The aches went into both ears as if she had a toothache.
This was unusual to her because she had had surgery on her ear and I really
dont suffer from earaches or anything. And with false teeth, I dont usually
have a problem with the bone in my mouth, she explained. Normally, if she
feels this way, she puts on oxygen and uses her inhalers to feel better. By one
oclock Saturday morning, neither the inhalers nor the Tylenol with codeine
relieved her symptoms. But, it was differentshe said in a questioning tone of
voice, it wasnt the breathing... I mean it was the pain. It wasnt lack of oxygen.
It was the pain. The pain was different, yeah.
2


She considered taking Theraflu but did not because she started getting
nauseous. She thought she had to throw up but did not. She just had dry
heaves. This up and down pattern lasted through the night. Mrs. Kaufman
thought she had indigestion and a simple remedy like crackers or peppermint
would make my indigestion go away. In lieu of finding peppermint, she ate an
Altoids but they didnt help. She never woke her husband up to tell him how
she was feeling because of his work.
Saturday morning she woke up and didnt feel as sick as she did Friday
night. But I just didnt feel well," she emphasized, I was just out of sorts. She
was no longer cold like the night before. She connected feeling cold to her
pneumonia symptoms. When she called her sister to tell her she was up all night,
her sister told her not to visit, but Mrs. Kaufman insisted. It seemed like it took
me forever to drive, she said, Probably because I didnt feel good. I... I just
didnt feel good. She suddenly felt like she needed the oxygen. I didnt have the
pain like I did. However, she just ached from head to toe. She thought, Man,
I got the flu!" but was surprised because she just had a flu shot. She mentioned
how she felt to her granddaughter, who then told Mrs. Kaufman she was worried
about her. However, because Mrs. Kaufman sees herself as a very stubborn
person, once I have my mind made up, its usually made up, and she does not
listen to others if she disagrees.
By 1:30pm, her oxygen still did not work and she told her sister she still
did not feel well. Again, her sister told her she should go home. Mrs. Kaufman
felt like she had to do Christmas she explained, I mean, it still wasnt
anything to think was a heart attack. About an hour later, the 'pain was back...
You know. It was just there. I was just hanging on my chest, like, just an
ache.. .it felt a little more like .. .heartburn almost. After returning home, Mrs.
3


Kaufman said, The longer I sat there, the worse I was feeling.. .1 like the pain
was more intense. The pain was solid.. .not a come and go pain at all.
She was sick of the pain and I needed to sleep is what I was thinking...
Ive got to go get something to relieve this pain so I can sleep. I didnt ever once
think of my heart, she recalled, Never did. Never. I thought it either had to do
with my breathing or indigestion because I always heard that heart attacks were
on the left side. And... all [my] symptoms were on the right side. The chest
pain, the arm pain, the jaw pain. My jaw was aching. It would come and go.
At 7:30pm, she decided to call the Ask-a-Nurse line. [I] couldnt make
up my mind if thats what I wanted to do... It either had to do with my breathing
or something that I ate.. .or something I didnt eat. She told the nurse her
symptoms and that she had not slept the night before. She begged, Please dont
suggest that I come in because Ive been hospitalized 10 days ago, she laughed,
Im not buyin into this. The nurse tried to encourage her further to go to the
hospital, You know you really need to be in here. After being convinced
Mrs. Kaufman said, Well, Ill drive in. But the nurse replied, Well no! We
dont want you driving m. She retorted back, Well yes... Ive been driving all
day, she argued back. The nurse argued further, We really would feel better if
you didnt.
Her number one reason for not contacting the ambulance was because of
the expense and her misperception that insurance doesnt cover it. She
considered calling a taxi but called her grandson instead because he lives closer
and she did not think she was having a heart attack. She thought her symptoms
were urgent enough to come into the hospital but not urgent enough to stay
overnight. She never thought her symptoms were life threatening; it never
once entered her mind. Had her symptoms been on the left, and scared her
4


enough to think she needed an ambulance, she would have thought she was
having a heart attack. To her, scared enough meant that she would have chest
pains and sweats. She got cold, not sweaty. I mean I was cold. I turned my
electric blanket on. I didnt get the sweats. So, if Id had the chest pain and if
Id had the sweats, and if I would have had a hard time breathing... with the
three of those, I would have called 911 in a heartbeat, she explained.
As the ER doctors examined her, Mrs. Kaufman remembered thinking,
This is really stupid... Its on the right side.. .What does this doctor know, its
on the right side. When her nurse confirmed, Yes, youve had heart attack,
Mrs. Kaufman was stunned literally, she laughed, Youve got to be kidding
me. Her MI experience concluded in a triple bypass two weeks later.
She recalled discussing her heart attack risk with her doctor. Well, Im a
diabetic, thats something that they tell you. All diabetics are at risk. And it runs
in my family. My father died from heart disease and my mother did too. The
heart attack surprised her only because that wasnt what I was thinking of.. .to
me the symptoms would have to be different. Ive always thought well if that
happens I know what a heart attack is and you know what, I didn 7. Her image
of heart attack was essentially the same as what she experienced but on the left
side. For some reason, I kept thinking, my heart was over farther. The idea of
having a heart attack surprised her because she had a stress test in the past, twice,
and nothing ever showed up. They always told me everything was good. This
MI confused Mrs. Kaufhian and she wondered and questioned why the blockages
did not show up in the past. And Im always afraid if I come in, like the little
boy crying wolf, you know, I dont want to come in and its nothing... especially
at [this hospital], they send you away sometimes.
5


Mrs. Kaufmans experience illustrates the confluence of factors that
configure womens treatment seeking behavior for heart attack symptoms. These
factors can expedite or impede the decision to seek care and are examined
throughout this dissertation. Our understanding of them is critical to reducing the
problem of delayed treatment seeking behavior among women. The in-depth
interviews granted by the forty-three women in this study provided insight on
treatment seeking behavior for heart attack symptoms that have been
underappreciated in the literature.
The Problem of Delayed Treatment Seeking Behavior
for Myocardial Infarction Symptoms
Coronary Heart Disease (CHD) is the number one cause of death in
women in the United States. Acute Myocardial Infarction (AMI) is the leading
killer of women regardless of ethnicity (Meischke et al. 1998; American Heart
Association 1996; AHA 2003). In 1996,1.5 million people had an MI in the
United States and one third of them died. Women have their first AMI an average
of ten years later than men and at an age when co-morbidities are more likely to
be present. Women experience greater morbidity and disability, have a worse
prognosis than men, and are almost twice as likely as men to die within one year
after an AMI (AHA 1997; McSweeney et al. 2000; Schoenberg 2003; Turek
2000). Mortality for AMI is greater for minorities and the reasons for the
disparities remain unclear.
The term MI reflects a loss of cardiac myocytes (necrosis) caused by
prolonged ischemia (Alpert et al. 2000). Ischemia is the result of a perfusion-
6


dependent imbalance between supply and demand (Alpert et al. 2000).
Reperfusion therapy can significantly reduce morbidity and mortality when
administered within the optimal window of opportunity (defined in the clinical
literature as less than 3 hours from symptom onset) (Dracup et al. 1995). It is
also effective when administered within 12 hours of symptom onset but
effectiveness diminishes with each hour of delay (Alpert et al. 2000). The Global
Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded
Coronary Arteries (GUSTO) trial reported that longer presentation delays and
treatment delays were both associated with increased mortality rates (Newby
1996).
Thrombolytic therapy is effective in reducing myocardial damage that
occurs as a result of AMI. The Gruppo Italiano per lo Studio della Streptochinasi
nelT Infarcto Miocardico (GISSI-1) study found reductions in mortality of 47%
when streptokinase was administered within 1 hour of infarction and 23% when
administered within 3 hours of infarction (GISSI-1 1986). The effectiveness of
reperfusion therapy is inversely related to the time between AMI symptom onset
and therapy. Approximately 25% of patients wait more than 6 hours before
seeking care (Yarzebski et al. 1994).
Despite its effectiveness, only a fraction of patients potentially eligible for
reperfusion therapy actually receive it, primarily attributed to delay in seeking
medical treatment (Meischke et al 1998; Goldberg et al 2002; Lee et al. 2000).
According to findings from the Corpus Christi Heart Project, the use of
thrombolytic therapy was 34% less among Hispanics, limited by the delayed time
from onset of symptoms and arrival at the hospital (Canto et al. 1998). Several
reasons for delayed treatment seeking behavior have been cited, namely 1)
failure to recognize symptoms as significant medical problems that require
7


immediate attention, 2) failure to recognize symptoms as heart attack related, and
3) barriers associated with accessing the health care system, such as financial and
structural barriers.
Females (Goldberg et al. 2000; Scheifer et al 2000; Dracup et al 1995)
and ethnic minorities are most likely to delay seeking care promptly for AMI
symptoms and delay longer than males and Whites. While this important finding
has been known for some time, few research studies focus on the contextual
variables (Lee et al. 1998; Scheifer et al. 2000) associated with womens
experiences with delay and even fewer studies include Latinos/as. Moreover,
most studies of AMI patients examine Whites, males, and middle to upper class
populations (Lee et al. 1998; Scheifer et al. 2000; Dracup et al. 1995).
The Purpose of the Study
This research project explored how women of different ethnic
backgrounds and languages experienced, recognized, and responded to
symptoms of acute myocardial infarction. This included an emphasis on the
differences among women with delayed response times to symptoms versus
relatively rapid response times from symptom onset to arrival in the emergency
room. The study examined factors that influenced women along their trajectory
for seeking care. The main objective of the dissertation is to refocus what is
known about treatment seeking behavior for acute myocardial infarction and,
ultimately, to address how the problem of delay in seeking care can be improved.
8


Contribution of This Study:
Furthering the Understanding of the Problem
One strength of this dissertation lies in its inclusion of underserved
women and ethnic minorities. Differences in symptom perception and treatment
seeking behavior among women of various demographic backgrounds have not
been studied. The research on womens experiences of an AMI has been limited
by small sample size (Scherck 1997), ethnic homogeneity (McKinley et al.
2000), and a lack of diversity in socioeconomic backgrounds. Ethnic minorities
and economically disadvantaged women have been underrepresented in research
focused on AMI. This study included Latina women as well as non-Latina
women (Denvers population is one quarter Hispanic), to allow a comparison of
how culturally dissimilar women interpret and seek treatment for AMI
symptoms.
Emphasis was placed on the progression of symptoms, the presence of
prodromal symptoms, and how women responded to symptoms since early
recognition of AMI symptoms and the decision to seek treatment will improve a
womans chance of survival (Dracup et al. 1995). The limited data to date
suggest that womens special issues affect their actions and outcomes (Reigel
et al. 1995). Reigel and Gocka defined special issues to include the presence of
prodromal symptoms, atypical symptoms, a lower perception of personal risk,
meaning of symptoms, discussions about symptoms with others, and self-
treatment strategies (Reigel et al. 1995). This study identified and defined these
issues in greater depth.
Findings from patient interviews provided the foundation for a medical
record review of 43 Latina and non-Latina women. The chart reviews compared
patients descriptions (during the interviews) of symptoms and events leading up
9


to their hospital presentation, to the initial descriptions and times documented in
the emergency room. Additional data collected in the chart reviews included 1)
sociodemographic components, 2) social history, 3) clinical data including
patients and their families medical history, especially notes regarding symptom
history and previous treatment seeking behavior for these symptoms, 4)
utilization of guideline-based therapies, 5) delay time, and 6) the chief complaint.
This study provided the first steps in reevaluating the typical
experiences of women seeking care for AMI symptoms and previously
recommended strategies for recognizing and seeking care. It recorded AMI
experiences from populations who are underrepresented in the behavioral science
literature. Results of this study promise to improve the education of health care
teams and providers. They suggest ways for women to recognize attributes of
symptoms and effectively communicate symptoms to providers. Findings may
enhance the assessment and evaluation skills of health care providers who are
often the first point of contact for women seeking professional advice for their
symptoms. These include but are not limited to primary care providers and
nursing triage staff, emergency response teams, and clinicians working as advice
nurses on Ask-a-Nurse hot lines.
The Theoretical Framework
The study was influenced by theory drawn from multiple disciplines
including psychology, medical anthropology, sociology, behavioral science, and
biomedicine. Using narratives, the study took a largely inductive approach to
derive an understanding of how women experience heart attack symptoms. This
approach helped me to make sense of womens sensations, descriptions, and
10


actions which were useful to reflect on culture, family, and how others may have
influenced womens thinking as their symptoms evolved.
The results of this research reveal that several theories and constructs are
useful to describe how perceptions and beliefs affect the behavior that is critical
to a womans early recognition and favorable outcome in AMI. No a priori
theoretical model was assumed, but I was mindful of potentially relevant
theoretical paradigms that permitted extrapolation of relevant themes from
interview data.
The Organization of the Dissertation
The structure of this thesis is as follows. The rationale for this type of
study was briefly reviewed in this chapter. Chapter 2 is a review of the literature,
with a focus on treatment seeking behavior among women, non-Whites and
Latinos/as. Within treatment seeking behavior, the chapter emphasizes prodromal
symptoms, atypical/less typical symptoms, and perceptions and discussions
relating to treatment seeking strategies. The chapter concludes with a review of
competing demands.
Chapter 3 focuses on how the theory and selected constructs informed the
study. Chapter 4 describes the research design, selection of study sites, and
sample selection, and details the steps taken to conduct the research under
HIPAA regulations.
Chapter 5 discusses the data analyses from the medical chart reviews. The
results from the qualitative interviews are found in Chapter 6 which lends
support to the data presented in Chapter 5. Chapter 6 also introduces four
typologies and themes that emerged from the interviews. Chapter 7 elaborates on
11


the typologies introduced in Chapter 6 by reviewing representative cases chosen
for each typology. A vignette example and patient career timeline are included
for each case.
Chapter 8 begins with a discussion of a review of the main findings from
the project and concludes the dissertation with recommendations for providers
and patients.
12


