 

..

~
3.. ...n...

a... W. 3.. 1
4.. . . dﬁﬁs
9. ..

1 . .

.vﬁ. «.31 .
. t x . . . .
. ..umnmn
«a.

 

tW.‘
r manure.
1 (shall.

31%»...
. .
1 .. 4m»?
u. .........
’4‘

 

. . 6
I.” . «.4.» ,.
‘uuhnr! u. .21
5.34.?! awrnﬁk.‘ ..

 

 

 

 

 

.z
s
. .
......zmwemr Mé.....mi.
. fun...
35'” a Iv 4 “\LJA.A
I. s. is... ..:-$4... 7 v
. f. a. ...
3...???» .mﬁifﬂ
f u. . .- 5.1. .
. . e... ......nu». ..
a «$1.... ..
«amt N54»
. ti 9’.)
.433.
... ......
mm". .34..
9:, inf
X y
54" It; to
5am. ......1.....
3| J... ’n1 v

I
(1.3.1.! .
ml. - .1.» .l

.
‘ . ...). 5

€19.41 “up.“

2;. tr :1 T1...

 

 

 

LIBRARY
Michigan State
University

.1

This is to certify that the
dissertation entitled

PREVALENCE OF WORK-RELATED DISABILITIES USING
THE ICF DEFINITION IN AN EMPLOYED COHORT
DIAGNOSED WITH BREAST OR PROSTATE CANCER

presented by

KATHLEEN OBERST

has been accepted towards fulﬁllment
of the requirements for the

 

 

 

 

PhD. degree in Epidemiology
/ ,II/lajor Profesg’or’ 5 Signature
9/12/07
Date

MSU is an afﬁnnative-acﬁon, equal-opportunity employer

o-.---:-- cu-.-.-.---.-.-----------------..-u.-

.-.—.-.---1--

. ....-.-.-.-.-.--

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DAIEDUE

DAIEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/Dale0ue.indd-p.1

 

 

PREVALENCE OF WORK-RELATED DISABILITIES USING THE ICF
DEFINITION IN AN EMPLOYED COHORT DIAGNOSED WITH BREAST OR
PROSTATE CANCER
By

Kathleen Oberst

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Epidemiology
2007

-
D»
-

I: hr
'mP

we"-
trEet.

ion

I}.
m
(1 )

(T.
“'V

(I)

(n

prE‘Ja

__’7

 

 

ABSTRACT
PREVALENCE OF WORK-RELATED DISABILITIES USING THE ICF
DEFINITION IN AN EMPLOYED COHORT DIAGNOSED WITH BREAST OR
PROSTATE CANCER
By
Kathleen Oberst
Advances in cancer diagnosis and treatment have resulted in
improved survival. Increasing emphasis on early detection has resulted in
working individuals having to face the disease. Therefore, effects of cancer
treatment may result in employed patients requiring workplace accommodations
to minimize absenteeism. This investigation focuses on cognitive and physical
disabilities reported at twelve and eighteen months following diagnosis by 447
breast and 267 prostate cancer patients recruited from the Metropolitan Detroit
Cancer Surveillance System. Disability was deﬁned to include activity limitation
in speciﬁed job tasks with labor market participation restriction. Twelve and
eighteen month disability prevalence estimates were calculated and compared.
Furthermore, inﬂuences of personal and environmental factors on disability
development in the breast cancer cohort were tested using logistic regression.
Participants reported more cognitive work demands than physical demands at
both time points. However, respondents reported signiﬁcantly more physical
disabilities compared to cognitive disabilities at twelve and eighteen months
(p<0.05); yet estimates were below national averages. Women experienced
signiﬁcant reduction in physical (17% to 10%) and cognitive disability (9% to 5%)

prevalence estimates from twelve months to eighteen months (p<0.05). Men did

HOT

 

 

 

not enjoy signiﬁcant changes in estimates of physical (8% to 5%) or cognitive
(3% to 1%) disability during this time frame. Regression results indicated
personal factors such as income or age were not signiﬁcant predictors of
disability in women. However, African-American women were at higher risk for
cognitive disability. Much more important to the outcome of disability was cancer
stage and treatment. Women diagnosed at later stages had signiﬁcantly more
disability. The most signiﬁcant environmental factor associated with presence of
disability was the absence of paid sick leave as a beneﬁt. Since treatment is
reﬂective of cancer stage, the importance of screening and early detection must
be stressed to minimize disability. Also, adequate recovery time is necessary to
adjust to the affects of cancer treatments. Individuals who cannot afford to take
this time and work through treatment may experience longer lasting impacts.
Employers may wish to consider the cost of paid sick leave versus the costs
associated with lower productivity to re-evaluate their beneﬁt packages.
Limitations of this study included design as a secondary analysis, low power to
detect small changes and lack of data regarding accommodation and
rehabilitation available to participants. Also, future studies should include

mechanisms to track the nature and intensity of limitations and restrictions.

Copyright by
KATHLEEN OBERST
2007

 

TM

this

 

DEDICATION

This work is dedicated to my husband, Dan. Once again, your support has made
this degree possible. I could not have completed this without your love, patience,
and ﬁnely tuned house-husband skills. I hope this will lead to an even brighter
future and I look forward to putting all my energy and time back into us. I love

you very much.

 

anc

Mic

 

ACKNOWLEDGEMENTS

I owe a great deal of gratitude to my advisor, Joseph C. Gardiner, PhD
and my entire committee, Cathy J. Bradley, PhD, C. William Given, PhD, and
Michael R. Rip, PhD. A special thank you also goes to Zhehui Luo, PhD for her
helpful comments on various drafts along the way. Your ongoing patience with
my changes, distractions, stream of consciousness writing, and ongoing
revisions were much appreciated. This was a tremendous learning experience,
both personally and professionally, and your mentorship was invaluable. I look
fonNard to starting down a new career path and hope that we have opportunities
to continue to work together. I also want to thank the many friends and family

who supported me through the entire degree process.

vi

millhg‘th. a.

 

LIE
US
LIE

CF

CH;
2 .

A)

 

 

 

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................. xi
LIST OF FIGURES ............................................................................................ xiv
LIST OF ABBREVIATIONS ................................................................................ xv
CHAPTER 1: INTRODUCTION ........................................................................... 1
1.1 Theoretical Framework ............................................................................... 2
1.2 Overview of Disability...................................... ........................................... 6
1.2.1 Prevalence Trends ............................................................................... 6
1.2.2 Factors Driving Disability Prevalence Trends ...................................... 7
1.2.3 Social Role - Work Force Participation ................................................ 8
1.2.4 Increasing Government Attention and Action ....................................... 9
1.2.5 Characteristics of Transient Disabilities ............................................. 13
1.2.6 Disability versus Illness ...................................................................... 14
1.3 Overview of Breast and Prostate Cancers ............................................... 15
1.3.1 Prevalence Trends ............................................................................. 15
1.3.2 Factors Driving Prevalence Trends .................................................... 16
1.3.3 Treatment Options ............................................................................. ’17
1.4 Rationale for Associating Disability with Cancer ....................................... 17
1.4.1 Rehabilitation Potential to Ameliorate Adverse Effects ...................... 19
1.5 Study Objective ........................................................................................ 21
1.5.1 Study Proposal .................................................................................. 21
1.5.2 Study Rationale ................................................................................. 22
1.6 Summary .................................................................................................. 24
CHAPTER 2: LITERATURE REVIEW ............................................................... 25
2.1 Critiques Of ICF as the Selected Theoretical Framework ......................... 25
2.1.1 Activity versus Participation Deﬁnition ............................................... 25
2.1.2 Omission of Quality of Life Component .............................................. 28
2.1.3 Inﬂuence of Contextual Factors ......................................................... 28
2.1.4 Model Testing and Evaluation ............................................................ 29
2.2 Domain 1: Activity Limitations Associated with Cancer ............................ 31
2.2.1 Physical Activity Limitations ............................................................... 31
2.2.1.1 Physical Activity Limitations Associated with Cancer Diagnosis
and Treatments ........................................................................................ 32
2.2.1.2 Physical Activity Limitations Reported in Literature - Breast
Cancer ...................................................................................................... 33
2.2.1.3 Physical Activity Limitations Reported in Literature — Prostate
Cancer ...................................................................................................... 36
2.2.1.4 Physical Activity Limitations Reported in Literature — Cancer Site
Not Speciﬁed ............................................................................................ 37
2.2.2 Cognitive Activity Limitations ............................................................. 39

vii

 

 

 

3,4

A

“..HA

2.2.2.1 Cognitive Activity Limitations Reported in Literature - Breast

Cancer ...................................................................................................... 39
2.2.2.2 Cognitive Activity Limitations Reported in Literature — Prostate
Cancer ...................................................................................................... 40
2.3 Domain 2: Participation Restriction Associated with Cancer .................... 41
2.4 Employment Experience of Cancer Patients ............................................ 42
2.4.1 Return to Work for Cancer Patients ................................................... 42
2.4.2 Work Limitations Reported by Cancer Patients ................................. 44
2.4.3 Labor Market Outcomes of this Study Population .............................. 46
2.5 History of Disability Models ...................................................................... 47
2.6 Literature Review Summary ..................................................................... 54
2.6.1 Prevalence and Direction of Disability Change .................................. 54
2.6.2 Comparison of Disability Estimates using Two Deﬁnitions ................. 55
2.6.3 Association of Contextual Variables with Disability ............................ 56
2.6.4 Additional Contributions of Proposed Study ....................................... 57
CHAPTER 3: RESEARCH METHODOLOGY .................................................... 59
3.1 Human Subjects Review .......................................................................... 59
3.2 Data: Source Study Overview .................................................................. 59
3.2.1 LMOS as Source Study ..................................................................... 59
3.2.2 LMOS Subject Recruitment and Retention ........................................ 60
3.2.3 LMOS Interview Subject Matter ......................................................... 65
3.3 Data: Disability Study Subject Restriction ................................................ 67
3.3.1 Disability Study Descriptive Data ....................................................... 69
3.3.2 Disability Study Disease-Related Characteristics .............................. 73
3.3.3 Disability Study Employment-Related Characteristics ....................... 75
3.3.3.1 General Employment ................................................................... 75
3.3.3.2 Job Satisfaction and Employer Accommodation ......................... 77
3.3.3.3 Activity Requirement: Job Tasks ................................................. 79
3.4 Research Questions and Hypotheses ...................................................... 83
3.4.1 Prevalence and Direction of Disability Change .................................. 83
3.4.1.1 Methodology: Document Required Job Activities ........................ 84
3.4.1.2 Methodology: Create Indicator Variable ...................................... 89
3.4.1.3 Methodology: Generate Disability Variable ................................. 91
3.4.2 Disability Prevalence Disparities Based on Deﬁnition ........................ 93
3.4.2.1 Methodology: Compare Disability Prevalence ............................. 94
3.4.3 Contextual Factor Association with Disability ..................................... 94
3.4.3.1 Methodology: Multivariate Analysis using Logistic Regression ...94
CHAPTER 4: RESULTS .................................................................................. 101
4.1 Study Question 1: Prevalence and Direction of Disability Change ......... 101
4.1.1 Hypothesis 1: Comparison of Physical and Cognitive Disability ...... 101
4.1.1.1 Comparison of Physical and Cognitive Activity Requirements
Between 12 and 18 Months .................................................................... 101
4.1.1.2 Comparison of Domain 1: Activity Limitation ............................. 105
4.1.1.3 Comparison of Domain 2: Participation Restriction ................... 107

viii

4.1.2 Comparison of Physical Disability .................................................... 108

4.1.2.1 Physical Disability Status Change ............................................. 109
4.1.3 Comparison of Cognitive Disability .................................................. 111
4.1.3.1 Cognitive Disability Status Change ........................................... 112
4.2 Study Question 2: Disability Prevalence Disparity Based on Deﬁnition.. 1 14
4.2.1 Hypothesis 2: Comparison of Disability Prevalence ......................... 114
4.2.1.1 Physical Disability ...................................................................... 114
4.2.1.2 Cognitive Disability .................................................................... 115
4.3 Study Question 3: Association of Factors with Disability ........................ 116
4.3.1 Hypothesis 3: Inﬂuence of Personal Factors .................................... 117
4.3.2 Hypothesis 4: Inﬂuence of Environmental Factors ........................... 117
4.3.3 Physical Disability ............................................................................ 117
4.3.3.1 Twelve Month Regression Results ............................................ 117
4.3.3.2 Eighteen Month Regression Results ......................................... 120
4.3.4 Cognitive Disability .......................................................................... 122
4.3.4.1 Twelve Month Regression Results ............................................ 122
4.3.4.2 Eighteen Month Regression Results ......................................... 124
CHAPTER 5: DISCUSSION ............................................................................. 126
5.1 Study Question 1: Prevalence and Direction of Disability Change ......... 126
5.1.1 Hypothesis 1: Comparison of Physical and Cognitive Disability ...... 126
5.1.1.1 Comparison of Physical and Cognitive Activity Requirements ..127
5.1.1.2 Comparison of Domain 1: Activity Limitation ............................. 129
5.1.1.3 Comparison of Domain 2: Participation Restriction ................... 136
5.1.2 Comparison of Physical Disability .................................................... 138
5.1.3 Comparison of Cognitive Disability .................................................. 139
5.1.4 Factors Associated with Changing Disability from Twelve to Eighteen
Months ....................................................................................................... 140
5.2 Study Question 2: Variation in Reported Prevalence Due to Disability
Deﬁnition ....................................................................................................... 142
5.2.1 Hypothesis 2: Comparison of Disability Prevalence ......................... 142
5.3 Study Question 3: Association of Factors with Disability ........................ 143
5.3.1 Hypothesis 3: Inﬂuence of Personal Factors .................................... 144
5.3.2 Hypothesis 4: Inﬂuence of Environmental Factors ........................... 144
5.3.3 Physical Disability ............................................................................ 144
5.3.3.1 Twelve Month Regression Results ............................................ 144
5.3.3.2 Eighteen Month Regression Results ......................................... 148
5.3.4 Cognitive Disability .......................................................................... 148
5.3.4.1 Twelve Month Regression Results ............................................ 148
5.3.4.2 Eighteen Month Regression Results ......................................... 149
5.4 Limitations Associated with Project ........................................................ 151
5.4.1 Overview .......................................................................................... 151
5.4.2 Study Participants ............................................................................ 152
5.4.3 Evaluation of Selection Bias ............................................................ 156
5.4.4 Disease Aspects of Cohorts ............................................................. 157
5.4.5 Employment Aspects of Cohorts ...................................................... 158

CHAPTER 6: CONCLUSIONS ......................................................................... 160

6.1 Implications & Recommendations .......................................................... 160

6.2 Summary ................................................................................................ 164
Appendix A: Work Related Activity Limitation Questions ................................. 168
Appendix B: Participation Restriction Questions ............................................... 169
Appendix C: Selection Bias Review — Enrolled Subjects Compared to Potentially
Eligible Subjects ................................................................................................ 171
Appendix D: Selection Bias Review - Census Comparison of Enrolled to
Potentially Eligible ............................................................................................. 173
BIBLIOGRAPHY ............................................................................................... 176

LIST OF TABLES

Table 1: General Nature of Interview Variables Collected ................................. 65

Table 2: Employment Categories Reported by Employed Participants at 12 and

18 months ........................................................................................................... 69
Table 3: Study Participant Descriptive Statistics ................................................ 71
Table 4: Study Participant Disease Statistics ..................................................... 74
Table 5: Study Participant Employment Statistics at Baseline ........................... 75

Table 6: Employment Categories Reported by Employed Participants at 12 and
18 months ........................................................................................................... 76

Table 7: Participant Report of Job Satisfaction and Employer Accommodation at
12 months ........................................................................................................... 78

Table 8: Job Task Frequency Reported at Twelve Months by Breast Cohort....80
Table 9: Job Task Frequency Reported at Twelve Months by Prostate Cohort.81
Table 10: Job Task Frequency Reported by Breast Cohort at Eighteen Months82

Table 11: Job Task Frequency Reported by Prostate Cohort at Eighteen Months

............................................................................................................................ 83
Table 12: Rotated Factor Loadings for All Job Tasks ......................................... 85
Table 13: Rotated Factor Loadings for Targeted Job Tasks ............................... 86
Table 14: Summary statistics for factor scores ..................................... 87
Table 15: Composite Activity Requirements for Cohorts .................................... 88
Table 16: Activity Limitation Coding Example — Physical Task .......................... 90
Table 17: Activity Limitation Coding Example — Cognitive Task ........................ 90
Table 18: Disability Coding Example ................................................................. 92
Table 19: Variables considered for inclusion in modeling procedures ................ 97

xi

Table 20: Rotated Factor Loadings for Job Satisfaction Variables ...................... 98
Table 21: Rotated Factor Loadings for Targeted Job Satisfaction ...................... 99
Table 22: Rotated Factor Loadings for Job Tasks .............................................. 99
Table 23: Activity Requirements for Breast Cohort .......................................... 102
Table 24: Activity Requirements for Prostate Cohort ....................................... 104
Table 25: Activity Limitation for Breast Cohort ................................................. 106
Table 26: Activity Limitation for Prostate Cohort .............................................. 107
Table 27: Participation Restriction Reporting ................................................... 108
Table 28: Physical Disability Reporting in those Performing the Task ............. 109
Table 29: Cognitive Disability Reporting .......................................................... 111
Table 30: Physical Disability Prevalence Comparisons Among the Cohorts....115
Table 31: Cognitive Disability Prevalence Comparisons Among the Cohorts .. 1 16
Table 32: Regression Results for Physical Disability at Twelve Months Among
Breast Cancer Cohort (n=279 observations retained) ....................................... 119
Table 33: Regression Results for Physical Disability at Twelve Months Among
Tiff. if??? 3???? f. 371???? FIT??? .TTTT???.??§?TY?‘.'.‘1‘.‘§°1 .’.?‘i‘i‘.ef‘2’o
Table 34: Regression Results for Physical Disability at Eighteen Months Among
Breast Cancer Cohort (n=221 observations retained) ....................................... 121
Table 35: Regression Results for Physical Disability at Eighteen Months Among

Breast Cancer Cohort - Treatment Eliminated (n=222 observations retained). 122

Table 36:

Regression Results for Cognitive Disability at Twelve Months Among

Breast Cancer Cohort (n=277 observations retained) ....................................... 123

Table 37:

Regression Results for Cognitive Disability at Twelve Months Among

Breast Cancer Cohort — Cancer Stage Eliminated (n=374 observations retained)

........................................................................................................... 124

xii

Table 38: Regression Results for Cognitive Disability at Eighteen Months Among
Breast Cancer Cohort (n=221 observations retained) ....................................... 125

Table 39: Regression Results for Cognitive Disability at Eighteen Months Among
Breast Cancer Cohort — Cancer Stage Eliminated (n=304 observations retained)

.......................................................................................................................... 125

xiii

LIST OF FIGURES

Figure 1: Interactions between the components of the International Classiﬁcation

of Functioning, Disability and Health ..................................................................... 4
Figure 2: Study components nested within the ICF model ................................... 6
Figure 3: International Classiﬁcation of Impairments, Disabilities, and Handicaps

Linear Structure of Consequence of Disease ..................................................... 53
Figure 4: Data Collection Timeline ..................................................................... 61
Figure 5: LMOS Breast Cancer Recruitment Flowchart ..................................... 63
Figure 6: LMOS Prostate Cancer Recruitment Flowchart .................................. 64
Figure 7: Breast Physical Tasks ....................................................................... 132
Figure 8: Prostate Physical Tasks .................................................................... 133
Figure 9: Breast Cognitive Tasks ..................................................................... 134
Figure 10: Prostate Cognitive Tasks ................................................................ 134

xiv

LIST OF ABBREVIATIONS

International Classiﬁcation of Functioning, Disability and Health ...................... ICF
World Health Organization ............................................................................. WHO
Americans with Disability Act ........................................................................... ADA
New Freedom Initiative ..................................................................................... NFI
Centers for Disease Control and Prevention ................................................... CDC
Labor Market Outcomes Study ..................................................................... LMOS
Metropolitan Detroit Cancer Surveillance System ...................................... MDCSS
Surveillance, Epidemiology and End-Results ............................................... SEER
Health and Retirement Survey ........................................................................ HRS
International Classiﬁcation of Impairments, Disabilities, and Handicaps ....... ICIDH
Family Medical Leave Act ............................................................................. F MLA

XV

CHAPTER 1: INTRODUCTION

Twenty percent of the US. population is affected by a disability of some
kind and societal costs associated with these approach $400 billion dollars.
Cancer conditions affect a smaller proportion of the population yet more than
$200 billion are invested in care associated with these diagnoses. Potentially,
the combination of these conditions accounts for nearly 5% of the US. gross
domestic product. Therefore, it is ﬁtting that attention be focused on the extent to
which these conditions co-exist and factors that may inﬂuence their convergence.

This effort is focused on work-related disability in a cohort of breast and
prostate cancer survivors. Prevalence estimates of work-related physical and
cognitive disabilities at twelve and eighteen months post-diagnosis will be
calculated and compared among the enrolled study subjects. One of the unique
features of this work is that the term “disability” will be explicitly deﬁned and
quantiﬁed. After the prevalence estimates have been documented, the focus will
shift to identifying contextual variables on the occurrence of disability.
Speciﬁcally, variables of interest categorized as personal or environmental will be
evaluated. Personal factors include socio-demographic (age, gender, race) and
disease-related (stage, treatment) variables while environmental factors include
employment-related (job satisfaction, employer accommodation, physical nature
of job) variables.

This introduction will be organized in a deliberate manner. The foundation
for this study will be established with an overview of disability, cancer and work

force participation. After describing the preferred theoretical framework, the

signiﬁcance of disability in America in terms of prevalence trends and the relative
participation of persons with disabilities in the work force will be discussed. For
the interested reader, a review of alternative disability models is presented in
Chapter 2. Furthermore, the distinction of disability from perceptions of health or
illness will be described.

Next, information regarding breast and prostate cancers will be presented
since this study is concerned with disability in previously healthy cancer patients.
The description will communicate data regarding prevalence trends, current
treatment options and survival expectations.

Lastly, the study proposal will be brieﬂy described. The general nature of
the study questions that are the focus of this paper will be illustrated and the
overall approach to implementing the study will be explained. Speciﬁc details

regarding the study methodology will be reserved for Chapter 3.

1.1 Theoretical Framework

A single concept of disability is not universally accepted. Competing
theories emphasize physiological, personal, environmental, and societal
characteristics to varying degrees. The most comprehensive deﬁnitions embrace.
a biopsychosocial model that acknowledges the contribution of each facet on an
individual at different points in time.

One consistent trend in the concept of disability is the increasing belief in
the notion that disability does not equate to inability. Individuals may experience
limitations in particular activities, for example, ambulation, but enjoy full

participation with life events, such as maintaining full-time employment.

The lntemational Classification of Functioning, Disability and Health (ICF)
was selected as the overarching conceptual model to frame this study. The
World Health Organization (WHO) published the ICF in 2001 (62). This system
was released as a complete revision to their 1980 document entitled the
lntemational Classiﬁcation of Impairments, Disabilities, and Handicaps (ICIDH).
The ICF acknowledges “mind, body, and environment are not easily separable
but rather mutually constitute each other in complex ways.” (10, 62) Individual
and social characteristics interact between three speciﬁc dimensions to establish
disability.

The three dimensions of the ICF classiﬁcation system include: impairment,
activity limitation and participation restriction. Impairment refers to the loss or
“abnormality” of a body part or function and is conceptualized in a value-neutral
manner. According to ICF, “abnormality” refers only to “a signiﬁcant variation
from established statistical norms, i.e. as a deviation from a population mean
within measured standard norms.” (62) Activity limitation refers to the difﬁculties
a person may have in executing activity. Participation restriction incorporates the
problems or difﬁculties a person experiences with involvement in life situations.
Furthermore, the ICF recognizes these three dimensions exist within “contextual
factors” such as environmental or personal factors (10, 62). Environmental
factors include items such as social systems or services while personal factors
include race, gender, or socioeconomic status characteristics. According to the
ICF then, disability is established in an individual by the interaction of the three

domains: impairments in body functions/structures, activity limitation and

participation restriction. The expression of disability may be inﬂuenced by the
contextual factors. Thus, a key component of this system allows for variation in
disability depending on the factors present at any one time (Figure 1). For
example, a lack of social services may exacerbate disability but if these become

available at a later date, disability may be lessened.

 

Health Condition . ..
(disorder or disease) D'sablllty

 

 

 

 

 

4—> Activities <~——> Participation

It i

 

 

 

 

Body functions
and structures

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Environmental Personal
factors factors

 

 

 

 

 

 

Figure 1: Interactions between the components of the International
Classiﬁcation of Functioning, Disability and Health (adapted)

The ICF model deﬁnes disability as limitations in physical or mental
functions (e.g. activity limitation), caused by one or more medical conditions
(impairment), in carrying out socially deﬁned tasks or roles (participation
restriction) (1, 2, 62). Ultimately, disability may be considered as “the negative
aspect of the interactions between an individual (with a health condition) and that

individuals contextual factors, environmental and personal factors.” (62)

The dimensions included in the ICF will be used to describe work-related
disability reported by breast and prostate cancer patients twelve and eighteen
months after a cancer diagnosis. Work-related disability will be measured by the
presence of new limitations in a variety of work-related tasks in conjunction with a
decrease in work participation level that the respondents attribute to their cancer
diagnosis and treatment (impairment). Therefore, participants in the study will be
considered to have a disability if they admit to having a limitation with any of their
work-related tasks and they have decreased their level of participation in their
work setting from a baseline measurement. In terms of the ICF model, the
combination of these elements as a disability represents negative interactions
between the individual and their job situation. The various components included

in this analysis may be categorized into the ICF model (Figure 2).

 

Breast/Prostate

 

 

 

 

 

 

 

 

 

 

 

Cancer
ll
v
Alteration in Cognitive/Physical . , Work force
function due to v Work Task . participation: hrs
surgery, radiation Limitation worked, shift length
and/or chemo

 

 

 

 

 

 

 

 

 

 

A T

 

 

 

 

 

 

 

Environmental factors (perceived Personal factors
discrimination, accommodation, (race, age, income)
job satisfaction)

 

 

 

 

 

 

Figure 2: Study components nested within the ICF model

1.2 Overview of Disability

Disability has only recently been recognized as a signiﬁcant societal issue
in the United States. The prevalence of persons with disabilities appears to be
increasing. In response to this, the federal government has increased attention

to issues affecting these individuals.

1.2.1 Prevalence Trends

The prevalence of persons with disabilities is increasing substantially
according to Census and the Behavioral Risk Factor Surveillance System
surveys (8,42,96,97). Twenty percent of Americans over ﬁve years of age were

documented to have some level of disability according to the latest Census (96).

This increase comes with larger costs both to the individual and society arguing
for increased research into the disability experience.

According to 2001 Behavioral Risk Factor Surveillance System data, the
unadjusted prevalence of disability among adults in Michigan was 19.6%, (95%
CI 18.2-21.0) (42). The median prevalence for all 50 states was 18.0% with a
range of 10.5% - 25.9% (42). The Detroit Metropolitan Statistical Area
unadjusted prevalence of disability among adults was 20.4% (95% CI: 18.0, 22.8)
and was signiﬁcantly higher than the Midwest median prevalence, 17.4% (42).

Disabilities are disproportionately identiﬁed within disadvantaged groups
although they can occur in a person of any race, age and gender (44). White
and Asian Americans suffer less disability within each age strata compared to
other racial groups. This relationship may be confounded by socio-economic
status. Disability is often correlated with poverty, low education and few social
resources. However, the direction of inﬂuence continues to be debated.
Disability increases with increasing age regardless of race/ethnicity. Per US.
Census estimates, female disability rates exceed male disability rates after 15

years of age.

1.2.2 Factors Driving Disability Prevalence Trends

The prevalence increase is felt to be a result of several factors. The rise
in life expectancy for all persons, including those with disabilities causes the
proportion of disabled individuals in the total population to increase. As the baby-
boomer generation advances in age, the number of disabilities related to chronic

diseases is expected also to rise (8, 97). Finally, formerly fatal conditions such

as cancer are now often medically managed so individuals with these conditions
are maintained in the population.

One key limitation to published prevalence ﬁgures is the assumption that
the target audience can be readily identiﬁed. Complicating this seemingly
straightfonrvard endeavor is the realization that disabilities may be visible or
invisible. Disability may involve physical, cognitive, developmental, or sensory
components. They may occur acutely or develop over longer time frames. They
may be present at birth or acquired over the life span. They may represent
permanent conditions or be a transient circumstance. Moreover, the same
condition may impact different individuals to different degrees (8). Because of
these variations, it is hard to state with certainty what the published disability

ﬁgures actually represent.

1.2.3 Social Role — Work Force Participation

The social role of interest to this study is participation in the labor market.
The vast majority of participants in this study were working at the time of their
cancer diagnosis so the opportunity exists to chronicle how their social role
changed, if at all.

Nationally, persons with disabilities are poorly represented in the labor
market. Of the estimated 49.7 million disabled individuals, 61% are between 21-
64 years of age but just over half are employed (44). This pattern holds for
males and females. In addition to lower work force participation, men with

disabilities have lower median wages compared to men without disabilities,

$1857 vs. $2190, respectively (44). Women’s wages follow this pattern but are
signiﬁcantly lower than men’s wages, regardless of disability status (44).

Participation restriction in working, the usual social role for this age group,
comes at a signiﬁcant cost to society. Lost productivity attributed to disability
accounts for more than $155 billion (1, 8). In addition, government payments to
support unemployed persons with disabilities are estimated at over $230 billion
annually (92). As disability prevalence increases, ongoing failure to promote
integration or re-integration of persons with disability into the working sector will
result in higher societal costs.

Therefore, efforts to keep or return individuals to the labor market would
be anticipated to be beneﬁcial to the general public. However, the challenge to
do so arises from the tremendous variation in the notion of disability. Some
individuals may need permanent accommodations to participate. Others, like
these study participants, may only require assistance for a deﬁned period of time

or with speciﬁc elements of their jobs.

1.2.4 Increasing Government Attention and Action

National interest in persons with disabilities has expanded dramatically in
the past decade along with increasing awareness of the prevalence trends. Key
government reports (e.g. Healthy People 2010, Call to Action, New Freedom
Initiative) have focused attention on this population unanimously charging the
public to identify opportunities to promote the full integration of persons with

disabilities into society.

Legislation supporting persons with disability has been passed in the past
ﬁfteen years. The Americans with Disabilities Act (ADA) and the New Freedom
Initiative (NFI) promote equal access and inclusion for disabled individuals. The
US. Department of Health and Human Services has publicized issues related to
persons with disabilities through several ofﬁces, including the Centers for
Disease Control and Prevention (CDC) and the Ofﬁce of the Surgeon General.
In similar fashion, non-governmental agencies dedicated to supporting
individuals with disabilities have increased in number and scope over the past
decade, i.e. Paws with a Cause (8).

The ADA was signed into law July 26, 1990 with a goal of making
American society more accessible to people with disabilities (11, 13). Five major
title areas are covered under the ADA: employment, public services, public
accommodations, and telecommunications. The ﬁfth part prohibits retaliation
toward or threatening disabled individuals who avail themselves of the Act to
assert their rights or of those that assist disabled persons to assert their rights.

Title I, Employment, states “business must provide reasonable
accommodations to protect the rights of individuals with disabilities in all aspects
of employment.” Discrimination in most job-related activities is prohibited under
the ADA. Discriminatory practices include refusing to hire an individual with a
disability for a job, ﬁring a worker without cause, providing unequal
compensation, working conditions or beneﬁts, or screening out persons with

disability (13). The ADA applies to private employers with at least ﬁfteen

10

 

 

 

it»! ‘

 

 

 

employees as well as state and local governments, unions or employment
agencies.

