III III: III .IIIIIIIIIII III
“III lII'I'I ..
ll IIII‘IHl 'Illl:/IIII 'II III
I. II III :II II
II I: .I..II III II IIIIIIIIIIi-II,,
III? TIES-y: J 1.1.x” ‘ ‘
III llIlIlIl Illll IIIII “Ill 'I‘ II IIII .I‘ 1'“; I~ I?“ II;,’
II III III. '
“'I "III II ..
II III IllllI III'. l}: J“ '

III} lllI LIIIII IIII II}

VI .1: ”Vt? IIII
‘1: :‘I! W)" IA ..k-‘ ‘
l I ti :0? g! . I. .‘v;
I
- I

. I
II D I ' ' I 'lyl‘: .l‘Z'é‘ ‘, 1:
I I [I KIWI}ll'lIIIIIIIIIL; “III-'I'.
'I - n,
:23). ._ «‘I‘? 'I
III {27‘ “DH.

..-"'~‘qu
IIIIIIIIIII
. , '\ .'II

'I.;I’III III III:

I’. I
54 ‘I

‘HII' in}, il
II "III III I
.‘ *IIIIII III ’I’

. . _ 'I ~
. 'I.}:R“ 41;} - i.-.
1;". -, 3,qu . YI‘ ~
. Ul‘l‘ 3%“ '1“. (LI:
1-. . IIIQIII’IIL. .5; a“. £5" .
"I M33452;
9 1"" 4

        

 
 

    
 

  

Y , ll l' I
IIIIIIIIIIIII'III I
III III lIIIIIIIIII‘II
IIII I'II “I I

E“; III I ’I Ml IIIIII I'll}

“ll llllIlIlI"'I

- I, IIIIII III I
. III IIIIIIIIII IIII'I IIIIIII

: llI Ill'IlIII I
III-I. I III ’I‘IIIIII

IIIllI :l‘ .III.

 

 
    
   
       
    

   
  
  
 

      

II'

   

I I

  

III

     

'llII
III II” III II
' III’II‘ :
.IIIIII III‘

III III
. II 9"“llII. III II
INIII :lllI III llIIIIIIIIII

 

 
       
  
 
 
  
 

 

    
   
    

 

 

 

 

Ill:
I'lIIII, III llI I I
Illl I IIII I”...

::‘ IlII. III I IlIII' ”III :IIIl IIII III

I
llll ll llI l : ll: l'lIIlIll 'l:_:I:
l l‘.‘ IIIIIlIIf "' II‘I

 
  

{I

III
I}

I;

\
‘U
L'l‘: an,“
‘5

‘

Eh
”I I

I. "7"" I 9-"?! $52 ' .

I. III]. .. mm
3.3,»: I;I;I 1.. . ,. ;III' Inf-I'- "
l 341;“ .I I. 5.5". .°. ' I,” :(

. 'II :lllI I: . :‘Ifl I II'II ll' .
I Ml". Ill“ :
H I H II: I lII. II;I lllllll: IIIIII IIIIIIll:l IIlIIJ
I I ‘ I: :I::

  
 

  
 

i‘ :- I" A-. w 4
.. III-III ...n,;:xIc, I?
. ‘,|'1 I It In,“ u" a .t. 'I’lnflfl 1 ”f :1"...
l 1.129% .- . ‘ V. __ H :33" f]. :V: . '
I‘ . I II? -'.'.I -II.; .

. <
. I,“ ‘
. I
L ..
C
§l V
X ' L‘ o .
":I I‘
'I'? 1‘ v 'I'
..
I : l
I

 
 
 

   
  
 
 

.II' I. - Ill:
:‘ : :WI‘“ :IIIlI:Il:l I:Illl.:!. l ll! I“ llll .~.\. 1' VI:
‘ ".~ 'I Ill I "I :IIII'I'I‘ 'IIII III'IIIi m ’I‘ ”I IIIIIII‘I II‘?‘ .I -;.-1_=‘ig.r,s‘“«:§.:
l‘ : :1: ‘:I l ll Illlllll‘ U}: ' ’ (if; fi‘fi-V’I:
.I'I 'I'l‘ll I ll: :lllm' I':|:':l::: I'll: IllI III‘llill :IIII: I.” . h : $2: .3565?
‘l "I I III III Il‘I :l‘" 'lIIlIIII I [I'lll - »:- «fag -
”'I." ‘I’I: ' ‘I I ‘ | ' "" I ' ll” "M II IIlI II " - 3.31:?
I I" --

III II. M III llIllIl
IIIIIII: I' ”W IIIIIIIII I

 
   
 

fir—«=5

.{ 'I’“: :I' lI ll: ”WI: I:IIII : : .l:::::::

:3—bh

III ' III till.IIIlHM IIII‘lllll'Ill II . “I" I‘lll :‘jgfil‘fixéfl‘g ..
. [I III; I . [I‘M II: ”.I llII‘ IIIIIIIIIIII l 'I. :II'II' “ll :lIJlllllIIIlri. 'I‘lI" fi'lI ‘nIllIl‘t'zd .I'32:
:1: 1 l I: II.” ‘5; “

‘ I I _.I I I‘ “Ill WI
II. _ H : : . I; ‘I Ill II' I'I’I IIIII
I. :. '.'I",. I: I: . : ‘ I"'.' I II IIIII III: I “Illa” III I :::I:: :Illl IllII‘lhlII IIIIIII II III“ “III:
. I , . I: i ‘1. ': IIII :III IIMII :'II:'l‘::I:'::: I:I:l 'II| :IIIIlI I'l I I I III”: III 'III lWIIlIII III I by. Jull'l:

I : 1‘ I HE'IIIIJII II l’llIIIIlIllllllII': I I Il liII'lII- ll l”ll'HIIJI : I‘l IIII'II‘: III”:

” - II.I,,II,I II.‘. II' |J
.' . I I: I IIl"II.:I..I 'IIIII'll.I'.I‘IlIlIlIIIIIII'Il'II, II"‘ ”If“ I'm.” .IIIII I II I
. ‘ I‘ ‘ ‘ ‘ I‘ I , ' I.

,l ' '.'..
r I :Im‘m III“. M : ”MIC? :E.:':' PIIIII‘” ‘oII.
I H I .‘ 'lI: II III I I I ' .W: lllIl H ,I'I ' I‘
I‘ll: . I l l, I‘.‘ I ._ I: I' lI I I'lII III lIlII:lIl I I l: l II In) WI) IIII: MI
1",, IIII‘ I :,l:I.l,;II:ll:lllllIII I” lIIIIIJIIH llIlIIIIIII II I ‘ I IIII: .
I j H: I: :III ll: ::I’I::II::: ”IVA 'l'I llI'IlI IIII III. III II lllllII thI “II II film.’ V'Y'III ”Ill: ’I“”?
III I llII::‘v'I'IIII"le;:I: '..I."‘ I III

III': :- I J I ”I.
.I IIIIII....I .I III VIII III I

“134'“: : IIIII ' n-I '. ..II.IIIII.III.IIIIIIIIII'IIIIMIII .Iilllnlnl lllllllll l:: ll

 

 

 

IILI]lilliiizilfljiiilwlmum”Willi

LIBRARY

.’ Michigan Sun:
" Univemty' ‘

 

This is to certify that the

thesis entitled

Patient Reporting Behaviors

in Cases of Colon and Rectal Cancer
presented by

Walter R. Gregg. Jr.

has been accepted towards fulfillment
of the requirements for

Masters Jegree in Anthr0poloqy

O/wfiw

Lajor professor

 

Date M‘ léD/ /q 7‘?
(r I

0-7639

 

 

OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

 

 

 

PATIENT REPORTING BEHAVIORS IN CASES
OF COLON AND RECTAL CANCER

BY

Walter R. Gregg, Jr.

A THESIS

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

MASTER OF ARTS

Department of Anthropology

1979

ABSTRACT

PATIENT REPORTING BEHAVIORS IN CASES
OF COLON AND RECTAL CANCER

BY

Walter R. Gregg, Jr.

Medical practitioners have postulated that the time
period between symptom recognition and diagnosis of cancer
is directly proportional to disease staging and prognosis.
With longer time intervals considered as maladaptive for
the patient's well-being, there has been considerable
research to enhance early detection. Although this research
attempted to better understand cancer patient behavior,
sub-cultural differences were not considered in any detail.
The present study explores sub-cultural factors associated
with cancer patient care-seeking behavior and involves a
retrospective record review of 135 deceased low-income
Black and White male colorectal cancer patients drawn from
four Mid-Western V.A. hospitals. In this sample, Blacks
experienced cancer-related symptoms for a shorter period of
time before seeking care than did Whites, contrary to
expectations. Further, analysis of the effect of various
symptom combinations on time between symptom onset and first
physician visit indicates differences between racial groups
consistent with published ethnographic accounts of folk

medical beliefs.