CHAPTER 2
REVIEW OF THE LITERATURE
The two main perspectives that address the research problem are
discemable in the literature. The first is the medical perspective which
concentrates on how people present clinically for AMI. This perspective focuses
on how to identify patients and what information can be drawn upon from the
medical record. The second perspective in the literature discusses the AMI
experience from the patients perspective. It includes their thoughts, decision
making processes, recognition and response to symptoms. My review of the
literature sought to understand both sides of the problem, as the problem is a co-
construction of both perspectives. My project was oriented more towards
representing the patients perspective.
Recent literature addresses womens unique risk factors for AMI, but
there has been little attention focused on the symptoms women experience prior
to an AMI. The widely publicized symptoms indicative of AMI are based
predominantly on studies of men and have been generalized to women
(Schoenberg et al. 2003). There is a lack of congruence between the symptoms
experienced by men and those experienced by women, which has complicated
diagnosis. Classic symptoms include severe pain in the center of the chest,
which radiates to the shoulders or arms, accompanied by nausea and shortness of
breath (AHA 1995). Atypical manifestations include neck and back pain, jaw
pain, nausea, or symptoms that have prodromal behavior (symptoms that come
and go prior to an MI and change after an MI) (McSweeney et al. 2000;
Goldbergl998). Although limited, previous qualitative and descriptive studies
about womens symptomatology prior to an AMI suggest that women have
13


unique symptoms and experiences (McSweeney 2000; LaCharity 1999). What
proportion of women experience classic as opposed to atypical symptoms is
unknown. It is possible that the atypical perception and description of symptoms,
and associated treatment seeking behavior, contribute to the delay in the decision
to seek care, and thus contribute to poorer outcomes for women with acute and
chronic heart disease (McSweeney et al. 2000; Zerwic 1998, Dempsey et al. 995;
Meischke et al. 1998).
This chapter begins with an introduction to prodromal symptoms which,
although commonly experienced by women, are recognized less often as an
important feature of MI symptoms. The chapter continues with an exploration of
other symptoms considered to be atypical or less typical followed by an
examination of how symptom recognition translated to action and treatment
seeking behavior, with an emphasis on non-Whites. The acute MI and its
outcomes are examined for Latinos. The chapter concludes with an exploration
of common strategies and the impact of competing demands on seeking
treatment.
Prodromal Symptoms for AMI
The prodromal phase is a symptomatic, initial period which precedes the
onset of acute signs and symptoms (Alonzo 1986). Prodromal signs and
symptoms are thought to be a consequence of compromised arterial flow to the
myocardium and therefore it is not surprising that prodromal symptoms reported
are almost identical to those reported during the acute phase of an AMI (Alonzo
1986). The symptoms experienced in this initial period are usually intermittent,
relatively mild, and frequently non-specific (Alonzo 1986). About 50% of
14


patients with AMI present with intermittent chest symptoms usually not
perceived as chest pain (Bahr 2002). Such symptoms have been called
preinfarction angina, prodromal unstable angina, waxing and waning angina,
winking and blinking angina, premonitory angina, and so on (Bahr 2002).
Approximately 50% of AMIs are heralded by unstable angina pectoris (Bahr
2002). If this is accurate, then the identification of patients with prodromal
symptoms may be useful (Bahr 2002).
Dracup et al. (Dracup et al. 1995) observed from patient interviews that
approximately one third of AMI patients do not report an abrupt onset of
symptoms and frequently have difficulty identifying the time of symptom onset.
Rather, these patients may report vague symptoms or symptoms that wax and
wane over time and sometimes disappear completely (the prodromal period).
This period of milder symptoms usually does not prompt an individual to seek
care (Dracup et al. 1995). Some say this is due to the vague nature of the
symptoms and that women do not perceive symptoms in this period to be
debilitating.
Women generally lack awareness of their risk for CHD and often ignore
both prodromal and acute symptoms. Womens symptoms often do not match
their expectations and hence they do not associate their symptoms with heart
problems (McSweeney et al. 2000). For example, Hofgren and colleagues asked
men and women to describe their symptoms for suspected AMI, how they
assessed the situation, and how they made the final determination to seek medical
help (Hofgren et al. 1995). Some experienced chest pain within the 48 hours
before the onset of symptoms that brought them to the hospital. Women reported
more prodromal symptoms than men. However, the non-specific character of the
symptoms reported, especially a generalized feeling that something is wrong,
15


made the severity of the situation difficult for women to assess them
(McSweeney et al. 2000; Hofgren et al. 1995). Because of the mildness and
ambiguity of prodromal symptoms, women may be less likely to tell family and
lay others about these symptoms (Alonzo 1986) and when they do make others
aware, family and lay others may not provide assistance in coping with the
symptoms (Alonzo 1986).
The occurrence of prodromal symptoms in women is a dimension of
symptomatology not clearly understood or emphasized in the treatment seeking
literature (Goldberg et al. 2000) even though prodromal symptoms are frequently
experienced by women (Hofgren et al. 1995; Alonzo 1986). McSweeney and
Crane interviewed 40 predominantly White, English speaking women from
Arkansas and found 92.5% of the participants experienced at least one prodromal
symptom beginning from a few weeks to two years prior to their AMI, and 26
participants experienced five or more prodromal symptoms (McSweeney et al.
2000). These symptoms included unusual fatigue, shortness of breath, and
discomfort in the shoulder blade area or chest sensations.
In a Swedish study, approximately sixty percent of patients with
confirmed AMI (N=454 men and women) experienced prodromal symptoms
other than chest pain before the onset of symptoms that brought them to the
hospital (Hofgren et al. 1995). Using a self-administered questionnaire, most of
these patients reported that prodromal symptoms lasted more than one week
(Hofgren et al. 1995). Prodromal symptoms occurred not only in patients with
AMI but also occurred in patients who were admitted with other ischemic heart
disease. Prodromal symptoms other than chest pain that were reported most
frequently in this study were tiredness, chest discomfort, pain in the arms, and
gastrointestinal symptoms. Although the process of seeking care was not the
16


emphasis of the study, it was interesting that 33% of patients reported visiting a
physician before being admitted to the hospital. This behavior is not surprising
since prodromal symptoms tend to be vague and non-specific. However, the
time and decisions associated with visiting a doctor do contribute to total
prehospital delays which could be reduced if prodromal symptoms were better
understood and recognized as possible warning symptoms of a heart attack.
Identification of the prevalence of prodromal symptoms in women may
also help practitioners to recognize and trace the frequency of symptoms that
may ordinarily go undetected. Understanding prodromal symptoms may prompt
the decision to seek care sooner rather than waiting for the arrival of acute
symptoms, the point in time when many patients decide to seek medical care. It
is speculated that their existence might explain why women do not recognize
AMI symptoms. Some experts are skeptical about the potential to empower
women to act upon prodromal symptoms without provoking malignant
attention (Alonzo 1986) where ones life centers around bodily sensations.
Also, the social and economic costs of acting upon prodromal symptoms should
be considered. Striking a balance between the extremes of informing the public
adequately and preventing over-utilization of emergency medical response
systems and emergency rooms needs further consideration (Alonzo 1986).
Knowledge learned about prodromal symptoms may be more appropriately
utilized first among health care professionals. Health care professionals could
recognize this period of prodromal symptoms, and advise women to seek care
promptly.
17


Atvpical/Less Typical Symptoms Experienced
by Women
Women commonly experience AMI atypical symptoms (McSweeney et
al. 2000; LaCharity 1999; Meischke et al. 1998; Meschak et al. 1998). The
American College of Cardiology (ACC)/AHA 2002 Unstable Angina Guideline
Update (Braunwald et al. 2000) notes special considerations for identifying acute
coronary syndromes in women, diabetic patients and elderly patients:
Women may present more frequently than men with atypical chest
pain and symptoms. Diabetic patients may have atypical
presentations due to autonomic dysfunction. Elderly patients may
have atypical symptoms such as generalized weakness, stroke,
syncope, or a change in mental status (Braunwald et al. 2000).
Beyond this, even typical presentations of AMI may be represented by a
myriad of symptoms, with complexities as depicted in the consensus document
of the Joint European Society of Cardiology/American College of Cardiology
Committee for the Redefinition of Myocardial Infarction (2000). This document
describes the clinical presentation of possible ischemic symptoms. Upon
reviewing this definition, it is understandable that women, known to spend
excessive time recognizing and acting upon AMI symptoms, delay seeking
prompt medical care. Furthermore, it supports the rationale to focus on womens
description, recognition, and treatment seeking behavior for AMI symptoms:
Possible ischemic symptoms include chest, epigastric, arm, wrist
or jaw discomfort with exertion or at rest. The discomfort
associated with acute MI usually lasts at least 20 minutes, but may
be shorter in duration. The discomfort may develop in the central
or left chest and then radiate to the arm, jaw, back or shoulder.
18


light-headedness. This discomfort can develop in the epigastrium
(often confused with indigestion), arm, shoulder, wrist, jaw or
back, without occurring in the chest, but such a pattern is atypical.
This discomfort is not affected by moving the muscles of the
region where the discomfort is localized, nor is it worsened by
deep inspiration. The discomfort is not positional in nature.
Symptoms can also include unexplained nausea and vomiting,
persistent shortness of breath secondary to left ventricular failure
and unexplained weakness, dizziness, lightheadedness or syncope,
or a combination of these. These symptoms may be noted in
association with chest discomfort or they may occur in the
absence of chest symptoms (Alpert et al. 2000).
Although many patients have symptoms such as those just described, these
complaints may go unrecognized or may be erroneously labeled as another
disease entity, such as indigestion or a viral syndrome (Alpert et al. 2000).
Although this definition was intended for health care providers and not
patients, there is clearly potential for miscommunication and misinterpretation of
symptoms.
In a Massachusetts study of 550 women and 810 men, Goldberg et al.
found that women were more likely than men to complain of 1) pain in the back
(38% vs. 22%), 2) neck pain (37% vs. 24%), and 3) to experience jaw pain,
nausea, and shortness of breath (Goldberg et al. 1998). Atypical symptoms that
have been studied differ in their anatomical location and characteristics. The
words used to describe atypical symptoms vary in quality, intensity, location, and
frequency (Dempsey et al. 1995; Hofgren et al. 1994). For example, seventy
percent of the episodes of myocardial ischemia caused by CAD are estimated to
be silent or unnoticed by [women] (Warner 1995). When symptoms are
atypical mid prodromal, women may think that no symptoms were experienced
but in fact, they may just not have been recognized as heart attack symptoms.
19


Women with confirmed AMI have also been found to use stronger word
descriptors than men (Hofgren et al. 1994).
It is not clear whether Latinas experience atypical symptoms since
Spanish-speaking women have been excluded or underrepresented in many
studies. Inclusion criteria often require that participants speak English
(McSweeney et al. 2000; Dempsey et al. 1995; Murray et al. 2000; McKinley et
al. 2000), have transportation (Murray et al. 2000), or be enrolled in a health
maintenance organization (Dempsey et al. 1995). These criteria would exclude
many Latinas and women of lower socioeconomic status who cannot afford
insurance, personal transportation, etc. The Corpus Christi Heart Project
compared AMI symptoms in 296 Mexican Americans (107 were women) versus
293 non-Hispanic White Americans (94 were women) (Meschak et al. 1998). In
support of the research proposed here, the results showed greater differences
between the sexes than between ethnicities. While the authors methods
suggested that data were collected on the symptoms preceding the coronary
event, those data were not reported (Meschak et al. 1998).
Women and Treatment Seeking Behavior
Biomedical models applied to this phenomenon quantify demographics,
clinical characteristics, and time delay, and assign risk status to patients. While
important, these models only offer partial explanations, and have methodological
limitations and generalizability issues. Methodological problems include missing
data, reliability, inconsistent definitions of delay time, inconsistent measurements
of symptom onset time, and different interpretations of chest pain symptoms.
20