The NFI is an executive order signed by President George W. Bush in
2001 intended to “eliminate the barriers that prevent people with disabilities from
participating fully in community life.” (8, 86) The overriding emphasis of the order
is to promote community-based alternatives for persons with disabilities and to
ensure that community-based programs promote independence and community
participation. NF I focuses on six areas, one of which is employment. Included
as part of this executive order were funds in excess of ﬁve million dollars to help
small businesses comply with the ADA and encourage them to hire persons with
disabilities (86). An additional 145 million dollars were dedicated to promoting
transportation programs (86). Lastly, funding for the Ofﬁce of Disability
Employment Policy at the Department of Labor was doubled in an effort to
increase the number of persons with disabilities employed in the labor market
(86).

The CDC included a chapter focused on people with disabilities in the
Healthy People 2010 document. Healthy People 2010 is the national plan that
promotes government, private and community health promotion and wellness
activities and policies to improve the health of Americans and eliminate health
disparities (12, 88). Disability status has now been included as a descriptor
variable in this latest edition. Available data suggest disability is a risk factor for

other health-related conditions (12). As such, Healthy People 2010 argues that

11

 

 

“I

persons with disability warrant focus, similar to persons of minority racial groups,
to reduce and eliminate health disparities (8, 88).

Lastly, the US. Surgeon General, Dr. Richard Carmona, MD, MPH,
FACS, published a Call to Action to Improve the Health and Wellness of Persons
with Disabilities in 2005 (8). The document supports the ability of persons with
disabilities to work, learn and engage in society by identifying barriers and
suggesting direction to improve health and wellness. Dr. Carrnona also
emphasized that research on disability must continue (39). A distinctive feature
of the Call to Action was that it was written with the input of persons with
disabilities, not simply from the health care provider perspective.

Another example of the national focus on persons with disabilities and
involvement in social roles is found in the establishment of the Committee on a
National Agenda for the Prevention of Disabilities. This committee has been
convened through the Institute of Medicine and has issued a number of
recommendations including provision of comprehensive vocational rehabilitation
(92).

Although most of the government initiatives portray the impression that all
disability is a permanent condition, models such as the ICF remind us that the
interaction of the individual within their physical and social environments can
change. Therefore, it is important to understand that not all disabilities represent

lasting situations.

12

1.2.5 Characteristics of Transient Disabilities

One important distinction of the persons included in this study is the
presumably transient nature of their disability. Most of these individuals did not
have pre-existing conditions that adversely affected their ability to maintain paid
employment. Individuals who only worked part-time were asked the reason for
this and only 3 (0.6%) respondents in the breast cohort and none of the prostate
cohort participants acknowledged health limitations at the baseline interview
encounter.

While disability statistics tend to emphasize permanent conditions,
transient conditions are increasingly legitimized. The Family Medical Leave Act
(FMLA) of 1993 requires employers who have at least 50 employees to provide
eligible workers up to twelve weeks of leave for their own medical illness among
other qualifying reasons (120). While this leave need not be paid, if the employer
offers medical coverage, they are obligated to continue this during the leave
(120).

Thus, the F MLA legislation can be considered to be disability legislation.
Individuals who are sick or injured are known to experience restrictions in
maintaining participation with working which is one of their social roles. The
beneﬁt of this Act is that it permits people to step away from this role for a time to
focus their energies on recuperation in the hopes that they are able to return to

work.

13

1 .2.6 Disability versus Illness

The emphasis of this study on disability in the presence of a cancer
diagnosis may inadvertently implicate a one-to-one relationship between
disability and disease. However, disability must be differentiated from the
concept of illness. Per the biopsychosocial model, disability does not equate to
illness. The cancer diagnosis each participant in this study has received does
not automatically confer disability. Recall that disability is the interaction of
impairment, activity limitation and participation restriction. Illness often refers to
some abnormal physiologic function. Illness, then, can be considered to be just
one piece of the puzzle that contributes to disability.

However, it is important to also discuss disability in terms of health.
Disability is not simply the opposite of health. Rather, the concept of health
encompasses a much broader interpretation of contributing factors. The World
Health Organization deﬁnes health as a “state of complete physical, mental and
social well-being and not merely the absence of disease, or inﬁrmity.” (62) Krahn
(2003) has deﬁned health as “achieving and sustaining an optimal level of
wellness, both physical and mental, that promotes a fullness of life.” (94)

Deﬁnitions such as these point to the importance of the respondent’s
Perspective on the concept of health. Persons with disability can seek their own
levels of wellness optimal to their personal situation. It is entirely possible that
Persons with a disability enjoy well-being in their physical, mental and social
GHdeavors and believe themselves to enjoy good health. Therefore, a participant

in this study may be labeled as “disabled” based on their responses to questions

14

about activity limitation or evaluation of work participation restriction. However,
this label should not be interpreted as comparable to how the participants” would

evaluate their own health status.
1.3 Overview of Breast and Prostate Cancers

1 .3.1 Prevalence Trends

Cancer prevalence estimates approach 4% of the population, but the
condition receives substantial investment of health care dollars in terms of
screening and treatment of disease. The National Institutes of Health reported
nearly $74 billion were spent on direct medical costs in 2005 while the overall
costs for care in 2005 were estimated at nearly $210 billion (57). The price of
lost productivity due to cancer illness is not well deﬁned. Some estimates
approximate the cost at $17.5 billion (57).

An estimated 213,000 new cases of female breast cancer were expected
for 2006 with nearly 41,000 deaths. New prostate cancer cases were estimated
to be nearly 235,000 with over 27,000 deaths (57). New Michigan cases for
female breast and prostate cancer were approximated at 7070 and 7370,
respectively, in 2006 (57). The incidence rate of female breast cancer in
Michigan between the years of 1998-2002 was 132.4 cases per 100,000 while
the prostate incidence rate for the same timeframe was 200.6 cases per 100,000.
Over half of incident breast cancer cases are expected in women between 20

and 64 years of age while nearly 40% of prostate cancer cases are expected in

15

men of the same age range (111). Thus, the potential impact of disability related

to a cancer diagnosis on persons in the work force is signiﬁcant.

1.3.2 Factors Driving Prevalence Trends

Cancer is one medical condition that has enjoyed signiﬁcant improvement
in treatment efﬁcacy. Screening practices also enable early detection. The
combination of early detection with better treatments offers more promising
prognoses. These circumstances have resulted in increasing numbers of breast
and prostate cancer survivors.

Breast cancer is largely asymptomatic with most cases being diagnosed
during screening mammography. Women aged 40 and up are recommended to
undergo screening mammography on a regular basis. Under certain
circumstances, as with a positive family history of cancer, screening is initiated at
even younger ages. The end result of early screening is that diagnosis will occur
during a woman’s working years.

Prostate cancer is the most frequently diagnosed cancer in men (57).
Prostate cancer screening is initiated in most men at age 50 unless personal or
family health histories call for earlier screening. Screening involves a blood test
for prostate-speciﬁc antigen along with digital rectal examination. Most men are
employed at age 50 so the potential for impact on work force participation is
great.

Available treatments for these cancers are aggressive even when
detected at early stages. The breast cancer survival rate (all stages combined)

at ﬁve and ten years is 88% and 80%, respectively (57). However, the ﬁve and

16

ten year survival rate for prostate cancer is 99% and 92%, respectively (105).
The number of cancer survivors in the United States is estimated to exceed ten
million persons with fourteen percent diagnosed more than two decades earlier

(63).

1.3.3 Treatment Options

Treatment of breast cancer includes surgical options as well as adjuvant
therapy. Surgery may include lumpectomy, local removal of the tumor, or
mastectomy, removal of the affected breast along with some lymph nodes.
Adjuvant therapy may include chemotherapy, radiation therapy or hormone
therapy (57). Treatment is generally concluded within six months of diagnosis
however some hormone therapy may be life long.

Treatments for prostate cancer vary and may include surgery and/or
adjuvant therapy as well. Surgery may include transurethral resection of the
prostate or prostatectomy. ln rarer cases, orchiectomy is another option.
Additional therapies include brachytherapy (implantable radiation seeds),
hormone therapy, chemotherapy and traditional radiation. In contrast to some

breast cancer treatment, men do not maintain therapy on a lifelong basis.

1.4 Rationale for Associating Disability with Cancer

Cancer is now being medically managed and is considered by some a
chronic condition. However, the aggressive treatments that help confer the
added survival may be accompanied by untoward side effects. As a result,

persons surviving their immediate cancer experience may ﬁnd themselves

17

experiencing restrictions in their usual activities subsequent to diagnosis and
treatment. Unfortunately, these limitations may be present for long periods of
time or even permanently. Since these persons may be of working age, it is
reasonable to suspect these limitations may affect their work force participation.

Increased survival has ushered in a new research focus termed
survivorship. The National Cancer Institute has identiﬁed interest in “studying
adverse Iong-tenn or late effects of cancer and its treatment.” (63) Potential
adverse effects may include work task activity limitation resulting in labor force
participation restriction. An exploration of work-related disability and how these
vary over time based on race, socio-economic status, age, stage of disease and
cancer treatment may lend insight into cancer’s disease burden for working-age
individuals.

Similarly, the US. National Occupational Research Agenda is considering
research on cancer survivorship. musculoskeletal disorders and work disability
as primary subjects for research (121). Feuerstein and Harrington (2006) argue
“Research is needed to identify the barriers that affect cancer survivors at work
and evaluate innovative interventions to assist these workers in both their return
to work and long term success at work" (121). They go on to describe the
following belief that is contributing to the public health burden of cancer with

respect to the American workplace.

18

“Our culture continues to perpetuate the view that an individual with
cancer is somehow now defective. While at this point limitations in
function often represent the sequelae of cancer and its treatment, the
question we need to be asking is not can he or she do the work, but rather
can the cancer survivor perform the essential tasks of his or her job and, if
not, can he or she be reasonably accommodated to minimize the impact
of the illness on work productivity? Yet employers and supervisors
continue to perceive cancer survivors as poor risks for advancement (124)
and cancer survivors are at high risk for job loss. These outcomes can
regrettably lead to a cascade of problems for the survivor, the workplace

and society (121).”

Feuerstein’s suggestion to identify speciﬁc tasks and then look for
reasonable accommodations when necessary is a rationale approach. However,
even greater worker productivity may result from speciﬁc rehabilitation in
conjunction with accommodation. Rehabilitation has been shown to produce
improvements in other functional aspects of life, such as ambulating or self-care.
Once speciﬁc work tasks have been described as being sensitive to cancer and
associated therapies, rehabilitation professionals could then identify practices to
minimize the complications and support an individual’s return to optimal

functioning.

1.4.1 Rehabilitation Potential to Ameliorate Adverse Effects

As mentioned, cancer rehabilitation is not a novel concept and has been
recommended to address competence with daily tasks. Some of the earlier work
comes from Lehmann and colleagues in 1978 focusing on dressing, feeding,
personal hygiene, ambulating and transferring (114). The improvements made
by patients were so remarkable that referrals for rehabilitation more than doubled
and the number of visits for rehabilitative services more than quadrupled.

O’Toole and Golden (1991) described an increase in independent ambulation

19

from 14% to 80% at discharge from a rehabilitation hospital while more than half
maintained this improvement for 90 days post-discharge (115). Similarly, bladder
continence improved from 38% to 87% (115). Yoshioka (1994) reported similar
encouraging results from their six-year prospective study of terminal cancer
patients in Japan with signiﬁcant improvements in transfer and locomotion scores
(116). These researchers further documented family satisfaction with the
rehabilitation efforts with over 75% reporting satisfaction (26,116).

As cancer becomes a chronic condition, expansion of rehabilitation to
additional environments (i.e. the work setting) is a reasonable concept yet
requires focused study. Yadav (2007) reports occupational therapists at MD.
Anderson Cancer Center facilitate return to work by simulating work activities for
patients (122).

The nature of work activities that seem to be most at-risk needs to be
deﬁned. However, it is unclear if the characteristics of the work activities are the
sole contributor to disability. Employer related factors likely contribute to
disability. Ganz (1990) reported vocational issues were often more serious for
individuals in blue-collar occupations compared to white-collar occupations (80).
This may reﬂect the differences in the physical demands usually associated with
these types of jobs. Or, this may be due to the fact that white-collar workers
generally have the beneﬁt of higher levels of education or employment status
with greater ﬂexibility in work assignments and scheduling.

Because health insurance is linked with employment in the US, the ability

of those with a cancer diagnosis to remain employed is often a ﬁscal necessity.

20

Since improvements in functional status subsequent to rehabilitative services
have been documented, it is reasonable to expect that continued presence in the

workforce is a viable option with vocational rehabilitation.

1.5 Study Objective

The objective of this study is to establish and describe how prevalence
estimates of physical and cognitive work-related disabilities vary between twelve
and eighteen months in an employed cohort of breast and prostate cancer
patients. Furthermore, for individuals participating at both timeframes, the
change in physical and cognitive disabilities between twelve and eighteen
months will be described. Logistic regression will be used to identify variables
that may inﬂuence disability. Disability will be deemed present when both activity
limitation and participation restriction are observed.

A brief narrative of the key facets of this study is presented below. A more

detailed description including speciﬁc hypotheses is provided in Chapter 3.

1.5.1 Study Proposal

This study will take advantage of an existing dataset from the Labor
Market Outcomes Study (LMOS), Pl: Cathy J. Bradley, PhD, NCl R01 CA86045-
01A1. This study enrolled an inception cohort of breast and prostate cancer
patients from the Metropolitan Detroit Cancer Surveillance System (MDCSS), a
participant in the National Cancer lnstitute’s Surveillance, Epidemiology and End-

Results (SEER) registry and followed them for eighteen months after diagnosis.

21

Participants were interviewed approximately six, twelve and eighteen months
after diagnosis.

Questions regarding work tasks and new limitations attributed to the
cancer diagnosis and treatment were asked of participants during the twelve and
eighteen-month interview encounters. Using these responses in conjunction with
documented participation restrictions represented by decreases in weekly hours
worked, shift lengths or withdrawal from the work force attributed to the disease
and treatment, it was possible to identify participants who may be said to have a
disability according to the WHO deﬁnition.

The prevalence of work-related disability was calculated from those who
reported activity limitation with participation restriction. Overall prevalence
estimates represented the ﬁrst cut to describe work-related disability in this
population. Once these were calculated for the twelve and eighteen-month
timeframes, the change in disability from twelve to eighteen months for
participants who had data available at both timeframes was described. Finally,
the focus turned to determining which contextual factors (e.g. environmental or
personal characteristics) inﬂuenced work-related disability through the regression

modeling process.

1.5.2 Study Rationale

Increasing attention to disability coupled with the increased detection of
cancer in a working population provide motivation for this study. Speciﬁcally,

literature discussing the impact of cancer on an individual’s ability to continue in

22

the workforce is limited. Recent articles published on this cohort document
changes in hours worked and wages during treatment and recovery phases.

The next step is to document what factors, if any, contribute to the labor
market patterns that have been observed. For instance, do physical or cognitive
tasks prove to be more difﬁcult to sustain during the treatment and recovery
period? Does treatment type affect the duration of disability? Do individuals with
higher socio-economic status experience less disability than persons with lower
socio-economic status?

Furthermore, disability is a term that is not well deﬁned. Studies have
been published that describe various disability rates. However, disability as a
characteristic does not necessarily affect all of life’s activities equally.
Additionally, the facets that are incorporated into a deﬁnition of disability can
vary. This study will provide preliminary information regarding disability in work-
related tasks using a disability deﬁnition that is explicitly deﬁned.

Understanding the relationships between speciﬁc factors and disability
could set the stage for conﬁrmatory testing through new study proposals
designed speciﬁcally to address these questions. Ultimately, the recognition of
factors that could potentially ameliorate or exacerbate disability could be
considered for rehabilitation theory, practices and policies.

This study has the necessary information to address the goals.
Participants were followed longitudinally with 90% retention of the original cohort

at the eighteen-month interview encounter. Questions on work-related tasks

23

were asked of all participants at twelve and eighteen months regardless of their

employment status at that time.

1.6 Summary

This investigation will contribute to the existing research on labor market
participation of cancer survivors by describing work related disabilities as deﬁned
by the ICF. Disability is reﬂected by activity limitation in conjunction with
participation restriction per this model. Both disability and cancer are signiﬁcant
issues for society and these conditions are expected to exert greater impact as
the prevalence of each increase. The enhanced survival rates for breast and
prostate cancer argue for more emphasis on the long-term effects of the disease
and treatment.

This study supports national cancer and occupational research programs.
The intersection of work-related disability in a cancer survivor cohort falls under
the cancer survivorship research agenda proposed by the National Cancer
Institute and the plan of the US. National Occupational Research Agenda.
Moreover, the participants in the proposed cohort have been followed over time
so that changes in reported disabilities will be evaluated.

This study addresses gaps in the existing literature by focusing on a
younger, employed cohort, documenting variation in work-related disability over
time, and providing opportunities to assess the impact of a variety of personal

and environmental contextual factors on work-related disability.

24

CHAPTER 2: LITERATURE REVIEW

This review will summarize the literature that has been published
regarding the selected conceptual model. Additionally, publications regarding
disabilities associated with cancer and employment characteristics of persons
with cancer will be studied. A review of disability model evolution is included as
well. The advantages of the ICF have been described in Section 1.1 as support

for selection of the ICF model as the theoretical model.

2.1 Critiques of ICF as the Selected Theoretical Framework

The ICF system is a comprehensive framework that is useful for
identifying key components that may contribute to disability. As mentioned in
Section 1.1, the model embraces the biopsychosocial aspects and attempts to
represent the varying nature of disability. ICF represents an advantage over the
medical model that predominated decades earlier. The medical model tended to
focus on physiologic derangement(s) of an individual and the activities they were
incapable of performing. The medical model charged an individual to
accommodate their behaviors and actions to the norms of society and their
surroundings. However, a key component to the ICF includes awareness of the

environmental and societal inﬂuences on conferring disability.

2.1.1 Activity versus Participation Definition

As mentioned in Chapter 1, the ICF evolved from the ICIDH. Yet despite

the improvements, the ICF is not without detractors. Some critics express

25

concern regarding the vague concepts of activity or participation (7,10,45,104).
During the lCF’s development, consensus regarding how activities are
distinguished between these concepts was not achieved despite the inclusion of
deﬁnitions of these terms. Generally, activity was thought to refer to an
individual’s ability while participation represented a societal role (104). The ICF
defined activity as “the execution of a task or action by an individual” while
participation was deﬁned as “involvement in a life situation.”

Regardless of providing different deﬁnitions for these concepts, the
document unfortunately catalogued the same basic areas under both terms.
Nine speciﬁc elements were linked to each: learning and applying knowledge,
general tasks and demands, communication, mobility, self-care, domestic life,
interpersonal interactions and relationships, major life areas and community,
social and civic life. By relying on this single list, the ICF failed to present
evidence for how these concepts vary which unfortunately blurred the notions.
Breaking down mobility into “ability to ambulate without assistance” and “ability to
move about a location such as a store” as examples of activity limitation and
participation restriction, respectively, could potentially offer more clarity into the
distinction between these concepts. Admittedly, the work involved to ensure
every contingency is covered would be burdensome.

Some suggest that the underlying concepts of the terms activity and
participation more accurately represent a measure of action in terms of “capacity”
and “opportunity” and recommend this terminology as a better reﬂection of the

concepts incorporated into the model. The critical distinction becomes whether

26

someone has the capacity to carry out some speciﬁc task. Then, presuming they
do have the capacity, the other element asks if they are afforded the opportunity
to accomplish these tasks.

Irrespective of the terminology debate, some critics argue that a method to
measure the individual’s desire for action is missing from the model. The
possible combinations of the terms activity, participation, capacity or opportunity
still fail to account for a person’s intent or “will” to carry out an activity (45).
Nordenfelt suggests that if an individual has no intention of carrying out some
action, they cannot be said to be disabled with respect to that action (45).

Nordenfelt’s argument is understandable yet operationally cumbersome.
Many of life’s activities are not as uniquely speciﬁc as this belief requires. Similar
cognitive or motor skills can be used for a multitude of tasks and simply because
an individual is not interested in performing some activity does not mean they
would not be hampered in their attempts to perform tasks that require similar
skills. For example, arthritis in the ﬁngers may affect a person’s ability to play the
piano. Nordenfelt argues that if they do not want to play the piano, they should
not be considered disabled in this action. However, the motor skills necessary to
play piano also impact the ability to write, type, deal with buttons on clothing, hold
a fork and many other activities. Therefore, while the particular activity of piano
playing is mute, the impacts are much further reaching. The attempts to
discriminate between every possible task would be onerous at best.

It is regrettable that the ICF authors did not continue the effort to better

differentiate these terms. This may be addressed in future evolutions of the

27

model. This study will deﬁne activity limitation as an individual’s capacity while
the notion of participation restriction will reﬂect involvement in social role.
Speciﬁcally, activity limitation will reference each particular cognitive or physical
task of interest (e.g. analyzing data or lifting heavy objects) performed by an
individual while participation restriction will represent a decrease in paid
employment status (e.g. fewer hours worked, shorter shifts or withdrawal from

the workforce) providing a measure of the individual’s social role.

2.1.2 Omission of Quality of Life Component

Some authors feel the most egregious omission in the ICF is the lack of a
quality of life category (104). Even at this time, there is no mechanism to include
a subjective assessment from the individual with a disability. WHO supports a
broad interpretation of quality of life deﬁned as “the perception of individuals of
their position in life, in the context of the culture and value systems in which they
live and in relation to their goals, expectations, standards, and concerns”(110).
An optimum quality of life is not dependent on the fact that an individual would be
without any medical conditions (62). A mechanism to incorporate an individual’s
assessment of his or her own quality of life would provide a more complete

measurement of disability.

2.1.3 Inﬂuence of Contextual Factors

More work is needed regarding a uniform understanding of the contextual
factors, eg. environmental and personal factors, which may inﬂuence the

domains of the ICF. Research regarding the role of environmental barriers and

28

participation is relatively sparse and indicates that the relationship is not as
unidirectional as believed. Whiteneck (2006) reviewed several studies and
concluded both negative and positive correlations were noted between
environmental barriers and participation suggesting multiple factors at work
(104). Regarding personal factors, current studies rarely go beyond
demographic measures (104). The reluctance to venture into areas of
personality, psychology, motivation or compliance may be related to the fear of
returning to a medical model of disability where problems are blamed on the

individual rather than society and/or environmental surroundings (104).

2.1.4 Model Testing and Evaluation

The ICF model as it exists has, as yet, not undergone rigorous testing
(10,45,104). It has been available forjust a few years and has not been
accompanied by tools to help standardize its use. Once tools are developed,
they could be validated and then applied to different populations to determine
generalizability. In order to be useful beyond just framing a classiﬁcation system,
the model should also provide information for designing interventions and policy
to increase the involvement of persons with disabilities in life activities.

The breadth of consideration of external and social factors as contributors
to disability is appreciated despite remaining concerns about the utility of the
system for classiﬁcation and enumeration purposes. The addition of domains
such as quality of life or intent/desire for activity would certainly enhance the
model and should be considered in future revisions. No doubt, the context of

disability will continue to evolve as additional research is conducted. However, at

29

this time, the ICF model offers a solid framework with which to investigate work-
related disability associated with cancer treatment and recovery (4,11,12). It
provides a comprehensive structure by acknowledging the biologic and social
(including environment) aspects surrounding the individual (10,62). Furthermore,
personal factors as Verbrugge and colleagues recommend are included.
Therefore, the lCF’s adaptation of a biopsychosocial model and its’ inclusion of
contextual factors will be used to structure the activities of this investigation.

Despite the concerns raised regarding the ICF conceptual model, the
beneﬁt comes in the form of a method to standardize the notion of disability.
Currently, authors use the term “disability” without explicitly stating which
domains, activity limitation, participation restriction, both or neither, they include
in their assessment. Disability rates reported in literature vary widely yet it
remains unclear if the variation reﬂects a real difference in how individuals are
affected by some pathology or rather simply a difference in what the researchers
considered “disabled.”

Since this study is interested in physical and cognitive work-related
disabilities according to an ICF deﬁnition that requires activity limitation with
participation restriction, the literature will be arranged as follows. First, the
common untoward effects of cancer and its treatment that affect and limit
physical or cognitive activities will be described. Then, the participation of

persons with cancer in the labor market will be reported.

30

2.2 Domain 1: Activity Limitations Associated with Cancer

This section will review common side effects associated with cancer
treatment. The goal of this section is to give an overview of the nature of
adverse events patients experience due to treatment that may impact an ability to
work.

Cancer is among the top ten conditions causing activity limitation despite
its’ relatively small prevalence compared to other chronic conditions (40).
Cancer related limitations currently occur in greater number than limitations
resulting from visual and hearing impairments or cerebrovascular disease such
as stroke (2,40). Even with early-stage disease, current treatment protocols are
more aggressive thus expanding potential for limited activity. As a result of the
increased ﬁve and ten year survival rates for most cancers, over ten million
Americans with a cancer history are in the population (57). Therefore, it is
reasonable to anticipate the number of cancer-associated activity limitations will
increase as the overall number of persons diagnosed and treated increases.
However, the presence of activity limitations in conjunction with social role

participation is not as fully documented.

2.2.1 Physical Activity Limitations

Literature describing general physical functioning is available and has
documented variables such as age, comorbidity, treatment and symptoms to be
associated with decreasing function secondary to a cancer diagnosis (125).
Furthermore, recovery of function has been shown to be inﬂuenced by cancer

site and treatments as well (126, 127). What has not been as carefully

31

 

 

SCHH‘

«ﬂat

lenns
asses
I
rune
l
lypma
physk

mese*

féalure

Heath;
chSe,
mmsJ
oilﬁe(I
Iﬂostatg
We.

12.1.1
TIEatm‘

IS
Unfontlr
have?

0bIIIIOUS

 

(Swelling

scrutinized is the afﬁliation of general physical functioning with respect to job-
related work tasks.

The literature on general physical functioning is not always comparable in
terms of study population or speciﬁc environment in which tasks are being
assessed (125, 126, 127). The inclusion of participation with work outcomes is
more limited since existing literature frequently describes older cohorts that have
typically left the work force (17, 23, 24, 125, 126, 127). While broad categories of
physical tasks may be replicated among studies, for example, lifting heavy loads,
the setting (i.e. the workplace versus the home environment) may include other
features that determine whether or not this presents as an activity that is limited.

Speciﬁc impairments attributable to a cancer diagnosis and associated
treatment vary. Research regarding the nature of “cancer disabilities” has often
focused on physiological limitations (i.e. activity limitation) and measured this in
terms of impact on activities of daily living (i.e. participation restriction) or quality
of life (17.31.32). We now focus on the speciﬁc treatments used for breast or
prostate cancer and the untoward effects these may have on other aspects of
life.

2.2.1.1 Physical Activity Limitations Associated with Cancer Diagnosis and
Treatments

Surgery is one treatment option for prostate and breast cancers.
Unfortunately, patients often complain of a variety of symptoms after undergoing
this treatment. Since the surgical sites are different, the side effects are
obviously different. Women with breast cancer often complain of Iymphedema

(swelling). heaviness, tightness, numbness and stiffness with the arm on the side

32

of surgical intervention (17,31). These symptoms are often more prevalent in the
year following treatment but a signiﬁcant proportion, 40%, continue to describe
these effects ﬁve years later (17). Men however, report sexual and urinary
dysfunction after undergoing surgery for prostate cancer. Urinary incontinence is
one of the most common complaints and certainly may interfere with ability to
work (32).

Both cancer diagnoses may also be treated with adjuvant therapy
including radiation and chemotherapy. These therapies are not without their own
untoward effects including anorexia, fatigue, and bowel dysfunction including
diarrhea, urgency or bleeding (35). All these conditions have the capacity to
adversely affect an employee’s presence at or performance of their job.
Additionally, many women and men report pain after undergoing cancer
treatments.
2.2.1.2 Physical Activity Limitations Reported in Literature - Breast Cancer

As mentioned earlier, surgery as a form of breast cancer therapy is not
without side effects. Axillary node dissection accompanying surgical therapy
may result in chronic chest walllaxilla pain or Iymphedema that can be “disabling”
(15,57). Kitamura and associates (2005) projected the number of breast cancer
patients in Japan that would suffer disability from surgery in their article (15). Of
note is the fact that the age-standardized incidence of breast cancer in Japan is
less than Europe or North America, 24.3/100,000 in Japan vs. 103.7/100,000 in
non-Hispanic white women in Los Angeles (15). Even among this population at

lower risk, the estimates of women with disability due to chest wall pain between

33

2000-2020 range from over 42,000 to 72,500. Similarly, the number of women
expected to report disability due to Iymphedema is not trivial, 22,400 in 2000 and
38,600 in 2020 (15).

Limitations of this analysis include the inability to account for
improvements in cancer treatment over time or institution of new methods to
minimize these adverse effects. The authors estimated cases of breast cancer,
proportion of surgical operations, frequency of associated disabilities and crude
survival rates to build their models (15). Furthermore, duration of disability was
not incorporated into their model. Also, while this article refers to “disability", the
ability to quantify how many of these women would end up with limited activity
and in turn, a reduced ability to participate with their job is limited.

Upper body tasks have been documented to be especially problematic for
women following breast cancer treatment (17). Hayes and associates (2005)
asked a sample of women about the frequency and physical demand of 48 daily
tasks requiring upper body function. These authors used factor analysis to
conﬁrm the general nature of activities that were difﬁcult. Whole body, ﬂexibility,
carrying/upper-body strength, hand and weighted ﬂexion tasks were reported to
be most burdensome (17). Lymphedema and poor ﬁtness were associated with
reports of difﬁculty.

As with many of these studies, the authors relied on a cross-sectional
design and included a convenience sample of respondents to a mailed survey
that achieved a response rate of 63%. Of these responses, less than 50% were

complete. Time since diagnosis ranged from 0-45 years. The tasks that were

34

 

incluc

 

lnstrt.
work I
data r

COITIO.’

abduc‘
contrlt
GWden
They n
Consec

anticip;l

leducﬂq
Ilmltatic

phl’Sica

Dan,'T,

factOrs.

SOCial SE
llCle the

Mdm0na

 

included in the survey were predominantly activities of daily living (ADLs) or
instrumental activities of daily living (IADLs). Activities directly related to possible
work environment demands were omitted. Finally, these authors did not collect
data regarding potential confounders such as education, income, or
comorbidities.

As indicated by the prior literature, Iymphedema and restricted shoulder
abduction subsequent to surgery are acknowledged to be the primary
contributors to breast cancer disability reports (28). Voogd, et al (2003) provide
evidence that these clinical conditions are not necessary to confer disability.
They measured arm circumference and shoulder abduction in a cohort of
consecutive women undergoing axillary lymph node dissection (24). As
anticipated, patients with a difference in arm circumference of >2 cm or a 20%
reduction in shoulder abduction reported severe physical complaints and
limitations in tending to daily chores. However, more than 15% without these
physical signs also reported adverse conditions as well as moderate to severe
pain. This research provides rationale for an inclusive approach to investigating
the untoward effects of treatment even in the absence of clinically apparent risk
factors.

McWayne and Heiney (2005) conducted a review of the psychological and
social sequelae of Iymphedema (29). The authors reported pain was a
signiﬁcant predictor of morbidity resulting in compromised quality of life. They
note the lack of research focusing on disadvantaged populations and call for

additional efforts to describe baseline data. Reitman and colleagues (2004) also

35

focused on shoulder impairments and quality of life and conﬁrmed pain was
associated with disability and to a lesser degree, poorer quality of life (28).

Ganz and associates (2002) conducted one of the few longitudinal studies
on a breast cancer cohort (30). They took advantage of a cohort assembled for
an earlier study by inviting women who had survived at least ﬁve years without
recurrence to participate in a mailed survey. A response rate of 61% was
achieved with responders being predominantly white, better educated and with
higher income. Eighty percent of these respondents reported no change in
employment status although younger (<60 years at diagnosis) patients reported
negative impact on work life or career and ﬁnancial situation (30). Current
physical and emotional well-being and age exerted the greatest impact (30).
These authors attributed changes in physical function noted among some
respondents to normal age-related changes. Limitations of this study included a
modest response rate and the restriction to two urban areas. Furthermore, with
respect to work-related disabilities, the authors did not speciﬁcally focus on tasks
required by an individual’s job.