To my wife

Deborah J. Gregg

ii

ACKNOWLEDGEMENTS

This study was conducted through Michigan State
University, College of Human Medicine, Department of
Anthropology, and I therefore wish to thank the University
and especially the Department of Anthropology for the
privilege of conducting this study under their auspices.

To the administrators of the Veteran's Hospitals of Ann
Arbor, Allen Park (Detroit), North Chicago, and Hines
(Chicago); I would like to express my appreciation for
their patience and cooperation in providing the necessary
data for this research. Without their help this study
could not have been conducted. Also, I would like to
express my gratitude to the many persons in various Depart-
ments who generously provided assistance in the proper
administration of this research. To Dr. Dan English,
Chairman of the Department of Surgery of Michigaantate,
for his enthusiasm in this research and his helpful letters
of introduction to the various hospitals. And, especially
to Dr. William Frey, International Rehabilitation Special
Education Network, and David West, College of Osteopathic
Medicine, for their generous assistance in the development

of the statistical analyses.

iii

To the members of my thesis committee and the reader
of my thesis, I owe a special debt of gratitude--to Dr.
Raulet who deepened my insights and integration of anthro-
pological concepts; to Dr. Snow for her helpful comments as
reader and for her useful insights into Black medical
beliefs; and especially to Dr. Cheryl Ritenbaugh, who as
chairperson of my committee has guided my progress and
development in the field of anthropology to this point, I

am very grateful.

iv

We live in an information gap. Between what our body tells
us and what we have to know in order to function, there is
a vacuum we must fill ourselves, and we fill it with

information provided by our culture (Geertz 1968: 27).

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . . . . . . . . .
Chapter

I. A THEORY OF CULTURE AND ILLNESS . . . . . . .

II. A
SY

An Ethnomedical Approach for the Study of
Medicine . . . . . . . . . . . . . . . .
Medical Belief Systems in Modern Society . .
The Culture Concept and Systems of Medicine
An Experiential Framework . . . . . . . . .

HISTORICAL BACKGROUND INTO THE STUDY OF

MPTOM DURATION . . . . . . . . . . . . . . .
Existence of Delay . . . . . . . . . . . . .
Variables in Symptom Duration . . . . . . .
Sociodemographic Factors . . . . . . . . . .
Cancer Related Factors . . . . . . . . . . .
Psychological Factors . . . . . . . . . . .
Physician and Medican Care Factors . . . . .
Sociocultural Factors . . . . . . . . .
Comments on Past Studies of Delay . . . . .

III. COLORECTAL CANCER AND SYMPTOM DURATION . . . .

APPENDIX

APPENDIX

The Sample . . . . . . . . . . . . . . . . .
Methodology . . . . . . . . . . . . . . . .
Analysis . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . .
The Sample Distribution . . . . . .
Symptom Duration as a Continuous Variable .
Symptom Duration as a Discrete Variable . .
Discussion . . . . . . . . . . . . . . . . .
Conclusions . . . . . . . . . . . . . . . .
A . . . . . . . . . . . . . . . . . . . . .
B . . . . . . . . . . . . . . . . . . . . .

vi

viii

xi

11
14
21
30

40

43
45
46
52
57
65
67
72

75

80
82
96
105
105
130
156
164
177

181

184

FOOTNOTES . . . . . . . . . . . . . . . . . . . . . . 188

LIST OF REFERENCES . . . . . . . . . . . . . . . . . . 191

vii

I.
II.
IIIa.
IIIb.
IIIc.
IV.
V.
VI.
VII.
VIII.
IX.
X.
XIa.

XIb.
XIC.
XId.
XIe.
XIf.
XIg.

XII.
XIIa.

XIIb.
XIIc.
XIId.
XIIe.
XIIf.
XIIg.
XIIh.

XIIi.
XIIj.

LIST OF TABLES

List of Variables . .

Primary Site of Cancer

Ethnicity of Patient
Ethnicity of Patient
Ethnicity by Site . .

Occupation of Patient . . . . . . . . . . .
Marital Status of Patient . . . . . . . .
Population Size of Birth Place . . .

Age at Diagnosis . .
Stage at Diagnosis .

Differentiation at Diagnosis . . . . . . .

Family Medical History
Personal Medical History:

Problems Non-Similar
Personal Medical History:

Problems Similar .

Personal Medical History:
Problems Non—Similar
Personal Medical History:

Problems Similar .

Personal Medical History:

Problems . . . . .

Personal Medical History:

Problems . . . . .

Personal Medical History:

Similarity . . . .
Symptoms Presented at
Symptoms Presented at

Bleeding . . . .
Symptoms Presented at

Stools . . . . .
Symptoms Presented at

Pain at Defecation
Symptoms Presented at

Persistant Pain . .
Symptoms Presented at

Abdominal Pain . .
Symptoms Presented at

Body Pain . . . . .
Symptoms Presented at
Symptoms Presented at
Symptoms Presented at
Symptoms Presented at

of Breath . . . . .

' ' éeésénél'médicél'
. . Personal Medical.
biégfioéié I I I I I I
Diagnosis: Rectal

oiégfioéié: Blood in '
Diagnosis:.Abdominal.
Diagnosis: Dull . . .
biégfioéié:'c£aip§ ' '
biégfioéiég'cénérél' '
Diagnosis: Diarrhea I
Diagnosis:

Diagnosis: Shortness

viii

Occasional Medical
Occasional Medical
Persistant Medical

Persistant Medical

No Reported Unusual

Constipation
Diagnosis: Weight Loss

84
107
107
107
111
111
113
113
113
117
117
117

120
120
120
120
121
121

121
124

124
124
124
124
124
124
124
124
124

125

XIIk.
XIIl.

XIIm.
XIIn.
XIIo.

XIIp.
XIIq.

XIII.
XIV.
XV.
XVa.
XVb O

XVI.

XVIa.
XVIb.

XVII.

XVIIIa.
XVIIIb.

XIXa.

XIXb.

XXa.

XXb.

XXI.

XXIIa.

XXIIb.

Symptoms Presented at Diagnosis: Weakness . .
Symptoms Presented at Diagnosis: Palpable

Mass . . . . . . . . . . . . . . . . . .
Symptoms Presented at Diagnosis: Distention .
Symptoms Presented at Diagnosis: Nausea . . .
Symptoms Presented at Diagnosis: Poor

Appetite . . . . . . . . . . . . . . . .
Symptoms Presented at Diagnosis: Malaize . .

Symptoms Presented at Diagnosis: Decreased

Caliber of Stools . . . . . . . . . . . . .
Symptom Distribution by Race . . . . . . . .
Reported Symptom Duration Prior to Diagnosis
Results of Oneway Analysis of Variance on

Reported Duration of Symptoms by Ethnic

Background . . . . . . . . . . . . . . . .
Results of Multiple Range Test, Tukey-B

at an Alpha of .05 . . . . . . . . . . . .
Reulsts of Multiple Range Test, Duncan

at an Alpha of .10 . . . . . . . . . . . .
Results of Oneway Analysis of Variance on

Reported Duration of Symptoms by Ethnic

Background . . . . . . . . . . . . . . . .
Results of Multiple Range Test, Tukey-B

at an Alpha of .05 . . . . . . . . . . .
Results of Multiple Range Test, Duncan

at an Alpha of .10 . . . . . . . . . . . .
Results of Oneway Analysis of Variance on

Reported Duration of Symptoms by Ethnic

Background . . . . . . . . . . . . . . . .
Results of Oneway Analysis of Variance on

Occupation by Reported Symptom Duration .
Results of Multiple Range Test, Tukey-B

at an Alpha of .05 . . . . . . . . . .
Results of Oneway Analysis of Variance on

Material Status by Reported Symptom

Duration . . . . . . . . . . . . . . . .
Results of Multiple Range Test, Tukey-B
at an Alpha of .05 . . . . . . . . . . . .

Results of Oneway Analysis of Variance on
Reported Duration of Symptoms by Age at
Diagnosis . . . . . . . . . . . . . . . . .

Results of Oneway Analysis of Variance on
Age at Diagnosis by Ethnic Background . . .

Results of Oneway Analysis of Variance on
Reported Duration of Symptoms by Stage
at Diagnosos . . . . . . . . . . . . . . .

Results of Oneway Analysis of Variance on
Population Size of Birth Place by Reported
Symptom Duration . . . . . . . . . . . .