The greatest deficit in these studies is the repeated analysis of homogenous
populations, resulting in interpretations that raise questions about generalizability
to women and Latinos/as. Fewer studies assess symptom attribution or patient
motivation, particularly with chart abstraction studies (Meischke et al. 1998;
Canto et al. 1998). Other quantitative methods do not provide subjects the
opportunity to elaborate on their responses to interviewer questions. Education
strategies designed to reduce delay may continue to be inadequate unless
biomedical models are integrated with theoretically-based behavioral models.
Behavioral models have the advantage of elaborating on somatic and emotional
awareness, perceived vulnerability, and socioenvironmental variables not
considered in biomedical models (Dempsey et al. 1995; Dracup et al. 1995).
Schoenberg et al. (Schoenberg et al. 2003) reviewed the limitations of the
biomedical perspective. Their criticism of the biomedical model included
minimal acknowledgment of the contribution of social, structural, individual, and
cultural complexities, and the ongoing need to explore factors and circumstances
that shape time to treatment and individual treatment decisions. As they
suggested, there is an inextricable linkage to broader social and structural
constraints, including the social construction of standard cardiac symptoms
based on male norms that ultimately confuse symptom detection (Schoenberg
et al. 2003). Another example compounding the social constructions with
treatment seeking behavior is womens negative encounters with healthcare
providers who discount their knowledge, competing social demands, and
structural barriers to accessing health care (Schoenberg et al. 2003).
Dempsey, Dracup and Moser addressed some of these deficiencies
(Dempsey et al. 1995). These investigators focused on the psychosocial
processes and found that self-treatment strategies applied by women (N=16, 6
21


Hispanic, 8 White, 2 Black), in an attempt to maintain control of the situation,
extend the delay process. While this study contributed significantly to the
knowledge on treatment seeking behavior, the sample selection weakened the
generalizability of the studys findings; only English speaking women were
enrolled from a health maintenance organization. However, in their proposed
Control Theory model, the investigators list influencing factors during the
maintaining control phase which they suggest are responsible for the longest
period of prehospital delay time (Figure 3.3). This phase is worth exploring
further in the proposed study. Examples of the influencing factors include
attributing symptoms to a benign cause, past symptom experience, beliefs about
women and CHD, commitments, and concern for others. The second phase of
the Control Theory model is called relinquishing control during which females
seek advice from lay consultants- spouse, family, friends, co-workers. Others
have described the impact of lay consultation on care seeking behavior for cancer
among Latinos/as (Larkey et al. 2001). It is this step in the model where there is
potential to distinguish behavior between Latinas and non-Latina women. We
may be able to fill a gap in the literature by describing how culture, perceptions
about and barriers to accessing care affect Latinas prehospital delay in seeking
care for AMI symptoms.
Treatment Seeking Behavior in Non-White Populations
The contribution of ethnicity in treatment seeking behavior has not been a
major emphasis in any studies seeking to draw firm conclusions about how
patients evaluate symptoms prior to seeking care (Dempsey et al. 1995;
22


Meischke et al. 1998; McKinley et al. 2000), particularly in Hispanics. Lee et al.
(Lee et al. 2000; Lee et al. 1998) provided an interesting comparative analysis of
Blacks and Whites by demonstrating the significance of ethnicity in symptom
perception, evaluation and treatment seeking behavior. A convenience sample of
132 study participants, 69 men and 63 women, with a first or recurrent AMI were
interviewed in a large Midwestern medical center and two community hospitals.
Significantly more Black patients than White patients were female (63% vs.
39%, respectively). The investigators found that 67% of all patients reported that
they made their own decision to seek treatment and there was no difference
between ethnicities. Forty percent of all patients came directly to the ER; 28%
sought medical attention by calling 911. About 60% of all patients were driven to
the hospital by a spouse, a family member, or a friend. Seventy-one percent
evaluated their symptoms as a minor problem or attributed the symptoms to other
problems such as stomach upset, indigestion, influenza, or stroke. Blacks (16%)
evaluated their symptoms as less serious than did the sample of Whites (35%).
More Blacks (29%) experienced intermittent symptoms than Whites (9%). Lee et
al. concluded that AMI manifests differently in Blacks vs. Whites (Lee et al.
2000; Lee et al. 1998). The decision time for Blacks was at least twice as long as
the decision time for Whites, measured in both mean and median times (Lee et
al. 1998). The authors found ethnicity had a greater affect on delay time than the
presence of atypical symptoms and suggested this effect may result from the way
patients of different ethnic backgrounds evaluate their symptoms in relation to
treatment seeking action (Lee et al. 1998).
In a subset analysis of the same sample (N=132), Lee et al. analyzed
clinical symptoms and treatment seeking behavior by examining the interactive
effects of different clinical symptoms and race on delay time among Blacks and
23


Whites (Lee et al. 2000). The investigators found that delay times for Blacks
with chest pain were longer than for Blacks without chest pain, whereas delay
times for Whites with chest pain were shorter than for Whites without chest pain.
Among Blacks, dyspnea at the onset of MI was associated with a shorter delay
time in seeking treatment, whereas chest pain was associated with longer delays.
The major finding from this analysis was that small percentages of patients
presented within the 3-hour time window of opportunity for effective
thrombolysis: 54% of Whites and 40% of Blacks. Only 13% of Blacks came to
the ER within one hour, whereas 30% of Whites came to the ER (Lee et al.
2000). These findings suggest that the window of opportunity is missed by the
majority of all AMI patients.
The work by Lee et al. shows that ethnicity accounts for some of the
differences in the patients symptom perception and evaluation to seek treatment
(Lee et al. 2000). Although the findings from this research add justification for
the proposed study on Latinas treatment seeking behavior, one weakness in their
work (Lee et al. 2000; Lee 1997; Lee et al. 1998) was limiting their analysis to
biomedical interpretations rather than examining ethnicity within a
sociobehavioral context. Behavioral theory would have provided an in-depth
perspective and enhanced their interpretations. Several studies note longer delay
times among Hispanics in comparison to non-Hispanic Whites (NHW). This
supports the need to investigate the context of sociocultural factors, particularly
in Latinos/as.
24


Latinos/as and AMI
Several terms are used to describe the Hispanic population. Hispanics
also identify themselves as Latino/a, Hispanic, Mexican, Mexican-American, or
Chicano/a, for example. Since there are inconsistencies in referring to this
population, the term Latina is used in this dissertation as an umbrella term for
women who self-report Mexican, Puerto Rican, Cuban, Central and South
American, and Spanish origin (Espinoza 2002). One exception to using the term
Latina is when previous literature is referenced and the terminology chosen by
the investigators of each study is used.
Canto et al. (1998) recognized the gap in the literature reporting on
demographics, cardiovascular risk factors, cardiovascular procedures, and
mortality among non-Black minorities. These investigators used the Second
Registry of Myocardial Infarction data (NRMI 2) for a chart abstraction
comparative analysis (N=275,046; 35% women) of the demographic and clinical
characteristics, treatment patterns, and clinical outcomes among Hispanics,
Asian-Americans, and Native Americans in comparison to Whites (Canto et al.
1998). Hispanics (N=6,896) represented 3% of the study population. Even
though the study did not focus on the process of seeking care for AMI, several
findings are worth noting (Canto et al. 1998). Hispanics were significantly
younger and male. Regarding prior cardiac history, Hispanics were less likely
than Whites to have angina pectoris, previous AMIs, percutaneous transluminal
coronary angioplasty (PTCA), or coronary artery bypass graft surgery (CABG).
However, in comparison to Whites, Hispanics had significantly more diabetes
mellitus and hypertension and less smoking, hypercholesterolemia, and a
25


reported family history of CAD. Despite this, Hispanics less than 40 years old
had a fourfold higher mortality risk than similarly aged Whites.
Other findings in Canto et al. that are worth noting, relevant to this study,
include the characteristics about patient presentation and pharmacologic therapy.
Hispanics had significantly longer delay times from AMI symptom onset to
arrival at the hospital in comparison to Whites (135 minutes vs. 122 minutes,
respectively). Hispanics used ambulance services less than Whites (44% vs. 47%,
respectively). However, it is interesting that these differences were only
attributed to the limited availability of ambulances or having to travel longer
distances to the hospital. The authors did not acknowledge that delay times and
transportation issues are also related to individual and cultural beliefs and
socioeconomic barriers in seeking care. For pharmacological therapy
administered within the first 24 hours after admission, Hispanics received beta-
blockers less often than Whites (42% vs. 46%, respectively) and lidocaine (15%
vs. 17%, respectively). At discharge, beta blockers and calcium channel blockers
were prescribed significantly less often to Hispanics than Whites, while ace-
inhibitors were prescribed significantly more often to Hispanics than Whites
(Canto et al. 1998). Delay times and discharge medications were collected in the
chart review of this study.
Using admission profiles and medical records to determine ethnicity was
one weakness identified in the research by Canto et al. (Canto et al. 1998). It is
not known whether patients self-reported ethnicity or if hospital staff subjectively
assigned ethnicity based on surnames or physical features. Reasons for
disparities in delays in receiving treatment beyond thrombolytic therapy are not
as well known but may be related to treatment bias or other socioeconomic
factors that affect patients ability to pay for treatment.
26


Canto et al. compared their findings to other studies examining racial
disparities in the process of care and found conflicting results. Another analysis
of administrative data in California found less angiography, PTCA, and bypass
surgery among Latinos (Carlisle et al. 1995; Giacomini 1996). Canto et al.
attributed the differences found in these earlier studies to smaller databases used
with less clinical detail available for analysis (Canto et al. 1998).
Given that females are more likely than males to delay seeking care for
AMI symptoms, it was unfortunate that a large study like Canto et al. (Canto et
al. 1998). failed to investigate gender separately. Finally, no interpretation was
provided to explain the disparities in patients receiving discharge medications.
This is important to interpret because guidelines have been published, i.e. the
AHAs Get with the Guidelines Program, that outline the benefits of prescribing
the optimal medications at discharge. In our study, I collected chart review data
about discharge medications which could be examined by ethnicity.
Canto et al. recognized that education, culture, and socioeconomic factors
influence each ethnic groups views about AMI symptoms (Canto et al. 1998).
They recommended targeting non-Black minorities in policy initiatives aimed at
reducing hospital delay via local and national campaign efforts, public service
announcements, and better doctor-patient relations. These recommendations
should be taken with caution, as current evidence does not elucidate the interplay
of ethnicity and MI symptoms. Launching educational campaign efforts with
messages that may not resonate with Latinos/as or other non-Black minorities is
unlikely to reduce prehospital delay time in AML Other public health initiatives
(Wong et al. 2002; Wigton et al. 2002) are finding that campaign strategies have
little impact on shifting health beliefs and behaviors among audiences who are at
different stages of behavioral change from the intended intervention.
27


Interventions aimed to reduce prehospital delays for AMI symptoms need to
understand how symptom interpretation and subsequent treatment seeking
behavior among different cultures impact delay time.
Cohen et al. identified that the time from symptom onset to presentation
and the time from presentation to treatment for myocardial infarction were longer
for Hispanics than non-Hispanics (Cohen 1999). Clark et al. and others found
that inner city Hispanics had a longer prehospital delay than White patients with
suspected AMI, and that most of the observed delay was due to the time it took
for patients to decide to seek medical care following the onset of symptoms
(Clark et al. 1992; Canto et al. 1998). Factors influencing the prolonged
decision-making among Hispanics include cultural differences in the perception
of disease, suboptimal health insurance coverage, limited health care access, and
lower education levels (Cohen 1999; Clark et al. 1992;Canto et al. 1998).
It seems that the disproportionate risk Hispanics have for diabetes,
hypertension, and obesity, (and excessive risks resulting from prehospital delay
time for MI symptoms) would negatively impact health outcomes and would be
reflected by higher cardiovascular disease (CVD) mortality rates in Hispanics.
However, the Hispanic Paradox offers an explanation for longer AMI delay
times yet similar clinical outcomes compared to Whites. The Hispanic Paradox
is the observation that Hispanics have lower all-cause and cardiovascular
mortality, despite increased diabetes and obesity, lower socioeconomic status
(SES), and barriers to health care (Hunt et al. 2002). Examining diabetic
participants from the San Antonio Heart Study, Hunt et al. challenged this
observation, hypothesizing that if it were true, diabetic Mexican Americans
would have fewer all-cause and CVD deaths than diabetic NHW. They found
evidence contradicting the Hispanic Paradox by comparing all-cause and
28


cardiovascular mortality among diabetic NHW, US bom Mexican Americans,
and Mexico bom Mexican Americans (Hunt et al. 2002). Compared to NHW,
US bom Mexican Americans were at 66% greater all-cause and cardiovascular
mortality. Hunt et al. concluded that it is important to differentiate Mexican
Americans by birthplace. Based on these conclusions, birthplace and
acculturation levels were identified and established during the interviews.
The high awareness of the Hispanic Paradox draws attention away from
the fact that CVD remains the leading cause of death for Hispanic men and
women (AHA 2003). As a result it is possible that a misperception developed
that Hispanics are not at major risk for CVD events, including AMI.
Overemphasis on one disease in a population can foster the development of
misperceptions. Consider the example of women and public health efforts
centered on breast cancer. Because of the high media attention given to breast
cancer, women often conclude incorrectly that they have minimal risk for AMI
compared to men. Although the incidence of CVD is higher in men, CVD (not
breast cancer) remains the leading cause of mortality among women in the
United States (AHA 2002).
Recent studies examining incidence and prevalence show that Hispanics
may not be as protected from heart disease as previously suggested. The Corpus
Christi Heart Project provided the first report showing that Mexican-Americans
have greater incidence of hospitalized AMI than NHW (Goldberg 2000). The
age-adjusted AMI odds ratios comparing Mexican Americans with NHW were
1.59 and 1.31 among women and men, respectively. These studies and others
(Hunt et al. 2002) help to reposition heart disease as a significant health problem
that needs more attention in the Mexican-American population. The Latino/a
29