2.2.1.3 Physical Activity Limitations Reported in Literature - Prostate
Cancer

Literature focusing on the experience of prostate cancer patients is limited
compared to reports regarding breast cancer. Prostate cancer treatment may
result in urinary dysfunction as a result of surgical intervention. Additionally,
radiation therapy has been associated with bowel dysfunction. Although these
sequelae may reduce quality of life and potentially interfere with work activities,

Giesler and colleagues (2005) report these untoward effects may be amenable to

36

intervention (21). Patients enrolled in the intervention arm of a randomized
clinical trial reported moderate improvements in urinary bother over control
patients. If urinary bother had presented as a disability for participants, the
evidence that improvements were attainable offer hope that the disability could
be reduced. These authors did report some limitations in the power of their study
as well as a relatively low participation rate.

Watchful waiting is a viable option for older men diagnosed with prostate
cancer. The side effects of treatment associated with surgery and radiation may
be too much burden for a man who experiences a localized cancer that is
asymptomatic. Steineck and colleagues (2002) compared quality of life
measures in men randomized to surgery and watchful waiting (34). These
researchers documented statistically more frequent urinary leakage in the
surgical group however less frequent urinary obstruction with weak voiding
stream. Both conditions may inﬂuence ability to carry out work-related tasks and
suggest that “no” treatment may contribute to disability as well.

2.2.1.4 Physical Activity Limitations Reported in Literature - Cancer Site
Not Speciﬁed

Increasing emphasis on cancer related disability accompanies the
transition of cancer to a chronic condition. Ness and colleagues (2006) used the
1999-2002 National Health and Nutrition Examination Survey (NHANES) to
evaluate physical performance limitations in recent and long-term cancer
survivors (1). They also compared the experience of these survey participants

with those with no history of cancer (based on self-report). These investigators

37

deﬁned a recent cancer survivor as an individual with less than ﬁve years since
diagnosis (1 ).

Physical performance limitations were statistically signiﬁcantly more
prevalent in the respondents with cancer compared to those without cancer
regardless of duration since diagnosis, greater than 50% versus 21%,
respectively (1). The nature of the limitations most frequently cited included
sustained muscle activity (walking 1/4 mile or standing two hours) or large whole
body movements (e.g. stooping, crouching, lifting, etc). No work speciﬁc
activities were included in the range of examples for respondents.

Individuals with a recent diagnosis experienced greater activity limitation
than those with distant diagnosis. These analyses reported inverse relationships
between income and activity limitation while a positive correlation was noted
between age and this outcome. These patterns support the relationships
between advancing age and greater disability reported in general disability
literature as well as the association of persons with fewer social resources or
lower socio-economic status experiencing greater disability rates than persons
with more resources or higher socio—economic status.

Several limitations of this study were the cross-sectional design and the
inability to control analyses for cancer treatment. The population-based nature
however was important to provide preliminary estimates of the proportion of
persons with a positive cancer history experiencing activity limitation. Ness
(2006) however, did not present disability according to the deﬁnition suggested

by the ICF framework. Although activity limitation and participation restriction

38

were reported, an assessment of the proportion of individuals who had both

conditions simultaneously is absent from the analysis (1).

2.2.2 Cognitive Activity Limitations

Several researchers emphasize the importance of understanding cognitive
functioning because of the increasing numbers of persons diagnosed with cancer
and the potential impact of cognitive dysfunction on education, career and quality
of life. Cognitive dysfunction is garnering attention as patients complain of
“chemo brain.” This phenomenon has been described by patients as an inability
to maintain attention or remember things (112).

2.2.2.1 Cognitive Activity Limitations Reported in Literature — Breast
Cancer

Silverman and colleagues (2006) demonstrated differences in brain region
activity during short term recall tests between patients who had undergone
chemotherapy 5-10 years prior and those who did not (112). The chemotherapy
group scored 13% lower on average than the non-chemotherapy group during
short term memory tasks although results were not statistically signiﬁcant.
However, the association of chemotherapy and lower recall test score followed
the trends identiﬁed in a larger cohort studied by these same investigators in
which the difference in test scores was statistically signiﬁcant (112).

Other studies documented a majority (75%) of breast cancer patients
undergoing systemic chemotherapy experienced lower scores in memory and
concentration within six months of treatment (38,109). In addition to short term

impacts, some reports document a cumulative effect of chemotherapy lasting for

39

 

 

Wome

report

 

systen
radiatr
magnc
Uﬁderj
COgntl

BMUSK

 

 

DGHErn

22.2.2
Canoe

”0i Wel
SUQQES
maYre(
Ca”Ger,
amOng.

years after the acute phase. Van Dam, et al (1998) demonstrated greater
cognitive impairment in women undergoing high dose chemotherapy compared
to standard-dose chemotherapy. Women receiving the standard dose regimen in
turn had greater impairment than women undergoing local treatment only.
Women included in this study averaged two years post-treatment (113).

Ahles and colleagues (2002) focused on neuropsychologic impact in their
report (38). They compared cognitive dysfunction in women who received
systemic chemotherapy to women who received local therapy (surgery or
radiation) and were more than ﬁve years post diagnosis. The average time since
diagnosis for their cohort was ten years and even for this duration, women
undergoing regimens including chemotherapy scored signiﬁcantly lower on global
cognitive scores than women receiving local therapy. These researchers also
adjusted for potential confounders including age and educational level and these
patterns persisted.

2.2.2.2 Cognitive Activity Limitations Reported in Literature - Prostate
Cancer

The literature reporting cognitive effects among prostate cancer patients is
not well established. The science presented among the breast cancer cohort
suggests the adverse effects occur subsequent to chemotherapy. Since men
may receive chemotherapy to a lesser degree in the treatment of prostate
cancer, perhaps this adverse effect is not as signiﬁcant in this group. Even
among women, cognitive dysfunction has not been completely described yet.

However, thorough comprehension of this experience is necessary so that

40

patients may make informed decisions regarding treatment decisions and

rehabilitation approaches may be designed.

2.3 Domain 2: Participation Restriction Associated with Cancer

Ness and colleagues (2006) also evaluated participation restrictions in
addition to activity limitations among recent and long-term cancer survivors in the
study referenced earlier (1).

Participation restrictions followed similar patterns as activity limitations
with persons having a positive history of cancer experiencing more restrictions
than those without a cancer history, 30% compared to 13% respectively.
Activities requiring sustained participation (e.g. shopping, movies, sporting
events) were again most problematic. No work speciﬁc activities were included
in the range of examples for respondents.

Similar with activity limitation, individuals with a recent diagnosis
experienced greater participation restriction than those with distant diagnosis.
These analyses reported inverse relationships between income and participation
restriction while a positive correlation was noted between age and participation
restriction. These patterns support the trends provided in general disability
literature.

Although both domains deemed necessary by the ICF model were
reported in this study, Ness did not discuss the sample that represented the
intersection of these domains, that is, those with activity limitation and

participation restriction. Furthermore, since the ICF model is relatively recent,

41

the literature specifying activity limitation in conjunction with participation
restriction is not as readily available.

This study attempts to contribute to the scarcity and is focused on activity
limitation with participation restriction. Moreover, the environment of interest is
the labor market. Therefore, literature regarding the employment experience of

cancer patients is presented in the next section.

2.4 Employment Experience of Cancer Patients

A review of the literature focused on the employment experience of cancer
patients follows. Outcomes of included studies did not necessarily identify an
outcome of “disability.” This information is provided to provide a background
against which the disability measure of the present study may be contrasted.

Twenty years ago, employed persons diagnosed with cancer often faced
discrimination. Cancer was a debilitating illness with fatal outcomes. Hoffman
(2005) documented concerns regarding lost proﬁts and productivity by insurers
and employers (13). In response to prejudicial hiring and employment practices,
state and federal legislation such as the ADA, the Federal Rehabilitation Act, the
F MLA and the Employee Retirement and Income Security Act were passed to
protect workers. Increasing advocacy by patients themselves also promoted

additional ﬂexibility in medical care schedules (13).

2.4.1 Return to Work for Cancer Patients

Patients historically and currently report the desire and ability to return to

work after treatment (13). More than thirty years ago, over 75% of employees of

42

Bell Telephone diagnosed with cancer returned to work after diagnosis and
treatment (117). Moreover, a majority of patients remain in the workforce beyond
ﬁve years. Bradley and Bednarek (2002) reported 67% employment 5-7 years
after diagnosis among a cohort of individuals with varying cancer diagnoses (66).
A cohort of young women with breast cancer reported similar employment rates
at diagnosis and ﬁve years after diagnosis per Bloom (13,118).

A variety of factors can impact work force participation including physical,
psychological and social factors (3). These same factors can also inﬂuence the
decision to remain at work or return to work after cancer treatment. Age, cancer
stage, ﬁnancial status, education, health insurance, transportation access,
physical demands of job and comorbid conditions have been identiﬁed to be
signiﬁcantly associated with return to work (13). A variety of psychological
beneﬁts such as empowerment and maintaining a sense of normalcy and control
also contribute to the decision to return to work (13,66).

Health care providers may also inﬂuence decisions regarding return to
work indirectly. They may assist the patient to develop reasonable expectations
of the course and potential side effects of treatment and anticipated speed of
recovery (3). Similarly, the workplace itself plays a substantial role in attempts to
return to work. The response of the employer to an employee’s medical
challenge may support or derail the individual’s motivation. A supportive, pro-
active environment dedicated to assisting the patient to return to work has been
identiﬁed as a beneﬁcial item. Spelten and colleagues (2002) conducted a

literature review to identify factors associated with return to work of cancer

43

survivors (36). They conﬁrmed the negative relationship of a non-supportive
work environment with return to work.

In a prospective study, Spelten, et al (2003) identiﬁed lower fatigue levels
as a predictor of return to work at eighteen months (65). This association was
independent of diagnosis or treatment but not of other cancer symptoms. This
suggests that adequate symptom management may promote workforce

attachment.

2.4.2 Work Limitations Reported by Cancer Patients

Although they express the desire to return to work, some cancer patients
acknowledge a limitation in their ability to continue as evidenced by Yabroff and
associates’ (2004) analysis of the 2000 National Health Interview Survey (60).
Working-age respondents with cancer experienced three times the inability to
work compared to matched controls, 16% vs. 5% respectively (13,60). Sasser,
et al (2005) identiﬁed a cohort of employed women with breast cancer and
compared their direct medical costs and imputed lost productivity costs to a
cohort of employed women without the diagnosis (14). The cancer group missed
signiﬁcantly more days due to medically related absenteeism than the control
group, 42 days and 15 days, respectively. The work-loss costs for the cancer
group was estimated $5900 greater than the control group, a statistically
signiﬁcant amount.

Estimates of limitations at work hover around 20% for most tasks
according to earlier studies. Short, et al (2005) evaluated work-related disability

in a cohort of employed persons with a variety of cancers (19). Their study was

44

mostly descriptive and followed up patients from 1-5 years post—diagnosis.
Participants were asked retrospectively about disability and if reasons for quitting
work were related to cancer. While the overall results documented 20%
disability, neither the breast nor prostate cancer sub-samples had statistically
signiﬁcant disability. The authors reported 21% and 16% of women and men,
respectively, experienced limitations due to cancer (19). Recall bias may be an
issue for participants in this study due to the retrospective nature of the
investigation.

The belief is that cancer exerts a greater impact on physical tasks over
mental tasks (13). Bradley’s (2002) results support this assertion with just 11%
describing limitations in cognitive skills while 18% reported limitations in physical
tasks (66). Spelten, et al (2002) also supported the belief that physical demands
are more compromised than cognitive demands by describing a negative
relationship between return to work and manual labor (36).

Greenwald et al (1989) published on work disability in cancer patients and
used both job attachment and level of function as measures of disability (71).
Patients diagnosed with prostate cancer were included in this study along with
patients diagnosed with lung, pancreatic and cervical cancer. Among this cohort,
physical demands of work predicted work function disability to a signiﬁcant
degree although prostate cancer was not signiﬁcantly related to either
employment or function outcome. Socio-demographic factors were not

statistically signiﬁcant predictors of either outcome.

45

In contrast to Greenwald’s study, Schultz, et al (2002) documented that
soclo-demographic variables, eg. age, gender, ethnicity, as well as cancer type
did impact working status (84). Eligible participants were recruited from a major
cancer treatment center with diagnosis at least ﬁve years prior to recruitment.
Breast cancer patients were included in the overall cohort although the study was
not powered for this cancer type speciﬁcally. This subgroup did experience a

statistically signiﬁcant reduction in employment at follow-up.

2.4.3 Labor Market Outcomes of this Study Population

Bradley and colleagues have published several papers on the cohort that
will be used for the proposed study (49,52,53,82). The overarching topic of these
papers has revolved around workforce attachment issues rather than
investigating the disability associated with the disease and treatment and how it
evolves over the duration of the follow-up. They demonstrated in their 2004 work
that compared to a control group assembled from the Current Population Survey,
women with breast cancer were 17% less likely to work six months following
diagnosis with the exception of women diagnosed with in situ cancer (52).
Moreover, this effect was noted to vary based on race with minority participants
experiencing twice the consequence. Women who remained in the workforce at
six months worked seven fewer hours per week than the control group.

Likewise, a 2005 article described the prostate cancer cohort experience;
a ten percent reduction in employment at six months compared to a similarly
constructed control group (53). Employment was evaluated at twelve months

with the ﬁnding that many men had returned to work at this time. Some

46

descriptive analysis of twelve-month activity limitation was included in this report
for these patients although further analysis of predictors of disability or the nature
of change between twelve and eighteen months was not undertaken (53).
Bouknight, Bradley and Luo (2006) investigated correlates of return to
work for the breast cancer cohort identiﬁed for the proposed study (49). This
analysis validated the importance of the employer on decisions to return to work.
Perceived employer accommodation was positively associated with return to
work at twelve and eighteen months after diagnosis while perceived employer
discrimination was negatively associated with return to work at twelve months
(49). The critical distinction between this analysis and the proposed study is the

outcome variable, return to work versus work-related disability.

2.5 History of Disability Models

A review of the evolution of disability models is presented in this section
for the interested reader. Consensus regarding the deﬁnition and the concept of
disability remains elusive. Disability is used interchangeably with terms such as
handicap, impairment or sickness. Sixty-seven different laws offer varying
deﬁnitions of disability in federal regulations (8,95).

Older theories relate disability to dysfunction of body or organ systems or
some underlying biologic issue. These theories were framed as medical or bio-
medical models. These models generally predominated pre-19603, although
some still persist. The underlying pathology, be it a result of illness or injury,
requires medical care provided by trained health care professionals (8,62).

Disability is viewed as a problem of the person with goals of treatment to cure or

47

minimize subsequent pathology. Take as an example, an individual working as
an accountant sustaining a spinal cord injury resulting in lower limb paralysis.
While cure is not an option, medical treatment would be necessary in order to
minimize complications such as pressure sores or pneumonia. Interest in the
person’s “function” is largely restricted to the physiologic functions of the body,
e.g. maintaining good skin integrity and clear lungs. Minimal efforts are

‘ expended on other activities, such as social functioning, of the individual.

Interest in rehabilitation began to swell during the 19603 and 19703 (61).
Popular advocacy began to promote integration of persons with disabilities into
society and to reduce institutionalization or segregation that had been the
predominant disability management technique. This advocacy coincided with
development of alternative models of disability. Dr. Saad Nagi was one of the
ﬁrst to pay attention to the functional aspect in the disability process (104). He
stated “disability is the expression of a physical or a mental limitation in a social
context.” (128)

Nagi’s disablement model distinguished the environment as an entity apart
from the individual that could be studied and felt that disability represented a “gap
between the person’s capabilities and the demands created by the social and
physical environment.”(61, 92, 128, 129) This led to the expansion of disability
theory into models that focused on “...limitations in physical or mental functions,
caused by one or more medical conditions, in carrying out socially deﬁned tasks
or roles.” (8,92) Nagi supported the idea that an impairment or activity limitation

does not automatically confer disability, including work disability and he focused

48

on three factors that contribute to but are distinct from disability: active pathology,
impairment and functional limitation. Active pathology referred to a disturbance
in the normal cellular mechanisms of a body. Impairment was then deﬁned as an
abnormality at the tissue, organ or body system level. Finally, functional
limitation incorporated restrictions in the performance of the person (129).

A person’s role was acknowledged to be multi-faceted thus integration
could occur on different levels. For persons 18-65 years of age, which is the age
group that roughly corresponds with the study population of this investigation, the
“normal” social role would be as a student or working individual (61). Integration
areas would reasonably be assumed to be higher education and/or a work
setting. According to the social role framework, the treatment of the hypothetical
person with the spinal cord injury would be to minimize physical sequelae of the
condition along with attempts to return that individual to their position as an
accountant. Theoretically, the fact that the person is reliant on a wheelchair for
mobility should not interfere with the normal daily activities of an accountant that
may include writing, key-stroking, reasoning, performing calculations, etc.

Unsuccessful attempts to position persons with disabilities within their
usual social milieu supported Nagi’s thesis that obstacles beyond the person’s
biologic dysfunction contributed to disability. Prior to the contemporary
acceptance of telecommuting, the accountant would be unable to return to
his/her position if he could not physically enter the ofﬁce. There could be
impediments to transporting himself between home and the ofﬁce, the ofﬁce

building or the physical layout of the office might not be wheelchair accessible.

49

M;¢m:m~1
i.

 

Therefore, because of environmental constraints, he would be unable to fulﬁll his
usual social role.

Over time, researchers began to appreciate the existence of environments
beyond the physical structure. This realization led to the further evolution of
current disability theories that began to include social interaction, rather than
social role, as a component of disability (62). Verbrugge and Jette (1994)
expanded on the work of Nagi during the 19905 with the Disablement Process
(50, 129). Their model incorporated elements of pre-existing risk factors (such
as biological factors), and behavior and attributes (50,129). They identiﬁed both
intra-individual (e.g. lifestyle, coping skills, psychosocial attributes, etc) and extra-
individual (rehabilitation services, medications, social support, physical
environment, etc) factors that could increase or decrease the risk of disability
(41,50,104). Accordingly, perceptions of the individual and others around
him/her could accelerate or slow down the path between pathology and disability
(41).

These expanded theories essentially focused on two factors that could
result in a disabled condition, activity and participation (8). Activity, or the actual
performance of some task, is a necessary consideration for disability. Simply
because the legs do not work does not necessarily mean the accountant is
disabled since he can perform his work. His work activity need not be restricted
or limited as a result of his physical condition. The second critical concept is
participation within some environment (8). For example, our accountant is

capable of the task yet unable to perform it because he cannot access the built

50

environment surrounding his ofﬁce. lnaccessibility is merely due to the fact that
building norms rely on multi-story construction. In essence, he is not disabled
but rather, society has placed him at a disability.

While the progression from a solely biological model was largely positive,
some advocates supported the social construct to the exclusion of biology.
These individuals argued that disability was a social phenomenon used to
oppress individuals that were different (10,106). Despite the best intentions of
these proponents, failure to include the biologic components of disability
operated as a disservice to persons with disability. Not recognizing the biology
underlying a physiologic dysfunction eliminates incentives to invest in medical
research or develop new prevention and/or treatment strategies. Therefore,
models failing to acknowledge all contributing facets of life offer nothing more
than incomplete assessments of a disability concept. Deﬁcient appraisals then
result in imprecise recommendations or solutions to problems associated with
disability.

WHO published their original classiﬁcation system entitled the
lntemational Classiﬁcation of Impairments, Disabilities, and Handicaps (ICIDH) in
1980 as one of the ﬁrst attempts to deﬁne and enumerate disability (107).
Although lauded for the effort to provide a systematic method to track disability
worldwide, it was criticized for the heavy reliance on the medical framework
despite the inclusion of social role fulﬁllment.

According to the ICIDH, the focal point for determining disability was the

limitation(s) of an individual (9,10). Impairment was deﬁned as “any loss or

51

abnormality of psychological, physiological, or anatomical structure or function"
(107). Impairment led to disability via a unidirectional pathway. Disability was
deﬁned as “any restriction or lack (resulting from impairment) of ability to perform
an activity in the manner or within the range considered normal for a human
being” (107). Finally, disability could result in handicap. Handicap was deﬁned
as “a disadvantage for a given individual, resulting from impairment or disability,
that limits or prevents the fulﬁllment of a role that is normal (depending on age,
sex and social and cultural factors) for that individual” (107).

Despite mounting awareness of the inﬂuence of environment on conferring
disability, the ﬁrst classiﬁcation system had failed to incorporate this element. As
Figure 3 shows, the root of disability remained with the individual and the
pathologic condition rather than some combination of pathology and
environment. Critics charged that another signiﬁcant omission was the failure to
“acknowledge the presence of social barriers in inﬂuencing disability” (10,108).
Social barriers included a variety of conditions from a lack of public transportation
that was accessible to an individual in a wheelchair to the failure of agencies to
provide TTY/T DD service. A further criticism was that the terminology used was

pejorative in nature, i.e. handicap (Figure 3).

52

~ as

 

._ "‘1‘;

 

Disease

Disorder

 

 

 

—* Impairment —’ Disability —' Handicap

 

 

 

 

 

 

Accident

 

 

 

Abnormality

 

 

 

Figure 3: International Classiﬁcation of Impairments, Disabilities, and Handicaps
Linear Structure of Consequence of Disease

Considering the participants in this study, this earlier version of a
classiﬁcation system is less ﬂexible than the ICF model. If the subjects were
unable to carry out their activities “normally” because of some impairment arising
from their cancer treatment, they would be considered disabled. This system
appears to emphasize activity limitation as the major determinant of disability.
The notion of participation restriction may be more closely aligned with the
“handicap” terminology. However, this model does not allow for someone to
receive help or accommodation with a particular task which may essentially move
them back along the continuum so they are no longer considered disabled. It
also fails to explicitly show the importance of other characteristics such as
personal or environmental factors that can inﬂuence the expression of disability.

In recognition of the importance of a broader context, WHO spent nearly a
decade revising the ICIDH and released the lntemational Classiﬁcation of
Functioning, Disability and Health (ICF) in 2001 (62). Perhaps as an attempt to

distance the new classiﬁcation system from the earlier version, the ICF

53

advocated “the presence of impairment does not necessarily indicate that a
disease is present or that the individual should be regarded as sick” (62).

Details regarding the ICF have been presented in Section 1.1. However,
as has been discussed, the interpretation of disability continues to generate
discussion. Valid criticisms of the ICF have been published and key issues are
currently omitted from the ICF (e.g. quality of life or will/intent to carry out some
activity). Tools to assess various domains of the model are not yet available.
The ﬁeld of disability is still relatively young and tremendous opportunities exist to
contribute to the theoretical foundations. It will be interesting to observe the

nature of ongoing evolution.

2.6 Literature Review Summary

The main research questions included were: the prevalence and direction
of disability change over time, the degree to which prevalence estimates varied
using two common disability deﬁnitions, and evaluating the association of
contextual factors on disability. The contextual factors included personal factors
such as socio—demographlc variables (e.g. race, income) and disease-related
variables (stage, treatment). Environmental factors included employment-related
variables (e.g. physical nature of job, occupational category and job satisfaction).

This study will discriminate between physical disability and cognitive disability.

2.6.1 Prevalence and Direction of Disability Change

The literature presented has discussed short and long-term disabilities

associated with cancer diagnoses. Hayes and colleagues (2005) reported mostly

54

physical disabilities in patients 0-45 years beyond diagnosis. Ness, et al (2006)
documented more physical disabilities in cancer patients compared to a
population of non-cancer individuals both within 5 years of diagnosis (short term)
and beyond 5 years of diagnosis (long term). Moreover, Ness provides one
estimate of the change in disability over time by noting a decrease in reported
physical disability in those with a recent diagnosis compared to a more distant
diagnosis. Silverman, et al (2006), van Dam, et al (1998) and Ahles and
colleagues (2002) also documented cognitive disability in women undergoing
chemotherapy for breast cancer.

A common difﬁculty with all these studies is the lack of deﬁnition provided
for the term disability. The reader cannot be certain what aspect(s) of disability
are being assessed. Activity limitation appears to be a driving force for a majority
yet the authors infrequently explicitly describe their outcome variable. Moreover,
these studies were cross-sectional in nature and were unable to document the
change in disability over the same sample. Finally, several of these studies
failed to collect data on potential confounders such as education or income. This
study however, deﬁnes the outcome of interest and includes collects data on a

number of potential confounders collected at multiple timeframes.

2.6.2 Comparison of Disability Estimates using Two Deﬁnitions

Disabilities associated with cancer diagnoses are fairly widely reported in
various sources yet the prevalence estimates vary. Inconsistencies in estimates
may be valid due to real differences in cancer sites or populations under study.

However, the failure of many authors to clearly describe what characteristics

55

make up their “disability” variable prevents the reader from evaluating to what
degree changes in estimates may be real. This study will report disability
prevalence estimates on the same population using two deﬁnitions. This
comparison will illustrate to what degree a change in deﬁnition may impact

results.

2.6.3 Association of Contextual Variables with Disability

Analysis of BRFSS and census data provide evidence of associations
between personal factors (such as socio-demographic characteristics) and
disability. White and Asian-Americans experience less disability than all other
racial groups. Furthermore, disability is often correlated with poverty and low
education. However, the direction of inﬂuence regarding these variables is not
well-established. Conversely, some authors including Greenwald, et al (1989)
fail to document an association of disability with socio-demographic factors. This
study will contribute by giving some evidence of the evolution of disability among
different groups.

Various disabilities have been documented in persons diagnosed with
breast cancer and Ahles and colleagues (2002) documented greater disability in
women undergoing chemotherapy compared to women having localized
treatments. However, no literature discussing the impact of chemotherapy on
men with prostate cancer was available and therefore any association remains
unknown. This study takes advantage of SEER data as well as patient report to

obtain a comprehensive overview of all therapies received for each individual’s

56

diagnosis. Speciﬁc attention on the relationship between these variables may
provide some preliminary estimates for men.

The second type of contextual factor is environmental. Regarding
disability and work demands, Bradley, et al (2002) and Spelten and colleagues
(2002) report greater limitations with physical tasks and manual labor compared
to cognitive tasks. This study contains data regarding speciﬁc physical or
cognitive tasks and will be able to examine if these patterns hold for these

participants and if/how they evolve over time.

2.6.4 Additional Contributions of Proposed Study

Research regarding the nature of cancer-related disabilities has often
focused on physiological limitations and is measured in terms of impact on
activities of daily living and quality of life. This study will expand the environment
of the patient to include a work setting. By examining factors associated with
employment, this also provides a younger-aged cohort than that usually targeted
for these particular cancer sites.

Available data permits the focus on the time period beyond the acute
treatment phase when adverse effects of therapies may be maximized.
Participants were followed at twelve and eighteen months post-diagnosis. This
study would provide some sense of the longer term adverse effects of breast and
prostate cancer on the ability to maintain pre-diagnosis employment patterns.

Ultimately, this study will contribute to the topical areas currently being
emphasized nationally. The National Cancer Institute has identiﬁed priority

cancer survivorship topics (58,63). Bradley (2005) notes effects of treatment

57

options on work-related measures are absent from most electronically available
cancer information sites (22). Steiner and colleagues (2004) argue “more and
better research on the work impact of cancer is necessary to inform the decisions
of cancer survivors and their treatment providers” (58). Employment matters,
such as potential for disability with job functions, are key to the topic of “chronic

and late effects of cancer and its treatment” (63).

58

CHAPTER 3: RESEARCH METHODOLOGY

3.1 Human Subjects Review

The LMOS that served as the source study for this disability analysis was
reviewed and approved by the Michigan State University and Wayne State
University institutional review boards. Karmanos Cancer Institute also reviewed
and approved the protocol followed for the study. Informed consent was
obtained from all participants. The focus of this paper, the disability study, was

smeitted separately to the Michigan State University institutional review board

and approved.

3.2 Data: Source Study Overview

This study is a secondary analysis of the LMOS. The following sections
describe the characteristics of that source study along with participant

recruitment and retention.

3.2.1 LMOS as Source Study

As mentioned earlier, the LMOS, Pl: Cathy J. Bradley, PhD, NCI R01
CA86045-01A1, will serve as the information source for the proposed study. This
study enrolled an inception cohort of women diagnosed with breast cancer and
men diagnosed with prostate cancer. The goal of the LMOS was to investigate
work force participation including hours worked and hourly wage of participants

and their spouses as they moved through the diagnosis, treatment and recovery

59

process of their condition. Sample size estimations were powered for these
research questions.

The LMOS sample was drawn from the MDCSS, a participant in the
National Cancer lnstitute’s SEER registry. The MDCSS was chosen for several
reasons including the high standards for its case ascertainment and reporting
accuracy as a member of the NCl’s SEER network. The location of the registry
in Metropolitan Detroit also provides data on a greater African-American
population than other registries. Eligible participants included those with newly
diagnosed, incident breast or prostate cancer aged 30 to 64, English-speaking

and employed or had an employed spouse at the time of diagnosis.

3.2.2 LMOS Subject Recruitment and Retention

Potential participants were identiﬁed through the MDCSS Rapid Reporting
system that identiﬁed cases within four months of diagnosis. The four-month
restriction was necessary in order to secure the provider and patient consents by
the targeted six-month interview. Telephone interviews were conducted
approximately six, twelve, and eighteen months post-diagnosis. However, the
six-month interview also included a retrospective component that asked the
participants about job characteristics three months before their diagnosis, i.e. the
baseline interview. Subjects were offered a $25 incentive payment to complete
the study. Nearly 500 (n=495) women and 300 (n=294) men enrolled between
June 2001 and April 2002 resulting in recruitment rates of 83.2% and 76%,
respectively. Figure 4 shows the timeline of the interview encounters throughout

the LMOS. Time zero (T=0) refers to diagnosis date. The baseline (Wave 1) and

60

Wave 2 interviews were usually conducted at the same encounter. However, the

time commitment for both interviews could exceed thirty minutes. Therefore, if

the participant requested to postpone Wave 2 to another day, this request was

 

 

 

accommodated.

T=0 T=6 mos T=12 mos T=18 mos T=24 mos
Case identiﬁcation Wave 3 - SEER data
and recruitment participants asked to ﬁles accessed
through MDCSS answer questions as for cancer
up through 4 of current date stage,
months post- diagnosis and
diagnosis treatment data

Wave 4 -
participants

Baseline (Wave 1) and Wave 2

interviews conducted

. Wave 1 - participants asked to
recall and report answers
based on status 3 months

prior to diagnosis

. Wave 2 — participants asked
to answer questions as of

current date

Figure 4: Data Collection Timeline

asked to answer
questions as of
current date

Reminder notices were mailed to participants to remind them of the Wave

3 (twelve month) and Wave 4 (eighteen month) interview appointments. Efforts

to maintain the interest of participants were successful as evidenced by the

retention rates at both twelve and eighteen months, 92.5% and 89.6%

respectively.

61

Refer to Figure 5 for the LMOS Breast Cancer Recruitment Flowchart.

The LMOS Prostate Cancer Recruitment Flowchart is displayed in Figure 6.

62

 

 

. . I.I.II.P.I:5.I.-.I.:....rl..