Results of Multiple Range Test, Tukey- B
at an Alpha of .05 . . . . . . . . . . . .

ix

125
125
125
125

125
125

125
126
129
131
131

131

133
133

133

134
134

137

137

137

138

139

140

140
141

XXIII.

XXIV.

XXVa.

XXVb.

XXVIa.

XXVIb.
XXVII.
XXVIII.

XXIX.
XXX.

XXXI.

Between Group Variances of all Variables
Selected for Study as Generated by
T- Test Procedures . . . . . . . . . . .

Statistical Relationship of Variable Means
for the Continuous Dependent Variable of
Symptom Duration, Generated by T-Test
Procedures . . . . . . . . . . . . . . . .

Results of T-Test Procedures on Variables
Demonstrating a Significant Relationship
with Continuous Duration, and All Variables

Included in Hypotheses (*). Between Group
Variances for Within Ethnic Group Means by
Symptom Duration . . . . . . . . . . . .

Results of T- Test Procedures on Variables
Contained within Hypotheses. Between
Group Variances for Between Ethnic
Group Means by Symptom Duration . . . . . .

Results of T-Test Procedures on Variables
Demonstrating a Significant Relationship
witn Continuous Duration, and All Variables
Included in Hypotheses (*). Within Ethnic
Group Means by Symptom Duration . . . . . .

Results of T-Test Procedures on Variables
Contained within Hypotheses. Between
Ethnic Group Means by Symptom Duration . .

Effects of Normal Appetite and Loss of
Appetite with Weight Loss on Symptom
Duration Time . . . . . . . . . . . . . . .

Results of Cross-Product Ratio Analysis:
Duration of Symptoms (Above/Below
Median) by Ethnic Group . . . . . . . .

Main Effects for Variable Occurrance by
Ethnicity . . . . . . . . . . . . .

Main Effects for Variable Value by the
Discrete Variable Symptom Duration . . . .

Interaction Effects between Ethnicity and
Listed Variables by the Discrete Variable
of Symptom Duration . . . . . . . . . . .

142

144

149

150

151

152

155

158
159

161

164

11.
12.
13.
14.
15.

16.
17.
18.

Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution
Distribution

by Percent

of

LIST OF FIGURES

Cancer Sites . . . . . . . .
Ethnic Groups . . . . . . . .
Subjects by Race . . . . . . .
Cancer Site by Race . . . .
Occupation . . . . . . . . . .
Marital Status . . . . . . .

Population by Birth Place . .
Age at Diagnosis . . . . . . .
Duke's Stage at Diagnosis
Differentiation at Diagnosis
Reported Cancer in Family .
Personal Medical Problems .
Reported Symptoms . . . . . .
Duration of Symptoms in Months
Duration of Symptoms in Months
Ethnic Group . . . . . . . .

Symptom Profile for Detection . . . . . . . .
Non-Symptom Profile for Detection . . . . . .
Discrete Variable Profile for Detection . . .

xi

106
108
108
110
110
112
114
114
116
116
118
118
123
128

136
145
145
160

INTRODUCTION

The primary purpose of this study is to contribute
to the understanding of people's responses to cancer. It is
suggested that this understanding may be enhanced by pro—
viding information derived from research based on concepts
from the science of human culture and society, with the help
of research methods appropriate for the study of people as
social beings. This study is based on a Veteran's
Administration hospital record review of Black and White
male cancer patients who postponed1 seeking medical advice
for symptoms resulting from either colonic or rectal
carcinoma. The main focus here centers around the role of
sub-cultural differences in response to illness, with the
assumption that these responses present identifiable
regularities patterned along cultural models common to
groups of similar ethnic origin. The major criterion for
this analysis is symptom duration, defined as that reported
time period between the first appearance or recognition of
symptoms (in this case related to cancer) and medical inter-
vention as initiated by the patient.

With reported symptom duration as the dependent
variable representing a measure of secondary preventative
health behaviors, special attention is given to group mean

duration times for analysis. Four general hypotheses are

presented to test for the existence of interethnic group
differences related to reported symptom duration. 1) The
mean reported symptom duration times for patients repre-
senting different ethnic backgrounds will be significantly
different. 2) Persons with a positive family medical
history for cancer (i.e., some suspected occurrence on the
part of the patient) will exhibit significantly different
between ethnic group mean reported symptom duration times.
It is also hypothesized that the within ethnic group mean
reporting times will be significantly different for persons
with a positive family medical history when compared to
those with a negative family medical history. 3) Individ-
uals with prior experience of nosologically similar (i.e.,
cancer related) conditions to the one under study will
exhibit significantly different between ethnic group mean
reporting times. 4) Of the symptoms found to be signifi-
cantly related to symptom duration for the sample as a
whole, those symptoms reported most often for the present
cancer episode will differ in content between ethnic sub-
groups.

Although several variables were used in the
analysis, this investigation is particularly concerned with
the five main variables of 1) reported symptom duration
time, 2) ethnicity, 3) family medical history, 4) personal
medical history, and 5) the symptoms presented during a
patient's initial diagnostic visit for cancer. The premise

linking these variables together maintains that behavioral

.5

responses to cancer will differ according to ethnic back-
ground, and that a discovery of these differences is pos-
sible through an analysis based on group mean reported
symptom duration times. The differences in these mean
duration times is thought to result from ethnically molded
medical beliefs and social orientations. These beliefs and
orientations integrate illness information obtained through
personal and familial medical experience and generate
behavioral responses appropriate to an individual's cultural
milieu. These responses, when analyzed in the context of
reported symptom duration prior to diagnosis, should reveal
themselves in the form of significantly different mean
duration times for a given condition and ethnic group.

The above stated hypotheses are somewhat general in
their prediction of the behavioral reponses surrounding the
phenomenon of symptom duration. This is in large part due
to the fact that there have been no similar studies per-
formed. Considerable research was undertaken on this issue,
then labeled patient delay, between the mid-1930's and
1960's in response to concern with only a few isolated
social variables such as individual personality, role
expectations, and socioeconomic status. Also, few used
preventative health behavior as a dependent variable.
Although it would seem that culturally oriented research
into the differential effectiveness of preventive health
behaviors would have been a productive approach to use,

"good data do not exist" (Kegeles 1976:104).

Anthropological studies have demonstrated that all
cultural systems provide some knowledge of diseases, their
classification, and etiology (Rubel 1960, Frake 1961,
Atkinson 1962, Price-Williams 1962, Hallowell 1963, Fabrega
1971, 1974). Studies such as Zola (1966) and Fantl and
Schiro (1959) on Irish and Italian patients, Hetherington
and Hopkins (1969) on U.S. Whites, Non-Whites, Scandinavians
and Poles, and Fabrega and Zucker (1977) on Ladinos,
Mestizos, and Indigenas of Mexico have also demonstrated
the differential effects of culture upon the interpretations
of various symptomatologies. However, even though such
research involving the relationships between ethnicity and
certain diseases states has been the concern of medical
sociologists and anthropologists for sometime, as yet there
have been no definitive studies concerning ethnicity and
cancer. While it may be prudent to imply that the illness
response mechanisms for non-malignant and malignant dis-
orders are essentially the same, this assumption still
remains untested. Furthermore, before illness behavior
information can be translated into effective practical
applications for cancer control programs, its functional
components need to be established in relation to malignant
disease. Once we can identify what illness behavior
modifers exist for particular ethnic subsets of the popu-
lation, we can begin to gain a general understanding about
why people behave the way they do. With this understanding

we should be able to identify what social groups are more

5
at risk than others for longer symptom evaluation periods
(i.e., symptom duration times) and then modify educational
and detection emphases accordingly for those groups.

The present study, like much of anthropological
research, has developed out of observations of human behav-
ior within a social context. This research grew out of
observations on how people have responded to the knowledge
that they have cancer and how this response has lead to
changes in the nature of social relationships between the
patients, their families, and friends. It was observed that
people who were near the patient socially were influenced
not only in terms of their relationship with the patient but
also in terms of how they viewed illness. It seemed that
although the people involved spoke and generally understood
the vernacular of modern medical science, they tended to
interpret and respond to the situation in quite emotional
ways. These observations and others lead to general
questions about how an illness situation could influence the
medical behaviors of these people surrounding the patient.
Is it generally the case for people, when confronted with
some disease like cancer, to ignore their basic understand-
ing about "scientific medicine" in favor of more emotional
responses? Are such emotional responses a reflection of a
person's social attitudes and beliefs, and if so, would such
a response lend itself to being significantly influenced by

that person's sociocultural milieu?