community is currently the largest and fastest growing minority population in the
United States, particularly in Western States such as Colorado.
In 1997, the Colorado Hospital Discharge Database showed Hispanics
(males and females combined) represented 23.4% of the total hospital discharges
with an ICD-9 410 (AMI) or 411 (other ischemic heart disease) (Cardiovascular
Disease Surveillance On-Line 2003), second only to Whites (66.8%) (see
Appendix I) (Colorado Hospital Discharge Database 1997). Male and female
Hispanics were discharged with these diagnoses more than any other ethnic
minority in both 1995 and 1996 as well. In 1997, females represented 35.3% of
the discharges for ICD-9 410 or ICD-9 411; males represented 64.7% (see
Appendix J) (Colorado Hospital Discharge Database 1997). Of the 35.3% of
females discharged with an ICD-9 410 or ICD-9 411 in 1997, 7.9% were
Hispanic women and White women represented 23.6% of all females discharged
with these diagnoses (see Appendix J). This data supports the importance of
including Hispanic women with AMI in my proposed study. Additionally,
Hispanics are the second largest population in Colorado. According to the
Census 2000 Statistics Hispanics/Latinos represent 17.1% of the total population,
of which 10.5% are Mexican origin. Whites represent 82.8% of Colorados
population. By sex, Hispanic females represent 280,475 (13.4%) of all women in
Colorado and non-Hispanic White females represent 79.1% of all women in
Colorado (Colorado Health Statistics Section 2001).
30


Perceptions of AMI Risk, Meaning of Symptoms,
Discussions about Symptoms with Others
and Self-Treatment Strategies
Perceptions, meanings, discussions about symptoms, and self-treatment
strategies influence how symptoms get recognized. An AHA study found that
women believe that AMI symptoms are the same for men and women, yet the
reality is that less common symptoms may be more likely in women (AHA
1998). The AHA found only 8% of women recognized that heart disease and
stroke were the leading cause of death and killed more women each year than the
next 16 causes of death combined- including diabetes, all forms of cancer, AIDS,
and accidents (Lancet 1997). Although most women (91%) are receptive to
talking about health prevention and treatment options with their doctors, few of
their doctors (30%) have discussed heart disease when discussing their health
(AHA 1998).
Meanings ascribed to symptoms can affect a womans illness-related
behavior prior to and following the event (Rosenfeld et al. 2000). In one study,
women acknowledged that they had physical symptoms but initially judged them
as benign or insignificant and, as a result, self-treatment strategies were
employed as a way of maintaining control of the situation (Dempsey et al. 1995).
Women may deny the severity of symptoms or treat their symptoms based on
past experiences with other medical problems. Prior to seeking care, many
women seek validation from family and friends, for example, to confirm their
suspicions. When a woman perceives that family or healthcare workers do not
take her symptoms seriously (by expressing doubt or disbelief to her), she may
minimize her symptom perceptions and delay or not seek healthcare.
31


Alonzo emphasized the lay evaluation phase which may have particular
relevance to Latinas and their families (Alonzo 1986). The lay evaluation phase
is the period between the patient seeking lay advice and the decision to seek
medical evaluation. This phase follows the prodromal or warning phase and the
self-evaluation phase (the period between acute symptom onset and seeking
advice from lay or medical providers) (Alonzo 1986). These 3 phases occur
within a social situational context. It is within this context that family and lay
others play a critical role in symptom recognition for AMI. Families have the
potential to expedite symptom recognition and decrease the delay in seeking
care. The lay evaluation phase may be exacerbated in the Latina population
given the emphasis of assigning meaning to symptoms and treating an illness in
the family before seeking medical attention.
Cavillo & Flaskeruds literature review on culture and pain in adults
focused on Hispanics and summarized key concepts relevant to treatment seeking
behavior (Cavillo et al. 1991). Traditional beliefs affect the perception of health
and illness. These include cultural expectations for behavior in response to
illness and pain (Cavillo et al. 1991). Beliefs are at the core of culture and
precede behavior. For many Hispanic cultures, the value of suffering and the
concept of fatalism are accepted beliefs with religious undertones. Many in the
Hispanic culture believe that life has many difficulties which must be accepted
without complaints (Cavillo et al. 1991). Cultural expectations may specify
different pain reactions according to age, sex, and occupation. Examples of these
expectations are the type of treatment to seek, the intensity and duration of pain
to be tolerated, expected pain responses, who to report to when pain occurs, and
what types of pain require attention (Cavillo et al. 1991). There is reason to
believe that these models would apply to ischemic cardiac pain.
32


Zborowski and Zola are well known for their work focused on culture and
pain (Zborowski 1952; Zola 1981). Their work centered on emotional and
physical responses to pain among Anglo American, Italian, and Jewish cultures.
According to Anglo-Americans, being a good patient means one who
minimizes pain, avoids complaining, does not provoke pity, and does not annoy
anyone with his or her pain experience (Zborowski 1952). Among Mexican
Americans there are many intragroup differences, including degree of
acculturation, that affect responses to pain (Cavillo et al. 1991). Some cultures
do not acknowledge signs and symptoms of pain; it may be considered a lack of
stamina and a sign of weakness. Functioning adequately means doing routine
everyday tasks even if symptoms are present (Calatrello 1980). Exhibiting self-
control is a common practice and an important element in the Mexican culture
and among Mexican Americans experiencing pain (Cavillo et al. 1991). Self-
control (controlarse) includes 1) the ability to withstand stress in times of
adversity (aguantarse)\ 2) a passive resignation in which the person accepts his
or her fate (resignarse)\ or 3) a more active cognitive coping which means
working through a problem (sobreponerse) (Cavillo et al. 1991).
Zola proposed that it would take different symptoms to bring patients to
the doctor, and the presence of the same symptom in two populations may lead to
vastly different courses of action from immediate concern to infinite delay
(Zola 1981). Sixteen years later Lee et al.(Lee et al. 1997) demonstrated this
assumption in a comparison of symptom experiences and treatment seeking
behavior between Blacks and Whites. Zola believed the reason why a patient
seeks help is not due solely to objective features of the event, as biomedicine
may assume (Morris 1998), but may be related to more global social factors such
as social class or ethnicity (Zola 1981). Patients make meanings out of their
33


pain, most often unknowingly. The result of assigning meaning may be
confusion and uncertainty in a persons assessment of their symptoms (Morris
1998). In women the perception of pain may simply be conveyed as something
is wrong (McSweeney et al. 2000, Dempsey et al. 1995). Extended time allotted
for self-treatment strategies and lay evaluations (Alonzo 1986) may exclude
women as candidates for reperfusion therapy, and may have a pronounced effect
on those who live or work long distances from the nearest hospital.
Symptom uncertainty is at the center of womens narratives. Many
women describe symptom uncertainty in conjunction with problematic health
care provider interactions (Schoenberg 2003). Negative encounters, e.g. frequent
attribution of womens symptoms to stress, psychosomatic causes, or simply age,
have been demonstrated to cause women to doubt their own knowledge and
bodily sensations and deem their symptoms unworthy of a clinic visit or call to
the physician (Schoenberg 2003).
Previous negative encounters with healthcare personnel (e.g. a provider
who hurries a patient, does not listen, or discourages patient participation) or
perceived personal limitations (e.g. lack of time, knowledge, support, fear,
embarrassment, or problems with making decisions) are barriers reported by
women in other studies and may apply to women assessing and seeking care for
AMI symptoms (Kroll et al. 2000). One study examining predictors of
participation in healthcare at menopause (N= 252, primarily White women ages
40-65), found that women who perceived or experienced more barriers to
participation were less likely to participate actively in their health care (Kroll et
al. 2000).
Historically, womens complaints have often been attributed to mental
rather than physical problems; it is suggested that gender bias in diagnosis and
34


treatment of coronary artery disease continues to exist due to clinical models and
diagnostic tests based on male subjects and a lack of research that focuses
womens experiences with cardiovascular disease (Beery 1995).
Competing Demands
Social obligations play a central role in the lives of women (Schoenberg
2003). Attending to illness is merely one of many needs that confront women
and influence symptom appraisal and treatment seeking behavior (Schoenberg
2003). Women often give primacy to the care of others and this often results in
diminished attention to their own concerns (Schoenberg 2003). When men and
women are asked why they did not inform family and lay others of their acute
symptoms, about one third say they did not wish to disturb others with health
matters, especially if it was the same old thing or they needed to complete
ongoing social obligations (Alonzo 1986). In addition to social obligations and
caring for others, Latinos/as describe familialism as a value that distinguishes
them from other ethnic groups. Two aspects of this concept are relevant to
treatment seeking behavior. Seeking advice from family is respected and a lack
of communication is a sign of disrespect (Niemann et al. 1999).
Evidence of the competing demand concept may also be reflected in the
household production of health model. In Clarks assessment, women provide
nearly 95% of all domestic care, also known as domestic medicine, self-care, self
management, or home healthcare (Clark 1993). Women reported that their
femaleness and generational position predisposed them to accept disproportional
responsibility for household health (Clark 1993). For children, women provide
35


the feeding responsibilities, maintaining and attending appointments, including
transportation and missing work to do these things. For their husbands, women
provide healthcare, especially as they age and are chronically ill. For their
elderly family members, women often attend to healthcare needs, fulfilling their
roles as daughters or daughters in law. Mexican and Mexican-American culture
emphasizes extended family participation in illness decision making (Clark
1993). Mexican American women may have a greater need to redistribute the
workload. Their health status is lower than Anglo women; they have more
underlying chronic health problems and lower perceived self health. They also
share responsibility in generating household income, have more partially absent
male partners, and have more children to care for than Anglo women. Together,
these factors increase the domestic workload in Mexican American women.
Some frame the household health production as the heart and soul of womens
work: the ability to take responsibility and initiative for knowing what need[s]
to be done, and knowing how to do it (Clark 1993). This model offers one
explanation why a Latina prioritizes family needs before her own needs.
Pre-existing co-morbidities may affect self-treatment strategies (Hofgren
et al. 1994). In a study including Mexican Americans, diabetic neuropathy was
identified as a potential explanatory variable of ethnic and gender differences in
reported symptoms because the prevalence of diabetes is greater in women and
Hispanics hospitalized with an AMI (Meschak et al. 1998). Symptoms
associated with diabetic neuropathy can alter the perception of myocardial
ischemia so that the diabetic patient is unaware that an AMI is occurring, which
leads to delayed care-seeking. AMI symptoms have been misrecognized as
symptoms of other illnesses like gastrointestinal disorders (indigestion), hiatal
hernia (chest pain), arthritis (joint pain), or cold and flu-like symptoms
36


(LaCharity 1999; Finnegan et al. 2000). Women also delay discussing their
symptoms with family and friends, when they 1) denied the significance of the
symptoms, 2) were concerned about worrying others, and 3) were concerned
about being unable to maintain commitments (LaCharity 1999; Dempsey et al.
1995).
Efforts to educate women on symptom recognition, however, seem
premature given the limited data about womens symptomatology and perception
of symptoms. Without further studies targeting underrepresented populations
such as Latinas, the effectiveness of education strategies remains questionable.
37


CHAPTER 3
THE THEORETICAL FRAMEWORK
Different perspectives informed this dissertation. An inductive approach
was used to derive typologies and themes from the data using the constant
comparison method (Crabtree and Miller 1999). Kleinmans explanatory models
using patient narratives were applied to understand patients experiences. This
approach was particularly acceptable among the Latinas who often use
storytelling in conversation.
This dissertation borrows ideas from the phenomenological perspective to
understand symptomatology in the context of a womans illness experience
(Kleinman 1998). Narrative analysis helped to examine how a woman and
members of her social network subjectively perceive, report, and explain
symptoms that subsequently are understood to signal an AMI. Kleinman uses the
term explanatory models to describe these individual cognitive models that a
person has about their own illness, and proposes that explanatory models affect
communication between the patient and the healthcare provider (Kleinman
1998). Acquiring additional in-depth knowledge about the symptoms and
experiences that influence how Latinas (English and Spanish speaking) and non-
Latina women process and act upon their symptoms could minimize the existing
gap in our understanding and improve the quality of pubic health awareness
efforts addressing heart attack symptoms and treatment seeking behavior
recommendations.
Explanatory models address the questions: 1) What do you think your
problem is? 2) What do you think caused it? 3) Why do you think it started when
38


it did? 4) How severe is your illness? Will it have a short or a long course? 5)
What kind of treatment do you think you should receive? 6) Who do you want
involved in the decisions about your care? 7) What are the most important results
you hope to receive from this treatment? 8) What are the chief problems your
sickness has caused you? 9) What do you fear most about your sickness? 10) Is
your financial situation an important consideration for you in deciding about
medical treatment? 11) What do your husband/family/friends think the problem
is? and 12) What do your husband/family/friends think should be done about
your illness? (Brown and Corbett 1995).
Figure 3.0 speaks to the complex web of factors influencing women who
experience a heart attack. Deciding to seek care for symptoms is a
multidimensional phenomenon. The complexity shows why interventions that
emphasize increasing knowledge of symptoms alone may be less effective at
producing behavior change. Cultural factors are represented by dotted lines to
indicate the piece of the model that has been underrepresented in the literature.
Figure 3.0 Multidimensional Phenomenon: the Decision to Seek Care
39