Flag-J

¢

 

 

Figure 5: LMOS Breast Cancer Recruitment Flowchart

 

 

Breast cancer subjects identiﬁed

2424

 

 

 

Previous cancer diagnosis
(552)

 

 

 

 

 

Physician refused initial contact
(43)

(439)

Resided out of county

 

 

 

 

 

 

(122)

 

No physician contact information

 

 

 

 

 

Physician letters mailed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1268
Physician refusals Excluded, accrual complete
(37) (19)
f Incomplete contact information
(6)
Subject letters mailed
1206
L I
Not screened Screened
(370) 836
l
l t t ct A l I t
unab?1§9‘;°" a H cc'"a(3°8‘;mp e e Ineligible Eligible
(241) 595
Refused f
(163) l
Enrolled Excluded, accrual completed Refused
495 (13) (87)

 

 

 

 

 

 

 

63

H—N=492 (99.4%) remaining at wave 2
-——N=461 (93.1%) remaining at wave 3
-—N=451 (91.1%) remaining at wave 4

Figure 6: LMOS Prostate Cancer Recruitment Flowchart

 

1871

 

Prostate cancer subjects identiﬁed

 

 

 

Previous cancer diagnosis

 

 

(318)

 

 

 

 

Physician refused initial contact

Resided out of county
(288)

 

 

 

 

(25)

 

 

 

 

No physician contact information
(30)

 

 

 

Physician letters mailed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1210
Physician refusals I Excluded, accrual complete
(62) I (145)
Incomplete contact information
(13)
Subject letters mailed
990
l I
Not screened Screened
(372) 618
J

 

 

Unable to contact ]__ Accrual complete

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ineligible Eligible
(174) (78) (223) J 395
Refused
(1 20)
Enrolled Excluded, accrual complete Refused
294 (9) (92)

 

 

 

 

 

 

 

 

64

N=291 (99.0%) remaining at wave 2
——N=269 (91.5%) remaining at wave 3
-—-N=256 (87.1%) remaining at wave 4

 

 

3.2.3 LMOS Interview Subject Matter

The interview structure remained relatively uniform throughout the four
interview waves of the study. The questions on work tasks and associated
limitations were ﬁrst asked during the Wave 3 (twelve month) interview. This
timeframe was selected because most participants would have completed their
treatments by this time and the acute effects of therapy might be avoided. The
job task and activity limitation question wording is presented in Appendix A.
Weekly hours worked and shift lengths were variables collected at all four waves
so that comparisons at twelve and eighteen-months against baseline would be
possible. Table 1 describes the nature of the interview variables that were

included in the interviews and the waves in which the questions were asked.

Table 1: General Nature of Interview Variables Collected

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Interview Data Wave 1 Wave 2 Wave 3 Wave 4
Demographic

Age x

Race/ethnicity x

Education x

Marital status x x x x
Children in home x x x x
General activity level x x x x
Perceived health status x x x x
Household income x x x
Participant Employment

Employment status and x x x x
why

Weeks worked in past x x
year

 

 

 

 

 

65

 

 

Table 1 (cont’d).

 

Hours worked per week

 

Why hours changed

 

Occupation

 

Employer description

 

Wages

 

Why wages changed

 

Retirement beneﬁts

 

Sick leave beneﬁts

 

Missed work due to
treatment

XXXXXXXXX

XXXXXXXXX

XXXXXXXXX

 

Last worked

><

 

Break types and length

 

Shift length

 

Job satisfaction

 

Job discrimination

 

Job tasks

 

Task limitations

 

 

 

XXXXXXX

 

XXXXXXX

 

Treatment

 

Count and type surgery

 

Chemotherapy

 

Radiation

 

Hormone therapy

 

 

XXXX

 

XXXX

 

XXXX

 

Health Insurance

 

Employer based
insurance available

 

Who covered by
insurance

 

Who carries policies

 

Available insurance
coverage source

 

Covered by government
insurance

 

 

Comorbidity

 

 

 

 

 

66

 

3.3 Data: Disability Study Subject Restriction

The disability study imposed additional eligibility restrictions upon the
participants recruited for the LMOS study. This was necessary because of the
focus on how work-related disability may present in these individuals.

Therefore, this disability analysis was restricted to participants that were
working or not currently working but had a job at the baseline interview (Wave 1)
and were retained at twelve and eighteen months. The interview ﬁrst asked
individuals if they worked for pay or proﬁt. Those who stated they were not
currently working were subsequently asked if they had a job either full or part-
time. If they responded they did not, they were excluded from the study. If they
did have a job, the reason for not working was assessed. Reasons for not
working but having a job may include being on layoff or participating in seasonal
jobs (i.e. a teacher not working during the summer months). If the participant
stated they did not work at baseline because of a medical condition, they were
excluded from the study. Individuals who reported they were retired, disabled or
unable to work at Wave 1 were excluded from this disability study since the
interest was in the possible development of work-related disability after the
cancer diagnosis and treatment. These criteria left n=447 employed women in
the breast cancer cohort and n=267 employed men in the prostate cancer cohort.

Overall retention rate for the duration of the study exceeded 85%. As
other research has documented, retention among women was greater than that
for men. The interviews conducted at twelve and eighteen months asked

participants about job tasks, limitations and perceived employer accommodations

67

regardless of their employment status at that time. Therefore, only the persons
lost to follow-up between the baseline interview and the twelve and eighteen -
month interviews were eliminated from analyses. This left 93% (n=418) of the
targeted breast cohort retained at twelve months and 91.5% (n=409) retained at
eighteen months. Slightly less, 91.0% (n=243) of men were retained at twelve
months and 86.1% (n=230) were retained at eighteen months. Table 2 displays
how the sub-samples of breast and prostate cancer participants, respectively,
were categorized during follow-up periods. Shading indicates participants in this

category were excluded from the disability study.

68

Table 2: Employment Categories Reported by Employed Participants at 12 and

 

 

 

 

 

 

 

 

 

 

 

18 months
Timeframe Does Doesn’t work Retired Disabled* Unable Lost to
work to follow-
work up
Has a No
job job
Breast (n=495 originally enrolled in LMOS)
Baseline 444 3 33 7 3 5 0
Employed (89.7) (0.6) (6.7) (1.4) (0.6) (1.0)
(n=447)
12 mos 342 20 18 17 11 10 29
(n=418) (76.5) (4.5) (4.0) (3.8) (2.5) (2.2) (6.5)
18 mos 336 11 21 21 18 2 38
(n=409) (75.2) (2.5) (4.7) (4.7) (4.0) (0.4) (8.5)
Prostate (n=294 originally enrolled in LMOS)
Baseline 267 0 3 20 2 2 0
Employed (90.8) (0.0) (1.0) (6.8) (0.7) (0.7)
(n=267)
12 mos 198 8 8 27 0 2 24
(n=243) (74.2) (3.0) (3.0) (10.1) (0.8) (9.0)
18 mos 194 1 9 24 2 0 37
(n=230) (72.7) (0.4) (3.4) (9.0) (0.8) (13.9)

 

 

 

 

 

 

 

 

 

*Participants were not provided a deﬁnition of “disabled” and it is unknown to
what extent they used the deﬁnition set forth in this study.

3.3.1 Disability Study Descriptive Data

Baseline socio—demographic characteristics of the breast and prostate

cohorts are displayed in Table 3. Race was obtained through a recoding of

reported race into categories of White/Other and African-American. The decision

to combine the other race category with the White/Caucasian category was due

to the fact that the majority of “other" reported their race as Asian. According to

cancer statistics, persons of Asian ancestry experience incidence rates more

69

similar to White/Caucasian individuals. However, the total number of participants
reporting “other” race was too small to maintain them as a separate category
(n=15). Education was categorized as less than or equal to high school diploma,
some college or college degree. Similarly, marital status was recoded into three
categories; currently married, never married, or formerly married
(widowed/divorced/separated).

Comorbidity count was tallied based on participant report. The nature or
perceived severity of the comorbid condition was not available so the decision
was made to simply count the number of conditions. Participants were also
asked to evaluate their own health status. Excellent and very good scores were
combined as were fair/poor scores. Perceived health status at twelve months is
also presented to indicate how health status had changed from baseline to
twelve months post-diagnosis and treatment.

Annual income levels were categorized as low (<20,000), mid-range
(20,000-74,999) or high (275,000). Similar to the health status, the twelve-month
reports are also presented. A general measure of activity level per week was
asked of participants at all waves and the baseline and twelve month reports are

included in Table 3.

7O

Table 3: Study Participant Descriptive Statistics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breast Cohort Prostate Cohort
Characteristic (n=447) (n=267)

Mean age (SD) 49.9 (7.7) 55.4 (5.9)
Age range 31-65 34-65
Race

White/Other 350 (78.3) 202 (75.7)

African-American 97 (21.7) 65 (24.3)
Education

3 High school 121 (27.1) 62 (23.2)

College, no degree 124 (27.7) 66 (24.7)

College graduate 202 (45.2) 139 (52.1)
Marital Status

Married 270 (60.4) 213 (79.8)

Formerly married 133 (29.8) 36 (13.5)

Never married 44 (9.8) 18 (6.7)
Comorbidities (measured at 12 mos)*

None 238 (56.9) 133 (54.7)

1-2 153 (36.6) 86 (35.4)

3+ 27 (6.5) 24 (9.9)
Baseline Health Status

Excellent/very good 314 (70.2) 194 (72.7)

Good 90 (20.1) 59 (22.1)

Fair/poor 43 (9.6) 14 (5.2)
Health Status (12 mos)*

Excellent/very good 251 (60.0) 161 (66.3)

Good 118 (28.2) 64 (26.3)

Fair/poor 49 (11.7) 18 (7.4)
Baseline annual household income

<20,000 31 (6.9) 7 (2.6)

20,000-74,999 222 (49.7) 90 (33.7)

275,000 179 (40.0) 161 (60.3)

 

 

 

71

 

Table 3 (cont’d).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ _ Breast Cohort Prostate Cohort
Characteristic (n=447) (n=267)
Annual household income (12 mos)*
<20,000 26 (6.2) 7 (2.9)
20,000-74,999 226 (54.1) 90 (37.0)
275,000 151 (36.1) 131 (53.9)
Engage in mild activity more than 3 79 (17.7) 52 (19.5)
days
Engage in mild activity more than 3 69 (16.5) 36 (14.8)
days (12 mos)*
Engage in moderate activity more 88 (19.7) 54 (20.2)
than 3 days
Engage in moderate activity more 86 (20.6) 45 (18.5)
than 3 days (12 mos)*
Engage in strenuous activity more 29 (6.5) 36 (13.5)
than 3 days
Engage in strenuous activity more 28 (6.7) 26 (10.7)
than 3 days (12 mos)*
Work up a sweat sometimes/always 316 (70.7) 228 (85.4)
Work up a sweat sometimes/always 284 (67.9) 184 (75.7)
(12 mos)*
”Breast cohort at 12 months, n=418, Prostate cohort at 12 months, n=243

 

As has been cited, one of the reasons the MDCSS was selected as the
patient source pool was the over-representation of minority cases. Unfortunately,
the participants recruited into the study for both the breast and prostate cohorts
were predominantly white, 78% and 76%, respectively. Moreover, approximately
half of the participants were well educated with college degrees. Forty percent of
the women reported an annual income greater than $75,000. A larger proportion
of men, 60%, documented this level of household income. Approximately ﬁve
percent of both cohorts in the high-income group shifted to the mid-range income

group at twelve months. This decrease was statistically signiﬁcant for women but

72

not statistically signiﬁcant for men. No statistically signiﬁcant changes occurred
in household income between twelve and eighteen months (data not shown).
More than sixty percent of both cohorts were married, 60% and 80% for the
women and men, respectively. Although the same age range was targeted in
both cohorts, the men were statistically signiﬁcantly older than the women, 55.4
years vs. 49.9 years (ttest p<0.01), respectively.

Participants appeared to be relatively healthy other than their cancer
diagnosis. More than half of each cohort reported no comorbid conditions (57%
and 55%) for the women and men, respectively. At least seventy percent of
respondents in both cohorts rated themselves as having very good or excellent
overall health status before they were diagnosed and 60% maintained this
opinion at twelve months post-diagnosis. However, the shift from very
good/excellent health status to good or fair/poor was statistically signiﬁcant for
women (p<0.01). Health status category shifts among the men were not

statistically signiﬁcant.

3.3.2 Disability Study Disease-Related Characteristics

Cancer stage information was obtained from SEER data ﬁles. Most
individuals were diagnosed before the cancer had metastasized. Treatment
information came from both the patient and the SEER data. Neither source may
be considered a gold standard. SEER treatment data may not include all the
treatments a patient received, especially in cases where individuals sought

treatment outside the SEER catchment area or if treatment was delivered in

73

outpatient settings. Alternatively, patients may not accurately report their
treatment protocols.

Therefore, SEER and patient reports were combined to determine the
individuals receiving the following treatments. In the event that the SEER report
and the patient report contradicted, the positive report of treatment was
accepted. That is, if SEER data suggested that a particular treatment was not
provided but the patient claimed to have received the treatment in question, the
individual was coded as having received the treatment. Medical record review to
conﬁrm treatment was not an option. The hormone category of treatment was
restricted to hormones used as a therapeutic option rather than prophylactic and
was asked only of the men. Table 4 presents the disease-related information of

the participants.

Table 4: Study Participant Disease Statistics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breast Prostate
Characteristic Cohort Cohort
(n=447) (n=267)
Stage
In-situ 118 (26.4) 0
Local 192 (43.0) 209 (78.3)
Regional 126 (28.2) 55 (20.6)
Distant/unknown 11 (2.5) 3 (1 .1)
Treatment
No treatment 1 (0.2) 9 (3.4)
Surgery only 85 (19.0) 171 (64.0)
Radiation (+/- surgery) 96 (21.5) 67 (25.1)
Chemotherapy (+/- radiation, surgery) 265 (59.3) 8 (3.0)
Hormone therapy (+/- surgery) -- 12 (4.5)

 

74

 

Slightly more than two-thirds of each cohort was diagnosed early in their
cancer pathology. Sixty-nine percent of women had either in-situ or locally
staged disease at diagnosis while 78% of men had locally staged disease at
diagnosis. The majority, 59%, of women underwent treatment regimens that
included chemotherapy. In contrast, a minority of the men, just 3%, had
chemotherapy. In fact, the proportion of men not having treatment of any sort
exceeded the proportion that received chemotherapy. The treatment of choice

for the majority of the men was surgery without additional therapy. 64%.

3.3.3 Disability Study Employment-Related Characteristics

3.3.3.1 General Employment

The purpose of this study was to evaluate the impact of cancer diagnosis
and treatment on the participants’ job-related tasks. A variety of job related
variables that could impact disability were included in the interviews. Baseline
information related to general employment characteristics and type of occupation

are presented in Table 5.

Table 5: Study Participant Employment Statistics at Baseline

 

 

 

 

 

 

 

 

 

 

 

Breast Prostate
Characteristic Cohort Cohort

(n=447) (n=267)
Sick leave available 288 (64.4) 205 (76.8)
White collar occupation 304 (68.0) 175 (65.5)
Mean hours worked per week 37.7 (11.3) 44.6 (11.6)
Mean shift length 8.0 (1.7) 8.8 (1.8)
Union member 1 17 (26.2) 76 (28.5)
Mean percent of shift spent sitting 45.4 (30.2) 47.7 (30.1)

 

75

Employment status at twelve and eighteen months was reviewed.
Essentially three-quarters of both cohorts remained employed at twelve months
with just a small percent of the prostate cohort dropping out of the workforce at
eighteen months. Table 6 presents the employment categories for breast and
prostate cohorts at twelve months. The categories were based on participant

self-report. Shading indicates that individuals were not included in analyses.

Table 6: Employment Categories Reported by Employed Participants at 12 and

18 months

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Timeframe Does Doesn’t work Retired Disabled* Unable Lost to
work to follow-
work up
Has a No
job job
Breast (n=447 employed at baseline)
12 mos 342 20 18 17 11 10 29
(n=418) (76.5) (4.5) (4.0) (3.8) (2.5) (2.2) (6.5)
18 mos 336 11 21 21 18 2 38
(n=409) (75.2) (2.5) (4.7) (4.7) (4.0) (0.4) (8.5)
Prostate (n=267 employed at baseline)
12 mos 198 8 8 27 0 2 24
(n=243) (74.2) (3.0) (3.0) (10.1) (0.8) (9.0)
18 mos 194 1 9 24 2 0 37
(n=230) (72.7) (0.4) (3.4) (9.0) (0.8) (13.9)

 

 

 

 

 

 

 

 

 

 

“Participants were not provided a deﬁnition of “disabled” and it is unknown to
what extent they used the deﬁnition set forth in this study.

As mentioned earlier, several plausible explanations exist to explain why
an individual did not work yet had a job. Of these participants, women were
more likely to identify their illness as the reason for not working. Fourteen (70%)
of the twenty women at twelve months who had a job but weren’t working stated

the reason was their cancer. At eighteen months, seven (63.6%) of the eleven

76

blamed their illness as the reason they were not working. Less than forty percent
of the men who had a job but were not working at twelve months blamed their
illness (3 of out 8, 37.5%). At eighteen months, the one individual remaining with

a job but not working did not identify his illness as the reason.
3.3.3.2 Job Satisfaction and Employer Accommodation

Participants were asked to evaluate job satisfaction and employer
accommodation at twelve months. Satisfaction and accommodation questions
were asked using a 4-point Likert scale response option. Eligible responses
included strongly agree, agree, disagree, or strongly disagree. The options
strongly agree and agree were collapsed while strongly disagree and disagree
were collapsed. Information related to these employment characteristics is

displayed in Table 7.

77

Table 7: Participant Report of Job Satisfaction and Employer Accommodation at

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

12 months
Breast Prostate
Characteristic Cohort Cohort
(n=418*) (n=243*)
Job Satisfaction Statement
Satisfaction in life from job (agree) 219 (52.4) 132 (54.3)
Most important things involve work 126 (30.1) 86 (35.4)
(agree)
Perfectionist about work (agree) 367 (87.8) 209 (86.0)
Live, eat and breathe job (agree) 66 (15.8) 47 (19.3)
Involved personally in work (agree) 334 (79.9) 207 (85.2)
Interest in new job if assured of 171 (40.9) 73 (30.0)
comparable beneﬁts (agree)
Employer Accommodation
Employer accommodating (agree) 364 (87.1) 197 (81.1)
Employer discriminated (agree) 28 (6.7) 4 (1 .6)

 

 

*All participants remaining at twelve months were asked these questions
regardless of their employment status at twelve months.

A review of the participants’ ratings of employment characteristics
indicates most are attached to their existing jobs. At least half of both cohorts
reported their satisfaction in life came from their job, 52% for women and 54% for
men. Over eighty percent, 88% and 86% for women and men respectively,
considered themselves to be perfectionists about their work. More than three-
quarters of the women, 80%, and 85% of the men were personally involved in
their work. When asked if they would look for a new job if they would be
guaranteed similar beneﬁts, 41% of the women and 30% of the men, reported

they would be interested in a new position. In contrast to what has been reported

78

historically, the majority, 87% for women and 81% for men, reported their
employers to be accommodating to their illness.

Although participants were attached to their jobs, it appears that job
attachment was not all consuming. Less than twenty percent acknowledged that
they “live, eat, and breathe” their jobs, 16% for women and 19% for men.
Similarly, the minority expressed the opinion that the most important things

involved work, 30% for women and 35% for men.

3.3.3.3 Activity Requirement: Job Tasks

Twelve months was the timeframe at which speciﬁc job tasks were ﬁrst
assessed. Participants were asked if particular activities were required for their
jobs regardless of their current employment status at the time of the twelve and
eighteen month interviews. Table 8 displays the proportion of breast cancer
participants reporting a speciﬁc task was required for their job at twelve months.

Table 9 shows similar information for the prostate cancer cohort. I

79

Table 8: Job Task Frequerg Reported at Twelve Months by Breast Cohort

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Job Task Job Task Frequency Response Options
All or Most of Some of None or
almost the time the time almost
all of the none of
time the time
Breast Physical effort 76 62 1 1 8 1 51
(n=418)* (18.2) (14.8) (28.2) (36.1)
Heavy lifting 24 20 89 274
(5.7) (4.8) (21.3) (65.6)
Stooping/crawling/ 43 37 139 188
kneeling (10.3) (8.8) (33.2) (45.0)
Keep up with pace 113 83 75 136
set by others (27.0) (19.9) (17.9) (32.5)
Concentrate 176 126 73 32
(42.1) (30.1) (17.5) (7.7)
Analyze data 125 83 105 94
(29.9) (19.9) (25.1) (22.5)
Learn new things 96 73 182 56
(23.0) (17.5) (43.5) (13.4)

 

*All participants remaining at 12 months were asked questions regardless of
employment status at that time.

80

 

Table 9: Job Task Frequency Reported at Twelve Months by Prostate Cohort

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Job Task Job Task Frequency Response Options
All or Most of Some of None or
almost the time the time almost
all of the none of
time the time
Prostate Physical effort 31 30 65 113
(n=243)* (12.8) (12.4) (26.8) (46.5)
Heavy lifting 6 8 60 165
(2.5) (3.3) (24.7) (67.9)
Stoopinglcrawling/ 18 1 5 86 1 19
kneeling (7.4) (6.2) (35.4) (49.0)
Keep up with pace 44 45 60 90
set by others (18.1) (18.5) (24.7) (37.0)
Concentrate 85 87 49 18
(35.0) (35.8) (20.2) (7.4)
Analyze data 81 62 55 41
(33.3) (25.5) (22.6) (16.9)
Learn new things 51 47 115 26
(21.0) (19.3) (47.3) (10.7)

 

*All participants remaining at 12 months were asked questions regardless of

 

employment status at that time.

These same questions were asked at the eighteen-month interview.
Tables 10 and 11 present the range of responses to the job task questions asked
at this time. As with the twelve-month responses, these questions were asked of
all participants, even if they had withdrawn from the workforce by this interview

encounter.

81

Table 10: Job Task Frequency Reported by Breast Cohort at Eighteen Months

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Job Task Job Task Frequency Response Options
All or Most of Some of None or
almost the time the time almost
all of the none of
time the time
Breast Physical effort 58 45 1 13 152
(n=409)* (14.2) (1 1.0) (27.6) (37.2)
Heavy lifting 1 1 25 86 246
(2.7) (6.1) (21.0) (60.2)
Stooping/crawling/ 29 19 158 162
kneeling (7.1) (4.6) (38.6) (39.6)
Keep up with pace 90 69 76 133
set by others (22.0) (16.9) (18.6) (32.5)
Concentrate 1 15 1 30 80 43
(28.1) (31.8) (19.6) (10.5)
Analyze data 86 88 101 93
(21 .0) (21.5) (24.7) (22.7)
Learn new things 84 80 162 42
(20.5) (19.6) (39.6) (10.3)

 

*All participants remaining at 18 months were asked questions regardless of

employment status at that time.

82

 

Table 11: Job Task Frequency Reported by Prostate Cohort at Eighteen Months

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Job Task Job Task Frequency Response Options
All or Most of Some of None or
almost the time the time almost
all of the none of
time the time
Prostate Physical effort 22 19 70 92
(n=230)* (9.6) (8.3) (30.4) (40.0)
Heavy lifting 4 7 53 139
(1 .7) (3.0) (23.0) (60.4)
Stooping/crawling/ 1 3 17 75 98
kneeling (5.6) (7.4) (32.6) (42.6)
Keep up with pace 33 39 51 80
set by others (14.4) (17.0) (22.2) (34.8)
Concentrate 57 74 54 18
(24.8) (32.2) (23.5) (7.8)
Analyze data 50 73 52 20
(21.7) (31.7) (22.6) (12.2)
Learn new things 43 49 94 17
(18.7) (21.3) (40.9) (7.4)

 

*All participants remaining at 18 months were asked questions regardless of

employment status at that time.

3.4 Research Questions and Hypotheses

3.4.1 Prevalence and Direction of Disability Change

The aggressive nature of cancer treatment is expected to result in

disability with work-related tasks. Furthermore, if the nature of these disabilities

represents a transient condition, it is reasonable to presume prevalence

estimates would decrease as time since treatment increases. The study will test

this presumption with the following hypothesis applied to physical and cognitive

disability:

83

 

0 H1: Cancer related physical and cognitive disabilities will decrease as the
timeframe from diagnosis increases from twelve months after diagnosis to

eighteen months after diagnosis.

Since disability was not measured directly through the interview process, it
was necessary to construct this variable using existing data elements. Activity
limitation and participation restriction are the two domains that make up the
outcome of disability per the ICF. The following sections describe the
methodology used to generate the activity limitation and participation restriction
indicator variables and then describe how these were combined to generate the
outcome variable of interest.

3.4.1.1 Methodology: Document Required Job Activities
3.4.1.1.1 Classify Job Activities as Physical or Cognitive

A review of the particular job tasks included in the interviews indicated
they might reasonably be collapsed into two categories such as physical or
cognitive. Several methods were employed to test if this were the case. STATA
v9.2 was used for all analyses.

Factor analysis was performed to assess the degree to which these
distinct tasks reﬂected an underlying work response dimension (59). After the
initial factor structure was determined, orthogonal (i.e. varimax) rotation was
carried out to further simplify the factor structure. Orthogonal rotation was
selected rather than oblique rotation as the factors were presumed to be

uncorrelated.

84

The factor analysis results suggested two main dimensions were
represented by the seven tasks. Three of the items loaded heavily on the ﬁrst
dimension while three loaded heavily on the second dimension. The ﬁrst
dimension Included the physical effort, heavy lifting and
stooping/crawling/kneeling tasks. The second dimension included the individual
tasks of concentration, data analysis and learning new things. The interpretation
of these dimensions were categorized as physical (Factor 1) and cognitive
(Factor 2).

The remaining job task, “keeping up with others” loaded slightly more
heavily on Factor 2 yet the split between the dimensions was not to the degree of
the other job task variables. Similar results were obtained for the wave 3
activities and the wave 4 activities. Table 12 gives the rotated factor loadings of

all seven variables for both factors.

Table 12: Rotated Factor Loadings for All Job Tasks

 

 

 

 

 

 

 

 

 

 

 

Wave 3 Activities Wave 4 Activities

Job Task Factor 1 Factor 2 Factor 1 Factor 2

Loading Loading Loading Loading
Physical effort 0.7629 -0.0790 0.7682 -0.0913
Heavy lifting 0.7129 0.0309 0.6908 -0.0205
Stoop/crawl or 0.6677 0.0231 0.6245 0.0296
kneel
Keep up with 0.2041 0.4114 0.2025 0.3170
others
Concentrate 0.0382 0.5596 0.0214 0.5876
Analysis -0.2222 0.5993 —0.2810 0.5850
Learn new 0.0581 0.5319 0.0186 0.5843
things

 

 

 

 

 

85

Since the variable “keep up with others” did not load as strongly, further
attempts to categorize this variable were conducted. Sensitivity analyses were
conducted which compared this variable with the other work-related task
variables. The highest sensitivity values were obtained with the lifting (70.5%)
and physical effort (69.4%) variables as the “true" status. This pattern was
observed for the twelve-month and eighteen-month accrual points. Due to the
apparent conﬂicting results between the factor analysis and the sensitivity
analysis, the decision was made to exclude the “keep up with others” variable as
a component of either the physical or cognitive dimensions.

Factor analysis was then re-run excluding the “keep up with others”
variable with similar results. Table 13 gives the rotated factor loadings for the six

retained variables for both factors.

Table 13: Rotated Factor Loadings for Targeted Job Tasks

 

 

 

 

 

 

 

 

 

Wave 3 Activities Wave 4 Activities

Job Task Factor 1 Factor 2 Factor 1 Factor 2

Loading Loading Loading Loading

(n=645) (n=645) (n=571) (n=571)
Physical effort 0.7471 -0.1263 0.7619 -0.1155
Heavy lifting 0.7161 0.0122 0.6972 -0.0259
Stoop/crawl or 0.6775 0.0228 0.6231 0.0118
kneel
Concentrate 0.0530 0.5429 0.0312 0.5727
Analysis -0.1950 0.6221 0.2562 0.6109
Learn new 0.0660 0.4971 0.0257 0.5639
things

 

 

 

 

 

These loadings were then used to generate factor scores that were saved

as new variables. As expected, due to their nature as standardized variables,

86

 

 

the means of each factor score were essentially zero with standard deviations
close to one (59). The physical dimension at both waves was somewhat
negatively skewed while the cognitive dimension was more symmetrical. Table

14 shows the summary statistics for each factor score.

Table 14: Summay statistics for factor scores

 

 

 

 

 

 

Wave 3 Activities Wave 4 Activities
Factor Score Factor 1 Factor 2 Factor 1 Factor 2
characteristic (Physical) (Cognitive) (Physical) (Cognitive)
Mean (1 SD) 0.0 (0.85) 0.0 (0.74) 0.0(0.85) 0.0 (0.76)
Skewness -1.12 0.11 -1.04 0.03
Kurtosis 3.66 2.24 3.48 2.34

 

 

 

 

 

3.4.1.1.2 Generate Composite Physical or Cognitive Variable

A physical task composite variable was created from the three speciﬁc
tasks identiﬁed from the factor analysis. Participants were classiﬁed as having to
perform a physical task if they claimed to perform any combination of physical
effort, heavy lifting and/or stooping/crawling/kneeling at least some of the time.

A similarly constructed composite cognitive task was created from the
three particular tasks (concentration, analysis and learn new things) shown to
load on the second factor.

It should be mentioned the scoring associated with the factor analysis
provided a mechanism to create composite variables that would have been
continuous in nature. However, in order to facilitate outcome interpretation, the
composite variables were retained as dichotomous measures. A value of zero

was associated with “no disability” while a value of one was associated with

87

 

“disability”. Table 15 displays the prevalence rates of the composite job task

variables at twelve and eighteen months for both cohorts.

 

 

 

 

 

 

 

Table 15: Composite Activity Rec uirements for Cohorts
Activity Requirement 12 months 18 months
Breast (n=418) (n=409)
Composite Physical Task 298 (71.3) 261 (63.8)
Composite Cognitive Task 398 (95.2) 357 (87.3)
Prostate (n=243) (n=230)
Composite Physical Task 154 (63.4) 134 (58.3)
Composite Cognitive Task 398 (95.2) 357 (87.3)

 

 

 

 

 

3.4.1.1.3 Compare Physical or Cognitive Job Activities between 12 and 18
Months

These comparisons were conducted to act as a reference for the
forthcoming analyses regarding activity limitation and disability that will address
the hypotheses. Several individuals were lost to follow-up during this study and
others changed their employment status between twelve and eighteen months.
Therefore, it was necessary to determine if the prevalence of having to perform
these activities changed over this timeframe. If the prevalence of a task changed
signiﬁcantly, it would be difﬁcult to evaluate any changes in activity limitation that
might be detected during the forthcoming analyses. Recall that the variable
“keep up with others” was dropped after the conﬂicting ﬁndings regarding its

association as either a physical or cognitive task.

88

 

Two comparisons were carried out for each task. The overall prevalence
reported by all available respondents at twelve and eighteen months was
compared ﬁrst. The subsequent comparison was then restricted to the
individuals who had paired data available for both timeframes. Stated another
way, the second comparison was conducted only on individuals reporting at both
twelve and eighteen-months. STATA v9.2 was used for all analyses.
3.4.1.2 Methodology: Create Indicator Variable

The ICF model of disability includes two domains that contribute to
disability, activity limitation and participation restriction. In order to establish if
these two elements were present for the participants, indicator variables were
created for each domain. The indicator variables were established as
dichotomous variables with 0=domain absent and 1=domain present. This
coding scheme would then allow for multiplication to establish the outcome of
interest, i.e. disability. The resulting outcome would then also be a dichotomous

variable with similar interpretation of 0=disability absent and 1=disability present.
3.4.1.2.1 Domain 1: Document Activity Limitation in Job Activities

Assessing activity limitation for speciﬁc job tasks was relatively
straightfonrvard. If the participants stated they performed each task at least some
of the time, they were immediately asked if their ability to carry out this task was
limited subsequent to their diagnosis and treatment. However, summary
variables representing overall activity limitation for the composite physical and
cognitive task variables were created for the twelve and eighteen-month

interviews.

89

The composite variable for “limitation in physical task” was based on the
responses to the activity limitation responses from the three speciﬁc physical
tasks. A similar strategy was carried out to generate a cognitive task limitation
variable. The presence of limitation in any one or combination of speciﬁc tasks
denoted presence of limitation in the related composite variable. Activity
limitation in the speciﬁc and composite variables were established as
dichotomous variables with 0=absence of limitation, 1=presence of limitation. An

example of one participant’s coding is displayed in Tables 16 and 17.