6

With the heavy emphasis that the American medical
establishment has been giving to alerting people to the
benefits of early diagnosis for cancer, symptom duration
prior to diagnosis seemed to be a perfect situation to ask
some of these questions about illness behavior. Considering
the history of this war on late diagnosis it was assumed
that the relevant literature would be replete with research
on human behavior and cancer. Although some studies were
found to deal with preventive health behaviors in light of
sociocultural variability, the vast majority of them did so
in a quite superficial manner, with the most prevalent topic
of concern being on the nature of the patient's responsi-
bility for delay in seeking medical advice. Unfortunately
this problem was almost always approached in ways that in
part ignored how the patient came to make the decisions he
did. It was almost as if the medical researchers were
viewing behavior in the same light as they did disease;
that is, as if both were pathologies to be categorized and
manipulated within the confines of a laboratory methodology.
In their report entitled "The American Health Empire"
Ehrenreich and Ehrenreich have summed up rather nicely the
manner in which many medical professionals have treated
patients and in turn the problems these patients present for
medical intervention. "Everything about the American
medical system seems calculated to maintain the childlike,
dependent, and depersonalized condition of the patient"

(1971:11). To be sure, there are members of the medical

7

system who are trying to change this attitude, but in the
case of studies on symptom duration, such attitudes have
had a dampening effect on the explanatory value of medical
investigation into human behavior. For example, the
depersonalizing manner used by researchers to approach the
problem served to eliminate possible insights into why
people reacted as they did. Such a pathogenically oriented
approach can only offer limited insights into the exam-
ination of medical behavior, and needs to be well balanced
by sufficient concern about the social context from which
such behaviors arise.

Although the issue is not free from disputez, there
are strong feelings within the professional medical commu-
nity that this period of symptom duration has a direct
relationship with poor prognosis in cancer patients such
that the longer a person waits the greater the chance of
having a poor prognosis (Blackwell 1963, Rogers 1974,

Gerard 1975, Copeland 1976, 1977). Since the application

of current means of treatment such as surgery and chemo-
therapy have only a limited ability to improve on colorectal
cancer patient prognosis (Muldoon 1977), every effort should
be made to bring people to medical care as soon as possible.
In doing this we can at least maximize the effectiveness of
such treatment regimens.

Research such as that outlined here is potentially
valuable on both a theoretical and practical level. On the

theoretical level such a study can provide additional

8
information for the existing data base on disease and ill-
ness behaviors from which further questions may be gener-
ated. Also, the discovery in this study that ethnically
related differences in behavioral response to cancer exist
implies that data from non-malignant studies may be used to
further enhance an understanding of behavioral response and
malignant conditions. On a practical level such a study
has obvious applications for the development of more
efficient health planning programs through the identifi*
cation of cultural groups at risk of taking longer evalu-
ation periods relative to particular symptoms and medical

conditions.

CHAPTER I

A THEORY OF CULTURE AND ILLNESS

Disease and illness have been universal experiences
for human beings in all times and places. These experiences,
although only a small reflection of an individual's total
human experience, are the result of a fundamental interface
between humankind's physical and social environments.
Throughout time, these experiences have exerted an influence
on the developmental character of human societies (Dubos
1973). In other words, disease as an environmental stress
and illness as the social response to this stress have
interacted throughout humankind's evolutionary history.
Considering the theoretical importance of this interface, it
is not surprising that anthropologists have found it most
informative to investigate societal aspects related to
disease and health. These researchers have noted a number
of ways in which this interaction between the physical and
social environments has influenced the development and
maintenance of various social systems. For example, such
interrelationships can serve in the maintenance of social
control (Rubel 1960; Lieban 1962, 1973; Hallowell 1963;

Paul 1963; Adams and Rubel 1967), and can affect the mode of

10

enculturation and interpersonal relationships (Price-
Williams 1962, Currier 1966, Adams and Rubel 1967, Ingham
1970).

Anthropologists interested in the "why" of human
behavior have used this interface between disease and social
systems as a stepping stone for research into a variety of
health issues. Although a number of anthropologists have
focused on contemporary health issues such as health policy
implementation (Paul 1955, Firth 1957, Miner 1960, Foster
1961, Imperato 1969, Hochstrasser and Tapp 1970), it has
only been within the last decade or so that specific
research into issues of this nature have been pursued with
any emphasis within the discipline. Until the early 1960's,
the majority of anthropological research that provided
information on how societies handled health problems was of
a holistically enthographic nature. That is, few
researchers went into the field with specific questions
about health issues and their management. If these issues
were discussed at all, it was often in the context of a
general description of a society's particular cultural
characteristics. Social systems that dealt with health
issues were often described, but not analyzed in depth
unless they were related to the investigator's particular
research problem. Since this time, more and more research
has focused on specific medical issues such as concepts of
illness and health maintenance practices. While these

studies often included analyses of how traditional medical

11

systems interacted with orthodox Western medical practices,
attention was usually given to non-Western medical beliefs
with little emphasis on those systems existing within more
modern industrialized nations.
An Ethnomedical Approach for the Study
Medicine

A research approach common to many of these investi-
gations involved an ethnomedical perspective. Ethnomedicine
can be defined as the investigation of medical problems by
an emphasis on how social and cultural factors affect their
perception, expression and consequences (Fabrega 1975). An
ethnomedical approach is therefore concerned with the socio—
cultural context in which medical problems arise and are
handled. As noted above, the vast majority of ethnomedical
research has been concerned with non-Western societies and
has been rooted in such cultural domains as religion, magic,
and witchcraft (Seijas 1973: 544). Such emphases are not
surprising when one considers that these domains are, in
comparison to orthodox Western medicine, performing tasks of
similar importance in regard to illness and diseaSe. The
underlying assumption here is that the occurrence of
disease creates certain sociocultural needs (e.g., identifi-
cation, explanation, care and management) which require
fulfillment. A cultural response apparently common in all
societies has arisen to meet these needs in the development
of medicine as a cultural system. And, illness behavior is,

in part, an operationalization of the premises that exist

12

within such a system. Unlike Western medical beliefs, such
as a cultural system need not be limited to secular issues
alone and can include the sacred as well. In general, a
medical system can be viewed as having four basic functions:
1) a social construction of illness as a human experience;
2) a cognitive response to illness such as labeling,
explaining, or interpreting; 3) therapeutic action; and

4) management of death and dying (Kleinman 1974, 1975). For
the purposes of this thesis, only the first three functions
will be considered in any detail in the following
discussions.

In traditional and modern societies alike, such a
medical system can often be grounded in a number of other
cultural domains from the realm of interpersonal relations
to science and magic. Also, in any given society, there can
be a number of distinct systems of medical care in operation
depending on the various social and epidemiological factors
at play. An ethnomedical approach is therefore not limited
to traditional societies and can provide a useful framework
for viewing health behavior phenomena in more modern
industrialized settings such as in the United States.

A short discussion of the medical belief system of
low income Blacks in the United States can best illustrate
how such a system can interact with a set of alternative
beliefs (orthodox medicine) and at the same time retain a

considerable degree of internal consistancy.

13

Several authors have noted that low income Blacks
and often highly educated and socially mobile Blacks tend
to lump phenomena which are kept conceptually separate by
practitioners of orthodox medicine. For example, while
modern Western medicine makes a distinction between science
and religion, no event is considered purely secular by
Black folk standards. "Importance is placed on oppositions
between good and evil, natural and unnatural and all events,
including illness, can be classified along such lines"

(Snow 1978: 70). In this particular folk system, all
illnesses are theoretically considered preventable if only
care is taken by the individual. Therefore, each individual
is responsible for knowing what positive actions are
required as well as which conditions need to be avoided in
order to achieve and maintain good health. Action is
essential in this system, a failure to act appropriately
will result in an illness for which responsibility must
usually be accepted by the ill person.

Illnesses then are classified as either natural or
unnatural (Cameron 1930; Wintrob 1973; Staiano 1974; Snow
1974, 1977, 1978a; Hillard and Rockwell 1978). This system
of medical beliefs is quite coherent and is not just a
random collection of isolated superstitions. "If the under—
lying premises are accepted, it makes just as much sense to
the believer as the principals of orthodox medicine do to

the health care professionals" (Snow 1974: 83). In this

14

system, most problems in health status are thought to occur
when the sick person is no longer in harmony with nature.

In such a case, the person would be considered to be suffer-
ing from a natural illness. Since a person's good health is
primarily based on such harmony, the rules of nature must

be known and followed. One of the most common reported ways
for a Black person to come down with an unnatural illness is
by way of hexing, voodoo, crossing-up, root work or sorcery.
Although a hex can be administered in several ways it is
usually administered by placing some magical substance in
the victim's food or drink. In sum, natural illness is the
result of some disturbance in the natural order (which is
often equated with good); whereas, unnatural illness is the
result of supernatural forces often emanating from evil or
malicious sources.