Appropriate theories or models to guide the research included the Health
Belief Model, the Self-Regulation Model of Illness Behavior, and Control
Theory (Dempsey 1995). These were used in previous examinations of risk
perceptions and treatment seeking behavior for AMI symptoms (Dempsey 1995;
Alonzo 1986; Dracup et al. 1995; Meischke et al. 2000; Kukafka et al. 1999).
The Health Belief Model
The Health Belief Model frames how perceived susceptibility of having
an AMI and perceived severity of her symptoms influence a woman to take
action. The original application of this model was used to explain participation
in prevention programs. Since then, the model has been expanded to understand
illness behaviors and sick-role behaviors. Now, the model supports the idea that
individuals take action if they perceive themselves to be susceptible to the illness,
perceive the illness to have serious consequences, believe that actions (i.e.
seeking help) benefit them by reducing the severity of the condition, and believe
that anticipated barriers with seeking help (i.e. cost) are outweighed by its
benefits (Strecher et al. 1997). A complete diagram of the Health Belief Model
is shown in Figure 3.1.
A criticism of the Health Belief Model (HBM) is the failure to test the
model in cultures other than mainstream US and European cultures (Larkey et al.
2001). To address this criticism, Larkey et al. (2001) applied the model to a
cancer screening study among Hispanics in Arizona to elicit a deeper
understanding of the fatalismo response to symptoms. Furthermore, issues
important to Hispanics, i.e. the need for confianza (trust) andpersonalismo
40


(caring) in medical interactions, are not explicitly accounted for in most medical
models of health-seeking behavior. Consider the example of the indirect path
taken from the detection of illness to going to the doctor (Larkey et al. 2001).
This path consists of first self-managing symptoms followed by consultation with
family and friends if self-management does not work. The next step may be
consulting traditional or alternative medical resources followed by the last resort-
accessing Western medicine. The seriousness of symptoms featured strongly as
a variable in the decision to go to the doctor indicating that providing accurate
information about warning signs and how those signs relate to serious disease
may be especially important in the Hispanic population (Larkey et al. 2001). The
significance of their work showed that applying the HBM to a Hispanic
population can uncover new variables in existing models, such as expectations of
negative experiences, faith in God, and need for trust and closeness with the
provider.
Individual Modifying Likelihood
Perceptions Factors of Action
Age, sex, ethnicity
Personality
Socioeconomics
Knowledge
Perceived benefits minus
perceived barriers to
behavior change
Perceived
susceptibility/severity
of disease
Perceived Threat of | Likelihood of Behavior
Disease
Change
Cues to action
Education
Symptoms, illness
Media Information
Figure 3.1 The Health Belief Model
41


Critics of the HBM say that rational decision models fail to provide
insight to the process of decision making and the automatic, non-reasoned
aspects of behavior (Dempsey et al. 1995; Leventhal et al. 1987). The model
lacks patients appraisal of symptoms and the events that follow. It cannot
distinguish between women who delay seeking care for MI symptoms and those
who do not. The Self-Regulation Model of Illness Behavior is an alternative used
to analyze womens behavior around MI symptoms (Dempsey et al. 1995).
The Self-Regulation Model of Illness Behavior
The Self-Regulation Model of Illness Behavior proposes that illness-
related behavior is determined by a (womans) cognitive representation, coping
strategy, and assessment of that particular illness (such as AMI symptoms)
(Cameron et al. 1995). This is a three stage dynamic process model that accounts
for the influence of emotions. Individuals can move back and forth between any
stage.
Cultural differences create different representations (Leventhal et al.
1987); people construct different representations of the same illness threat and
different action plans are seen as appropriate for containment of the threat
(Leventhal et al. 1987). Constructs of interest for the proposed study are: illness
representation (labeling), coping strategy (self-treatment or treatment from
others), and symptom appraisal. Dempsey, Dracup, and Moser applied concepts
of this model to demonstrate that women do not go through a systematic analysis
of their symptoms (the costs and benefits of seeking care), but instead focus on
42


emotions related to their ability to maintain control and self-manage their
symptoms (Dempsey et al. 1995).
Control Theory
Control Theory proposes that a womans desire to maintain control over
oneself and ones environment affects decision to seek care for symptoms of
AMI (Dempsey et al. 1995). Dempsey Dracup and Moser developed the model
describing Control Theory (Dempsey et al. 1995). This model integrates the self
regulation model and concept of the interactionist role (symbolic interactionism)
to explain decision making by a woman and the influence of members in her
social network. Overall, the study demonstrated a good use of grounded theory
and the self regulation model to reveal womens experiences even though the
model did not account for cultural variation (Dempsey et al. 1995).
Figure 3.2 The Self-Regulatory Model of Illness Behavior
43


The diagram of the Control Theory shown in Figure 3.3 represents how women
responded to acute myocardial infarction symptoms.
Self-Efficacy
Banduras concept of self-efficacy relates to a persons belief that they
can perform the behaviors required to produce a given outcome (Bandura 1997;
Kroll et al. 2000) and can be thought of as self-perceived competence. The
central construct of self-efficacy was used in this project. Failed self-treatment
strategies and actions taken to seek validation from others may signal that
women have low self-efficacy in successfully identifying MI symptoms and
acting immediately to get medical attention. On the other hand, engaging family
and lay others in symptom assessment may reflect high self-efficacy in ethnic
cultures that endorse family participation in determining when medical attention
is needed beyond home or self-treatments applied by the person and her family.
44


Maintaining Control
Relinquishing Control
INFLUENCING FACTORS
Attribution to Benign Cause
Beliefs About Ml
Past Symptom Experience
Denial
Beliefs About Women & CHD
Commitments
Concern for Others
Figure 3.3 Control Theory
Participation in ones health care can include a) seeking information, b)
preparing and asking questions, and c) participating in decision making (Kroll et
al. 2000). The constructs discussed above, the Health Belief Model, Self-
Regulation Model of Illness Behavior, and Control Theory, are reflected in the
questions in the interview guide (Appendix A).
Outcome Certainty
INFLUENCING FACTORS
Continuing Symptoms
Anxiety
Perceived ------ Lay ______________^ Hospital
Significance Consultation Treatment
45


CHAPTER 4
RESEARCH DESIGN AND METHODOLOGY
Methodology is the way of thinking about and studying social reality
(Strauss & Corbin 1998). In order to answer the research question, What are the
symptoms Latinos and non-Latina women experience related to an AMI and what
are the processes used to seek care for these symptoms? , the study included
methods defined by Glaser and Strauss (1967), Morse, (1994) and Crabtree and
Miller (1999). These are acceptable methods for assessing narrative or
qualitative information when a research project seeks to dig deeply into meaning
and perspectives in order to reveal new themes that have not been fully explored
in previous research. The transcripts, vignettes, and timelines served the purpose
of assessing the data listed in Table 4.0. To be able to compare my sample and
findings to data reported in the literature, the methods were mixed. In some
instances, using mixed methods provided a way to triangulate and corroborate
data such as was the case with the timelines and analysis of symptoms.
The Case for In-Depth Interviews
I used focused, qualitative interviews to understand heart attack
symptoms and experiences associated with those symptoms. The interview is a
research-gathering approach that seeks to create a listening space where meaning
is constructed through an interexchange/cocreation of verbal viewpoints in the
interest of scientific knowing (Crabtree and Miller 1999, p.89). Pain is a
46


physiosubjective experience and understanding pain requires asking patients
about their experiences with pain (Crabtree and Miller 1999, p.90). Rather
than setting up the respondents answers with language expressing the
researchers concerns, as with structured interviews, this approach allowed and
encouraged the interviewee to voice their own perceptions and meanings
surrounding the pain experience.
Table 4.0 Multi-Method Approach
Data Source: Data analyzed Quantitative Qualitative
Medical Chart X
Reviews (N=43)
Demographics X
Delay times X
Troponin values X
Symptom analysis & description of the chief X
complaint
Interview Thematic development X
transcripts (N=84) (Each person had two interviews except 2)
Demographics X
Acculturation scale X
Symptoms identification and analysis X X
Vignettes (N=43) Thematic development X
Establish time sequence of events X
Patient Career Thoughts, meanings, and symptoms associated X
Timelines (N=43) with treatment seeking events over time.
Each timeline assigned to a typology X
The interview is a type of partnership that is a conversational research
journey (Crabtree and Miller 1999, p.91). Part of the context of an interview is
the communication event. This includes the form of the message (verbal,
47


gestures, facial expressions), the actual message, and the participants in the
interview. Viewing the interview as a partnership is a reminder that there are at
least two active participants involved in meaning-making work (Crabtree and
Miller 1999, p.92). For the interviewer, learning how to ask first requires
understanding the different ways in which people communicate and then
deciding on the appropriate methods (Crabtree and Miller 1999). For this study,
in-depth interviews were selected to probe topics in greater depth compared to
previous research using a structured approach.
In-depth interviews are a powerful tool to use when the focus of the
inquiry is narrow, the respondents represent a clearly defined unit with an already
known context, and the respondents are comfortable and familiar with the
interview as a means of communication (Crabtree and Miller 1999, p.90). The
goal of in-depth interviews, a data gathering process, is to generate themes and
narratives that focus on a fairly specific research question. Also, this format for
interviewing is a cultural preference for Latinas who prefer to express themselves
through story.
In-depth interviews use mainly open, direct verbal questions that elicit
stories and case-oriented narratives. The women in this study were comfortable
and familiar with an interview approach. Discussing health and illness in a
conversational style with women is an acceptable means to obtain information.
Most women are eager to elaborate on their experiences. This style also works
well with participants in which limited literacy skills might otherwise prohibit
them from participating in a study requiring reading and writing skills. In-depth
interviews were also chosen because the study was exploratory. This interview
format allowed the opportunity to learn insights and revelations as well as to
48


discover individual perceptions and narrative understandings about experiencing
MI symptoms (Crabtree and Miller 1999, p.93).
Several steps were taken before deciding on the in-depth interview
approach. First, the literature review identified existing descriptive and
theoretical categories about treatment seeking behavior for MI symptoms. Key
conceptual domains guided the development of an interview guide. In the review
of the literature on MI symptoms, I discovered that most of the articles focused
on categorizing symptoms by location and quality and identifying predictors of
the types of patients most likely to delay seeking care for MI symptoms. Other
articles focused on determining the clinical presentation characteristics of
patients who were having an MI. A third approach focused on eliciting the
perspective of the medical provider and the misdiagnosing of patients presenting
with MI symptoms. The fourth category of articles from the literature review
took a qualitative approach to explore reasons for delayed treatment seeking
behavior. These turned up problems with recognition of symptoms, the need to
take care of others before taking care of themselves, and trying to maintain
control and to self-treat symptoms as long as possible before seeking medical
care.
While important findings were discovered from social science oriented
literature, the findings raised additional questions regarding the populations
represented in their samples. My goal evolved into one that reproduced their
research questions but addressed them with a larger sample of women than what
is typically used in a qualitative study. The focus was also on enrolling an
underrepresented sample in the literature: Latinas, particularly Spanish speakers.
This facilitated comparisons between the findings in my study sample and the
findings in the literature.
49