Table 16: Activity Limitation Coding Example - Physical Task

 

 

 

 

 

 

 

 

 

Speciﬁc Task Required by Participant P;£:T?£itt:;a;:s
Job Response Treatment
Physical effort Some of the time Yes
Heavy lifting None of the time n/a
Stoop/crouch/kneel All of the time No
Composite Physical Scon'ng
Physical Task required Yes (1)
Activity Limitation Present Yes (1)

 

 

 

Table 17: Activiy Limitation Coding Example — ngnitive Task

 

 

 

 

 

 

 

 

Speciﬁc Task Required by Participant ngltiﬁm‘ittfrlt)?
Job Response Treatment
Concentrate Some of the time No
Analysis None of the time nla
Learn new things All of the time No
Composite Cognitive Scoring
Cognitive Task required Yes (1)
Activity Limitation Present No (0)

 

 

 

90

3.4.1.2.2 Domain 2: Document Participation Restriction in Labor Market
Participation

Participation restriction was deﬁned as an indicator variable focusing on
decreases in speciﬁc measures of employment attachment that were conﬁrmed
by the participants to be a result of their diagnosis and treatment. Since a global
question regarding participation restriction was not asked, other variables that
could be used as proxies for restricted work participation were combined.

The variables used to create a proxy variable for participation restriction
included the following: number of hours worked weekly, reason for new
unemployment, reason for working only part time and reason for earnings
decreasing. After determining if these conditions were present, the survey
further queried the participants if the decrease in hours, earnings, or job status
was a result of their cancer and treatment. The participation restriction variable
was coded as “1” if the patient stated the participation restriction was due to
diagnosis and treatment. Alternatively, participation restriction was coded as “0”
if the participant did not report participation restriction or if the respondent
identiﬁed the participation restriction to be related to some cause other than their
cancer. Appendix B contains the wording of the questions used to generate the
participation restriction variable.
3.4.1.3 Methodology: Generate Disability Variable

After creating the indicator variables for the domains of interest (e.g.
activity limitation and participation restriction), these were multiplied to create the
outcome variables for physical disability and cognitive disability. Due to the OM

coding convention selected, simple multiplication allowed the disability variables

91

 

to reﬂect 1 if the participant reported an activity limitation with participation
restriction and 0 if one or both of the domains were not impacted. Table 18

shows the possible outcomes from the multiplication of both indicator variables.

 

Table 18: Disability Coding Example
Activity Limitation Participation

Status Restriction Status Disability Status

 

No participation

No activity limitation (=0) restriction (=0)

No disability (=0)

 

Activity limited (=1) :gtfiigfrngg No disability (=0)

 

Restricted work

No actlvrty Ilmltatlon (=0) participation (:1)

No disability (=0)

 

Restricted work
participation (=1)

 

Activity limited (=1) Disabled (=1)

 

 

 

 

3.4.1.3.1 Evaluate Prevalence and Change of Disability

The overall change in disability prevalence between the time periods was
compared using two methods. First, comparisons of the overall rates between
twelve and eighteen months including all possible respondents (not just those
with paired data from twelve and eighteen months available) were tested using
the Pearson chi-square. Secondly, the change for respondents in each cohort
that were included in both time periods was calculated using paired analyses, i.e.
comparing each individual’s status from twelve months to eighteen months.

After the disability status for each participant at twelve and eighteen
months was established, the direction of disability status change was
categorized. This required that the participant have a valid assessment of

disability at both twelve and eighteen months. If the participant was not disabled

92

 

at twelve months but was disabled at eighteen months, they were categorized as
“worsen”. If a participant was disabled at twelve months but not disabled at
eighteen months, they were classiﬁed as “improved”. Finally, if a participant was
disabled at twelve and eighteen months, they were classiﬁed as “unchanged”.
Individuals who were not disabled at twelve and eighteen months were excluded
from this evaluation. Methods to investigate characteristics of participants who
improved or worsened from twelve to eighteen months were considered including
ordered logistic regression. Unfortunately, small cell sizes prohibited this form of .
analysis. Unadjusted tests for association with potentially inﬂuential variables

were conducted using Fisher’s Exact Test.

3.4.2 Disability Prevalence Disparities Based on Deﬁnition

The second study question was to evaluate and compare the physical and
cognitive disability estimates obtained from an activity limitation deﬁnition with the
estimates obtained from the ICF (activity limitation with participation restriction)
deﬁnition. Due to the more comprehensive nature of disability supported by the
ICF model, it is expected that this deﬁnition would result in a lower prevalence.
The second hypothesis will be tested for both physical and cognitive disabilities.

0 H2: Cancer related physical and cognitive disabilities will be more
prevalent at twelve and eighteen months using an activity limitation
deﬁnition of disability compared to an ICF deﬁnition requiring activity

limitation and participation restriction.

93

 

 

3.4.2.1 Methodology: Compare Disability Prevalence

The difference between reporting a disability using just an activity
limitation deﬁnition and a combined activity limitation/participation restriction
deﬁnition as per the ICF model was compared using Pearson Chi-Square in an

unadjusted analysis.

3.4.3 Contextual Factor Association with Disability

The last study question seeks to describe associations between a number
of contextual factors with disability. The contextual factors may be categorized
as personal or environmental. Although cancer treatment is not necessarily a
personal characteristic, it was included as a personal factor because of the
association with cancer stage. Patterns of inﬂuence will be tested using the
following hypotheses against both outcomes, physical and cognitive disability.

0 H3: Personal factors that will place individuals at risk for physical and
cognitive disability include minority race, low annual household income,
chemotherapy and later staged disease.

0 H4: Environmental factors (associated with employment) that will place
individuals at risk for physical and cognitive disability include blue-collar
employment, less sedentary job, no employer accommodation and low job

satisfaction.

3.4.3.1 Methodology: Multivariate Analysis using Logistic Regression
The goal of the multivariable analysis was to predict which contextual

factors inﬂuenced the presence of disability in the participants. Some variables

94

 

may be amenable to intervention that could assist patients to select among
treatment options or plan for rehabilitation. Logistic regression was used since
the dependent variables of interest were dichotomous in nature, i.e. a
physical/cognitive disability was either present (=1) or absent (=0). As the
hypotheses suggest, personal factors such as minority race, low income, late
staged disease or chemotherapy were of particular interest (hypothesis #3). The
remaining factors were environmental with respect to the work setting. These

included job classiﬁcation, physical demands of the job, employer

 

accommodation and job satisfaction (hypothesis #4). ‘
Logistic regression permits users to “assess the probability of a disease or
health condition as a function of a risk factor and covariates.” (130) Otherwise
stated, logistic regression is used to estimate the probability that a person has a
particular outcome (131). Another feature of logistic regression is that the slope
coefﬁcients can be transformed into odds ratios (131). The odds ratio obtained
for each variable represents its inﬂuence when controlling for all other
independent covariates included in the model (131, 132). The Iogit function used
in logistic regression allows the values of independent variables to be linked to
the probability of the dependent or outcome variable. The logistic regression
model can be written as follows where Y is a binomial variable with probability of

success = p :

 

Logit(E(Yi)) = Iogit (p) = In (1 fp )= (30 + (3.x1 + (32x2 + + (3.x.

The X’s represent the independent variables of interest while the

coefﬁcients are constants representing unknown parameters, which may be

95

estimated by using the dataset (110). The coefﬁcient Bo is said to give the log of
the “baseline” or “background” odds, i.e. the odds occurring if there were no X’s
(110). The interpretation of the B coefﬁcients is understood as the “change in the
log odds that would result from one unit change in the variable X when all other
X’s are ﬁxed” (110).

Selection of potential variables to include in the modeling procedure was
guided by the variables discussed in existing literature. As described in Chapter
2, personal variables that are socio-demographic in nature (e.g. age, race,
income, and poor ﬁtness) have been associated with activity limitation or
participation restriction. Similarly, disease-related covariates including treatment
and stage have been linked with activity limitation. Finally, items that would
indicate a supportive work environment, eg. sick leave beneﬁt, perception of
employer accommodation were included for their documented association with
return to work. The physical nature of the job was assessed by the duration of
shift length and percent of shift length spent sitting. Table 19 displays the

variables considered for inclusion in the modeling process.

96

Table 19: Variables considered for inclusion in modelin procedures

 

Personal Contextual Factor

Environmental
Contextual Factor

 

Socio-demographic

Disease

Employment

 

 

 

 

Marital status

 

Annual household
income

 

 

Routine physical activity

 

Age Stage at diagnosis Sick leave beneﬁt

Race Treatment category Occupational
classiﬁcation

Education Employer
accommodating*

 

Satisfaction in life from
job*

 

Most important things
involve work*

 

Perfectionist about
work*

 

Live, eat, and breathe
the job*

 

Involved personally in
work*

 

Interest in a new job*

 

Duration of work day

 

 

Percent work day
seated

 

 

*Variables focused on job satisfaction

However, the full complement of potentially explanatory variables was

considerable, numbering close to twenty. Therefore, in the interest of parsimony,

factor analysis was performed to assess the degree to which the seven distinct

variables asking about job satisfaction reﬂected a fewer number of underlying

dimensions (59).

The factor analysis results suggested one main dimension was

represented by the variables based on Eigenvalue (1.75 for factor 1 and 0.34 for

97

 

factor 2). Three of the items loaded heavily on the ﬁrst dimension. The ﬁrst
dimension included satisfaction in life from job, most important things involve
work and live, eat and breathe the job.

Three of the remaining satisfaction variables, perfectionist about work,
involved personally in work and interest in approaching a new job loaded slightly
more heavily on Factor 1 yet the split between the dimensions was not to the
degree of the other satisfaction variables. The remaining variable, employer
accommodating, actually loaded more heavily on factor 2 yet again, the split was
not extreme. Table 20 gives the rotated factor loadings of all seven variables for

both factors.

Table 20: Rotated Factor Loadings for Job Satisfaction Variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Wave 3 Satisfaction

Satisfaction Factor 1 Loading Factor 2 Loading

(n=579) (n=579)
Employer accommodating 0.0689 0.3915
Satisfaction in life from job 0.7038 0.0972
Most important things 0.7315 0.0226
involve work
Perfectionist about work 0.2496 0.0085
Live, eat, and breathe the 0.6603 -0.0548
job
Involved personally in work 0.4653 0.1411
Approach a new job -0.0465 -0.3968

 

 

Factor analysis was then re-run retaining just the three satisfaction
variables that reﬂected the same dimension. Table 21 gives the rotated factor

loadings for the retained variables for Factor 1.

98

Table 21: Rotated Factor Loadings for Retained Job Satisfaction

 

 

 

 

Wave 3 Satisfaction
Satisfaction Factor 1 Loading (n=651)
Satisfaction in life from job 0.7205
Most important things involve work 0.7554
Live, eat, and breathe the job 0.6400

 

 

 

 

These loadings were used to generate a factor score that was saved as a
new variable. As expected due to their nature as standardized variables, the
means of the factor score was essentially zero with standard deviations close to
one (59). Table 2 shows the summary statistics for the satisfaction factor score.
This score was included in the regression modeling in lieu of the individual
satisfaction variables. However, employer accommodation was retained as a
separate variable for the modeling procedures since the factor association

appeared to contradict the other satisfaction related variables.

Table 22: Rotated Factor Loadin s for Job Tasks

 

 

 

Wave 3 Satisfaction
Factor Score characteristic Factor 1
Mean (1 SD) 0.0 (0.85)
Skewness -0.19
Kurtosis 3.02

 

 

 

The inﬂuence of socio-demographic, disease-related and employment
covariates on the presence of physical or cognitive disability at twelve and
eighteen months was tested using backward selection logistic regression.
Separate analyses were conducted for the breast and prostate cohorts, therefore

gender was not included as a covariate in the regressions. Robust standard

99

 

 

errors were used as these are able to provide variance estimates and conﬁdence

intervals for models that are not well speciﬁed.

100

CHAPTER 4: RESULTS

4.1 Study Question 1: Prevalence and Direction of Disability
Change

4.1.1 Hypothesis 1: Comparison of Physical and Cognitive Disability

The ﬁrst hypothesis set down for this study was stated as:
0 H1: Cancer related physical and cognitive disabilities will decrease
as the timeframe from diagnosis increases from twelve months

after diagnosis to eighteen months after diagnosis.

In order to evaluate the changes in disability, it was necessary to ﬁrst
investigate patterns associated with the components that made up the outcome
variable. Speciﬁcally, activity requirement, activity limitation and participation
restriction needed to be compared from twelve to eighteen mOnths to inform
interpretation of the disability results.

4.1.1.1 Comparison of Physical and Cognitive Activity Requirements
Between 12 and 18 Months

Seventy percent of the breast cohort and 64% of the prostate cohort
reported their jobs required physical activity at twelve months. However, more
than 95% of men and women described a cognitive role to their job. The
proportion of individuals requiring cognitive tasks was statistically signiﬁcantly
greater than those requiring physical tasks for women and men.

Table 23 displays the proportions of breast cohort respondents
documenting speciﬁc and composite task requirements at twelve and eighteen

months. The differences between overall proportion of speciﬁc physical tasks

101

and the composite physical task from twelve to eighteen months were not

statistically signiﬁcant. Similarly, the apparent decreases In two of the speciﬁc

cognitive tasks, analysis and learning new things, as well as the composite

cognitive task were not statistically signiﬁcant. However, the decrease in the

concentration requirement from twelve to eighteen months reached statistical

signiﬁcance at the p<0.05 level for the sub-group of women with paired data

available.

Table 23: Activity Requirements for Breast Cohort

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Task 12 months 18 months p-value p-value
(n=418) (n=409) (paired)

Physical effort 256 (61.2) 216 (52.8) 0.23 0.28
Heavy lifting 133 (31.8) 122 (29.8) 0.89 0.48
Stoop/crawl or kneel 219 (52.4) 206 (50.4) 0.54 0.32
Composite Physical Task 298 ( 71.3) 261 (63. 8) 0.48 0. 54
Required
Concentrate 375 (89.7) 325 (79.5) 0.07 0.04*
Analysis 313 (74.9) 275 (67.2) 0.48 0.40
Learn new things 351 (84.0) 326 (79.7) 0.33 0.11
Composite Cognitive Task 398 (95. 2) 357 (87. 3) 0.50 0. 54"
Required

 

”Statistically signiﬁcant at p<0.05

 

"Fisher Exact Test used due to cell size<5

One possible explanation for a signiﬁcant change in activity requirement
would be changes in the proportion of women working between twelve and
eighteen months. It is important to note the activity requirements were asked of
all eligible participants without the condition they be employed at the time of the
interview. This introduces the possibility that recall bias may affect the proportion

of respondents who report having to perform the tasks, i.e. it is possible that

102

persons who retired or were no longer employed may remember their job tasks
differently then someone who is still employed.

In order to assess if recall bias could have affected the concentration task,
employment status was categorized as working, not working or left the job
market and compared against concentration task performance using F isher’s
Exact test due to small cell sizes. Reporting of concentration as a job
requirement was not signiﬁcantly associated with employment category at either
twelve or eighteen months. Moreover, concentration was not signiﬁcantly
associated with a change in employment status from twelve to eighteen months
(e.g. leaving the workforce, no change or returning to the workforce). A
statistically signiﬁcant association between these characteristics was not
documented with the majority, 84%, remaining employed at both waves.
Therefore, the women who no longer worked for some reason were not
responsible for the decline in task requirements.

Another possible reason for the decrease in concentration tasks might be
because the woman changed positions within the same employer and this
activity was simply not required to the degree it once was. The interview
prompted for additional industry or occupation speciﬁcs only in the event that the
individual changed employers from wave to wave. Therefore, this impact cannot
be quantiﬁed.

However, since the decrease in the concentration task among those

women who reported at both twelve and eighteen months was statistically

103

signiﬁcant, the characteristics of the women who reported a change in their
concentration requirement were evaluated in greater detail.

A total of 39 individuals reporting at both timeframes changed their
answers from the twelve to the eighteen-month interview. However, the majority
(n=32, 82.0%) of these respondents remained employed. This provides some
reassurance that the change was probably not due to a large group of

respondents leaving the workforce and introducing some bias to their reporting

although this potential cannot be ruled out.

Table 24 presents similar data for the prostate cohort. None of the

changes in proportion of men performing the individual and the composite tasks

were statistically signiﬁcant at the p<0.05 level.

Table 24: Activity Requirements for Prostate Cohort

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Task 12 months 18 months p-value p-value
(n=243) (0:230) (paired)

Physical effort 126 (51.8) 111 (48.3) 0.68 0.31
Heavy lifting 74 (30.4) 64 (27.8) 0.90 0.53
Stoop/crawl or kneel 1 19 (49.0) 105 (45.6) 0.71 0.59
Composite Physical Task 154 (63. 4) 134 (58. 3) 0.73 0. 54
Required
Concentrate 221 (91.0) 185 (80.4) 0.61 0.64
Analysis 198 (81.5) 175 (76.1) 0.33 008"
Learn new things 213 (87.6) 186 (80.9) 0.38 0.12
Composite Cognitive Task 234 (96.3) 199 (86.5) 1.0" 1.0"
Required

 

I‘Fisher Exact Test used due to cell size<5

104

 

 

4.1.1.2 Comparison of Domain 1: Activity Limitation

Once the proportions of tasks were calculated and compared, the focus
turned to presence of activity limitation for each task caused by the cancer
diagnosis and treatment. Activity limitation was measured directly for the
discrete tasks included in the survey. As would be expected, limitation in any
task was only possible among those individuals who reported having to perform
the task. Table 25 displays the results for the proportion of breast cohort
respondents who report experiencing a limitation. Heavy lifting was most
problematic for women followed by general physical effort. This pattern was
observed both at twelve and eighteen months. Regarding cognitive tasks,
concentration had the most limitations among the women at both time points.

The decline in the composite physical and cognitive activity limitations
from twelve to eighteen months for the individual items and the composite tasks
were statistically signiﬁcant at p<0.01. Thus, while the relative proportion of
these required tasks did not change signiﬁcantly from twelve to eighteen months
(as evidenced by Table 23), the limitations that women reported decreased over
this timeframe. The signiﬁcant declines were observed in the samples of all

respondents and the participants with paired data.

105

Table 25: Activity Limitation for Breast Cohort

 

 

 

 

 

 

 

 

 

Task 12 months 18 months p-value p-value
(paired)
Physical effort 133 (52.0) 71 (32.9) <0.01“ <0.01**
Heavy lifting 88 (66.2) 53 (43.4) <0.01" <0.01**"
Stoop/crawl or kneel 76 (34.7) 30 (14.6) <0.01" <0.01**
Composite Physical 1 76 (59. 5) 94 (36. 3) <0. 01 ** <0.01"
Task Limitation
Concentrate 121 (32.3) 66 (20.3) <0.01** <0.01**
Analysis 92 (29.4) 37 (13.4) <0.01“ <0.01**"
Learn new things 71 (20.2) 42 (12.9) <0.01“ <0.01**
Composite Cognitive 136 (34. 3) 78 (22. 0) <0.01“ <0. 01 **
Task Limitation

 

 

 

 

 

 

 

* Statistically signiﬁcant at p<0.05
”Statistically signiﬁcant at p<0.01
"Fisher Exact Test used due to cell size<5

Similar to the direction of limitations noted among the women, the
proportions of limitations in the individual and composite tasks decreased from
twelve to eighteen months among the prostate cohort. The nature of the affected
tasks was similar to that documented in the breast cohort. Heavy lifting and
physical effort encountered the most limitations for the physical activities while
the concentration presented as the most limited task among the cognitive tasks.
Overall, fewer of the speciﬁc task decreases were statistically signiﬁcant.

The men only experienced signiﬁcant decreases in the composite physical
limitation (p<0.05). The decrease observed in the composite cognitive limitation
was not statistically signiﬁcant. These same patterns were detected in the group
of men with paired data. The results of the limitation analyses for the prostate

cohort are displayed in Table 26.

106

Table 26: Activity Limitation for Prostate Cohort

 

 

 

 

 

 

 

 

 

 

Task Limitation

 

 

 

 

Task 12 months 1 8 months p-value p-value
(paired)

Physical effort 33 (26.2) 16 (14.4) 0.04* 0.06"
Heavy lifting 22 (29.7) 11 (17.2) 0.08 0.18"
Stoop/crawl or kneel 26 (21 .8) 6 (5.7) 0.01**" 0.02“"
Composite Physical 44 (28. 6) 23 (1 7. 2) 0. 02* 0. 04*
Task Limitation
Concentrate 26 (1 1.8) 13 (7.0) 0.10 0.15"
Analysis 17 (8.6) 6 (3.4) 0.04* 0.09"
Learn new things 11 (5.2) 5 (2.7) 0.16" 0.22"
Composite Cognitive 28 (12. 1) 14 ( 7. 0) 0. 08 0. 07

 

”Statistically signiﬁcant at p<.05
“Statistically signiﬁcant at p<0.01
"Fisher Exact Test used due to cell size<5

4.1.1.3 Comparison of Domain 2: Participation Restriction

The second domain in our disability deﬁnition included participation

restriction. Women in the breast cohort reported more participation restriction

compared to the men. At twelve months, eighteen percent of the women

indicated participation restriction vs. nine percent of the men. Fifteen percent of

the women continued to report participation restriction at eighteen months

compared to 7% of the men. Both groups experienced a decline in participation

restriction from twelve to eighteen months. In both cohorts, however, the

declines were not statistically signiﬁcant at p<0.05. The decline remained not

statistically signiﬁcant for the samples with paired data. Table 27 shows the

proportion of respondents in both cohorts reporting participation restriction.

107

 

Table 27: Participation Restriction Reporting

 

 

 

 

 

 

 

 

 

Cohort 12 months 18 months p-value p-value
(paired)
Breast 75 (17.9) 60 (14.7) 0.20 0.10
Prostate 22 (9.0) 16 (7.0) 0.40 0.30

 

Now that the results of the constituents making up the outcome variable,
i.e. disability, have been presented, focus can turn to the results of the disability

analyses.

4.1.2 Comparison of Physical Disability

Physical disabilities decreased from twelve to eighteen months for both
cohorts. However, only the disability decline in the breast cohort from 18% to
10% was statistically signiﬁcant at p<0.05. The conﬁdence in this pattern of
decline increased with the paired sample as the signiﬁcance level increased to
99%. Overall, women experienced signiﬁcantly greater physical disability
compared to men. This trend was anticipated for several reasons. First, women
experienced greater proportions of the domains that make up disability, e.g.
activity limitation and participation restriction. The second reason for expecting
more disability was the more aggressive treatment regimens the women
experienced both in terms of receiving chemotherapy and in terms of undergoing
multiple treatment types (surgery in addition to radiation and/or chemotherapy).
Table 28 displays the proportion of individuals in each cohort having physical

disability per the ICF deﬁnition (activity limitation with participation restriction).

108

 

 

 

Table 28: Physical Disability Reporting in those Performing the Task
Variable 12 months 18 months p-value p-value
(paired)
Breast 51 (17.2) 25 (9.6) 0.01* <0.01**
Prostate 12 (7.8) 6 (4.5) 0.25 0.70

 

 

 

 

 

 

 

*Statistlcally signiﬁcant at p<0.05
"Statistically signiﬁcant at p<0.01
4.1.2.1 Physical Disability Status Change

Forty-one women experienced a change in physical disability status
between twelve and eighteen months. Of these, 19 (46%) reported their
disability improved or resolved from twelve to eighteen months. Sixteen (39%)
identiﬁed no change in their disability while a lesser proportion, 15% (n=6) stated
they developed the physical disability between twelve and eighteen months.

Just nine men experienced a change in physical disability between twelve
or eighteen months. Nearly half (44%) of these men enjoyed improvement in
their disability status. Unfortunately, the next larger proportion identiﬁed a
decline in disability status (33%) while the fewest experienced no change (22%).

It became of interest to identify variables that could be associated with
worsening or improvement in disability status over time. No association between
change in physical status (worsen, stay the same, resolve) and the following
variables were noted for the women: cancer stage, treatment, race, physical
nature ofjob as measured by shift length and percent of shift seated, occupation
type, availability of sick leave, age or baseline perceived health status. However,
baseline income level was signiﬁcantly associated with the change in physical

disability (p<0.05). Women with the lowest income levels were more likely to

109

report physical disability worsened from twelve to eighteen months while women
with the highest incomes reported improvement in their physical disability. Also
of interest was the degree to which changes in physical disability status
accompanied changes in perceived health status reports from twelve to eighteen
months. Despite reporting physical disability status changes from twelve to
eighteen months, these women did not report statistically signiﬁcant changes in
perceived health status evaluations.

In similar fashion, no association between change in physical status
(worsen, stay the same, resolve) and the following variables were noted for the
prostate cohort: cancer stage, treatment, race, income level, physical nature of
job as measured by shift length and percent of shift seated, occupation type, sick
leave beneﬁt, age or baseline perceived health status. Also, no signiﬁcant
associations were observed between changes in perceived health reports with
disability status change among the prostate cohort.

When evaluating the relationship between disability reporting and
employment status at twelve months, a statistically signiﬁcant association was
noted. Individuals not identiﬁed to have physical disability at twelve months were
more likely to remain employed at twelve months (93%) compared to not working
(4%) or leaving the workforce altogether (3%), (p<0.01). Women who did report
disability were more evenly split with only 1/3 remaining employed, 30% not
being employed and 37% leaving the workforce. Similar patterns were observed
at the eighteen month timeframe. Fewer individuals with disability worked

compared to individuals without disability (96% vs. 24%).

110

The relationship between disability and employment among the men was
similar to what had been observed in the women. A majority (85%) of men
without disability were more likely to be employed at twelve months compared to
those who did not work (6%) or left the workforce (8%). However, only 42% of
those with a disability were employed while 33% did not work and 25% left the
workforce. These trends were continued at eighteen months. Most of the men
without disabilities (97%) worked compared to just half of those with disabilities

(50%).

4.1.3 Comparison of Cognitive Disability

Cognitive disabilities decreased from twelve to eighteen months for both
cohorts in a similar fashion to the physical disabilities although the decline from
9% to 5% was only statistically signiﬁcant for women (p<0.05). Again, women
experienced signiﬁcantly greater cognitive disability compared to men, which is
likely a result of the same factors identiﬁed previously (i.e. greater activity
limitation, participation restriction and aggressive treatment). Table 29 displays
the proportion of individuals in each cohort having cognitive disability per the ICF

deﬁnition (activity limitation with participation restriction).

 

 

 

Table 29: Cogitive Disability Re porting
Variable 12 months 18 months p-value p-value
(paired)
Breast 39 (9.3) 20 (4.9) 0.03* 0.02*
Prostate 7 (2.9) 2 (0.9) 0.19" 0.25"

 

 

 

 

 

 

 

“Statistically signiﬁcant at p<0.05
"Fisher Exact Test used due to cell size<5

111

Physical disability was more prevalent than cognitive disability in the
breast and prostate cohorts. This relationship is contrasted to the relative
proportion of required activity classiﬁcation where cognitive skills were in more
demand than physical skills. Women experienced statistically signiﬁcantly
greater physical disability compared to cognitive disability at twelve months
(p<0.01). However, although more physical disability was reported at eighteen
months compared to cognitive disability by the women, the difference was not
statistically signiﬁcant.

Similarly, men reported more physical disability than cognitive disability at
both twelve and eighteen months. As the women experienced, the proportion of
physical disability was statistically signiﬁcantly greater than the proportion of
cognitive disability at twelve months (p<0.05). The difference between these was

no longer statistically signiﬁcant at eighteen months.

4.1.3.1 Cognitive Disability Status Change

Fewer individuals complained of cognitive disability over the duration of
follow-up compared to physical disability. Thirty-six women experienced a
change in cognitive disability status between twelve and eighteen months.
Nearly half (47%) beneﬁted from an improvement in this type of disability by the
eighteen-month interview. Fourteen individuals (39%) had no change while 5

(14%) declined.

112

Five men complained of cognitive disability at twelve or eighteen months
however none worsened over time. Most (60%) improved while the remainder
was unchanged.

No associations between changes in cognitive status for the breast cohort
and the following variables were detected: cancer stage, treatment, income, race,
physical nature of job, occupation, sick leave, age, or baseline perceived health
status. Like the pattern observed for physical disability status change, the
evolution of cognitive disability status was unaccompanied by signiﬁcant changes
in perceived health status.

Similarly, the prostate cohort had no apparent associations between
cognitive disability status change and cancer stage, treatment, income, race,
physical nature of job, occupation, sick leave, age, or baseline perceived health
status. No association between cognitive disability status change and change in
perceived health status was observed for the prostate cohort.

Breast cancer participants with cognitive disability were less likely to be
employed compared to participants without cognitive disability at twelve months
(33% vs. 90%, p<0.01) and eighteen months (40% vs. 94%, p<0.01).

The relationship between disability and employment among the men was
similar to what had been observed in the women. A majority (84%) of men
without disability were more likely to be employed at twelve months compared to
only 43% of those with a cognitive disability (p<0.01). These trends continued at

eighteen months with neither (0%) of the two individuals identiﬁed with a

113

cognitive disability working at eighteen months compared to 96% of those without
a cognitive disability (p<0.01).

4.2 Study Question 2: Disability Prevalence Disparity Based on
Definition

4.2.1 Hypothesis 2: Comparison of Disability Prevalence

The hypothesis regarding variation in prevalence estimates of disability
depending on deﬁnition was written as:
0 H2: Cancer related physical and cognitive disabilities will be more
prevalent at twelve and eighteen months using an activity limitation
deﬁnition of disability compared to an ICF deﬁnition requiring

activity limitation and participation restriction.

4.2.1.1 Physical Disability

Activity limitation as a measure of disability results in greater proportions
of participants having the label disability compared to a deﬁnition requiring a
combination of activity limitation with participation restriction. This pattern is true
for physical tasks at both twelve and eighteen months. The difference in
proportions for physical tasks at both timeframes was statistically signiﬁcant at
p<0.01. These results were unchanged when the samples with paired data
available were used. Table 30 summarizes the proportions obtained using each

deﬁnition for the cohorts.

114

 

Table 30: Physical Disability Prevalence Comparisons Among the Cohorts

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Activity Activity p-value p-value
Limitation Limitation & (paired)
Participation
Restriction
Breast
12 Months 176 (59.5) 51 (17.2) <0.01** <0.01"
18 Months 94 (36.3) 25 (9.6) <0.01** <0.01"
Prostate
12 Months 44 (28.6) 12 (7.8) <0.01" <0.01“
18 Months 23 (17.2) 6 (4.5) <0.01" <0.01“

 

“Statistically signiﬁcant at p<0.01

4.2.1.2 Cognitive Disability

Again, the deﬁnition used to establish disability signiﬁcantly affects the

reported prevalence. An activity limitation deﬁnition provides a higher estimate

than the ICF deﬁnition. Both the breast and prostate cancer cohorts experienced

statistically signiﬁcant decreases (p<0.01) in reported disability between the

estimate calculated by the activity limitation deﬁnition and the estimate compared
to the ICF deﬁnition, regardless of timeframe. This relationship held when the

individuals with paired data were tested separately. Table 31 demonstrates the

reported prevalence estimates for the cohorts based on disability deﬁnition.

115

 

Table 31: Cognitive DisabilityPrevalence Comparisons Amorg the Cohorts

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Activity ICF: Activity p-value p-value
Limitation Limitation & (paired)
Participation
Restriction
Breast
12 Months 136 (34.3) 39 (9.3) <0.01" <0.01**
18 Months 78 (22.0) 20 (4.9) <0.01“ <0.01“
Prostate
12 Months 28 (12.1) 7 (2.9) <0.01“ <0.01“
18 Months 14 (7.0) 2 (0.9) <0.01" <0.01”

 

 

 

 

 

"Statistically signiﬁcant at p<0.01

4.3 Study Question 3: Association of Factors with Disability

Although two hypotheses were established to address the contextual
factors of interest, the variables categorized as personal or environmental were
included in the regression models simultaneously. Of note, the employer
accommodation variable was eliminated from the regression models after initial
data runs and no longer appears as a covariate. Although the variable appeared
to inﬂuence both types of disabilities initially, the conﬁdence intervals were very
large. Therefore, the hypotheses are reiterated here yet the results presented
will discuss both categories concurrently. All modeling procedures were
successful for the breast cancer cohort. Furthermore, all four modeling
equations had signiﬁcant (p<0.01) chi-square values for the models containing
the variables compared to a null model. The null model is that which states there

is no effect of the independent variables on the dependent variable (133).