In this medical system, if a particular set of
symptoms are considered natural in origin, help tends to be
readily sought from the professional sector; whereas, if the
symptoms are unnatural, they are not considered amenable to
modern medical practices and traditional folk healers such
as root doctors, herb doctors, spiritualists, or conjure

men and women may be approached for aid.

Medical Belief Systems in Modern Society

 

Despite the strong reliance upon scientific para-
digms in Western industrialized societies, one should not

View modern professional or biomedical concepts of health

15

and disease as a total reflection of Western medical
beliefs. The modern medical profession is only one aspect
of the total Western health system with folk medical
beliefs and popular medical culture also representing
important sectors for concern.

For heuristic purposes these three sectors of
professional, folk, and popular medical beliefs can be
viewed as a continuum of concepts which, although inter-
related, tend to retain their somewhat distinct systems of
classification and action. The folk and professional
medical systems occupy opposite ends of this continuum
because of their lack of congruency. The popular systems
falls somewhere between the two. The professional or bio-
medical sector represents the major institutionalized system
of medical beliefs for Western industrialized societies and
has as its base a taxonomy and mode of action which stems
from Western scientific thought. The folk sector has
pragmatic aspects similar to orthodox medicine and incor-
porates those aspects of biomedical theory which are
conceptually compatible with it. However, folk medical
beliefs often transcend the theoretical limitations of
Western medicine to embrace additional domains of experi-
ence. These domains which folk beliefs focus upon as
important for the resolution of medical problems are often
considered by orthodox practitioners to be in the realm of
the sacred or the supernatural. As such, orthodox

practitioners often profess neither understanding,

16

competence, or even interest (Rubel 1964). Popular medical
culture can be defined as a system of medical beliefs that
are held by the medical laity. While this system tends to
draw rather heavily on the terminology and general paradigm
of biomedical thought, it translates them into more social
and human terms, creating an amalgam of various folk and
biomedical concepts. In commenting on folk and popular
medical systems Kleinman has noted that
we have hardly begun to examine their crucial inter-
actions, which seem to be responsible for when patients
are labeled sick, how they regard their illness, when
they seek care, the type of care they seek, how they
utilize health care facilities, whether they comply with
treatment programs, and how they evaluate treatment
(1975: 591).

Although the cultural domains of folk and popular
medical beliefs are amenable to an ethnomedical perspective,
the nature of the present research question on preventive
health behavior and cancer requires that additional research
perspectives be considered. This is necessary because of a
lack of significant behavioral science analyses of the
response of modern industrial peoples to such diseases as
cancer with their medical, social, and personal ramifica-
tions for preventive health care (Kegeles 1976: 104). The
present study uses a research orientation somewhat similar
to past ethnomedical investigations, albeit with some
modifications. These modifications result primarily from
two differences between the present study and past ethno-

medical research. The most important difference is that a

different set of research questions are being asked of the

17

data. Secondly, the abundance of non-anthropological
investigations on symptom duration and cancer necessitates
their inclusion into the overall discussion of the phenom-
enon. As will be apparent in the literature review, the
major contributors to the study of symptom duration and
cancer have been from the fields of medicine, psychology,
psychiatry, and sociology.

Although some of the reviewed investigators demon-
strated an awareness of a functional relationship between
culture and health behaViors, most did not find it pertinent
to their particular research questions. I do not mean to
imply that medical, psychological, or sociological
approaches are not valid for the study of disease-behavior
interactions, but a strict adherence to these traditional
perspectives can serve to present only a partial picture of
what may be occurring. Fabrega (1977: 379) has voiced a
similar caution regarding the use of traditional perspec-
tives of cultural anthropology and psychiatry. In a sense,
this noted lack of emphasis on the explanatory power of
cultural factors resulted from certain inherent differences
in the theoretical orientations of medicine, psychology, and
sociology.

In part, these differences in research approaches
result from the historical development of the respective
disciplines. Although a thorough discussion of this history
is beyond the scope of this presentation, a slight

digression will serve to clarify why they do not emphasize

18

the culture concept as an analytic tool. The discipline of
medicine in Western industrialized societies is based on
what has popularly been labeled the "biomedical model."
Science with a capital "S" has become the belief system
central to the Western medical model. Disease as a "patho-
logical" state defined in terms of somatic parameters is the
major premise for this system. Disease is seen as a process
that begins with symptoms, and then proceeds to clusters of
symptoms, to syndromes and ultimately to a specific patho-
genesis. "It leaves no room within its framework for the
social, psychological, and behavioral dimensions of illness"
(Engel 1977: 130). Because of this non-social orientation
there develops a gap between what the patient is asking for
and what the physician has to offer. As Scotch (1963: 35)
has noted, modern medicine tends to forget that the patient,
in order to be treated successfully, needs to be treated and
viewed within his frame of reference and cultural values
regarding disease.

Psychology seems to have taken a tack similar to
that of professional Western medicine and although there is
some distinction made between illness as a process and
disease as an event, it is still within a pathological or
aberrant context (Preston 1966: 1109). The most important
distinction that can be made between an anthropological
approach and the one used by the psychologists and psychia-
trists who have written about symptom and cancer is problem

orientation. While there are a number of psychiatrists

19

(notably Fabrega 1971, 1974; Kleinman 1974, 1977; and Kiev
1964) who have emphasized a sociocultural perspective in
their research, the psychiatrists and psychologists who have
worked on symptom duration have not taken this approach.
These researchers, much like the physicians who have also
written about this phenomenon, have formulated their
research questions from the vantage point of the profes-
sional medical sector.

Sociologists on the other hand have become somewhat
isolated from the culture concept due to their tendency to
emphasize analyses of the internal evolution of Western
institutions (Wax 1970). In general they seem to be
concerned with a finer grain of difference than anthro—
pologists. Although sociologists do look at differences
between subcultural groups and topics such as the sociology
of popular medical knowledge, for most, these questions are
not a major topic for concern. With this tendency to focus
on such fine grain topics as socioeconomic factors that
affect preventive health behavior in one kind of culture,
certain aspects of the social context become blurred or
invisible. This exclusion of some of the broader aspects of
the social context does not hinder the sociologist's goal of
explaining within culture differences and similarities.
However, such an exclusion precludes the culturally holistic
approach traditional to anthropological research. To be
sure, there are sociologists that can be said to be doing

anthopology and anthropologists doing sociology, for there

20

are no real hard and fast boundaries that can be drawn
between the two perspectives. Sociologists and anthropolo-
gists often use similar research methodologies on quite
similar problems, but the questions they ask are usually
from two different contextual levels. When presented with
the same observations on differences and similarities
between two social groups, sociologists would tend to ask

a different set of questions than anthropologists about why
the differences are important.

In sum, while the approaches used by physicians,
psychiatrists, psychologists, and sociologists in the study
of symptom duration and cancer are quite valid, the exclu-
sion of the patient's cultural point of view presents only
a partial picture of the total stream of events. While a
person's cultural mileau is not a major topic of concern
for these researchers' it is a logical outgrowth of antho-
pological theory. By augmenting these investigations with
research that poses questions based on this anthropological
sensitivity, the potential for filling in such gaps in the
total picture is enhanced. Thus, in order to add to the
explanatory potential of the above approaches, the concept
of culture has been included for an investigation of the

phenomenon of symptom duration and cancer.

21

The Culture Concept and Systems of Medicine

 

Although most anthropologists would agree that an
awareness of human cultures and cultural differences is a
fundamental postulate for the research questions they are
asking, there would be some disparity as to what the concept
of culture implied. To be sure, one needs to take only a
brief glance at the literature to see that considerable
variability exists about what constitutes culture (cf.
Kroeber and Kluckhohn 1952). This variability is due to the
abstract nature of the term. In being an abstract concept,
the term is open to as many definitions as there people who
feel a need to define it. As Hallowell has so aptly put it,
culture as an abstraction "is our abstraction, a convenience
adopted to the kind of analysis we wish to make of the
problems we wish to pursue" (1953: 611). Culture, like the
concept of the gene, is an analytical heuristic tool that
enables the investigator to focus in on those relationships
which are considered most pertinent to the problem at hand.