The in-depth interviews were organized around an interview guide
(Appendix A) (see subsection Interview Guide in this chapter for further
description) consisting of open-ended and closed-ended questions. Although the
grand tour (Spradley 1979) questions were standardized, they were
somewhat modified as needed for patients who could more easily answer the
questions when the language was simplified.
The role of the interviewer is to assume a low profile, put the informant at
ease, and reinforce the continuance of the conversation to facilitate storytelling
and, with this framework, customize the interview (Crabtree and Miller 1999,
P-101).
Open-ended questions offered more advantages than disadvantages to
answer my research question. Open-ended questions allowed allowed the
Latinas to have the opportunity to share their way of knowing and symptom
experiences not identified in the literature. The disadvantage of open-ended
questions is the long length of time allotted to each subject and the difficulty of
tabulating responses, which is more easily done when using a survey with closed-
ended questions.
This study explored how women of different ethnicity, language, and
cultural backgrounds recognize and respond to heart attack symptoms. My
challenge was to design a study that enabled me to reach women as close to the
hospital presentation time as possible in order to obtain the richest level of detail
in their stories. Interviewing women immediately following their acute event was
the best way determined to capture detailed stories and to minimize recall bias
among the participants.
The data collected describe the heart attack experience through the
perspective of the patient and sometimes family members involved with the
50


patient during the event. Objective data collected through the medical chart
reviews served several purposes: to corroborate findings from the interview, to
establish the time of presentation, and to do a limited exploration of the
narratives written by medical care providers in their assessment of the patient and
her symptoms.
The Sites
After the study protocol received approval from the University of
Colorado Human Subjects Committee and the Colorado Multiple Institutions
Review Board (COMIRB), I proceeded to secure approvals from the hospitals
recommended as good study sites. My physician mentors were integral in helping
me gain entree. They made recommendations and provided insights on how to
proceed and with whom I should speak before the study began. Nine Colorado
hospitals participated. Two sites that were poised to enroll patients but did not
enroll any during data collection included one rural hospital and one teaching
hospital in Denver. Seven sites actively enrolled patients. They included one
rural hospital, one teaching hospital, one which serves as the safety net hospital
in metro Denver, and four regional hospitals serving rural areas.
The selection process was both purposeful and convenient. I selected
hospitals that were located in communities with a strong representation of
Latinos (Greeley, Denver, and Pueblo) (U. S. Census Bureau 2000). With
assistance from the University of Colorados High Plains Research Network, one
of my physician mentors helped me access additional hospitals in North and
Northeast Colorado.
51


Gaining Entree
Six of the nine hospitals were accessible as study sites because my
physician mentors had patient privileges or had pre-established research
relationships with the sites. To gain entree, the physicians initiated contact with
key administrators, determined their interest, and obtained verbal commitments.
Conducting Research Under the
Health Insurance Portability and Accountability Act
(HIPAA)
The Standards for Privacy of Individually Identifiable Health
Information (Privacy Rule) establishes, for the first time, a set of national
standards for the protection of certain health information (U.S. Department of
Health and Human Services 2003). This rule addresses the disclosure of
protected health information (PHI), in any form of media including paper,
electronic, or oral, and standards for an individuals privacy rights. Upon the
implementation of this rule, patients control how their health information is used.
According to the U.S. Department of Health and Human Services, ...
individually identifiable health information is demographic and other information
that deal with 1) the individuals past, present or future physical or mental health
or condition, 2) the provision of health care to the individual, or 3) the past,
present, or future payment for the provision of health care to the individual, and
that 4) identifies the individual or for which there is a reasonable basis to believe
can be used to identify the individual. Examples of identifiable health
information include name, address, birth date, and social security number.
52


In January 2003,1 met with a hospital administrator at each site to discuss
conducting this research study and to discuss the assistance needed from the
nursing staff. A process for identifying and enrolling patients was tailored to
each facility but generally involved the same steps. Before meeting with
administrators, I was aware that accessing patients under the new HIPAA
regulations would be difficult for a non-employee with no patient care
relationship. Additionally, I was sensitive about existing demands on nursing
staff. With that in mind, I suggested patient recruitment methods that minimized
the time needed from the nursing staff and wherever possible, set up the process
to be in line with their regular nursing responsibilities. These processes adhered
to the HIPAA regulations.
Several of the hospital administrators quickly agreed to allow me to
conduct the study because the area of research was important to them and
relevant to their patient population and health care providers. Their enthusiasm
drew from personal experiences with female family members who had difficulty
recognizing MI symptoms. Administrators also reviewed past admission
statistics to get a sense of an average of monthly female MI admissions. This
clearly identified that the study was relevant to their patients. Another reason
cited for supporting my study was the interest in maintaining their pre-existing
research relationship with the High Plains Research Network of which one of my
physician mentors directed.
I met with different key administrators at each site including the Director
of the Emergency Room Services, Director of Cardiac Risk Reduction Services,
Director of Nursing Administration, Supervisors of the nursing units, and a
manager of Cardiovascular Services. Sometimes these meetings required follow-
up meetings with staff identified as my daily contact throughout the study.
53


The next step to gaining entree was applying to the local Institutional
Review Boards in each facility. The timeframe to complete this step varied in
each hospital from one to eight months. In some cases, that process was limited
to submitting an application packet or review. In other cases, that process
involved presenting the study to the Review Boards and/or conducting in-
services for the nursing unit staff.
The in-services ranged from ten to thirty minute presentations to the
nursing staff on units where eligible patients for my study were most likely to be
identified. To maximize attendance, they were scheduled at times when shifts
overlapped. Flyers were posted in the units, and unit supervisors informed the
staff about the presentations and arranged time for staff to participate. The in-
services were scheduled in a nearby conference room on the unit. I brought food,
set up presentation materials, and repeated the presentation several times over a
three or four hour period. The presentation included a study overview,
information on my background, highlights from the literature review, potential
benefits to the hospital, and an outline of the enrollment process. Nurses asked
questions and offered solutions to make the enrollment process work smoothly
with their units. Field notes were taken during the in-service period.
Four positive outcomes resulted from the in-services. First, it was an
opportunity to educate nurses about the problem of delayed treatment seeking
behavior and symptom presentations for women. Some of the nurses were well
aware of the problem but many did not realize the impact of the problem for
hospitals and patients. Second, feedback from the nurses helped refine the
enrollment process. Nurses identified the best location for the nurses recruitment
binder to be visible as a daily reminder about the study, and the best type of nurse
(i.e., team leader, unit clerk, or charge nurse) that had access to the data needed
54


to determine if patients met inclusion criteria. They made suggestions about
what would help them remember the study and critical information to give to the
patient. The apparently minor changes to the enrollment process were critical
elements to make the enrollment process successful.
The third positive outcome, particularly beneficial to me, came from the
in-service in Pueblo. The majority of the nursing staff was Latina. They shared
concerns about how Latinas were to be identified, and asked questions that might
concern Latina patients and their families. They made suggestions to improve
the wording on the study flyers and nursing index cards to maximize the
possibility of getting Latinas to participate. We had lengthy discussions during
the in-service about defining ethnicity and terminology used to identify Latinas.
The nurses were very supportive and gave their approval of the study in their
discussion of how the information learned from the study would benefit the
patients in their community.
Finally, conducting an in-service was a way for the nurses to connect the
study to the investigator. Having a face to connect to the study made the study
memorable to them and helped them remember me when I called on a daily basis
looking for patients. To maintain good relationships with the nurses at each
hospital, I brought them bagels, candy, and baked goods throughout the
enrollment period with a flyer thanking them for helping me recruit patients.
They were very appreciative of my efforts to acknowledge their support and
participation in the enrollment process.
55


Selection of the Study Population
The recruitment process over-sampled women from diverse socio-
economic backgrounds and Latina ethnicity; at the beginning of this study the
ethnic breakdown among men and women in Colorado was 82.8% White and
17.1% Hispanic (U.S. Census Bureau 2000). Women self-reported the ethnic
category to which they belong. The study population included women bom in the
United States and women bom in foreign countries. Twenty-five non-Latina
White women, one African American woman, and seventeen Latinas underwent
semi-structured interviews. Interviews continued until saturation was reached,
i.e., no unique symptoms or themes were identified, and variation was accounted
for and understood (Morse 1994).
Inclusion and Exclusion Criteria
Criteria to participate included adult women living in Colorado, Latinas
and non-Latina women, who were diagnosed with an AMI. Participants had to
be able to communicate in Spanish or English. There was no exclusion by age,
to facilitate case finding, since the Denver metropolitan area has a relatively low
incidence of coronary heart disease compared to the rest of the United States
(AHA 1999). Study participants were clinically stable before they were
interviewed. This was as early as four hours after admission of the hospital stay
for uncomplicated patients or day 7 for patients who initially had complications
which resolved. This allowed me to capture those patients who experienced
56


prolonged prehospital delay times who had more complications during their
hospital stay as well as patients who sought prompt medical care.
The main criteria used to identify patients with confirmed MI included a
positive Troponin value and, in the absence of Troponin data, CK-MB and CK
values. In the absence of Troponins, CK-MB or CK values, the enrollment was
supplemented by reviewing the physician notes indicating a positive MI.
Exclusion criteria included: 1) males, 2) presence of delirium or
dementia, 3) complicated status, 4) an unstable condition, 5) patients who were
not able to communicate in Spanish or English, and 6) those unwilling to give
consent.
Recruiting Methods and Enrollment Procedures
The estimated patient enrollment period was six months. Due to
challenges with HIPAA regulations and the nature of the study (i.e. waiting for
women to have heart attacks), the actual enrollment period was sixteen months.
Because the enrollment period took longer than expected with the original seven
sites, two additional sites were added to recruit additional Latina participants.
These required additional applications through local Institutional Review Boards
and, in a few cases, making presentations to the review boards and in-servicing
the staff who were volunteered to assist with patient enrollment.
A purposive sampling strategy was used to generate a list of women
diagnosed with an AMI using admission records. No records were accessed until
authorization was received from appropriate personnel. Recruiting procedures
varied among study sites based on available personnel to assist research staff
57


with enrolling newly admitted patients. Each site had a designated key contact or
team in the ICU, CCU or Step Down Units at each hospital to help disseminate
knowledge about the study to other ICU nurses, and to identify potential study
participants. The typical recruitment process was: The key contact nurse asked a
patient if they were interested in participating. Those interested in participating
received a recruitment packet briefly outlining the project.
The Impact of HIPAA on the Patient Enrollment Period
On April 14, 2003, HIPAA regulations went into effect. Because this
procedure and the regulations rolled out just as my study was beginning, research
offices at each institution and hospital staff were going through the learning
process of adopting the regulations. The impact this had on my study was
primarily a delayed approval process to begin patient enrollment. Forms were
new to the institutions and at times instructions and content changed which
resulted in resubmitting forms before I gained site approval. Some institutions
had one HIPAA form while others had two, one for the patient to authorize the
release of their name and the other to enroll patients into the study with
permission for me to access medical records.
Nursing support was critical to patient enrollment. Without their support,
it would have been extremely difficult to identify and enroll patients during
admission. The nurse was the first person to contact the patient with information
about the study.
58


HIPAA and Consent Procedures
Nurses and translators obtained HIPAA authorizations from and described the
study to potential subjects (see Figure 4.0), after training by me. The staff who
were assisting me had to be able to answer basic questions about being a study
participant. Nursing units were given index cards and recruitment binders that
stated what information should be given to the patient about the study. The
recruitment binders contained prepared recruiting packets in English and
Spanish. Each packet consisted of a study flyer, HIPAA authorization form, and
a copy of the subject consent for reference purposes only. A signed consent was
obtained at the time of the first interview. However, some patients, particularly
the Latinas and their families, were more comfortable with the study when they
had the consent containing a detailed description of the study before deciding to
release their name on the HIPAA form. Nursing staff was not required to obtain
signed informed consents. The cards explained a step by step procedure to
obtain the first HIPAA authorization which released the patients name and
contact information to me. This enrollment procedure was set up to minimize the
time required from the nurses to assist me. Study participants had adequate time
to review the consent and were encouraged to ask questions in the privacy of
their rooms to maintain confidentiality.
Prior to all interviews, the interviewer explained the studys purpose, how
the interview would be conducted, and how the data would be used. The
interviewer read the informed consent to the participant and provided a consent
form, in English or Spanish, to read and sign. A copy of the consent form was
given to the participant.
59