116

The disability models for the prostate cohort were problematic due to small
sample sizes. When disability was deﬁned to include activity limitation and
participation restriction, a mere dozen were identiﬁed to experience either the
physical or cognitive outcome. This provided unstable results with very large
conﬁdence intervals. Therefore, the multivariate analyses for the prostate cohort

are not reported.

4.3.1 Hypothesis 3: Inﬂuence of Personal Factors

The third hypothesis sought to assess the following:
0 H3: Personal factors that will place individuals at risk for physical
and cognitive disability include minority race, low annual household

income, chemotherapy and later staged disease.

4.3.2 Hypothesis 4: Inﬂuence of Environmental Factors

The last hypothesis sought to assess the following:
0 H4: Environmental factors (associated with employment) that will
place individuals at risk for physical and cognitive disability include
blue-collar employment, less sedentary job, and low job

satisfaction.

4.3.3 Physical Disability

4.3.3.1 Twelve Month Regression Results
Race, annual household income and treatment were not signiﬁcantly

associated with the presence of disability at p<0.05. Moreover, these variables

117

failed to achieve sufﬁcient power (<0.20) to be retained in the ﬁnal model. The
remaining personal factor (disease stage) was found to be signiﬁcantly
associated with the development of disability. As expected, in-situ disease was
associated with less physical disability compared to women diagnosed with local
disease.

Environmental factors had more impact on physical disability than
personal factors when holding variables constant. Shift length and percent of
time sitting represented the physical nature of the job. Both variables failed to
meet criteria of p<0.20 to be retained in the model. The job satisfaction variable
also was not included in the ﬁnal model. However, blue-collar occupation was
associated with a higher likelihood of developing physical disability compared to
white-collar occupation. Women in a blue-collar occupation were twice as likely
as women in white-collar occupations to develop physical disability at twelve
months. Another variable that had not been identiﬁed a priori for the hypothesis
had a signiﬁcant inﬂuence on physical disability. Women without sick leave as a
beneﬁt experienced twice the risk of physical disability. Occupation and
availability of paid sick leave were correlated with a larger proportion of white-
collar workers having paid sick leave available compared to blue-collar workers.
Table 32 shows the odds ratios and associated p-values for the variables that
remained in the model, i.e. those variables that had p-values <0.20 during the

stepwise process.

118

Table 32: Regression Results for Physical Disability at Twelve Months Among
Breast Cancer Cohort (n=279 observations retained)

 

 

 

 

 

 

 

 

 

Contextual Variable I Odds Ratio P-value 95% Cl
Personal
Never married 0.3 0.11 0.1 — 1.3
Some college 0.5 0.08 0.2 — 1.1
ln—situ disease 0.1 <0.01” 0.0 — 0.4
Environmental
Blue collar occupation 2.2 <0.05* 1.1 — 4.2
No sick leave 21 <0.05* 1.2 — 4.0

 

 

 

 

 

*p<0.05, **p<0.01
Reference categories include: married, college degree, locally staged disease,
white-collar occupation, and having sick leave

The failure of treatment to inﬂuence disability was considered more
carefully. Cancer stage and treatment were strongly correlated with women
diagnosed at later stages undergoing more aggressive therapy. Therefore, the
logistic regression was re-run dropping the cancer stage variable to assess the
possible contribution of treatment. This resulted in treatment having a signiﬁcant
association with physical disability at twelve months. Speciﬁcally, women
undergoing chemotherapy were 3.2 times more likely to experience physical
disability compared to women having just surgical therapy as displayed in Table

33.

119

Table 33: Regression Results for Physical Disability at Twelve Months Among
Breast Cancer Cohort — Cancer Sta e Eliminated (n=279 observations retained)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual Variable Odds Ratio P-value 95% CI
Personal
Never married 0.3 0.12 0.01 — 1.4
High school diploma or less 1.8 0.11 0.9 — 3.6
Chemotherapy 3.2 <0.01" 1.4 — 7.3
Environmental
Blue collar job 1.7 0.13 0.8 - .5
No sick leave 2.0 <0.05* 1.1 — 3.9

 

 

 

 

 

*p<0.05, **p<0.01
Reference categories include: married, college diploma, surgery, white collar job
and having sick leave
4.3.3.2 Eighteen Month Regression Results

At eighteen months only one variable, sick leave, signiﬁcantly inﬂuenced
physical disability. Once again, race, income, physical nature of job and job
satisfaction failed to have a signiﬁcant relationship with physical disability.
Occupational classiﬁcation just missed the cut-off for the stepwise procedure
(p=0.2065). Stage did not have an impact at this time and treatment just missed
the cut-off to be considered signiﬁcant. Although not statistically signiﬁcant, the
odds ratio associated with chemotherapy hinted at a three-fold increase in
likelihood of physical disability at eighteen months. Marital status remained in
the model yet was not signiﬁcant. At eighteen months, nothing appeared to
reduce the probability of developing physical disability. However, not having sick
leave was associated with a four-fold increase for the development of physical

disability compared to women who had sick leave as a beneﬁt when controlling

120

for the factors in the model. Table 34 presents the odds ratios for the variables

remaining in the model with p<0.20.

Table 34: Regression Results for Physical Disability at Eighteen Months Among
Breast Cancer Cohort (n=221 observations retained)

 

 

 

 

 

 

 

 

 

 

Contextual Variable I Odds Ratio l P-value 95% Cl
Personal
Formerly married 2.3 0.08 0.9 — 5.8
Chemotherapy 2.9 0.05 1.0 — 8.3
Environmental
No sick leave 4.0 <0.01 1.6 — 10.5

 

 

*p<0.05, **p<0.01
Reference categories include: married, only surgical treatment and having sick
leave

Despite the inclusion of both cancer stage and treatment in the model for
physical disability at eighteen months, the treatment variable was able to come
through as a covariate to be retained. This time, the regression was re—run twice.
Once after eliminating cancer stage to see if treatment would become more
signiﬁcant. The second run eliminated treatment to see if cancer stage would
appear in the model. With cancer stage out of the model, the evaluation of the
model estimates and statistics were unchanged. With treatment out of the
model, absence of sick leave continued to exert a strong inﬂuence on the

development of disability. Table 35 displays the outcome of the model revised by

eliminating treatment.

121

Table 35: Regression Results for Physical Disability at Eighteen Months Among
Breast Cancer Cohort - Treatment Eliminated (n=222 observations retained)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual Variable Odds Ratio P-value 95% Cl
Personal
Formerly married 2.3 0.08 0.9 - 5.8
Chemotherapy 2.9 0.05 1.0 - 8.3
Environmental
Blue-collar 1.8 0.20 0.7 — 4.2
No sick leave 3.8 <0.01" 1.4 - 9.9

 

 

 

 

 

*p<0.05, **p<0.01
Reference categories include: married, surgery, white-collar and having sick
leave

4.3.4 Cognitive Disability

4.3.4.1 Twelve Month Regression Results

As hypothesized, race did signiﬁcantly impact cognitive disability at twelve
months. African-American women were nearly three times as likely to develop
cognitive disability compared to women whose race was classiﬁed as
White/Other. Income, job satisfaction and physical nature of job once again were
not retained in the model. Occupational classiﬁcation did not meet criteria to stay
in the cognitive disability model as it had for physical disability. Women with later
staged disease were three times more likely to develop cognitive disability at
twelve months compared to women with locally staged disease. Marital status,
education and vigorous activity were retained in the ﬁnal model yet not at
statistically signiﬁcant levels. Table 36 displays the results for the modeling

process.

122

Table 36: Regression Results for Cognitive Disability at Twelve Months Among
Breast Cancer Cohort (n=277 observations retained)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual Variable I Odds Ratio F-value L 95% Cl

Personal

African-American 2.7 <0.05* 1.2 — 6.1
Formerly married 1.8 0.14 0.8 — 4.0
Never married 0.1 0.13 0.0 — 1.8
High school diploma or less 1.9 0.08 0.9 — 4.2
No vigorous activity weekly 0.4 0.07 0.1 — 1.1
Regional/distant/unknown stage 3.0 <0.01** 1.4 — 6.4
Environmental

No sick leave 2.1 0.05 1.0 — 4.4

 

 

*p<0.05, “*p<0.01
Reference categories include: White/Other race, married, college diploma, some
vigorous activity weekly, locally staged disease and having sick leave

Cognitive disability at twelve months was signiﬁcantly related to cancer
stage when both cancer stage and treatment variables were included in the
model. With cancer stage present, treatment was not sufﬁciently strong enough
to be retained. As was done for physical disability, the cognitive disability
regressions were re-run after eliminating cancer stage to determine if treatment
would remain in the models. At this time, treatment did remain in the model and
was associated with a statistically signiﬁcant relationship with cognitive disability.
Women receiving chemotherapy experienced a six fold greater likelihood of
developing a cognitive disability compared to women receiving surgery when
controlling for other variables. The results from the revised model are presented

in Table 37.

123

Table 37: Regression Results for Cognitive Disability at Twelve Months Among

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breast Cancer Cohort — Cancer Stage Eliminated @374 observations retained)
Contextual Variable I Odds Ratio I P-value—I 95% CI
Personal
African-American race 2.7 <0.05* 1.2 — 6.0
Formerly married 1.7 0.15 0.8 — 3.6
Never married 0.2 0.15 0.0 — 1.8
High school diploma or less 1.7 0.14 0.8 — 3.7
No vigorous activity weekly 0.4 0.05 0.2 — 1.0
Chemotherapy 6.4 <0.01** 2.1 — 19.0
Environmental
No sick leave I 1.8 I 0.10 0.9 — 3.8

 

 

“p<0.05, “*p<0.01
Reference categories include: White/Other race, married, college diploma, some
vigorous activity weekly, surgery, and having sick leave
4.3.4.2 Eighteen Month Regression Results

At eighteen months, race no longer inﬂuenced cognitive disability. Annual
household income, physical nature of the job, job satisfaction and occupational
status failed to achieve sufﬁcient association to even be retained in the modeling
process. Treatment was not kept in the model in the face of cancer stage.
Women with more extensive disease experienced greater cognitive disability as
expected. Lack of sick leave again was found to signiﬁcantly affect disability.
Women who were formerly married were three times more likely to report
cognitive disability holding all factors constant compared to married women.

Table 38 presents the results.

124

Table 38: Regression Results for Cognitive Disability at Eighteen Months Among

Breast Cancer Cohort (n=221 observations retained)

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual Variable I Odds Ratio I P-value 95% Cl

Personal

Formerly married 3.6 <0.05* 1.4 — 9.7
High school diploma or less 2.2 0.11 0.8 — 5.8
No vigorous activity weekly 0.3 0.10 0.1 — 1.3
Regional/distant/unknown stage 3.0 <0.05* 1.1 — 8.1
Environmental

No sick leave 4.9 1 <0.01** I 1.7 — 14.1

 

*p<0.05, **p<0.01
Reference categories include: married, college diploma, some vigorous activity
weekly, locally staged disease and having sick leave

Cognitive disability at eighteen months had similar ﬁndings whether
cancer stage and treatment were included in the regression or just the treatment
variable. Once stage was eliminated, treatment was able to remain sufﬁciently
strong to have a statistically signiﬁcant association with cognitive disability as
shown in Table 39.

Table 39: Regression Results for Cognitive Disability at Eighteen Months Among
Breast Cancer Cohort — Cancer Stage Eliminated (n=304 observations retained)

 

 

 

 

 

 

 

 

 

 

 

 

 

Contextual Variable I Odds Ratio I P-value 95% or
Personal
Formerly married 2.9 <0.05* 1.1 — 7.7
High school diploma or less 2.1 0.14 0.8 — 5.5
No vigorous activity weekly 0.3 0.07 0.1 — 1.1
Chemotherapy 6.1 <0.05* 1.4 — 28.3
Environmental
No sick leave 4.3 I <0.01** 1.6 — 11.8

 

*p<0.05, **p<0.01
Reference categories include: married, college diploma, some vigorous activity
weekly, surgery, and having sick leave

125

CHAPTER 5: DISCUSSION

The results of this study indicate that physical and cognitive work-related
disabilities occur in women and men subsequent to breast and prostate cancer
diagnosis and treatment. Both types of disability are generally greater at twelve
months post-diagnosis than eighteen months although the declines are not
always statistically signiﬁcant. Furthermore, this study reinforces the critical
importance of an explicit deﬁnition of disability. Statistically signiﬁcant variation
was observed in physical and cognitive prevalence estimates within the same
population using two common deﬁnitions. Prior work on disability associated with
cancer has identiﬁed a variety of contextual factors that have been shown to
signiﬁcantly impact disability. Results obtained from the logistic regression
performed in this study did not always support the ﬁndings that have been

reported in other literature.

5.1 Study Question 1: Prevalence and Direction of Disability
Change

5.1.1 Hypothesis 1: Comparison of Physical and Cognitive Disability

The ﬁrst hypothesis of this study was to test the direction of disability
prevalence changes.
. H1: Cancer related physical and cognitive disabilities will decrease as the
timeframe from diagnosis increases from twelve months after diagnosis to

eighteen months after diagnosis.

126

In order to evaluate the changes in disability, attention must be focused on
the components that contribute to this measure. Firstly, matters regarding
changes in the composite activity requirement covariates were considered
followed by examination of the indicator variables (activity limitation and
participation restriction) that were multiplied to obtain disability estimates.
5.1.1.1 Comparison of Physical and Cognitive Activity Requirements

The composite task measures were established as dichotomous
measures. The interviews did not have the capacity to ensure the responses
provided to each particular task were mutually exclusive of the other tasks.
Therefore, no effort was expended to generate a composite variable that would
take into account the cumulative degree of physical activity required by each
participant. The dichotomous variable was selected to minimize “multiple
counting” on the outcome of interest although this decision could underestimate
the extent of physical activity required in the workplace. As would be expected
from the additive nature of the composite measure, the proportion of those who
had performed the composite physical task exceeded the proportions carrying
out each component task. This occurred because some individuals might not
admit to performing a task such as heavy lifting but state they do another
physical effort and therefore contribute to the numerator for the composite
physical task. Therefore, the prevalence of heavy lifting would not be as
substantial as the composite report.

Ultimately, no statistically signiﬁcant difference between the two time

periods was observed in the proportion of respondents performing physical tasks

127

for either cohort. Similarly, the proportion of participants requiring cognitive
activity did not change from twelve to eighteen months. This pattern is signiﬁcant
for the subsequent work that will focus on comparing disability prevalence
estimates. The fact that the overall demand of physical or cognitive tasks did not
vary indicates that resulting observations of documented decreases in disability
are not simply due to changes in the work tasks.

Physical and cognitive activities were not rare events for the jobs held by
our participants. Cognitive tasks were in much higher demand compared to
physical tasks with nearly all participants, 90% or more, acknowledging cognitive
activity. Physical activity on the other hand was required by only 60-70%.
However, results suggest that physical disability was more prevalent than
cognitive disability.

Regardless, disabilities occurring with these tasks signiﬁcantly affect
employment. Focused analyses reveal that persons with physical and cognitive
disabilities were statistically signiﬁcantly less likely to remain employed at twelve
and eighteen months compared to those who identiﬁed to not have a physical or
cognitive disability. Speciﬁcally, individuals with physical or cognitive disability
were represented to a greater degree among the cohort of individuals who left
the workforce at both time points. Of those that continued to work despite a
disability, the average workday was signiﬁcantly less than those without
disability.

Since physical disability was more prevalent, additional focus was directed

toward occupation characteristics. Speciﬁcally, it was expected that blue-collar

128

workers would be more likely to be engaged with physically demanding jobs.
The analysis of association between occupation and physical task requirement
revealed a statistically signiﬁcant association between these variables for both
cohorts. As expected, the proportion of blue-collar workers having to perform
physical tasks at their jobs was statistically signiﬁcantly greater than the
proportion of white-collar workers for the women and men. The results obtained
for the analysis of occupation with cognitive task revealed white-collar workers
carrying out this type of activity to a greater degree than blue-collar workers.

The physical nature of a job as measured by the reporting of physical
tasks was conﬁrmed by analyzing the sedentary aspect of a job. The percentage
of time spent sitting during their work day was calculated for each participant.
The percentage of time spent sitting was statistically signiﬁcantly different for
participants having to perform physical tasks compared to those who did not as
was expected. Women who did not have to perform physical tasks spent 63% of
their workday sitting compared to women who did undertake physical tasks and
only sat 38% of their workday, (ttest p<0.01). Men experienced similar results
with those not having physical tasks spending 68% of their day sitting compared
to 37% of those performing physical tasks (p<0.01).
5.1.1.2 Comparison of Domain 1: Activity Limitation

Work related activity limitations appear to signiﬁcantly decrease as
elapsed time from diagnosis and treatment increases for women treated for
breast cancer. Limitations occur more with physical tasks than cognitive tasks in

women despite the fact that jobs often require more cognitive skills. The

129

proportion of women complaining of physical limitation at eighteen months was
nearly half that reported at twelve months. A smaller proportion, just 1/3 of the
women complained of cognitive limitation at twelve months which decreased to
22% at eighteen months.

- Men diagnosed with prostate cancer experience different degrees of
limitations. Although activity limitations decrease as time from diagnosis
increases for men, only the decrease associated with physical limitations was
statistically signiﬁcant. The inability to detect a signiﬁcant change in cognitive
limitations could be due in part to study power issues. As did the women, men
enjoyed nearly one-half of the physical limitations at twelve months compared to
eighteen months. Cognitive limitations were uncommon compared to physical
limitations within the prostate cohort and the decrease observed in these
limitations did not achieve statistical signiﬁcance. The difference in limitations
was likely due to the different treatment regimens between men and women.
Men were more likely to receive surgical intervention only for their diagnosis.

The fact that women had higher proportions of physical limitations was
somewhat expected because more were required to perform physical tasks
compared to men. However, the cognitive job demands for both cohorts were
equal yet the women experienced greater limitations in performing these tasks
than the men. This may reﬂect the “chemo-brain” phenomenon that others have
acknowledged. A statistically signiﬁcant association between cognitive limitation
and chemotherapy at twelve and eighteen months (p<0.01) was documented

with chemotherapy patients complaining more about cognitive limitations.

130

Finally, attention was turned just to those individuals who had paired data
available for the twelve and eighteen month task and limitation measures. This
subgroup of women also experienced statistically signiﬁcant decreases in
physical and cognitive activity limitation. The results for the men with paired data
were unchanged. The physical limitation variable changed signiﬁcantly over time

while the cognitive limitation variable did not.
5.1.1.2.1 Supplemental Activity Requirement Comparisons - Physical Task

The results of this study indicated activity limitations decreased over time
even when the relative proportion of individuals reporting the task did not.
Several explanations for this pattern exist. One explanation relates to the fact
that all participants were asked to answer questions regarding activity and
limitation regardless of their employment status at the time of the interview. It is
possible that recall bias associated with leaving the workforce may have altered
perceptions from twelve to eighteen months. However, analyses of activity
requirement and activity limitations did not vary signiﬁcantly by employment
category (employed, not employed, retired/disabled/unable to work) or change in
employment category (leaving/joining the workforce vs. no change). Similar
results were obtained for the physical and cognitive tasks and limitations.

Therefore, another option was that sub-groups of respondents were being
lost preferentially. Figures 7 and 8 depict the proportion of breast and prostate
respondents, respectively, that performed each task at the various categories.
The breast cohort did have a statistically signiﬁcant (p<0.01) change in the

relative proportions responding they had to stoop all of the time or some of the

131

time. However, since both these categories would be collapsed together for the
physical task composite variable, the overall impact of this shift was trivial. No
statistically signiﬁcant changes were noted in the relative proportion of prostate

cohort respondents reporting each physical task time commitment option.

 

100%
80% —
60% —
40% ~
20% —

0% —

 
 

III/III
I ’I””’ .2“

   

12
Phys. effort

 

 

 

 

II All the time El Most time till Some time I?) No timeI

 

Figure 7: Breast Physical Tasks

132

 

100% ~—

t/Ir/I/ [III/III

III/Ill III/II” III/Ill Ill/Ill Ill/Ill] III/III

I’ll/ll I’ll/Ill Ill/Ill Ill/Ill [Ill/Ill III/Ill

o [Ill/ll III/III] Ill/Ill III/Ill [/1111]!

80 /o ‘I III/Ill Ill/(Ill Ill/III III/III Ill/Ill] Ill/Ill
ll/t/I/ III/”II Ill/Ill Ill/Ill till/III [Ill/ll

[Ill/ll III/III! III/Ill Ill/Ill I’ll/Ill I’ll

Ill/III III/III! Ill/III Ill/Ill III/Ill! It’ll/l

o III/III III/”II III/Ill [Ill/ll III/Ill! Ill/III

60 /o ' // Ill/Ill It Ill/III
Ill/Ill III/til I’ll/ll I’ll/Ill III/III

 

”III/II Ill/Ill ll/I/I/I
”III/II III/III

 

 

 

o , 3:15;:
I
40 /° III/Ill III/III
20% -
0% -

12 18 12 18 12 18
Phys. effort Heavy lifting Stoop

 

 

 

 

 

Ill All the time l3 Most time [III Sometime No time

 

Figure 8: Prostate Physical Tasks

The next step was to compare the individuals reporting on the individual
tasks making up the composite cognitive task variable. The breast cohort did
experience some differences in the analysis task. Similar to the stooping
requirement, the signiﬁcant difference occurred among respondents reporting the
all of the time and most of the time categories. Since both these categories
contribute equally to the composite variable, the overall impact on the composite
measure would not be critical. The prostate cohort again experienced no
statistically signiﬁcant difference in the relative proportion of individuals reporting
each time commitment. Figures 9 and 10 display the relative proportions of each

response for the breast and prostate cohort, respectively.

133

 

 

 

 

Concentrate Analysis Learn new things

 

 

 

IAII the time l'_'l Most time Some time E No time

 

 

Figure 9: Breast Cognitive Tasks

 

100%
80%
60%
40%
20%

0%

 

I

l

I

 

 

 

 

IIAII the time IZIMost time ISome time a No time

 

 

Figure 10: Prostate Cognitive Tasks

134

 

However, the change in reporting categories among the breast cohort for
the concentrate task was statistically signiﬁcant (p<0.01) and did involve
respondents changing between the category for some of the time and none of
the time. This obviously, could impact the composite cognitive measure.

A review of the individuals contributing to the composite cognitive
measure identiﬁed 32 total individuals who had the composite measure
determined solely due to the concentration task at either twelve or eighteen

months. Recall that this would occur in the event that the individual did not

 

perform analysis or learning new things as part of their job. Further examination I:
of these individuals showed that the majority, n=20 (62.5%), had no change in
their performing the task and no change in reported activity limitation of the task
between twelve months and eighteen months. Six individuals, 19%, reported no
change in conducting the activity but ratherjust a change in the activity limitation
associated with the composite task. This left just six individuals, 19%, who
changed their status regarding performing the activity.

In an effort to apply a conservative approach to determining the impact of
these individuals who contributed to the composite task solely on the basis of
their concentration task, all 32 were omitted and unadjusted analyses re-run.
Although the actual p-values of some of the comparisons changed, the overall
interpretation of statistical signiﬁcance for the comparisons: 1) the proportion of
the composite cognitive task requirement, 2) the proportion of the cognitive task
limitation and 3) the proportion of participation restriction at the p<0.01 or p<0.05

levels were unchanged. The ultimate interpretation of the change in cognitive

135

disability remained statistically signiﬁcant but achieved greater levels of
signiﬁcance, i.e. moved from p<0.05 to p<0.01. Since the presence of these

individuals appears to have no effect on the components of disability and dilutes

 

rather than establishes signiﬁcance of overall disability ﬁndings, the more
conservative approach selected was to retain them in order to maximize the

number of respondents available for the multivariate analyses.
5.1.1.2.2 Activity Limitations Compared to National Estimates

We see that the individuals in this study had less overall limitation when

 

Fan-tar. L:.- - -
i

evaluating activity limitation against the 50+% national estimate provided by Ness
and colleagues (2006) (1). At twelve months, the breast cohort approached this
limitation proportion for physical activities only. As mentioned above, the
limitation among the men was much less than this at both twelve and eighteen
months. This disparity may be a result of the participants enrolled in this study
versus the general US. population. However, it could be an example of the
difﬁculty in comparing estimates without clear deﬁnitions of the outcomes
provided. The general pattern of greater physical disability (as deﬁned by
limitation) compared to cognitive limitation reported in the literature was
supported by the ﬁndings of this study. Despite the greater number of
participants having to carry out cognitive tasks for their job, fewer experienced
limitations in these activities compared to the physical tasks.
5.1.1.3 Comparison of Domain 2: Participation Restriction

A smaller percentage of the breast cancer cohort, less than 20%,

documented any participation restriction with their work in contrast to the larger

136

proportions experiencing activity limitations. This prevalence did not change
markedly from twelve to eighteen months. Similarly, less than ten percent of
men reported participation restriction at twelve months. The small decrease to
7% at eighteen months did not reach statistical signiﬁcance at p<0.05. The
participation restriction experienced by the individuals in this study was less than
the 30% national estimate obtained by Ness (2006) despite these participants
being much closer to their diagnosis.

In contrast to the activity limitation covariates, the participation restriction
variable was constructed from other data. Measures such as weekly hours
worked, working full or part-time, and reductions in salary from baseline to twelve
and eighteen months were used to generate the restriction variable. The beneﬁt
of these questions was that response options to these made it possible to identify
occurrences that were related to the cancer diagnosis and treatment. However,
it is likely that additional participation restriction may exist within these groups yet
the surveys lacked speciﬁc questions that could tease out this concept more fully.
This may explain the discrepancy between these ﬁndings and Ness’ results.

When evaluating activity limitation and participation restriction as the
necessary components of disability, these data indicate participation restriction
seems to be the more difﬁcult criteria to ﬁll. With few exceptions, the prevalence
of activity limitation for each task (including the composite measures) exceeded
the proportion of women documenting participation restriction regardless of time.
A similar relationship between these domains was noted for the tasks (speciﬁc

and composite) within the male cohort. Due to the nature of the survey questions

137

and the need to create indicator variables for participation restriction, the validity

of activity limitation data is likely superior to participation restriction.

5.1.2 Comparison of Physical Disability

Physical disability prevalence among the cohort of women statistically
signiﬁcantly decreased from 13% at twelve months to 7% at eighteen months,
p<0.05. A positive conﬁrmation for the sub-sample of women that had paired
data available documented statistically signiﬁcant decreases in physical (p<0.01)
disability from twelve to eighteen months. Thus, the women in this study enjoyed
decreasing disability as time elapsed since their diagnosis and treatment. One
interpretation of this relationship may be that although they appeared to endure
greater levels of disability initially, their conditions were more transient In nature.

Men also experienced a decrease in physical disability from twelve to
eighteen months yet this ﬁnding was not statistically signiﬁcant. When restricted
to the paired samples, the decrease remained not statistically signiﬁcant. This
could suggest then, that although men appear to experience less disability
secondary to their disease process, the conditions may represent more
permanent conditions compared to those that women experience.

It should be noted that disability occurred in a smaller proportion of
individuals than either activity limitation or participation restriction as expected.
For these purposes, disability was constructed based on two domains, activity
limitation and participation restriction, as recommended by the ICF model. While
the model actually requires a third domain, the health condition measure, this

was considered inherent in both cohorts as a result of their cancer diagnosis.

138

 

 

Regarding the disability outcome, it was not surprising that women
experienced greater disability related to their cancer diagnosis and treatment
compared to men. Women experienced greater prevalence of each disability
component (e.g. activity limitation and participation restriction) compared to the
men so it is expected that they would incur greater burden of the combined
measure. Also, women undenrvent more aggressive treatment regimens than did
men leading to greater adverse outcomes. These patterns conﬁrmed that
documented by census data regarding increased disability in females compared

to males after ﬁfteen years of age.

5.1.3 Comparison of Cognitive Disability

The decline in cognitive disability from 9% at twelve months to 5% at
eighteen months was also statistically signiﬁcant within the breast cancer cohort
(p<0.05). Again, the sub-sample of women with paired data also experienced a
statistically signiﬁcant decrease in cognitive (p<0.05) disability.

Cognitive disability also declined in men from twelve to eighteen months.
However, this was not a statistically signiﬁcant change. When restricted to the
paired sample, the decrease remained not statistically signiﬁcant.

As noted with the physical disability outcome, the experience of the
women may indicate a transient condition that improves over time. However,
these data do not contain information over a longer duration that could address
the possibility of complete recovery. It may be that some low level of disability is

a permanent condition.

139

The men however, did not show any improvements in physical or cognitive
disabilities. The small prevalence estimates may represent a baseline disability
that is present as a permanent condition. Or, it may be that complete recovery is
possible but the men were not followed for a sufﬁcient length of time to measure
this. Yet another consideration is that the study lacked the power to detect a
decrease in disability.

5.1.4 Factors Associated with Changing Disability from Twelve to
Eighteen Months

The results indicate that for most participants, disability status was a
relatively stable condition from twelve to eighteen months post-diagnosis.
Among the breast cancer cohort experiencing a change in disability, just slightly
more documented an improvement compared to no change for both physical and
cognitive types of disabilities. A similar pattern was noted for the men. A
promising observation among the women is that those reporting disability
worsened represented less than twenty percent of those having a change. A
higher proportion of men reported physical disability worsened yet none claimed
of worsening cognitive disability.

Small cell sizes prohibited regression modeling although ordered logistic
regression was considered to evaluate the disability change over time as a
categorical variable with three levels. Therefore, some unadjusted analyses
were performed to evaluate if associations were present. Changes in physical
disability over time had no associations with cancer stage, treatment, income,

race, physical demands of job and occupation. The only variable to have an

140

 

 

association was baseline health status (p<0.05) among the women. A greater
proportion (43%) of individuals who worsened over time claimed to have only a
fair/poor perceived health status compared to those reporting a good or very
good/excellent health status. This suggests that women who have compromised
health at the time of a cancer diagnosis may be at risk for ongoing physical
disability. This relationship would support that published in the scientiﬁc p ,
literature.

Since the physical disability measure was referenced to the work

 

environment, the lack of association between changes in disability and ”II
occupational status was somewhat surprising. However, when evaluating
disability change against employment status change between twelve and
eighteen months, women did experience a signiﬁcant association. Women who
reported a worsening of their disability status were more likely to have left the
workforce compared to women who reported their disability status had improved.
Changes in cognitive disability over time in the breast cancer cohort were
associated with cancer stage and age at diagnosis. As expected, those who
worsened over time were more likely to be diagnosed with regional disease
(80%). These associations reinforce the importance of early screening and
detection. Individuals able to receive treatment before cancer has spread are
then less likely to have their disability status exacerbated over time. Again,
occupational status was not associated with cognitive disability change.
However, unlike physical disability change, the change in cognitive disability was

not related to employment change between twelve and eighteen months.

141

No signiﬁcant associations were observed between men who changed
their physical or cognitive disability status over time and the variables listed
above. It is probable the study did not have sufﬁcient power to test these

relationships.

5.2 Study Question 2: Variation in Reported Prevalence Due to
Disability Definition

5.2.1 Hypothesis 2: Comparison of Disability Prevalence

- H2: Cancer related physical and cognitive disabilities will be more
prevalent at twelve and eighteen months using an activity limitation
deﬁnition of disability compared to an ICF deﬁnition requiring activity
limitation and participation restriction.

These analyses show the importance of a clear deﬁnition for disability.
When framed only by activity limitation, prevalence estimates of physical and
cognitive disability were signiﬁcantly greater than estimates generated by a
framework of activity limitation with participation restriction when measured on
the same subjects. The differences between approximations were on the order
of three to four-fold, and were all statistically signiﬁcant at p<0.01. The variability
reported within the same study sample shows how difﬁcult it might be to compare
disability rates among these participants to other groups in the literature.