In the case of this present study it has been
necessary to abstract and characterize sociocultural factors
in broad terms so as to permit the type of epidemiological
analysis employed. Although this form of analysis was
selected because of the characteristics of the data used, it
should not be overlooked that this analysis is informed by
an ethnomedical perspective designed to enhance an under-
standing of the fundamental nature of cultural reality. In

this case, culture as an abstraction is seen as being based

22

on the observation of individual members of some society
whose behavior is manifested in particular responses to
particular situations. These responses are, in a sense,
extrinsic representations of thought and action resulting
from an interplay between culturally determined behavioral
rules or instructions and the needs generated by particular
environmental circumstances. For example, if he suspects
that someone harbors malicious feelings towards him, a low
income Black would tend to interpret an otherwise non-
threatening symptom as evidence of sorcery and respond
accordingly. Since the occurrence of such responses are in
large part dependent upon the symbolic character of cultural
rules for behavior, the responses like the rules that foster
them are embellished with systems or complexes of symbols
(Geertz 1973). And, it is this interrelationship that makes
such responses analyzable in cultural terms.

A definition of culture which seems most appropriate
for this present research can be seen in the works of
anthropologists like Clifford Geertz. For him culture, in
terms of these symbols, is "the fabric of meaning in terms
of which human beings interpret their experience and guide
their action" (Geertz 1957: 34). Furthermore, Geertz feels
that "culture is best seen not as complexes of concrete
behavior patterns, customs or traditions, . . . but as a set
of control mechanisms . . . for the governing of human
behavior" (1968: 24). In this study, culture then is

considered as being dominated by various sets of regulatory

23

ideas (conveyed by systems of symbols which add meaning to
experience) and act as control mechanisms that permit humans
to transform an essentually chaotic universe into something
with meaning, order, and predictability. I agree with
Geertz that human behavior, without these regulatory ideas
for giving meaning to experience, would be a totally
unmanageable chaos of pointless acts embedded in an unpre-
dictable experience.
Culture, the accumulated totality of such (response)
patterns, is not just an ornament of human existence
but--the principal basis of its specificity--an
essential condition for it" (Geertz 1968: 24—25).
People can adapt themselves to any situation their imagina-
tions can cope with, but their flexibility in this regard
has limits depending on the nature of their particular world
view and the extraneous circumstances involved.

Put very succinctly, a peoples particular world view
entails a culturally oriented mind set for ordering the
universe. In commenting on the relationship between world
View and health related concerns, Pellegrino (1963: 10) has
stated that,

every culture has developed a system of medicine which
bears an indissoluble and reciprocal relationship to the
prevailing world view.
Through an interaction between a peoples' world view or
cognitive framework, their tone of life, and their external
environment arise particular behavioral responses that order

and reaffirm their existence. In these terms Western

professional medicine can be seen as being dependent upon

24

a world view that provides a particular explanation for
illness causation and management. This development is a
function of Western industrialized concepts of illness and
disease and in being a cultural system it is not the only
means by which people in industrialized societies can order
their health related experiences. The medical system of
American low income Blacks, with its classification of
illneses as either natural or unnatural, is a good example
of how people can maintain two classificatory systems with-
out conflict.

The point here is that similar conceptualizations
about particular aspects of the health experience do not
preclude their inclusion into more than one medical system.
The converse is also true in that fine grain differences can
be found to exist in the form of regional variations within
a single medical system. A case in point here involves the
hot and cold theory of disease which is widely spread
throughout Latin America. In this situation the system of
hot and cold classifications of illness, foods, medicinal
plants, and now even Western medicines can be seen to vary
while the premise of balancing hot with cold elements
remains essentially the same (Harwood 1971, Logan 1978).
When considering the low income Black and Latin American
systems as they interact with the Western professional
system, conflicts in what the patient expects and what the
Western physician has to offer can be noted (Harwood 1971,

Snow 1977, Logan 1978). For example, Whitten (1962: 322) in

25

his study of Black medical beliefs in North Carolina noted
that all of his sources reported that a physician using
scientific practices could not help an individual who was
under the influence of a spell. These conflicts can some-
times hinder the Western physician's attempts at effecting
"proper" treatment. And, these perceived problems in
patient compliance are often thought to result from the
patient's lack of knowledge about the importance of the
"required" treatment regimen. This way of looking at the
problem fails to recognize that the patient may feel that
there are other equally or more important measures that
should be taken or that the ones used should be employed in
a different fashion. When considering problems in the
delivery of health care to persons with world views quite
distinct from that of Western medicine, there is consider-
able potential for misconceptions and conflict of interests
to arise. To be sure, such problems can occur and be
analyzed within a single culture as well as between two
distinct cultural groups. However, in either case, the
misconceptions that can arise between health provider and
client can be especially confusing if no consideration is
given to the patient's particular cultural milieu.

At this point it is important to note that this
theoretical construct with its abstractions of culture,
world view, and response mechanisms does not have a direct
relationship with what may be happening on the ground of

human interaction. This construct is merely a heuristic

26

device to aid in the understanding of human interpersonal
relationships. By using such devices the investigator can
translate observed behaviors into meaningful terms relevant
to the research questions asked. This framework for
investigation is certainly not the only way of approaching
those data one wishes to analyze. A11 investigators should
remain aware of this caution and of the possibility that
their particular theoretical approach could lead them to
focus on domains of human experience where analyzable
differences do not exist and to exclude those where true
differences occur. For example, although in the present
study any observed intergroup differences in health related
behavior is considered to have evolved out of differences in
intergroup world views, an analysis of such a relationship
is beyond the scope of the research tools available. It is
always possible that different behaviors could be the result
of other less distinguishable or more subtle differences
that the analysis was not designed to uncover (e.g., body
image or concepts about the self).

From the discussion above it should not be assumed
that the various aspects of a culture need to be in complete
harmony at all times in order for the culture to persist.

To be sure, a change in any aspect bf a culture frequently
affects the relationships that exist between other cultural
components, but there is considerable flexibility. This can
be seen in the way the Puerto Ricans and Guatemalans have

incorporated Western medicines into their hot-cold system.

27

One of the main reasons behind this variability is the
nature in which culture is transmitted to new members of
the group. Culture, with its dependence on extrinsic
sources of information for development and maintenance,
relies upon a learning process (enculturation) that utilizes
symbols in the form of gestures, actions, and words. Since
there is always variability, cultural systems possess the
ability to change in response to different circumstances.
Regarding the continued existence of medical belief
systems Young has pointed out that
a people's medical beliefs and practices persist because
they answer instrumental and moral imperatives, and they
are empirically effective since they enable sickness
episodes to communicate and confirm ideas about the real
world (1976: 5).
This is also the case for Western professional medicine;
however, it is necessary to realize that in place of tradi-
tional forms of support for the medical system (e.g.,
religion and magic) has arisen a different belief system,
"scientific knowledge." While this source of support may
be substantively different from religious and magical belief
systems, in function it is the same, and as Engel (1977: 130)
has suggested, it could be considered a folk medical system
in its own right. The Western medical system, like tradi-
tional systems, can be seen as involving people on two
levels, as a group and as individuals. However, unlike most
traditional systems, the Western professional system inter-

acts with groups of a more culturally heterogeneous nature

and because of this fact certain problems arise that can

28

can reduce the efficacy of such medical practice. The point
here is not whether the Western system is more valuable in
reducing the impact of disease or illness than a more
traditional medical system, but that the majority of the
Western medical practitioners conceive it as being more
Valuable. Western physicians are more action oriented than
knowledge oriented (Freidson 1970: 168) and therefore, they
would be less inclined to even consider such discrepancies
because their prime concern is healing the patient.

Following Young's argument, people living in the
United States should have a marked tendency to respond to
illness in a biomedical manner because of the strong
presence of Western professional medicine in this society.
In other words, the majority of persons born in the U.S.
grow up with a set of biomedical terms that become inte-
grated into their medical experiences through various
social mechanisms of learning (e.g., peer groups, family
interrelations, books, radio, and television). The result
is that there is a strong tendency for such societal
members to somatize a variety of quite social experiences.
For example, people who have grown up under such a biomedi-
cal systems might tend to express anxiety or stress in
physical or biomedical terms, whereas, individuals from
other subcultures like the American low income Black might
speak of being hexed or rooted. People tend to live and
express their experience in terms relative to their

particular cultural or subcultural background.

29

In the U.S. society an individual's use of bio-
medical principles in response to illness has the potential
to vary depending on the person's position in the total
social system. For upper middle class Whites, biomedicine
tends to comprise a large part of their folk medicine,
whereas, lower class Blacks tend to have a more dissonant
folk system by virtue of their peripheral membership in the
society. With clinical medicine involving largely a
dramaturgical experience, a differential internalization of
biomedical principles should lead to different experiences
in clinical encounters. Thus a major problem develops in
regard to the different approaches to and conceptions of
health, disease and medical treatment used by both the
recipients and the suppliers of medical care. The results
of such differences can be readily seen in the problem
orientation and scope of investigations into symptom
duration and cancer. Such orientations can, and in many
cases do, lead to a'misuse of resources in the delivery of
health care by fostering unnecessary conflicts between the
medical practitioners and the people they serve. Also, as
mentioned earlier, these attitudes of the supplies of health
care tend to reduce their chances of efficiently resolving
various perceived problems in health care delivery because
they are aware of only part of the situation. If we could
accurately describe and understand health and illness

behavior from the patient's point of view we could begin to

30

formulate health services that do not conflict with such
cultural values and beliefs (Simmons 1953, Paul 1955,
Hochstrasser and Tapp 1970, Messing 1976).