Figure 4.0 Enrollment Process Diagram
When patients were approached, they were reminded that participation or refusal
to participate would neither enhance nor detract from the quality of the care
received and that participation was entirely voluntary. Participants were
reminded that they could withdraw at any point during the study without negative
consequences. Copies of the signed consents and HIPAA forms were placed in
the patient charts. Originals of the signed consents and HIPAA forms were
retained in the project files. The Informed Consent found in Appendix B
contains the Colorado Multiple Institution Review Board (COMIRB) approved
subject consent form. HIPAA A and B forms are located in Appendix C and D,
respectively. Similar informed consents were modified to fit local requirements
from hospitals not included under COMIRB.
All hospitals required the completion of a Health Insurance Portability
and Accountability Act (HIPAA) Authorization Form. Some hospitals had two
forms to complete the HIPAA process. This included HIPAA Authorization A:
Research Recruitment Form and HIPAA Authorization B: Enrollment into
Research. By signing these forms, the patient authorized the release of their
60


contact information. Examples of the forms are located in Appendix C and D.
By signing this form, the patient authorizes the release of their health
information. The HIPAA forms and, in some hospitals, the physician
notification letter as well as other research notification forms required by one
health system (Appendix E) were put into the patients medical chart.
If the patient was interested and signed the HIPAA form, the nurse
notified me. To set up the initial interview, I (or the Spanish translator) called or
visited the patient to arrange a time. This time was also approved with the nurse
who was caring for the patient to avoid conflicts with diagnostic tests, feeding
schedules, or nursing duties. The interview was scheduled as close to the initial
phone call as possible.
In a few cases, the nurse notified me too close to the time of the patients
discharge and there was not adequate time to drive to Pueblo or Greeley, for
example, to reach and interview the patient prior to discharge. In a few instances
the nurse informed the patient about the study and the patient was interested but
was discharged before I reached them. In those instances, the patient gave
permission to release their home phone number. These interviews were
conducted by phone (one patient from Wyoming) or in the patients home (two
patients). Each patient received a grocery certificate valued at $10 upon
completion of the first interview.
Protecting Patient Confidentiality
The recruiting process made every effort to protect patient confidentiality
and ensure that the patients decision was made without coercion. Identifiers (i.e.
61


name, social security number, patient account number, phone number) were
necessary to recruit patients and to obtain contact information for arranging the
follow-up interview. All personal information was stored in a locked cabinet and
a password protected computer. Transcriptions were stored in a password-
protected computer. Assumed names or code numbers were assigned to
participants so they are not identifiable in reports or publications resulting from
this study.
Adverse events were not anticipated as a result of participating in the
interviews. However, if the patient became physically or mentally unable to
proceed with the interview, I stopped the interview, notified the nurse of the
patients condition, and resumed the interview when the patient was ready.
There were a few instances were I waited 20 or 30 minutes until the patient was
feeling better. No first interviews were rescheduled to another date. Since this
was a minimal risk study, there was no need for an independent data safety
monitoring board. Monitoring was my responsibility. I consulted with the
physician mentors in cases where I was not certain of the safety of the patient. I
oversaw all recruiting, interviews, data transcription, and data analysis.
Hiring Spanish Interviewers
I advertised to hire Spanish interviewers through word of mouth. Studies
have documented that the quality of the information collected during an
interview is enhanced when the ethnicity and gender of the interviewer and the
patient are the same (Davis 1997, Dailey and Claus 2001). I interviewed five
interested translators and created an on call list. The first available person who
62


responded to the request for an interviewer was hired each time. A Spanish
speaking transcriptionist was hired to transcribe and translate the Spanish
interviews. Because some hospitals required a letter of Spanish certification, the
translators and transcriptionists participated in a certification interview and
received a letter attesting to their Spanish speaking ability. Additionally, each
signed a confidentiality statement to protect each subjects privacy.
All interviewers were trained by me about the recruitment paperwork, the
data collection instruments, and about the hospitals expectations for staff
accompanying me during the interview. I trained them in data collection and
interviewing techniques. Translators wore identification badges indicating they
were part of the research team.
A bilingual interviewer conducted the Spanish interviews while I was
present. To establish rapport with the patient, I introduced myself, the purpose of
the study, and the reason for having a translator accompany me. I took notes
while the interview was tape recorded. This arrangement allowed me to capture
outward expressions (observable illness behaviors) such as body language, facial
expression, and nonverbal vocalizations (sighs, moaning). These influence how
others judge the severity of the patients situation. Somatization (a form of
expression) is greater in disenfranchised populations (Osterweis et al. 1987).
These are influenced by familial and cultural norms (Osterweis et al. 1987; Zola
1981).
Before each interview, I reviewed the patients information and
communication needs, if known. After each interview, the translator and I would
meet immediately afterwards and discuss the patients story and identified follow
up questions for the second interview when data needed from the first interview
was not disclosed. The translator and I completed the Contact Summary Form
63


(Appendix H) to capture our reactions to the interview, prominent themes, and
unanswered questions.
Description of the Study Design
This study was multi-method; the primary method was qualitative, semi-
structured interviews, followed by medical record reviews for the interviewees.
According to Babbie (2001, pg 110), the best study design uses more than one
research method and takes advantage of their different strengths. The qualitative
component of the study was the set of semi-structured interviews. The
quantitative portion consisted of analysis of the data collected from the medical
record reviews. Although the aims selected for the study focused on qualitative
descriptions of the symptoms women experience and how decisions were made
to seek care, it was important to describe quantitatively how these findings
differed between Latinas and non-Latinas as well as giving consideration to other
demographic factors that affected care seeking behavior.
Analysis involved interpreting the qualitative findings independent of the
quantitative data and vice versa. The final analysis considered the contribution
that the medical record data had on patients assessment and evaluation of their
symptoms.
64


Study Instruments
The following instruments were used as aids to track enrollment, help the
nursing staff identify patients, and interview and consent subjects.
The Interview Guide
An interview guide facilitated building rapport and confidence with the
interviewee, elicited some biographical and sociodemographic information, and
also assisted the interviewer to address intended topics (Crabtree et al. 1999).
The interview guide consisted of identifying questions constructed in a closed-
ended fashion and seven open-ended questions otherwise known as grand tour
questions (Crabtree and Miller 1999, p.96; Spradley 1979) with associated
prompts, probes, and follow-up questions.
The open-ended structure of the questions permitted an environment for
the researcher to listen for the words women might choose in everyday
conversation to describe their symptoms and experiences.
The closed-ended questions about background and demographics
information were reserved until the end of the interview. Most of this
information was verified in the medical charts and the follow-up interview.
Acculturation was assessed by language preference. Additionally, each subject
completed a set of acculturation questions created by Marin and Marin
(Appendix A) which was collected but not analyzed for the dissertation. This
additional data was collected with the intention of further analyses planned for
future manuscript preparation.
65


The Contact Summary Form
After the first and second interview, I took notes to capture my thoughts
and reaction from the interview. During the data collection period, I created a
Contact Summary Form to collect the post interview information more
systematically. The purpose of the form is to jot down the main concepts,
themes, issues, and questions that I saw during the interview (Miles and
Huberman, 1994). The contact summary is a single sheet with summarizing
questions about the interview (Appendix H). After each interview, I reviewed the
field notes (the Spanish translator helped me complete this form with the Spanish
interviews as needed) and answered each question briefly to develop an overall
summary of the main points of the interview. The filled out sheet was used to
! guide planning for the next interview or to document final thoughts about the
patient following the second interview. The sheets also helped to reorient me
with the patient prior to calling her for the second interview (Miles and
Huberman, 1994).
The Medical Record Review
A Medical Record Review Form (MCR) was created to facilitate the
retrieval of chart information (Appendix G). The form was divided into
sociodemographics, social and family history, prehospital care reports, the
emergency department visit, medical history summary, and discharge
information. Data collected in the chart reviews included the following: 1)
sociodemographic components including age, education, insurance, occupation,
66


income, gender, marital status; 2) social history including drug and alcohol use;
3) clinical data including the patients chief complaint, medical history, symptom
history and previous treatment seeking behavior for these symptoms, and family
medical history; 4) utilization of guideline-based therapies documented by
medications reported in the discharge summary; 5) delay time-times of initial
symptom onset, acute symptom onset time, arrival, and times associated with
utilizing ambulance services; and 6) mode of transportation- self-transport,
ambulance, transportation provided by family, friends, or other source.
Structure of the Data Collection Process
Initial Interview
Initial interviews lasted between one-half and one hour. Follow-up
interviews, when the patient was no longer in immediate recovery, lasted longer
for some patients. The interviews began by asking the patient about their family,
which is a culturally acceptable way to engage in conversations with Latinas.
The interview continued with open-ended questions that invited the interviewee
to begin their descriptive narrative. The interviews were guided and open-ended
communication events that were co-created by the investigator and the
interviewee (Crabtree et al. 1999). All subjects received a ten dollar grocery
store gift certificate to thank them for participating in two interviews.
In addition to field notes, I used a memo-ing technique to capture
important insights during the interview. Memo-ing consists of brief hand-written
67


notes that capture abstract thinking about the data to formulate theory and
facilitate analysis (Strauss et al. 1990).
Some of the questions selected in the interview guide have been
employed in prior research (McSweeney et al. 2000; Dempsey et al. 1995).
Semi-structured interviews or in-depth interviews have been used in other
descriptive research studies that include Caucasian, or African-American women
(McSweeney et al. 2000; LaCharity 1999; Rosenfeld et al. 2000; Lee et al. 2000).
It is not known if in-depth interviews have been used with Latinas with respect to
questions focused on health beliefs and interpretation of AMI symptoms.
Appendix A contains the COMIRB approved questions used during the patient
interviews. Both interviews were audio-recorded with permission from the
patient. No subjects denied permission to audiotape the interview. Before the
interview began, I asked the participant for permission to contact them for the
follow-up interview.
Interview topics included womens personal attitudes, beliefs, and
experiences related to heart attack symptoms, social support networks within
which discussions about personal health issues occurred (specifically their heart
attack symptoms), and experiences with accessing medical care (both the
emergency medical response team, the hospital, and from the doctors office). I
also used category questions (based on Kleinmans explanatory model queries)
to probe interviewees for elaboration and/or clarification on all the relationships,
activities, settings, and relative value of the domain being discussed (Kleinman
1988; Crabtree et al. 1999).
68


Follow-up Interview
The follow-up interview allowed women to clarify symptoms, validate interview
data, or expand on their symptom experiences. Interviewing women again after
the event provided the opportunity to examine the symptoms that changed or
decreased in frequency or occurrence; this enhanced the identification of
symptoms as prodromes to their AMI. Interviewing women too close to their
event may not capture the changes in prodromal symptoms. It is important to
understand the full scope of prodromal symptoms before and after the event since
less is known about them in relation to women and AMI.
Follow-up interviews were scheduled approximately two weeks after
discharge from the hospital. Telephone contact was used to confirm data
collected from the first interview and to verify the transcribed information. The
second interview was conducted with the same interviewer who conducted the
first interview in the hospital. Participants were given the telephone number of
the study telephone (connected to an answering machine) if they wished to make
contact or to re-schedule an interview. All phone contacts were logged and
content noted.
The Medical Record Review
Details of the chart review were briefly described in the section titled,
Contribution of this Study. The purpose of the chart reviews was to compare
and corroborate interview data with the data documented in the admitting
records, i.e. emergency room intake sheet. As an additional analysis, I gathered
69


narrative descriptions documented by the emergency response teams, i.e., 911,
and the hospital emergency room staff.
The purpose of collecting narrative information was to examine how
nurses and physicians interpret the reported symptoms. For example, the
physician may only document acute chest pain but may not document vague or
less frequent symptoms reported by patients. I examined the medical chart to see
if an interpretation was documented rather than the symptoms reported by the
patient. I also examined to the emergency room records to see if only typical
ischemic heart disease symptoms were documented. The data were entered on a
coding sheet and transferred into Excel for analysis.
Analytic Methods for Quantitative Data
Because the sample size was too small for most statistical comparisons,
the analysis was limited to descriptive statistics, such as frequencies and
percentages, and chi square analysis where appropriate. First, I examined the
data describing the total sample followed by separate analyses for Latinas and
non-Latinas. Next, I described the differences among Latinas according to their
language preference for the interview (English or Spanish). The data analysis
identifies participants living in rural, urban, or mixed communities in Colorado.
A mixed community is a small town located near a larger community and is
assigned the zip code affiliated with the larger town (see Appendix K). The data
are displayed in tables showing sociodemographic data, clinical variables, delay
time, and a comparison of symptom descriptions reported by the patient and
recorded by medical personnel.
70


Data Reduction Processes
The MCR forms, interview guides, and vignettes served as data sources
for the quantitative analysis. Data were abstracted from these sources and
entered into the Subject Coding Form (Appendix O) followed by entry into an
Excel spreadsheet. The Subject Coding Form was comprehensive and designed
with variable descriptors culled from the transcripts and MCR forms. Once all
data were entered for each patient, all Subject Coding Forms were reviewed to
see if any patterns existed among the sample and categories were collapsed
where patterns were present.
An Excel spreadsheet was created using the SCF data, for the purposes of
running descriptive statistics and chi squares. Patients were given a unique
identifier on all forms and the Excel spreadsheet. Microsoft Visio Professional
(version 2002, Redmond, WA) was another analysis tool used to corroborate data
collected on the Subject Coding Forms to achieve Aim 1.
Measures. All data were described comparing the total sample, all
Latinas, all non-Latinas, Spanish speaking Latinas, and English speaking Latinas.
The sample characteristics included age, highest level of education completed,
health insurance status, marital status, whether the patient lived alone or not,
geographical location of patients residence, cardiovascular risk factors, whether
the patients co-morbidity factored in the patients assessment of their symptoms,
and method of transportation. Data were collected during the interviews and/or
verified using the medical record data.
71