One concern with the activity limitation deﬁnition of disability is the
tendency to emphasize the pathologic disruption of the individual over the

capability of performing a social role. An individual may be unable to stand

142

 

 

upright to perform surgery in the conventional manner. However, this does not
necessarily mean the individual is incapable of being a surgeon. Peter A. Galpin,
MD, FACS, successfully worked with a wheelchair manufacturer to design an
instrument that could support his body so that he could operate at the same level

as the rest of a surgical team. Therefore, although he experienced an activity

 

limitation in that he is a paraplegic and cannot walk, he pursued environmental P
assistance that allows him to operate within the conﬁnes of this limitation. This
individual would likely be considered disabled by the activity limitation deﬁnition I
but not necessarily by the ICF deﬁnition including participation restriction. I

Continued debate and discussion regarding a suitable disability deﬁnition
and model is necessary by the communities focused on “disability”. Perhaps the
ICF deﬁnition is too restrictive while the activity limitation does not sufﬁciently
discriminate the concept of disability as the negative interaction between
someone and their contextual factors. In the absence of a well-deﬁned and
widely accepted deﬁnition, disability prevalence estimates will likely vary even
within similar conditions and under similar environments. Furthermore, informing
policy, legislation or social campaigns and trying to evaluate their effectiveness
will be difﬁcult at best without a clear focus on the condition and what it

represents.

5.3 Study Question 3: Association of Factors with Disability

These study questions sought to identify the relationship between
contextual factors with the presence of disability using logistic regression.

Robust standard errors were used to provide estimates in case the models were

143

not well speciﬁed. Because of small numbers, modeling was unsuccessful on

the prostate cohort.

5.3.1 Hypothesis 3: Inﬂuence of Personal Factors

The third hypothesis was interested in evaluating the association between
disability and personal characteristics. I.
0 H3: Personal factors that will place individuals at risk for physical and

cognitive disability include minority race, low annual household income,

 

'1.’ as...

chemotherapy and later staged disease.

5.3.2 Hypothesis 4: Inﬂuence of Environmental Factors

The last hypothesis was interested in evaluating the association between
disability and environmental (employment) characteristics.
0 H4: Environmental factors (associated with employment) that will place
individuals at risk for physical and cognitive disability include blue-collar

employment, less sedentary job, and low job satisfaction.

5.3.3 Physical Disability

5.3.3.1 Twelve Month Regression Results

Both personal and environmental contextual factors inﬂuenced the ﬁnal
model for physical disability at twelve months yet neither race nor income

speciﬁcally appeared to inﬂuence the presence of physical disability. The

144

environmental variables that did signiﬁcantly inﬂuence physical disability included
blue-collar occupation and lack of sick leave.

Women with in—situ disease were less likely to experience physical

 

disability compared to women with local disease. This relationship was

anticipated; more extensive disease would generally require more aggressive

 

treatment. r
The interaction of stage and treatment however was of concern. Of note,

the women in the in-situ group underwent signiﬁcantly less chemotherapy for

treatment than women in the local stage group (p<0.01). Therefore, early u

detection is critical not only to enhanced survival but appears to be associated
with decreased likelihood of physical disability as well. Although treatment had
been included as a variable in all the modeling processes, it only was retained in
the ﬁnal model for the long-term physical disability. Because of the correlation
between stage and treatment, the models were re-run without cancer stage.
Subsequently, treatment did have an impact on physical disability as expected.
Chemotherapy was signiﬁcantly associated with increased risk of physical
disability at twelve months and just missed the cut-off for increasing risk at
eighteen months.

Women in blue-collar occupations were at twice the risk for physical
disability compared to women employed in white-collar occupations. Similarly,
women without a sick leave beneﬁt were at greater risk of physical disability than

women who had sick leave.

145

Several variables selected to represent the physicality of a job were
included in the modeling process. These variables were the duration of a
working shift (in hours) and the percent of that shift spent seated. Neither
variable was sufﬁciently inﬂuential on physical disability to be retained in the
models during the stepwise process. The variable related to job satisfaction also
failed to remain in the model.

However, it is possible that the inﬂuence of a more physical job was
incorporated within other variables such as occupation type. Occupational status
was shown to inﬂuence the likelihood of physical disability at twelve months.
Women who were classiﬁed as having blue-collar occupations had signiﬁcantly
less time during their work shift spent seated compared to women in white-collar
occupations. Similarly, women in blue-collar occupations were at increased risk
for physical disability. Although the physical demands of the job were not
retained in the ﬁnal models, women in blue-collar jobs were shown to have more
physically demanding jobs than those in white-collar jobs. This may explain why
they experienced greater physical disability. The increased likelihood may also
be due to the more stringent scheduling that blue-collar jobs may require.
Women in white-collar professions may enjoy more ﬂexibility with scheduling.
Therefore, although they may be experiencing the same adverse effects, i.e.
fatigue, they might be able to modify their work hours to lessen the impact of
participation restriction.

A similar argument may be made for the lack of sick leave being

associated with increased likelihood of physical disability. If women are not able

146

to take time off from work to recuperate from the cancer treatments, their
physical capacity will be compromised. Women suffering from fatigue or pain are
less likely to approach physical activity with the same zeal as if they did not have
these conditions. This in turn, may adversely affect their job performance with
respect to physical demands.

The continued persistence of paid sick leave as a signiﬁcant inﬂuential
variable in disability is intriguing. Paid sick leave is generally associated with
positions that would be considered white-collar or high-paying and these
variables were included in the modeling procedures. However, paid sick leave
may be associated with blue-collar positions if the individuals enjoy union
membership that has bargained for this particular beneﬁt. Among the
participants in the breast cohort who were classiﬁed as blue-collar workers, only
1l3 reported they belonged to a union. However, when comparing the sick leave
beneﬁt of blue-collar workers who were in a union versus those who were not
union members, a statistically signiﬁcant association was observed with nearly
90% of union members having sick leave while only 40% of non-union members
had sick leave. When the model was re-run after including the union variable,
the sick leave variable had been replaced by the union variable as a signiﬁcant
predictor. As expected, those who were union members had less risk of physical
disability at twelve months compared to those who were not union members

(OR=0.29, 95% Cl: 0.11 - 0.71, p<0.01).

147

5.3.3.2 Eighteen Month Regression Results

Prolonged physical disability, i.e. that present at eighteen months, was
associated with lack of sick leave. Women without sick leave appeared to have
nearly four times the risk for long term physical disability compared to women
with sick leave as a beneﬁt. The associations of sick leave at twelve and
eighteen months suggest that the impact on physical disability increases over
time. Prolonged adverse effects are signiﬁcant when individuals do not have the
optimum time away to recover fully. Of note, women without sick leave were
employed in more physically demanding jobs with 39% of their time sitting
compared to 49% for women with sick leave (ttest p<0.01). The inﬂuence of
union membership was evaluated in light of the strong association noted
between sick leave and long term physical disability. At eighteen months, union
membership did not affect the relationship between sick leave and physical

disability and was not signiﬁcant itself.

5.3.4 Cognitive Disability

5.3.4.1 Twelve Month Regression Results

Personal variables were signiﬁcantly associated with the presence of
cognitive disability at twelve months. In contrast to physical disability, race was a
statistically signiﬁcant predictor of cognitive disability with African-American
women having nearly three times the risk of white/other women for cognitive

disability. This pattern supports what has been described in disability literature.

148

This association may be due to the increased representation of African-American
in the subgroup of women diagnosed at later stages.

Cancer stage also impacted cognitive disability at twelve months for
women with breast cancer. Women diagnosed with regional/distant or unknown
disease were three times more likely to have cognitive disability compared to
women with local disease. This relationship evolved as expected. Women with
more invasive disease had more extensive treatment regimens and presumably
more untoward effects. In this sample, the proportion of women with local
disease undergoing chemotherapy regimens was less than the proportion of
women with widespread disease.

Paid sick leave was not signiﬁcantly associated with cognitive disability
and just missed the formal cut-off value. Therefore, the model was re-run adding
the union variable into the model to examine what impact this would have.
Similar to what was observed for physical disability, union membership was
protective against cognitive disability (OR=0.26, 95% CI: 0.1 — 0.8, p<0.05) and
the sick leave variable was no longer retained in the model.
5.3.4.2 Eighteen Month Regression Results

At eighteen months, extensive disease continued to be associated with an
increased risk for cognitive disability. Along with this, lack of sick leave again
became signiﬁcantly associated with increasing risk of cognitive disability. This
suggests the importance of sufﬁcient time for recovery from cancer treatments.
Women able to take advantage of paid sick leave experienced not only less long-

term physical disability but also less long-term cognitive disability. Union

149

membership did not impact the relationship observed between sick leave and
long-term cognitive disability.

Marital status became an important variable for long-term cognitive
disability. Women who were separated, divorced or widowed were over three
times more likely to suffer cognitive disability compared to women who were

married during their cancer episode. This may reﬂect a lack of social support

 

from which to draw among women who were formerly married. Alternatively, this

may be in part due to available resources. Married women might not rely solely

 

I‘HLu-Jn. .- -

on their own employment for ﬁnancial or health care coverage. Therefore,
formerly married women may be less willing to risk job loss and continue working
without adequate recovery periods.

The association between chemotherapy and cognitive disability was
apparent when cancer stage was removed from the model. In fact, treatment
exerted the greatest inﬂuence on cognitive disability in the short and long-term
among the independent variables. This supports the notion of a “chemo-brain”
phenomenon about which some patients complain. These observed
relationships also support what has been described in the literature regarding
impaired cognitive function subsequent to chemotherapy. Since dosing was not
included as part of data collection, a dose-response evaluation could not be

conducted.

150

5.4 Limitations Associated with Project

5.4.1 Overview

Several limitations of this study must be acknowledged. The primary
issue arises from using an existing data set to answer different study questions.
A sampling scheme powered to answer these questions was not carried out.
Based on the disability estimates obtained from this sample and the available
participants, the power estimates ranged from 66% down to 13%. All the
available respondents remaining in the subject pool at twelve and eighteen
months were included in these analyses yet the disability deﬁnition required the
presence of both activity limitation and participation restriction which resulted in
small numbers having the outcomes of interest.

Furthermore, issues related to framing of interview questions must be
recognized. Since this study was not the primary purpose of the funded study,
the most direct manner of obtaining data related to participation restriction and

disability was not available. Instead, indicator variables were necessary to

represent these domains. Activity limitation questions were not without their own

problems. No examples of activity were provided to subjects and no

measurement of capacity for physical activity was available.

Another limitation is the geographic restriction of these results. The scope

of this study is work-related disability as reported by individuals residing in
Wayne, Oakland and Macomb counties. Individuals residing in rural areas may

experience disability in varying amounts and patterns.

151

 

 

In addition to the issue of the limitation of the catchment area for
participants, there are also concerns about the representativeness of these
respondents to the larger population that may be diagnosed with breast or
prostate cancer. Despite the advantages of this particular registry system, the
cases recruited to the study were largely White and of higher income levels. The
employment eligibility and telephone criteria may have resulted in differential
participation.

While a great deal of baseline data was available for the multivariate
analyses, some of the employment measures were only ﬁrst addressed at the
twelve-month interview. This calls into question whether these employment
variables were independent variables or dependent variables in light of the
disability outcome.

Lastly, with the exception of SEER provided stage and treatment
information, all data regarding job tasks, limitations and participation restriction
were based solely on self-report. Veriﬁcation of data provided against medical or

employment records was not possible.

5.4.2 Study Participants

By design, the participants in this study were at least 30 years old and
were employed at the time of diagnosis. The men were signiﬁcantly older than
the women, 55 years vs. 50 years (p<0.01). Since disability has been shown to
increase as age increases, this might indicate that the men would experience
greater disability than the women. However, national disability statistics also

claim that women suffer more disability than men beyond ﬁfteen years of age

152

which would then argue for increasing disability prevalence among the female
cohort in this study.

Comparisons of employment-related factors between men and women are
not always valid and a key feature of this study was the focus on disability with
respect to physical or cognitive work activities. Therefore, attempting to compare
the disability estimates in the women with those in the men may be inappropriate.
Consequently, this study is unsuitable for assessing comparisons in disability
based on gender.

As mentioned earlier, one of the rationales for selecting the MDCSS as a
recruitment pool was the large inclusion of African-Americans because of its
geographic location. Unfortunately, the LMOS was unable to recruit African—
Americans to the same degree. Consequently, the proportion of minority
participants in the disability study was less than 25%. Preferential enrollment of
white individuals may have occurred because of other participation requirements
that simultaneously selected for a higher socioeconomic status group, e.g.
having a telephone or being employed.

The participants were relatively well educated and a good proportion had
high-income levels prior to their diagnosis. Even twelve months after their
diagnosis, more than half of the men still enjoyed a household income level
greater than $75,000 annually. Changes in income level were not statistically
signiﬁcant for the men between baseline and twelve months. However, the
women did experience a statistically signiﬁcant shift between from the high to the

moderate level income during this time.

153

The race, income and education distributions observed in the study
sample suggest persons enjoying higher socioeconomic status were included.
Therefore, the disability prevalence estimates might not accurately reﬂect what
may be expected from these particular cancer conditions in a randomly selected
cohort. However, physical and cognitive work activities were believed to
correlate with socioeconomic condition so estimates of disability associated with
each category may provide some insight to the experience of individuals of
different socioeconomic status. Physical tasks were speculated to be associated
with manual labor (i.e. lower socioeconomic status) while cognitive tasks were
considered to be associated with higher socioeconomic conditions. An
unadjusted look at the relationship between occupation and requirement for
physical activity documented a statistically signiﬁcant relationship with more blue-
collar workers performing physical tasks compared to white-collar workers
regardless of male or female.

Since ﬁtness level has been associated with disability in other studies, we
were interested in describing measures of usual physical capacity for these
cohorts. Fitness level may affect disability by contributing to an individual’s
resilience to health insults. Conﬂicting information regarding ﬁtness levels of the
participants were obtained. No more than twenty percent of either cohort
participated in regular activity at least three times per week at baseline
regardless of intensity. This suggests that the participants were more sedentary.
However, more than 2/3 of both groups declared they worked up a sweat either

sometimes or always. This result could be interpreted to mean that respondents

154

did some form of vigorous activity and perhaps were not as sedentary as
originally believed.

One reason for this discrepancy might be the lack of objective time frame
in the sweat question. Alternatively, respondents may have inferred the regular

activity questions to be asking about recreational activity in nature while the

 

sweat question may have included activities that would not be considered b
recreational, such as yard work or physical labor. Because of these conﬂicting

results, the ability to interpret the baseline tolerance for physical activity was

questioned. Subsequent analyses failed to implicate a signiﬁcant association of :L

activity with physical disability yet the variable was retained at a non-signiﬁcant
level in the modeling equations for cognitive disability. The direction of inﬂuence
suggested women not engaging in vigorous activity had lower probability of
developing cognitive disability compared to women who did work up a sweat.
Since the goal of the study was to evaluate disability in previously healthy
breast and prostate cancer patients, some measures of pre-existing medical
conditions were important to include. Individuals with multiple medical conditions
may experience less tolerance for the untoward effects of treatment compared to
persons with none or just a few. This may result in a greater likelihood of
subsequent disability. Several variables, including count of comorbid conditions
and perceived health status were taken as proxies of health. Results indicated
that both the breast and prostate cohorts enrolled in the study were relatively
healthy. Less than ten percent of each reported having three or more comorbid

conditions and over 70% claimed to have excellent or very good health status.

155

The disability estimates may then be underestimated in this population relative to

the general population.

5.4.3 Evaluation of Selection Bias

Because of the limited recruitment of minorities, a formal evaluation of the
potential selection bias that may have occurred in this source study population
was conducted in Bradley, et al’s 2005 methods paper (119). This analysis was
performed by comparing demographic and clinical elements on enrolled
participants to all the potentially eligible subjects identiﬁed in the MDCSS. In
addition, census data on variables at the block group level were obtained for all
potential subjects. Refer to Appendices C and D for details of the Selection Bias
Examination reported in that article.

These comparisons revealed that potential subjects who refused to be
screened for eligibility were somewhat older than the enrolled participants by
approximately two years for both cohorts. Also, the proportion of African-
American subjects were statistically signiﬁcantly (p<0.01) greater in the “unable
to contact” subgroup compared to the enrolled populations for breast and
prostate cancer patients. More African-Americans were also included in the
“eligible but refused” group among the prostate cases. Similarly, the “unable to
contact” and “ineligible” breast and prostate groups had statistically signiﬁcantly
(at least p<0.05) higher proportions of individuals with unknown cancer stage.
This was hypothesized to reﬂect individuals who may have not had their cancer
staged because of advanced disease at diagnosis, thus these individuals may

have been in poorer health (119). Since the outcome of interest is disability, the

156

omission of potentially sicker individuals would result in underestimating the

overall prevalence.

5.4.4 Disease Aspects of Cohorts

Over seventy percent of both cohorts were diagnosed early, in-situ or local
stage, in their disease process. Early diagnosis has been associated with
improved survivability thanks to aggressive treatment. This could impact
disability in two directions. If disability is associated with later staged cancer that
ultimately suffers greater mortality, the overall prevalence of longer-term cancer-
related disability in the population could remain low. However, if the aggressive
treatment causes some disability, the improved survival of persons with early
disease may result in increasing disability prevalence. Cancer stage and
treatment were strongly correlated in this study.

Data regarding cancer treatment were obtained from two data sources.
The SEER data set provided information on stage and treatment. The
participants also provided data regarding treatment. Limitations with SEER and
patient reporting have been documented and the degree to which each data
source was inaccurate for these participants cannot be deﬁned with these data.
As a result, the decision was made to accept all reports of treatment regardless
of source.

The breast cohort’s experience reveals the aggressive nature of treatment
provided even for early stage disease. Sixty-six percent of women with local
disease received chemotherapy compared to only four percent of women with in-

situ disease. The more extensive the treatment regimen, the more plausible it is

157

disability may result. In contrast, treatment of choice for the men was largely
surgical intervention without adjuvant therapy. Over 60% of the men received
just surgery. In fact, the proportion of men not receiving treatment of any form

exceeded the proportion undergoing chemotherapy.

5.4.5 Employment Aspects of Cohorts

Most, 2/3 of total participants, were employed in white-collar occupations.
This is not unexpected in light of the income levels that were reported among the
respondents. This may impact the disability prevalence estimates and therefore
these ﬁndings may not be representative of the entire employed population that
may be diagnosed with cancer. The mean shift lengths at baseline were 8 and
8.8 hours for the breast and prostate cohort, respectively, with nearly half the
shift being sedentary. The greater proportion of workers with these
characteristics (white collar, sedentary work) may result in greater cognitive
disability and less physical disability being reported in these cohorts.

More than 80% of both cohorts reported they had a supportive employer
who accommodated their illness. Thankfully, this stands in contrast to
discrimination experiences that persons with cancer historically faced. The
support of their employer may have had an inﬂuence over respondent loyalty
since a minority (less than 40%) of individuals stated they would be interested in
a new job if they could be guaranteed similar beneﬁts. At least 80% expressed
dedication to their work in the form of acknowledging being a perfectionist about

work or being personally involved in their work. Half of the breast and prostate

158

cancer individuals stated satisfaction in their life came from their job yet job
satisfaction ratings did not vary by employment status change.

The dedication individuals’ feel towards their work may impact disability in
both directions. Individuals who are dedicated to their job and obtain life
satisfaction from their work may minimize potential disability in order to maintain
the connection. Alternatively, those who are perfectionists may be more aware
of the limitations imposed by their disease and treatment and the inability to carry

on in their normal fashion resulting in higher reports of limitations.

159

 

 

CHAPTER 6: CONCLUSIONS

6.1 Implications & Recommendations

These results suggest options for further investigation. The ﬁrst research
question was interested in the level of physical and cognitive disability with work
tasks in these participants. Disability estimates were less than national averages
yet the comparability of the involved populations and results are questionable.
The individuals in this study were more likely to enjoy higher socioeconomic
status than the general population; this provides one rationale for the
discrepancy.

Evaluated only within the conﬁnes of the experience of the enrolled
individuals, physical disability appears to be more widespread than cognitive
disability for both men and women. This relationship occurs in spite of the fact
that physical tasks are not as prominent in the work setting.

Women appear to experience greater disability with their work tasks at
both time frames than men experience with their own jobs. However, women
enjoy a statistically signiﬁcant decline in reported physical and cognitive disability
from twelve to eighteen months. Therefore, the study was successful at rejecting
the null hypothesis that no difference would exist between the time frames for the
women. The men’s experience appears to be more stable and this pattern likely
reﬂects differences in the treatment protocols between breast and prostate
cancer. However, it may also be that this pattern could be an artifact of small cell
size. The null hypothesis of a decrease in prevalence could not be rejected

based on these data for the men.

160

 

 

These ﬁndings hint that physical demands may be more problematic than
cognitive and therefore, a reasonable primary target for accommodation or
rehabilitation efforts. Unfortunately, this study lacked the detail regarding the
degree to which activity was limited. Additionally, no information was available

regarding the nature of accommodation or rehabilitative services to which these

 

women had access. Perhaps the decreases observed reﬂect the availability of n
these services. Or, perhaps these occurred despite having no accommodation
or rehabilitation services. If the latter is the case, this offers some hope that a

'Efl

baseline long-term disability could be reduced to nearly zero with appropriate H

 

supports. Variables of this sort would be important to include in subsequent
studies.

The second study question focused on exploring the potential difference in
disability estimates within the same population using two common disability
deﬁnitions. Many authors report disability without explicitly deﬁning the term and
this presents obstacles to comparing published rates. One common reference
for disability is some measure of activity limitation. This deﬁnition tends to
minimize the importance of participating in a social role. The ICF deﬁnition
includes both activity limitation and participation restriction as necessary domains
for disability.

As anticipated, the disability prevalence estimates using activity limitation
were signiﬁcantly greater than the prevalence estimates using the ICF deﬁnition.
The variation was on the order of three to four-fold. These results rejected the

null hypothesis that no difference in prevalence estimates would be obtained

161

from the two deﬁnitions. Differences of this magnitude indicate prevalence
comparisons between studies could be suspect. However, the manner in which
participation restriction was computed in this study may play a role in these
discrepancies as well.

Lastly, the third study question sought to evaluate the inﬂuence of
suspected contextual factors on the presence of disability. The contextual
factors were categorized as personal or environmental. The environmental

factors were all employment-related variables.

 

Contrary to expectations, income level was not associated with disability.
Race played a lesser role than expected. African-American women only were
nearly three times more likely to experience cognitive disability compared to
Caucasian/Other women holding variables such as income or education
constant. Marital status played a role in cognitive disability at eighteen months.
Women who were formerly married were over two times as likely to have
cognitive disability compared to married women. This may reﬂect a lack of
support structure in dealing with the disease.

Cancer stage and treatment had the strongest impact of the personal
factors with both physical and cognitive disability. Women diagnosed with earlier
staged disease experienced less disability than those diagnosed later. Also,
those receiving chemotherapy experienced more disability than those undergoing
surgery. Stage and treatment were correlated and both wouldn’t remain in the
models together. This emphasizes the importance of early detection. Screening

is essential to early diagnosis where the treatments may be less severe.

162

Environmental factors had an impact on work-related disability as
expected. However, the degree of physical activity at a job and overall job
satisfaction did not seem to inﬂuence disability. Occupational category was only
inﬂuential for physical disability with blue-collar workers having more disability
than white-collar workers. Blue-collar workers had more physical jobs than
white-collar workers so it is likely the physical demands did play a role. The
availability of sick leave was consistently related to physical and cognitive
disability. Women who did not have a sick leave beneﬁt were more likely to

suffer. Legislation such as FMLA provide workers some assurance of

 

maintaining a job should they be off work for a medical condition yet no mandate
exists for having this be paid time off. Women who do not have sick leave may
not be able to afford lost wages associated with time off. Therefore, they have

no option but to compromise their health to continue to work.

 

Increasing attention to disability coupled with the increased detection of
cancer in a working population provided motivation for this study. The results
and limitations associated with this study identify elements that future work
should address. While the results obtained here support relationships
documented in existing literature (e.g. more physical disability, impact of stage
and treatment), other well-established associations (e.g. age, income level or
ﬁtness capacity), were not observed. This argues for the need to replicate this
work with a focus on these research questions and with sufﬁcient enrollment for

robust analyses.

163

6.2 Summary

The ﬁndings of this study may not be representative of the disability
experience that all employed persons with a cancer diagnosis may face. There
is evidence that the cohorts recruited for this study were mostly white and from a
higher income bracket. Furthermore, the disability prevalence estimates vary
within the same sample depending on the domain(s) used to establish disability.
Signiﬁcant work is yet to be done to completely describe the disability experience

of cancer patients, the outcome of which may be used to suggest interventions to

 

ameliorate disability. This work is offered as the ﬁrst step in establishing a
baseline disability framed by a comprehensive theoretical model.

These results provide evidence as to the importance of deﬁning disability

 

clearly when disability is an outcome of interest. The disability prevalence
estimates reported here were much lower than those published elsewhere.
Readers must use caution when evaluating disability estimates and making
comparisons. The outcome should be clearly deﬁned by the author. This study
used the deﬁnition of disability according to the ICF theoretical framework and
incorporated elements of activity limitation along with participation restriction.
These data showed that disability is a concern for employed breast and
prostate cancer patients at least within eighteen months after diagnosis.
Physical disabilities were more prevalent than cognitive disabilities for women
and men and occurred at a lower level than national disability estimates. Also,
women enjoyed signiﬁcant decreases in prevalence while men did not. It is

reasonable to suspect that women may suffer more disability due to the more

164

aggressive treatment regimens they undergo yet the affects may not necessarily
be permanent conditions. On the other hand, the stable estimates for men might
suggest that their distress is indeed permanent. Over time, a permanent
condition could be more problematic for the population of prostate cancer
survivors.

Based on these data, we were unable to test the determinants of disability
in the prostate cohort. The problems encountered serve as testament to the
problems encountered in using secondary data sources. The available sample
was simply not sufﬁcient for the very low prevalence estimates that appear to
afﬂict the population.

One goal of this study was to identify variables that were signiﬁcantly
associated with disability and might be amenable to intervention. Personal
factors such as income or age were not signiﬁcant in these participants.
However, African-American women were at higher risk for cognitive disability.
Much more important to the outcome of disability was cancer stage and
treatment. Since treatment is reﬂective of cancer stage, the importance of
screening and early detection must be stressed. Women diagnosed with in-situ
disease had signiﬁcantly less disability. Regarding variables such as race,
income level or education, it is important to provide educational materials
promoting screening and early detection using culturally competent methods.

The most signiﬁcant environmental factor for inﬂuencing presence of
disability was the absence of paid sick leave as a beneﬁt. Adequate recovery

time is necessary to adjust to the affects of cancer treatments. Individuals who

165

cannot afford to take this time and work through treatment may experience
longer lasting impacts. These results suggest legislation like F MLA do not go far
enough in protecting workers when they are diagnosed with a serious illness.
Employers may wish to consider the cost of paid sick leave versus the costs
associated with lower productivity to re-evaluate their beneﬁt packages.
Important elements related to disability were not included in this study
since it relied on existing data. Future studies should include mechanisms to

track the nature and intensity of limitations and restrictions. The dichotomous

 

measures used here are suitable for pilot investigations but complete analyses of
disability should strive for more discriminating data. It would also be necessary
to include information on the types of accommodations and/or rehabilitation

services patients use during their recovery period. Once these are accurately

 

described, the focus could turn to developing methods to reduce the associated

disability.

166

APPENDICES

167

Appendix A: Work Related Activity Limitation Questions

 

Question

 

Eligible Responses

 

Physical Tasks

 

My job requires lots of physical effort

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

My job requires lifting heavy loads.

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

My job requires stooping, kneeling, or
crouching

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

 

My job requires me to keep up with the
pace set by others

 

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

Cognitive Tasks

 

My job requires intense concentration or
aﬂenﬁon.

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

My job requires me to analyze data or
information.

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

My job requires that I learn new things.

 

All or almost all of the time

Most of the time

Some of the time

None or almost none of the time

 

Asked After Each Task if All/Most/Some response

 

 

Has cancer or its treatment limited or

interfered with your ability to do this part of

your job?

 

Yes
No

 

 

168

1"!" I

Appendix B: Participation Restriction Questions

 

Question

Eligible Responses

 

What is the main reason you (the
patient) do not work?

Own illness/cancer

 

Was this decrease (in hours)
related to your cancer?

Yes - due to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes - due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - due to all of the above

 

Some people work part-time
because they cannot ﬁnd full-time
work or because business is poor.
Others work part-time because of
family obligations or other personal
reasons. What is your MAIN
reason for working part-time?

Health/medical limitations due to cancer
and/or its treatment

 

Was this decrease (in hourly
earnings for main job) related to
your cancer?

Yes - clue to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes - due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - due to all of the above

 

 

Was this decrease (in total
earnings for main job) related to
your cancer?

 

Yes - due to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes - due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - due to all of the above

 

169

 

Appendix B: Participation Restriction Questions (continued)

 

Question

Eligible Responses

 

Was this decrease (in hourly
earnings for other jobs) related to
your cancer?

Yes - due to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes - due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - due to all of the above

 

Was this decrease (in total
earnings for other jobs) related to
your cancer?

Yes - due to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes - due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - due to all of the above

 

 

Was the reason you stopped
working related to your cancer?

 

Yes - due to treatment

Yes - due to cancer symptoms

Yes - due to other reasons

Yes — due to treatment and symptoms
Yes - due to treatment and other reasons
Yes - due to symptoms and other reasons
Yes - clue to all of the above

 

170

 

.Fo.v: .0 0.00.330 00..o::0 Ea... 00:85:90 .00..0..0.0
00.0:00 ...... 0:0 mov: .0 0.00.330 00:9:0 80:. 00:85:90 .00..0..0.0 00.0:00 .... .0 5:28 5 0.00.330 .80.:8

o. 030 .05. 0:. 0. 00:09:00 00.0 0.03 o 5:28 :. Lo. 00.:3800 0.00.330 0.0.9.05 0:0 0 5:28 332:. N :E28

0. 00:0:E8 0:03 .0 5:28. 0.00.030 00_.o::w ..mvuc. 0.59.05 0:0 ..m..u:. 00:00:00 .o: ..muc. 0.0020: .8715
.00.:8 o. 0.00 .0: .0 Fuc. .0030: :0.0.0>:: ”0.:E00 .0005 0:. 5 0.00 000: 0500.5. ..mruc. 0.59.05 ... Fuc. 00:00:00
.o: ...u:. 0.0020: ..tuc. .00.:8 o. 0.00 .0: ..nnc. .0020: :0.0.0>:: 0.:E00 .0005 0:. 5 0.00 000 0500.5. .0000
0:0 000: .0. 0000580: :. 56:0 0.0 00002090: .000 :0. 0000580: :. 550:0 .:o..0.>00 0:00:00qu 00.02

 

 

......0. 0. ..:. N. .00. N .....00: 0. 6.4. a ....0. t 585::
...N. 0 .00. 0 .00. 0 .00. 0 ...... a .00. 0. 20.0.:
.30. 00 6.3. as .000. 00 6.40. as .000. as .000. 00. 08.00:
.600. 00 .000. 00 6.00. 00 .600. .0 .00.... E 6.0... 000 :83
.000. .0 .0. E. 00 .20. a 6.00. as .... E. .0 .~.0~. 0m. 5.0 s
000.0
0.:0:0.I
1:0: £005
65. .0 .00.. 00 .30. .0 2.6.00. 40 6.3. v0 63. so. c2000.:
6.00. 00. .00.. 0: .000 .0 ......»0. 3 ... E. 0: 6.2. ~00 .050 .222,
000K
.00.“. .80.
.....000. 00.00 ..:..0. 2.00 00.00 00.00 .60. 0. .0 60.». 0.00 .00. 000 e002
S~uz 00....2 Buz 00Tz 00Tz 0001.2 .853 .0020
.0030:
8.0.0.3:
05....
00030.. cow—«EOE.
00:00:00 00020: «00.80 o. “03:00 02.0.50
0.0.9.2.. .02 0.0.9.0 use 82 5.0.0.3: cz 0.8.300

.0.

.0.

.0.

.0. .u.

...

0.83:0 03.9.0 60.058: 9 02380 0.86:0 00:220. 1 30.300 00.0 5.08.00 ..0 005:3

171

 

 

‘I n. 0. .. . Its—L.