Although much of the work done by anthropologists in
the domain of medical belief systems has been concerned with
the testing and development of anthropological theory, there
is a growing trend for current research to deal with the
applications of anthropological knowledge to medical
settings and particularly to the field of public health
(Vargas 1976: 441). As mentioned earlier, disease and
illness have contributed to the biological and behavioral
development of mankind, and the behavioral processes that
are involved with the occurrences of disease provide a most
informative area for investigation. Research in this
area involves the testing of anthropological theory in
regard to recurring behavioral responses and sociocultural

variability.

An Experiential Framework

 

The approach used for investigating behavioral
responses and disease in this presentation will involve
Fabrega's (1974: 142-148) suggested phenomenological or
experiential framework. This framework is firmly anchored
in an ethnomedical perspective and entails a concern with
past experiences along with a cultural awareness for an
analysis of such health related responses. A concept of

major importance to this framework is terms an illenss

31

behavioral model. This model entails "general responses to
stress and coping behaviors as well as more specific and
socially organized behaviors" (Fabrega 1971: 213), and as
such represents an operational aspect of the previously
defined concept of culture. Another key concept in this
framework is labeled illness recognition and defines the
phase or process in which ego alone or with alter(s) decides
that according to various criteria an illness or illness-
free state exists. "This is a crucial decision in the
health-111ness-medical-care cycle and should not be assumed
to follow an invariant pattern across cultures" (Fabrega
1971: 214).

Before proceeding further it will be necessary to
make a distinction between the concepts of disease and
illness. Disease is a biomedical category that heavily
implicates the individual with a system of biomedical
classification, and illness is a sociocultural category in
which much of a person's response is determined by the
attitudes and advice of people. Followers of the Western
biomedical model see disease as involving changes in
specific organs of the body caused by particular agents in
predictable ways. As Engel has noted,

(this biomedical model) assumes disease to be fully
accounted for by deviations from the norm of measurable
biological (somatic) variables . . . it also demands
that behavioral aberrations be explained on the basis

of disordered somatic (biochemical or neurophysiological)

processes (1977: 130).

Although there is a strong tendency in the United States to

32

incorporate some of these concepts into one's illness
experience (i.e., the emphasis on somatization of com-
plaints) a complete biomedical interpretation such as that
used by the modern medical professional would rarely occur
to the person experiencing the disease. For the patient,
altered bodily sensations are seen to fuse with the socio-
cultural interpretations and form a culturally valid
picture which symbolizes a disarticulation of the self.

It should be noted here that in the case of the patient as
opposed to Western medical practice there is no dichotomous
relationship between mind and body and the self represents
the total individual.

In sum, even though the individual may use bio-
medical terms, his response to this condition tends to be
very social and emotional in nature. The nature of this
difference lies in the fact that unlike the medical
practitioner who sees the problem in discrete biomedical
terms, the sick person feels that something is wrong with
him as a whole individual, and his sickness is likely to
be reflected in everything he does. His illness represents
both an individual and a social event where he attempts to
come to grips with his physical and mental well-being by
evaluating the meaning of any symptoms of which he has
become aware. This illness behavior response is broadly
defined as one dimension of a set of cultural roles and
strategies which extend from a person's social and cultural

environment and which have evolved to cope with the

33

potential chaos of disease situations. These rules and
strategies, which are under continual modification, first
take shape through the observation of illness models acted
out by members of the person's immediate social network
(family and significant others) through the process of
enculturation. The result of this process is the establish—
ment of a system of patterned discrimination and categori-
zation of illness symptoms and the appropriate responses
thought to be most efficacious in reducing such symptoms

or at least accepting them. This illness behavior response
in time should become modified depending on the nature of
additional personal experiences. The nature of these
experiences will depend on interactions with other people
(e.g., physicians, traditional healers, co-workers, friends
or the corner druggist) and on other sources of illness
information such as books, radio, and television.

This illness behavior process is composed of three
interdependent stages; recognition of some altered health
state, an evaluation of this state, and the initiation of
some ascribed mode of action in response to the evaluation.
With the concern of this project being on those influences
operating prior to the last stage, only the first two stages
will be considered. These two stages can be broken down
into analytical elements that are helpful in assessing
cultural influences on help seeking behavior in response

to disease.

34

The occurrence of a particular disease usually
involves a process that gives rise to certain organismic
changes which result in altered body sensations. Of course,
in populations where such symptoms or sensations are
ubiquitous, both the society and the individual could
perceive the condition in conventional terms such as being
quite normal as part of the ageing process or to be expected
with particular occupations (Koos 1954, Mechanic 1974). In
situations where this is not the case, people rely upon
other sources of information for defining the character and
importance of such sensations (Mechanic 1974: 61). Several
authors have noted that there is a cultural aspect to these
other sources of information and that these culturally
oriented factors can noticeably influence a person's
response to various symptoms (Zborowski 1952, 1969;

Opler 1961; Zola 1966; Hetherington and Hopkins 1969).
These influences are related to person's particular socio-
cultural milieu and can involve a number of factors such as
concepts about body function and image (Opler 1961) and
concerns about disease causation (Glick 1967: 49). This
implies that, within a culturally homogeneous group of
people, an analysis of the patterns of response to particular
symptoms or sensations should uncover clues for the
differential effects of a people's cultural milieu when
compared to another culturally homogeneous group. These
response patterns should reflect a medical taxonomy and

course of action that corresponds to particular cultural

35

categories specific to particular groups. If a person has
"gastrointestinal changes that are reflected in altered
abdominal sensations of some sort, they will be perceived
and given expression in terms of the relevant categories
that the person has at his disposal" (Fabrega 1977: 211).
Although an analysis of symptom response can give
some insights into how cultural factors can influence
illness behavior, the evaluation stage has even more
explanatory potential, especially when a phenomenological
perspective is used. The types of experiential categories
useful for analysis in this stage include such sources of
illness and disease information as: 1) personal medical
experience; 2) family medical experience; 3) interaction
with persons other than family members; and 4) exposure to
other extrinsic sources of information. An individual's
personal medical experience involves a first hand contact
with illness. Contrary to popular "scientific" opinion,
people do not experience medical situations as discrete
event but on a continuum. That is, they view their present
medical situation in relation to what their past personal
medical history has been and by comparing it with the
histories of others. For example, if an individual has had
a history of intermittent rectal bleeding due to peptic
ulcers or hemorrhoids, he may be less likely to interpret a
recent onset of bleeding as needing intervention than if he
had never had similar problems before. In the process of

evaluating our health status, we constantly weigh all of the

36

possible explanations for particular symptoms. While some
of these alternatives stem from past medical experience,
they still pass through a cultural filter and are perceived
accordingly.

Exposure to family medical problems presents a
situation with the addition of various behavioral models
provided by members of one's immediate social network
during the developmental and maturational process. Included
in the observation of these illness behavior models is
information provided in the form of informal discussion
which includes evaluations on how well or how poorly a
person may be managing their condition. For example,
Mechanic (1974: 61) has noted that children learn at very
young ages how to "respond to various symptoms and feelings
in terms of reactions of others to their behavior and
social expectations in general." O'Frake in his study of
the Subanun, has presented evidence in support of Mechanic's
observation (1977: 185) as have other researchers under
different social circumstances (Hallover 1972, Pless and
Satterwhite 1973, Litman 1974). The substance and method
of communication of illness behavior from one's societal
members is the same as for his familial source of informa-
tion, except in the temporarl sequence in which they occur.
As Friedson (1970: 290) has correctly noted, the degree of
cohesiveness of the social group of which a patient is a

member determines to an extent how that person will respond

37

to an illness episode. This information then is secondary
and in a sense less influential because of its peripheral
relationship in the enculturation process as compared to the
family.

Exposure to other extrinsic sources of information,
such as less personal social contacts, and (especially in
industrialized societies) modes of communication such as
radio and television, can provide behavior cues which have
an impact on a person's overall illness behavior. These
factors are less demonstrable than ones previously listed
because of their diverse nature and have a lower potential
for analytic application. This makes an analysis of such
influences a long and involved process that is more
suitable for follow-up studies performed after a pilot
study has been conducted on the more accessable factors.