Data were recoded for the purposes of calculating p values. Recoded
variables were education, health insurance, geographic location of patients
residence, and method of transportation. These variables were dichotomized to
form large enough cell sized to facilitate using chi square analysis.
To meet the eligibility criteria for the study, subjects had to have a
positive Troponin-1 (Tnl, ng/ml) which was reported by the nurses to me. For
the medical record reviews, three chart reported troponins were collected: the
first recorded troponin value, the first positive troponin value recorded, and the
peak troponin value recorded. Positive Troponin-I data reflected the reference
ranges from each recruiting site. The methods for assays and analyzers varied
among hospitals. A list of each hospitals Troponin criteria is listed in Appendix
L.
Delay time was divided into three values that were documented as
follows:
T1: Period between onset of clinical symptoms and
subjects decision to seek professional help
T2: Period between initiation of travel to the hospital
and arrival at the hospital
T3: Total time between onset of clinical symptoms and
arrival at the hospital (the sum of T1 and T2)
As demonstrated in Tables 5.9 and 5.9.1, T3, or total pre-hospital delay
time, is represented. Obtaining the value of T2 was less reliable than establishing
the other two values. Often, subjects could not remember details of the travel
time due to the acuity of their symptoms or they guessed the time it took to
travel. The time of the onset of clinical symptoms was established by the patient
and is indicated on the patient career timelines and the medical chart review
72


forms. For some patients there were two times that could be represented as the
onset of clinical symptoms. For these patients, T1 was subdivided into Tla and
Tib. Tla represents the earliest acknowledgement of symptoms in the prodrome
period.
Analysis
Analysis began with descriptive summaries of the data and chi square
analysis, a non-parametric test of statistical significance for bivariate tabular
analysis (Connor-Linton 2004). The chi square statistic tests whether or not two
different samples are different enough that the findings from the sample can be
generalized to the population from which the sample was drawn. The total
sample was divided into two groups only, Latinas and non Latinas, for
calculation ofp values. In Chapter 5, Tables 5.1 and 5.1.1 present summary
statistics and chi square statistics. Data are represented in frequencies and
percentages.
Figure 5.0 shows the results of a small exploration of the data to see if
there was significance between peak troponin and total delay time. The process
for producing this figure was the following: 1) Frequency distributions were run
for peak Troponin and total delay time. Since both were skewed, the basic
assumptions for correlations were not met (data need to represent a normal
distribution), 2) A log transformation was created using SAS (Version 9.1 Cary
NC 2005). The graphs below show the pre-log transformed data and the data
after the log transformation was done. Log transforming the data created a
73


distribution that was closer to normal and made the data appropriate for doing a
correlation analysis.
74


'J
FREQUENCY
0 80 K0 240 320 400 480
FREQUENCY
&
11
13
12
11
10
9
8
7
6
5
4
3
2
1
0
FVak Trouonbi
Figure 4.1 Peak Troponin Data Before and After Log Transformation
LogIPfcak TroponUi)


'J
On
FREQUENCY
30
0 200 400 000 800 1000
Total Delay
:ncy
h
is
12 :
U
10:
9
8 :
7 ;
6 -
5 :
4
3
2
1
0
Figure 4.2 Total Delay Time Before and After Log Transformation
Log( Total Delay)


Analytic Methods for Interview Data
Chapter Six describes the qualitative methods used to analyze the patient
interviews; provides information about the cases; and describes and discusses the
results of the qualitative analyses. The analytic methods described here were
adopted from those supported by Glaser and Strauss (1967), Morse (1994),
Strauss (1987), and Kleinman (1988). Chapter 4 addresses the methods for data
collection. These produced two data sources per patient to be analyzed, the
transcripts from the interview and the medical chart review forms, but the focus
here is on the results from the interviews.
A variation of Crabtree and Millers (1999) immersion/crystallization
(I/C) organizing style (Figure 4.1) was employed with elements of their template
analysis technique (Figure 4.2). The I/C style is intuitive and fluid and requires
the ability to listen deeply to the data and the individuals in order to find
patterns. It consists of cycles whereby the analyst immerses him or herself into
and experiences the text, emerging after concerned reflection with intuitive
crystallizations, until reportable interpretations are reached (Crabtree and Miller
1999). The I/C process begins with a reflection and awareness of the research
questions and insights by the interpreter prior to collecting and analyzing the text.
77


Immerelon/Crvstaltlzation
Interpreter
{Reflective Participant)
t
Report ----- TEXT
Corroborating/
legitimating
Crystallite J
Connections
Figure 4.3 Immersion/Crystallization Organizing Style
78


Template
Report *

Template/
code book
I
rp TEXT +-
1
Identify Units
1
Reviae/sort
categories
j
Connecting
Corroborating/
legitimating
J
Figure 4.4 Template Analysis Technique
Broad a priori codes were developed based on general understandings
from prior research or theory and after initial exploration of the data. Other
codes emerged from data collection and analysis. To achieve this, codes were
synthesized into a code manual in preparation for further analysis. These insights
were crystallized during the process of data collection, iterative analysis, and
immersion/crystallization throughout the data collection and immersion into the
data. Insights were captured in the form of memos (Strauss and Corbin, 1990). I
79


went back and reviewed all memos written throughout data collection and
analysis and incorporated them into the findings.
Table 4.1 A Priori Codes
From Behavioral Science Theory From Chanter 2 Literature Review
Cues to action Maintaining control
Perceiving susceptibility/severity of Relinquishing control
symptoms Influencing factors
Appraising the symptoms Awareness of the symptoms
Acting on symptoms Something was wrong
Explanatory Models Bothering others Competing Demands Coming and going symptoms Ignoring the symptoms Denying the symptoms Feeling foolish Working through the problem (sobreponerse)
After each audio recorded interview, I completed a contact summary form
(shown in Appendix H). This tool captured immediate memos, insights, and
themes from the interview and served as a preparation tool for the follow up
interview. Insights began to crystallize as the data were being collected and
assisted with the identification of early patterns. Patterns were documented in
memos and the Summary Contact Forms following each interview. Other
insights were gathered from discussions with nursing staff, research staff, and
colleagues or friends. At times, insights occurred during the interviews and
jottings were made for future reference. This is sometimes referred to in
process analyses (Crabtree et al., 1999) and allowed me to refocus subsequent
interviews to check the validity of the insights.
The interviews were transcribed verbatim from audiotapes, using
Olympus digital recording technology and WAVPedal software (version 5,
80


2002). The narrative data was entered and coded according to the constant
comparative method (Glaser and Strauss 1967). Interviews conducted in Spanish
were translated into English for coding and thematic development.
Data Reduction Processes
Each interview produced an average of 30 pages of transcription. The
conversational style of the interview made following the sequence of events
difficult at times. With a conversational style and permitting necessary
interruptions from the healthcare team to care for the patient, the interview was
spontaneously interrupted by questions from family members, visitors, and
nursing staff. Another type of interruption in the conversation occurred when a
patient became ill during the interview and the conversation was put on hold until
the patient felt well enough to continue.
Crystallizing and Corroborating/Legitimating Connections. To reduce the
transcript data into manageable formats, I used chunking approaches (Crabtree
and Miller 1999) to create a vignette for each participant and displaying
techniques to create a patient career timeline for each participant.
Vignettes. Chunking segments of text permits the identification of
themes within the text chunks found in the vignettes. Vignettes, a type of data
subset, are summaries of rich pockets of data that are especially representative
81


and meaningful of each case. They are written in a focused way for interim
understanding to mine the data from the transcripts (Miles and Huberman
1994). The structure is narrative or story-like to re-count the event as it unfolded
in each patients life. Each vignette opens with an introduction to the person to
establish how the MI fit within the context of her life, including general
background information and/or demographic information, to set the stage of
understanding the patients story leading up to the index MI. The story continues
by linking time points to symptoms, thoughts, care seeking events, and
communication. The vignettes aided in formulating themes, codes, and early
theories, and captured the essence of each persons story in narrative form.
Individualized themes and codes were generated at the end of each vignette and
transferred to a master list representing the 43 patients experiences. The
vignettes permitted cross comparisons of how the events enfolded.
Data Management and Analysis. Analysis involved identifying emerging
themes, interpretations, and ideas about participant experiences with AMI.
Analysis occurred throughout the data collection and analysis phases (Morse
1994; Huberman et al. 1998). Data management was implemented using Atlas ti
computer software. This software was chosen because it is designed to apply
grounded theory and analytic induction methods by its use of codes, memos, and
diagrams which were then developed into a conceptual understanding of the
process (Huberman et al. 1998). The software allowed for the identification of
variables and linkages between variables using an iterative approach.
Using content analysis, specific symptoms were identified and grouped
into symptom clusters (Glaser et al. 1967). The validity of the qualitative data is
82


the extent to which the phenomenon is captured and reflected in the data
(Dempsey et al. 1995). Triangulation is a method used to augment information,
discover inconsistencies between data from chart recordings versus data obtained
in relaxed conversational settings. Triangulation strengthens the credibility of
data collected when more than one form of data collection is used. Data from the
first interview were verified and cross-checked by comparing them to the data
from the second interview and to a lesser extent data in the medical records.
Triangulation served to verify the themes and codes captured. Coding supported
the steps taken to identify emerging theory (Strauss et al. 1998).
The vignettes, Patient Career Timelines (PCTs), and master list of early
themes and codes generated additional codes in preparation for analyzing the
transcripts and vignettes using Atlasti software (version 5.0, Berlin, Germany).
Atlas facilitated the organization of text segments and helped to search for
confirming/disconfirming evidence of the interpretations. Analytical induction
(Glaser et al. 1967) was the approach taken to develop ideas that emerged from
the data and to convert them to hypotheses.
Reliability and Validity Checks. Five steps were involved in checking the
data for reliability and validity. Adequacy refers to the amount of data collected,
not the number of subjects. Saturation of themes signaled that adequacy was
ensured. Adequacy was also ensured when variation in the data was accounted
for and understood. Appropriateness was ensured by the culling of information
according to theoretical needs. When repetition and saturation were reached,
this signaled the adequacy and appropriateness of themes. The third method to
ensure rigor was the use of an audit trail. This documented all research process
83


steps, including all decision points. This documentation allowed the peer auditor
to reconstruct and verify the research process and findings that led to the studys
conclusions (Morse 1994; Huberman 1998). Verification exercises also
supported the rationale for conducting the second interview. Processes for
seeking treatment were discussed with each patient during the second interview.
This confirmed the accuracy of the field notes and transcriptions from the first
interview and allowed patients to provide additional information.
The final step to ensure rigor was the use of a peer auditor to check the
data for adequacy and appropriateness through an audit trail and verification
exercises to strengthen the findings and interpretations. As principal
investigator, I attended all of the interviews and it was assumed that I had the
most knowledge compared to a second auditor who viewed the transcriptions
and other forms of reduced data but did not have access to the emotions, body
language, or other visual cues that can not be captured on typed formats.
Five interviews were randomly chosen to establish accuracy of the data
transferred from the transcriptions to the vignettes and PCTs. Peer audit led to
discussion of the discrepancies across data sources. These differences were
discussed until we came to an agreement.
84


CHAPTER 5
ANALYSIS OF DATA
FROM MEDICAL CHART REVIEWS
Chapter Five presents the results from the chart review data. The analysis
addressed the first aim of this study, to describe and compare the initial
symptoms that Latina and non-Latina women report for an MI. Results were
guided by the 7 Attributes of a Symptom (Bickley et al. 1999) and displayed in
tables. Results, shown as frequencies and percentages, represent the comparisons
between the total sample, whites, Latinas, and Spanish speaking Latinas. The
SCF and Excel tools are best for counting activities. The VISIO displays were
used mainly as a reference to corroborate data entered in the SCF and Excel.
The Sample
Of the 43 subjects who consented to participate in the study, 41
completed both interviews. The two subjects who did not complete both
interviews were Latina (one passed away; the other was lost to follow up).
However, all subjects were included in the quantitative analysis. Approximately
6 patients were eligible for the study but declined to participate. The reasons
patients chose not to participate (as recorded in field notes) included: feeling too
tired, feeling sick, not wanting to bother with the study, or being residents from
out of state and feeling it would be too difficult to participate in a research study.
85


Approximately 9 patients were missed due to the short length of stay for an
uncomplicated myocardial infarction.
Data for the 43 respondents are shown in Table 5.1. Non-Latinas
represented 60 percent of the total sample (N=26). Of these, 25 were White and
one subject was African American. Latinas represented forty percent of the total
sample (N=17) and of this group five of the seventeen preferred to speak Spanish
for their interview (16 were of Mexican American heritage). Nine out of 17
Latinas referred to self-reported their ethnicity as Hispanic. One Latina identified
herself as Latina. Three Latinas self-reported their ethnicity as Spanish. Three
reported their ethnicity as Mexican and one woman described herself as
Panamanian. The age range of the total sample was thirty-nine to eighty-nine
years old. There was no difference in mean age between the Latinas (62 11.4)
and the non-Latinas (62 12.5), although the Latinas who preferred to speak
Spanish had a slightly older mean age (68 9.9).
Overall, there were more socio-demographic similarities than differences
between Spanish speaking and English speaking Latinas. Education was the only
sociodemographic variable that showed a statistically significant difference (p
^>.001) between the Latinas and non-Latinas. Of the total sample who
completed less than twelve years of school, Latinas were much less educated
compared to non-Latinas (88% vs. 27%). In fact the non-Latinas were well
educated. Fifty four percent had a post high school education. This included
completing trade or business school beyond a high school education, and one or
more college degrees. However, when the ethnicity language variable was
considered, the contrast in education was more striking. For example, 80% (4
out of 5) of the Latinas who preferred speaking Spanish completed less than a
86