.....v: .0 080.030 00:98 :8... 00:85:90 .00..0..0.0

00.080 .... 0:0 mov: .0 0.00.300 00:98 .8... 8:005:90 .00..0..0.0 00.0:00 ... .0 5:28 5 0.00.330 .80.:8

0. 030 .o:.. 05 o. 88:58 00.0 0.0.... o 5:28 5 ..o. 00.:880 0.00.330 0.59.05 0:0 0 5:28 88.5 N 58.8

o. 00.0:Eo0 0.0.5 .. 8.28. 0.00.2.0 00._o.:m. ..mmnc. 0.59.05 .305 00:00.00 .0: ..muc. 0.0020. ...vnc. 80.80
o. 030 .o: ..o.u:. .0050. :0.0.0>:: 0.9.80 0.0.09: 05 5 0.00 000. 0500.5. .085 0.59.05 ..muc. 00:00.00

.8 ..muc. 0.0050. ..omuc. .00.:8 o. 0.00 .0: ..mu:. .0020. :0.0.0..:: .0.:E00 0.0.09: 05 5 0.00 000 0500.5. .0000
0:0 000. .0. 0000588: 5 550:0 0.0 0000.808: .000 .0. 0000588: 5 56:0 .:o..0.>00 0:00:50qu 00.02

 

 

 

:60. 00 .0... 0 .50. 0 2.6.0.. 00 6.0. 0 .0... 00 .3655
...0. 0 .00. 0 0 .... 0 .00. 0 .0.... . 20.0.:
.60.. 0.. .... .. ... ... .0. 00 .60.. 0. .5... 0. .40.. .0 .2200:
.000. 00. .0.... 00 .00.. .0 .000. 00. .00.. 00 .00.. 4.0 .83
0 6.0. . 0 0 0 0 5.0 s
000.0
0.88....

6.00. .0 .00.. 00 ....00. .0 :6... .0 .000. 00 .000. ... ...c: 0.00.0

...00. .0. .000. 00 ..0..0. 00 ......0... 0. .000. .0 .0.... 000 .050 .0....3
momm—

..00.
:60. 0.00 .65.. 0.0 0.00 .000. 0.0 .000. ...0 .000. .000 .00. 000 :00...
000.12 00.uz ~0uz ....uz ~0uz e082 8:8 3502....
.0030.—
:0.0.0..::
0:0
20.00.53...
00050.. 00030.. «00.80 8 «03:00 00:28

0.0.5.2.. 08020082 0.0.0.0. 0.0302 5.0.0:... 02 0.8.300
.0. .0. .... .0. .0. ...

.002580. 8.00:.E0xm 00E 8.80.00 .0 x.0:0::<

172

 

 

...o.v0 .0 0.00.300 00:25 So... 00:85:90
.00..0..0.0 00.0:00 ...... 0:0 mo.v0 .0 0.00.300 00:95 80... 8:005:90 .00..0..0.0 00.0:00 ... 0.00.300 ...00.:00 0.0.00
.0:.. 0:. 0. 00.0068 00.0 0.03 0.00300 0.0.9.05 0:0 0 5.5.00 :000.:. 0 5:58 0. 00.09.80 0.02. 0.00.300 00__0.:m.
..0>0. 000.0 ..00.0 0:. .0 0.0 0000.50.00 .050 _.< ..0>0. .00.. 000:00 0:. .0 0. 3.050 30.00 000.:00.00 0: .. ”00.02

 

 

0.00 0.00 0.00 0.0 0.00 0.00 0000.00

0. .0 0. .0 0.00 00.. 0.... 0.00 000.0.0v 0:0 000.0000

0.00 0.0. 0.... ..000 0.0. ..v. 000.000v
.0.... 0800:. 50:00:01
00. 0.. 0.0 :0... 0.0 ... ...... 0.0.60 30.00
00.0. .000.
.0.. 0.3. .0.. 0.0 0.3 0.3. 9.. ... .0.. .o 00.0.0505
..00 .00 0.00 0.00 0.00 .....0 .0._8 00.0
.0.... .:0E>0.0:.m
0.0 .. .0 0.00 0.0 0.00 0.00 .000... .0 00.000 002.00
0.00 0.00 0.00 0.00 0.00 ..00 000.8 00.00
0.00 0.00 0.00 0.00 ....0 0.00 00.0.0.0 .860 .0...
0.0. ..0. ..0. ..000 0.... 0.0. 00.0.0... .8000 :0... oz
...... .0050... 808000.
0.0.. 0.0.. 0.0.. .0.. .0.. 0.3V ...... 0.v 00.0.0.0
0.00 0.00 0.00 0.00 ..00 0.00 0.00.0
0.0 0.00 ..00 0.00 0.00 0.00 0030...; 00.0.83 .8320
00.. 0.0.. 0.0.. .000 0.00 0.00 0050.2
.0. 00.0.0 0.00....
0.0 ..0. ....0 ....00 0.00 ..00 ..8..0E<...8...<
0.00 0. .0 0.0. .000 0.... 0.0. 00.0.00. 0.22.
.0. 0.0.0.0280”.

0000.0. .00.:00
00:00.00 0000.0. .00.:00 0: 0000.0. 00..0.:0
0.0.0.3... .02 0.0.0.0. o. 0.00 .oz 5.0.0.... 0.00.000

.0. .0. .... .0. .0. ... .8000 .0020

 

03.9.5 3.002030. 0. 00:0.Em .0 =00...me00 000200 - 0.0.300. 00.5 =0...00.0m. ..Q 35:230.

173

...o.va um £00.33. 3:95 E0... 8:853?

60.6.66 .86ch AL new nova a 98.35 3:25 Eat 02.8536 .8232». 8.0ch 8 £80.33 .3350 B mBm
8.... m5 2 8.388 02m 995 $3.53 29.96:. was m 5:28 595.5 N 5:28 8 8.3:.8 2m; mﬁmﬁam 3:25
._m>m_ anew x003 m5 6 9m mmmﬂcmoaa 650 __< ._m>m_ “on: 96:8 9: «m w. >tm>oa .523 $3523 9:. .862

 

 

174

3.0. 3m 3m 53 3.0, gm 832m

Now o? 93 «on m. 5 m9. 833v .25 25.8?

new .92 «.2 .58 at 69 ooodmav
§v 9:8... 29.320:
99 3 :2: :oi .3: 3 30. 2.2.8. 26.8
8.8 .092
:8 m. 5 man :98 m. 5 N8 9.. c. .8 8 83358:
98 v.9 tom m. B 9% v.3. 5.8 35
3°. EmE>anm
0.3 2% m8 :5 9mm 2% .29.. .o 85% $2.8
v.3 3w «mm .F. a New New $2.8 9:8
EN ...8 98 new 3N 3N «522.0 .853 :9...
m8 mi wt :9 5 E: 02 «62...... .838. 3... oz
...... 8:838
q? a? a? 0.3 09. mi. ...... 8v c2220
.35 m8 v. 5 «.8 :8 Q8. 2me
met oi 0.2 .0.: v.9 a? 838.; .3983 3.95
E: «.8 9mm .99 v. B In 8.5.2
3°. «3% 3.5.2
3m 3m mam 2a.; new mum c8_.me<.cmo.=<
08 SK «.8 3m 3» ﬁt x0298: 628$: £55
30. 2.0.2.0.8mm

60969. Hum—:00
£09320» Danae «05:00 O: ..mmaoh 00:93.0

«3.2.2.. 3: 6.99m. 3 2% 3: 5.0%»... 38.3..» .850 338....

 

$035200. coauﬁﬁaxm mam cowoﬂmm .0 52.0%?

 

 

BIBLIOGRAPHY

175

10.

11.

12.

BIBLIOGRAPHY

Ness KK, Wall MM, Oakes M, Robison LL, and Gurney JG. Physical
performance limitations and participation restrictions among cancer
survivors: a population-based study. Ann Epidemiol 2006; 16(3): 197-205.

Rolland JS. Anticipatory loss: a family systems developmental framework.
Fam Proc 1990; 29: 229-244.

Franche RL, and Krause N. Readiness for return to work following injury r
or illness: conceptualizing the interpersonal impact of health care,
workplace, and insurance factors. J Occup Rehabil 2002; 12(4): 233-256.

Tate DG, and Pledger C. An integrative conceptual framework of disability:
new directions for research. American Psychologist 2003; 58(4): 289-295.

 

Kopec JA. Concepts of disability: the activity space model. Soc Sci Med
1995; 40(5): 649-656.

Costa AJL, Massad E, Ortega NRS, and Araujo A. Perception of disability
in a public health perspective: a model based on fuzzy logic. Int J Med
Inform 2004; 73: 647-656.

Dahl TH. International classiﬁcation of functioning, disability and health: an
introduction and discussion of its potential impact on rehabilitation
services and research. J Rehabil Med 2002; 34: 201-204.

U.S. Department of Health and Human Services. The Surgeon General‘s
Call to Action to Improve the Health and Wellness of Persons with
Disabilities. Available at http:/Mwwsurgeongeneral.gov, last accessed
1012005.

Kearney PM, and Pryor J. The international classiﬁcation of functioning,
disability and health (ICF) and nursing. Journal of Advanced Nursing
2004; 46(2): 162-170.

lmrie R. Demystifying disability: a review of the international classiﬁcation
of functioning, disability and health. Sociology of Health and Illness 2004;
26(3): 287-305.

Silvers A. Inequality, abnormality, and the americans with disabilities act.
The Journal of Medicine and Philosophy 1996; 21: 209-224.

Lollar DJ. Public health and disability: emerging opportunities. Public
Health Rep 2002; 117: 131-136.

176

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Hoffman B. Cancer survivors at work: a generation of progress. CA
Cancer J Clin 2005; 55(5): 271-280.

Sasser AC, Rousculp MD, Birnbaum HG, Oster EF, Lufkin E, and Mallet
D. Economic burden of osteoporosis, breast cancer, and cardiovascular
disease among postmenopausal women in an employed population.
Women's Health Issues 2005; 15: 97-108.

Kitamura Y, Ohno U, Kasahara S, Murata K, Sugiyama H, Oshima A,
Tsukuma H, Ajiki W, and Hasegawa T. Statistical estimation of the number
of breast cancer patients with disabilities resulting from surgery. Breast
Cancer 2005; 12(2): 130-134.

Rosenfeld B, Roth AJ, Gandhi S, and Penson D. Differences in health-
related quality of life of prostate cancer patients based on stage of cancer.
Psycho-Oncology 2004; 13: 800-807.

Hayes SC, Battistutta D, Parker AW, Hirst C, and Newman B. Assessing
task "burden" of daily activities requiring upper body function among

women following breast cancer treatment. Support Care Cancer 2005; 13:
25-265.

Rijken M, vanKerkhof M, Dekker J, and Schellevis FG. Comorbidity of
chronic diseases. Qual Life Res 2005; 14: 45-55.

Short PF, Vasey JJ, and Tunceli K. Employment pathways in a large
cohort of adult cancer survivors. Cancer 2005; 103(6): 1292-301.

Erickson P. A health outcomes framework for assessing health status and
quality of life: enhanced data for decision making. J Natl Cancer Inst
Monogr 2004; 33: 168-177.

Giesler RB, Given B, Given CW, Rawl S, Monahan P, Burns D, Azzouz F,
Reuille KM, Weinrich S, Koch M, and Champion V. Improving the quality
of life of patients with prostate carcinoma: a randomized trial testing the
efficacy of a nurse-driven intervention. Cancer 2005; 104(4): 752-62.

Bradley CJ. The need for online information on the economic

consequences of cancer diagnosis, treatment, and survivorship. J Med
Internet Res 2005; 7(3): e29.

Hodgson NA, and Given CW. Determinants of functional recovery in older
adults surgically treated for cancer. Cancer Nurs 2004; 27(1): 10-6.

Voogd AC, Ververs JMMA, Vingerhoets AJJM, Roumen F MH, Coebergh
JW and Crommelin MA. Lymphoedema and reduced shoulder function

177

 

 

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

as indicators of quality of life after axillary lymph node dissection for
invasive breast cancer. Br J Surg 2003; 90: 76-81.

Soothill K, Morris SM, Harman J, Francis B, Thomas C, and Mclllmurray
MB. The signiﬁcant unmet needs of cancer patients: probing psychosocial
concerns. Support Care Cancer 2001; 9: 597-605.

DeLisa JA. A history of cancer rehabilitation. Cancer 2001; 92: 970-4.

Marciniak CM, Sliwa JA, Spill G, Heinemann AW, and Semik PE.
Functional outcome following rehabilitation of the cancer patient. Arch
Phys Med Rehabil 1996; 77: 54-7.

Rietman JS, Dijkstra PU, Debreczeni R, Geertzen JHB, Robinson DPH,
and DeVries JAAP. Impairments, disabilities and health related quality of
life after treatment for breast cancer: a follow-up study 2.7 years after
surgery. Disability and Rehabilitation 2004; 26(2): 78-84.

McWayne J, and Heiney SP. Psychologic and social sequelae of
secondary Iymphedema: a review. Cancer 2005; 104(3): 457-66.

Ganz PA, Desmond KA, Leedham B, Rowland JH, Meyerowitz BE, and
Belin TR. Quality of life in long-term, disease-free survivors of breast
cancer: a follow-up study. J Natl Cancer Inst 2002; 94(1): 39-49.

Yap KPL, McCready DR, Narod S, Manchul LA, Trudeau M, and Fyles A.
Factors inﬂuencing arm and axillary symptoms after treatment for node
negative breast carcinoma. Cancer 2002; 97(6): 1369-75.

Clark JA, Talcott JA. Symptom indexes to assess outcomes of treatment
for early prostate cancer. Med Care 2001; 39(10): 1118-1130.

Staff l, Salner A, Bohannon R, Panatieri P, Maljanian R. Disease-specific
symptoms and general quality of life of patients with prostate carcinoma
before and after primary three-dimensional conformal radiotherapy.
Cancer 2003; 98(11): 2335-43. ’

Steineck G, Helgesen F, Adolfsson J, Dickman PW, Johansson JE, Norlen
BJ, Holmberg L. Quality of life after radical prostatectomy or watchful
waiting. N Engl J Med 2002; 347(11): 790-796.

Gotay CC, Holup JL, Pagano l. Ethnic differences in quality of life among
early breast and prostate cancer survivors. Psycho-Oncology 2002; 11:
103-1 13.

178

 

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

Spelten EF, Sprangers MAG, Verbeek JHAM. Factors reported to
inﬂuence the return to work of cancer survivors: a literature review.
Psycho-Oncology 2002; 11: 124-131.

Juster FT, Suzman R. An overview of the health and retirement study. The
Journal of Human Resources 1995; Supplement: S7-856.

Ahles TA, Saykin AJ, Furstenberg CT, Cole B, Mott LA, Skalla K, Whedon

MB, Bivens S, Mitchell T, Greenberg ER, Silberfarb PM. Neuropsychologic
impact of standard-dose systemic chemotherapy in long-term survivors of

breast cancer and lymphoma. J Clin Oncol 2002; 20(2): 485-493.

Carmona RH. Improving the Health and Wellness of Persons with
Disabilities: A Call to Action. Am J Public Health 2005; 95(11): 1883.

Churchill RE. Disabled or enabled. Am J Public Health 2005; 95(11):
1887-1888.

 

Clarke P, George LK. The role of the built environment in the disablement
process. Am J Public Health 2005; 95(11): 1933-1939.

Okoro CA, Balluz LS, Campbell VA, Holt JB, Mokdad AH. State and
metropolitan-area estimates of disability in the United States, 2001. Am J
Public Health 2005; 95(11): 1964-1969.

LaPlante MP. Health conditions and impairments causing disability.
Disability Statistics Abstract 1996; 16: 1-4.

Kraus L, Stoddard S, Gilmartin D. Chartbook on disability in the United
States. An lnfoUse Report 1996; Available at
http://www.pascenter.orgldisabilitydata/3_3.php, last accessed 9/22/2005.

Nordenfelt L. Action theory, disability and ICF. Disability and Rehabilitation
2003; 25(18): 1075-1079.

Arthanat S, Nochajski SM, Stone J. The international classiﬁcation of
functioning, disability and health and its application to cognitive disorders.
Disability and Rehabilitation 2004; 26(4): 235-245.

Wang PP, Badley EM, Gignac M. Exploring the role of contextual factors
in disability models. Disability and Rehabilitation 2006; 28(2): 135-140.

Chronister JA, Johnson EK, Berven NL. Measuring social support in
rehabilitation. Disability and Rehabilitation 2006; 28(2): 75-84.

179

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

Bouknight RR, Bradley CJ, Luo Z. Correlates of return to work for breast
cancer survivors. J Clin Oncol 2006; 24(3): 345-353.

Verbrugge LM, Jette AM. The disablement process. Soc Sci Med 1994;
38(1): 1-14.

Day R, Ganz PA, Costantino JP, Cronin WM, Wickerham DL, Fisher B.
Health-related quality of life and tamoxifen in breast cancer prevention: a
report from the national surgical adjuvant breast and bowel project P-1
study. J Clin Oncol 1999; 17(9): 2659-2669.

Bradley CJ, Neumark D, Bednarek HL, Schenk M. Short-term effects of
breast cancer on labor market attachment: results from a longitudinal
study. Journal of Health Economics 2004; 24(): 137-160.

Bradley CJ, Neumark D, Luo Z, Bednarek H, Schenk M. Employment
outcomes of men treated for prostate cancer. J Natl Cancer Inst 2005;
97(13): 958-965.

 

'7 I}. ‘ .

Caban ME, Nosek MA, Graves D, Esteva FJ, McNeese M. Breast
carcinoma treatment received by women with disabilities compared with
women without disabilities. Cancer 2002; 94: 1391-1396.

Lippman ME, Hayes DF. Adjuvant therapy for all patients with breast
cancer. J Natl Cancer Inst 2001; 93(2): 80-82.

Roberts AA. The labor market consequences of family illness. J Mental
Health Policy Econ 1999; 2: 183-195.

American Cancer Society, Inc. Cancer facts and ﬁgures 2006. Atlanta:
American Cancer Society: 2006.

Steiner FJ, Cavender TA, Main DS, Bradley CJ. Assessing the impact of
cancer on work outcomes. Cancer 2004; 101(8): 1703-1711.

Hamilton LC. Statistics with Stata: updated for version 9. Canada:
Thomson Brooks Cole: 2006.

Yabroff KR, Lawrence WF, Clauser S, Davis W, Brown ML. Burden of
illness in cancer survivors: ﬁndings from a population-based national
sample. J Natl Cancer Inst 2004; 96(17): 1322-1330.

Sussman MB. Sociology and rehabilitation. Washington DC: American
Sociological Association: 1965.

180

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

World Health Organization. International classiﬁcation of functioning,
disability and health: ICF. Geneva: World Health Organization: 2001.

National Cancer Institute. Cancer survivorship research 2004. Available at
http:/ldccps.nci.nih.gov/ocs, last accessed 10/11/2006.

Mellette SJ. The cancer patient at work. CA Cancer J Clin 1985; 35(6):
360-73.

Spelten ER, Verbeek JHAM, Uitterhoeve ALJ, Ansink AC, van der Lelie J,
de Reijke TM, Kammerijer M, de Haes JCJM, Sprangers MAG. Cancer,
fatigue and the return of patients to work - a prospective cohort study.
European Journal of Cancer 2003; 39: 1562-1567.

Bradley CJ, Bednarek HL. Employment patterns of long-term cancer
survivors. Psycho-Oncology 2002; 11: 188-198.

 

Hinman MR. Factors inﬂuencing work disability for women who have
undergone mastectomy. Women and Health 2001; 34(2): 45-60.

Stewart DE, Cheung AM, Duff S, Wong F, McQuestion M, Cheng T, Purdy
L, Bunston T. Long-term breast cancer survivors: conﬁdentiality,

disclosure, effects on work and insurance. Psycho-Oncology 2001; 10:
259-263.

Bushunow PW, Sun Y, Raubertas RF, Rosenthal S. Adjuvant
chemotherapy does not affect employment in patients with early-stage
breast cancer.. J Gen Intern Med 1995; 10(2): 73-76.

van der Wouden JC, Greaves-Otte JG, Greaves J, Kruyt PM, van
Leeuwen 0, van der Does E.. Occupational reintegration of long-term
cancer survivors. J Occup Med 1992; 34(11): 1084-1089.

Greenwald HP, Dirks SJ, Borgatta EF, McCorkle R, Nevitt MC, Yelin EH.
Work disability among cancer patients. Soc Sci Med 1989; 29(11): 1253-
1259.

Satariano WA, DeLorenze GN. The likelihood of returning to work after
breast cancer. Public Health Rep 1996; 111: 236-241.

Razavi D, Delvaux N, Bredart A, Autier P, Bron D, Debusscher L,
Strychmans P. Professional rehabilitation of lymphoma patients: a study of

psychosocial factors assocaited with return to work. Support Care Cancer
1993; 1(5): 276-278.

181

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

Andrykowski MA, Curran SL, Studts JL, Cunningham L, Carpenter JS,
McGrath PC, Sloan DA, Kenady DE. Psychosocial adjustment and quality
of life in women with breast cancer and benign breast problems: a
controlled comparison. J Clin Epidemiol 1996; 49(8): 827-834.

Peteet JR. Cancer and the meaning of work. General Hospital Psychiatry
2000; 22(3): 200-205.

Avis N, Stellato R, Lawrence R. Medical and psychosocial factors affecting
quality of life among newly diagnosed breast cancer patients age 50 and
under. ASCO Annual Meeting Proceedings. 1999.

 

Schag CAC, Ganz PA, Wing DS, Sim MS, Lee JJ. Quality of life in adult
survivors of lung, colon and prostate cancer. Quality of Life Research
1994; 3(2): 127-141.

 

Chirikos TN, Russell-Jacobs, A, Cantor AB. Indirect economic effects of
long-term breast cancer survival. Cancer Practice 2002; 10(5): 248-255.

 

Malone M, Harris AL, Luscombe DK. Assessment of the impact of cancer
on work, recreation, home management and sleep using a general health
status measure. J R Soc Med 1994; 87(7): 386-389.

Ganz, PA. Current issues in cancer rehabilitation. Cancer 1990; 65: 742-
751.

Houts PS, Yasko JM, Kahn SB, Schelzel GW, Marconi KM. Unmet
psychological, social, and economic needs of persons with cancer in
Pennsylvania. Cancer 1986; 58: 2355-2361.

Bradley CJ, Neumark D, Luo Z, Bednarek HL. Employment-contingent
health insurance, illness, and labor supply of women: evidence from
married women with breast cancer. National Bureau of Economic
Research; Working Paper 11304. Available at
http://www.nber.org/papersNV1 1304, last accessed 11/3/2006.

StataCorp. Stata statistical software: release 9. College Station, TX:
StataCorp LP; 2005.

Schultz PN, Beck ML, Stava C, Sellin RV. Cancer survivors work related
issues. AAOHN 2002; 50(5): 220-226.

U.S. Department of Justice. Americans with Disabilities Act of 1990.
Available at http:/lwww.ada.govladahom1.htm, last accessed 11/6/2006.

182

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

U.S. Department of Health and Human Services. New Freedom Initiative.
Available at httpzllwww.hhs.govlnewfreedom/, last accessed 11/6l2006.

Hernandez B, Balcazar FE, Keys CB. Disability rights: attitudes of private
and public sector representatives. Journal of Rehabilitation 2004; 70(1):
28-37.

Centers for Disease Control and Prevention Healthy People 2010. Health
People with Disabilities Fact Sheet. 2004; Available at httpzll0-
www.cdc.gov.mill1 .sjlibrary.org/ncbddd/factsheets/DH_hp2010.pdf, last
accessed 11l6/2006.

Centers for Disease Control and Prevention Disabilities and Health Team.
Disability and Health in 2005: Promoting the Health and Well-Being of
People with Disabilities. 2005; Available at http://0-

www.cdc.gov.mill1 .sjlibrary.org/ncbddd/factsheets/Disabi|ity_Health_AtAG|
ance.pdf, last accessed 11/6/2006.

Centers for Disease Control and Prevention, Marge M (editor). Healthy
people 2010 disability objectives: private sector and consumer
perspectives: conference proceedings and recommendations. 1998.
Available at http://0-
www.cdc.gov.mill1.sjlibrary.org/ncbddd/dh/publications/hpmarge.htm, last
accessed 11/6/2006.

Committee on Cancer Survivorship: Improving Care and Quality of Life,
National Cancer Policy Board, Maria Hewitt, Sheldon Greenﬁeld, and
Ellen Stovall (Eds). From cancer patient to cancer survivor: lost in
transition. Washington, DC: National Academy Press: 2005.

Committee on a National Agenda for the Prevention of Disabilities:
Andrew M. Pope and Alvin R. Tarlov (Eds). Disability in America: toward a
national agenda for prevention. Washington, DC: National Academy
Press: 1991.

Committee on Assessing Rehabilitation Science and Engineering, Edward
N. Brandt, Jr., Andrew M. Pope (Eds). Enabling America: assessing the
role of rehabilitation science and engineering. Washington, DC: National
Academy Press: 1997.

Krahn G. Changing concepts in health, wellness and disability, in Institute
on Disability and Development (Ed). Portland, OR: 2003.

Cherry Engineering Support Services, Inc. Federal Statutory Deﬁnitions of
Disability: Prepared for the lnteragency Committee on Disability Research.

183

 

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

2003. Available at httpzllwww.icdr.us/documents/deﬁnitions.htm, last
accessed 12l28/2006.

McNeil JM, Binette J. Prevalence of disabilities and associated health
conditions among adults — US 1999. MMWR 2001; 50(7): 120-125.

Federal lnteragency Forum on Age-Related Statistics. Older Americans
2004: Key Indicators of Well-being. Washington DC: US Government
Printing Ofﬁce: 2004.

Yang Q, Rasmussen SA, Friedman JM. Mortality associated with Down’s
Syndrome in the USA from 1983 to 1997: A population-based study.
Lancet 2002; 59(3): 1019-1025.

Centers for Disease Control and Prevention. Summary health statistics for
the U.S. population, National Health Interview Survey. National Vital
Health Statistics Report 2004; 10(220).

Henry J. Kaiser Family Foundation. Americans' views of disability. A
Kaiser Health Poll Report. 2004. Available at
http:l/www.kff.org/spotlight/disability/index.cfm, last accessed 12l29l2006.

Kinne SK, Patrick DL, Doyle DL. Prevalence of secondary conditions
among people with disabilities. Am J Public Health 2004; 94(3): 443-445.

Center J, Beange H., McElduff A. Mentally retarded individuals: a
population at risk? Medicine and Science in Sports and Exercise 1998;
23(5): 586-593.

Rimmer JH., Braddock D, Pitetti KH. Research on physical activity and
disability: An emerging national priority.Medicine and Science in Sports
and Exercise 1993; 28(11): 1366-1372.

Committee on Disability in America: A New Look. Workshop on disability
in America: a new Look-summary and background papers. Washington,
DC: National Academy Press: 2006.

American Cancer Society. Prostate Cancer Survival Rates. 2006.
Available at
httpzllwww.cancer.org/docroot/CRl/content/CRl_2_2_6x_Prostate_Cancer
_Survival_Rates.asp?sitearea=, last accessed 119/2007.

Oliver M. The Politics of Disablement. London: MacMillan: 1990.

World Health Organization. International classiﬁcation of impairments,
disabilities, and handicaps. Geneva: World Health Organization: 1980.

184

 

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

Bickenbach J., Chatterji S., Badley E., Ustun T. Models of disablement,
universalism and the international classiﬁcation of impairments, disabilities
and handicaps. Social Science and Medicine 1999; 48: 1173-1187.

Wieneke MH, Dienst ER. Neuropsychological assessment of cognitive
functioning following chemotherapy for breast cancer. Psycho-Oncology
1995; 4: 61-66.

Kleinbaum DG. Logistic Regression: a self-learning text. New York:
Springer-Verlag New York Inc.: 1994.

National Cancer Institute. Age distribution of incidence cases by site.
Available at http:/Iseer.cancer.govlcgi-
binlcsr/1975_2003lsearch.pl#results, last accessed 3/6/2007.

Silverman DHS, Dy CJ,Castellon SA, Lai J, Pio BS, Abraham L, Waddell
K, Petersen L, Phelps ME, Ganz PA.. Altered frontocortical, cerebellar,
and basal ganglia activity in adjuvant-treated breast cancer survivors 5-10
years after chemotherapy. Breast Cancer Res Treat 2006.

van Dam FSAM, Schagen SB, Muller MJ. Impairment of cognitive function
in women receiving adjuvant treatment for high risk breast cancer: high
dose versus standard dose chemotherapy. J Natl Cancer Institute 1998;
90: 210-218.

Lehmann JF, DeLisa JA, Warren CG, deLateur BJ, Bryant PS, Nicholson
CG. Cancer rehabilitation: assessment of need, development and
evaluation of a model of care. Arch Phys Med Rehabil 1978; 59: 410-9.

O’Toole DM, Golden AM. Evaluating cancer patients for rehabilitation
potential. West J Med 1991; 155: 384-7.

Yoshioka H. Rehabilitation for the terminal cancer patient. Am J Phys Med
Rehabil 1994; 73: 199-206.

Stone RW. Employing the recovered cancer patient. Cancer 1975; 36:
285-286.

Bloom J, Stewart S, Chang S. Then and now: quality of life of young
breast cancer survivors. Psycho-oncology 2004; 13: 147-60.

Bradley CJ, Neumark D, Oberst K, Luo Z, Brennan S, Schenk M.
Combining registry, primary and secondary data sources to identify the
impact of cancer on labor market outcomes. Med Decision Making 2005;
25: 534-547.

185

 

120.

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

Waldfogel J. The impact of the Family and Medical Leave Act. Journal of
Policy Analysis and Management 1999; 18: 281-302.

Feuerstein M., Harrington CB. Recommendations for the U.S. National
Occupational Research Agenda: Research on cancer survivorship,
musculoskeletal disorders and work disability. Journal of Occupational
Rehabilitation 2006; 16: 1-5.

Yadav R. Rehabilitation of surgical cancer patients at University of Texas
MD. Anderson Cancer Center. Journal of Surgical Oncology 2007; 95:
361-369. E"

Naeim A, Keeler EB, Reuben D. Perceived causes of disability added
prognostic value beyond medical conditions and functional status. Journal
of Clinical Epidemiology 2007; 60: 79-85.

 

Messner C, Patterson D. The challenge of cancer in the workplace. '-
Cancer Practice 2001; 9: 50-51. 5“"
Given CW, Given B, Azzouz F, Stommel M, Kozachik S. Comparison of

changes in physical functioning of elderly patients with new diagnoses of
cancer. Medical Care 2000; 38: 482-493.

Given B, Given CW, Azzouz F, Stommel M. Physical functioning of elderly
cancer patients prior to diagnosis and following initial treatment. Nursing
Research 2001; 50: 222-232.

Hodgson NA, Given CW. Determinants of functional recovery in older
adults surgically treated for cancer. Cancer Nursing 2004; 27: 10-16.

Committee to Review the Social Security Administration’s Disability
Decision Process Research. Survey measurement of work disability.
Washington, DC: National Academy Press: 2000.

Nagi S. An epidemiology of disability among adults in the United States.
Milbank Memorial Fund Quarterly: Health and Society 1976; 54: 439-467.

Centers for Disease Control and Prevention, National Center for Health
Statistics. Web tutorial: logistic regression. Available at
httpzllwww.cdc.govlnchsltutorials/currentnhameslnhanes/analyses/Iogisticr
egression/logistic__regression_intro.htm, last accessed 5/30/07.

Morton RF, Hebel JR, McCarter RJ. A study guide to epidemiology and
biostatistics (3" edition). Rockville, MD. Aspen Publishers, Inc. 1990.

186

132. Rosner, B. Fundamentals of biostatistics (4th edition). Belmont, CA.
Wadsworth Publishing Co. 1995.

133. UCLA Academic Technology Services. Resources to help you learn and
use STATA. Available at http://www.ats.ucla.edu/statlstata/, last accessed
8/24/07.

187

Memberships/Societies

2003-present Genetics Society of America

187

 

FTF f a.