Illness behavior information is transmitted by the
family, its kinship, and friendship networks within a
cultural context and influences the manner in which
individuals define and act upon symptoms or life crises.
Most of the researchers concerned with health care
utilization issues have given some attention to these
implications, albeit there have been few attempts to
specify their nature (Mabry 1964, Haggerty 1965, McKinley
1972). As in the investigations involving symptom duration,
these utilization studies have tended to look at all patient
behavior as a homogeneous set. Many have not given emphasis

to the fact that although biomedical conepts have provided

38

help for certain classes of medical problems, such concepts
are "not only irrelevant to others (classes of problems),
but misleading because they misdirect our efforts"
(Eisenberg 1977: 19). It is within the context of problems
such as these that anthropological theory has the potential
to not only test certain questions about human behavior
in regard to cultural factors, but to provide clues that
may aid in the implementation of more effective solutions.
The theoretical concepts of the biomedical model used so
often by Western medical practitioners and reserachers needs
to be appended with cultural concepts in approaching
problems of health behavior and medical care systems in
modern societies.

The present investigation involving U.S. Whites
and U.S. Blacks (both with low levels of income) draws on
the premise that certain subcultural differences exist
relative to preventive health behavior. A review of the
literature on low income Blacks indicates that the beliefs
previously discussed are fairly widespread among this U.S.
population (Wintrob, Fox and O'Brien 1971; Michaelson 1972;
Snow 1977; 1978b). Since this system appears to be fairly
homogeneous and shows some contrast with that of the
beliefs held by the White majority, a comparison of racial
differences in health behaviors would seem amenable to the
approach outlined above. One could suggest that the
congruence between biomedicine and subcultural factors may

be greater for Whites than for Blacks. Although this does

39

not imply any directional relationship for preventive health
behaviors and cancer, it does make a case for expecting some
differences to exist. However, any group difference
uncovered can not automatically be assumed to originate from
these basic differences. For example, observed behavioral
differences could be the result of more subtle variations in
body attitudes or certain aspects about bodily function or
even socioeconomic factors such as past patterns of health
care utilization. Research of this nature provides a basis
for examining some of these possibilities and clearly does
not depend upon the comparison of two separate cultures to

provide useful and meaningful information on such issues.

CHAPTER II

A HISTORICAL BACKGROUND INTO THE

STUDY OF SYMPTOM DURATION

Investigations into the phenomenon of symptom dura—
tion have been conducted since the early 1900's and have
resulted in hundreds of professionally published reports
(cf. Kutner et a1. 1957, Blackwell 1963, Antonovsky and
Hartman 1974). Most of these studies were designed and
carried out by medical and paramedical personnel. These
researchers viewed symptom duration in light of such patho-
logical implications as length of duration and its effects on
patient prognosis. This concern with pathological implica-
tion (as mentioned earlier) was due to the general somatic
orientation of the Western medical model. When this orien-
tation was combined with the premise that a longer duration
resulted in a lower survival potential and with the often
paternalistic stance that physicians took in regard to
patients, symptom duration became weighted with value judg-
ments. The length of symptom duration was seen as a biolog-
ically maladaptive phenomenon and the longer it tood for a
person to come in for care, the more that person's actions
were judged to be a negative value. The result of this

negative connotation was the creation of the concept of

40

41

"patient delay" in seeking medical care. With this orien-
tation investigations were structured to enable researchers
to fix some degree of culpability on the individuals
involved. By identifying individuals or classes of individ-
uals prone to delap it was assumed that the problem could
eventually be corrected through appropriate education. Also
by quantifying the actions of these individuals over time it
was held that various educational programs could be evalu-
ated for their effectiveness in altering unwanted behaviors.
Generally, delay was defined as that period of time
between the conscious recognition by a patient that a given
condition or sensations may be pathological (i.e., a symptom
of something wrong) and the point at which a therapeutic
action relating to that symptom or symptoms is undertaken.
This period of delap was divided into two components:
"patient delay," the time elapsed between a conscious
recognition of a pathological symptom or sign and the first
presentation and the initiation of therapeutic action.
Several terms have been noted in the literature which may
shed some light on the type of attitudes medical profession-
als used in the study of retarded or prompt health care
utilization. Terms like reasonable or undue delay (Pack and
Gallo 1938: 443) and avoidable or unavoidable delay (Kutner
et a1. 1958: 96) suggest that persons using such terms felt
that the physician or patient fully understood the problem
from the stand point of the medical system. While this is

certainly possible for the physician it would not seem very

42

likely in the case of the patient. Even simplistic terms
like delay or non-delay (Goldsen et a1. 1957:1, Henderson
et a1. 1958:27) tend to carry similar connotations. The
impartial term of lagtime suggested by Worden and Weisman
(1975:69) is perhaps an indication of some change in this
general attitude toward the patient who takes a little
longer than others to come in for advice or care.
Evaluations of delay have for the most part been
based upon some arbitrary point in time beyond which the
delayer was held clinically responsible. The most frequent-
ly used definition of this criteria of delay was originally
defined in a study by Pack and Gallo "delay on the part of
the patient is designated as reasonable when...it is under
three months; as undue delay when this interval is three
months or over" (1938:443). Pack and Gallo's study repre-
sented the first large scale investigation (1,000 patients)
into the phenomenon of symptom duration and as such influ-
enced many subsequent studies to also use a three month
marker (Leach and Robbins 1947, Gray and Ward 1952,
Henderson et a1. 1958, Cameron and Hinton 1968). However,
investigators have used time periods ranging from one week
(MacDonald 1947, Sugar and Watkins 1961) to four months
(Shedden 1939) with most being one month (Stearns 1950,
Guiss 1955, Soost and Thomas 1969). Although the use of
such time markers has become traditional in the study of
delay, current researchers have suggested using it with

caution (Shuval 1970) or not using it at all (Hackett et a1.

43

1973). Also, some investigators are beginning to question
the concept of delay as an effective means of looking at the

situation in general (Worden and Weisman 1975).

Existence of Delay

 

Despite the fact that there have been problems in
identifying delay in cancer diagnosis (with the exception of
heart disease, most delay studies have focused on cancer),
much effort has been made to demonstrate its existence.
Wainwright (1911) set the stage for later studies of delay
when he reported on the effects of "patient procrastination"
and physician's failure in recognizing cancer on patient
prognosis. After the publishing of Wainwright's research,
studies on the frequencies of delay became more widespread
(Gibson 1915, Farr 1919, Simmons and Daland 1920, 1924).
These earlier studies made no attempts at dividing delay
into the above mentioned classifications of avoidable and
unavoidable delay or as Blackwell (1963:6) has stated "bad
delay or not-quite-so-bad delay". Delay was divided into
patient and physician categories but the evaluation was
totally negative. Not until after the Pack and Gallo study
in 1938 did delay become classified as either a negative or
a not so negative action. Subsequent investigations reported
that for the most part, physician delay (defined as more than
one month) was reported from twenty-three per cent (Cameron
and Hinton 1968) to seventy-six per cent (Henderson 1966).

Unfortunately the only conclusion that can be drawn from an

44

examination of these data is that delay exists.

Except for the uniformity of using a three month
temporal marker for delay (which excludes many studies using
other definitions of delay) and a general focus on cancer,
these studies are difficult to compare for several reasons.
First, there has been no strong attempt at controlling for
specific cancer sites even though it has been common know-
ledge for some time that cancer sites have variable rates of
growth and development (even within single sites, Cooper and
Smith 1975, Hughes 1976). Second, by not controlling for
the cancer site, problems in comparability arise because of
the degree of variability in site related symptoms. Third,
care has not been taken to adequately identify the popu-
lations from which the samples have been drawn and most
populations that have been clearly outlined have not been
analyzed for secular trends. There are exceptions to this
aspect of sample identification; the research of Pack and
Gallo (1938), Leach and Robbins (1947), Robbins et a1.
(1950), and Robbins et a1. (1953) were all conducted on
patient populations from the same institutions over a period
of twenty years. But even with these studies one can not be
sure if the patients were drawn from comparable populations
since many factors can alter institutional utilization (e.g.,
changes in institutional status as treatment centers).

Last, as also pointed out by Antonovsky and Hartman (1974:
109) there is a methodological problem concerning the nature

of the questions asked of the patients and the setting in

45

which the questions were asked. "Quite conceivably, one
might obtain different answers from the same person, depend-
ing upon whether he is asked about a pain episode, a
persistant pain, or an unusual sensation" (Antonovsky and
Hartman 1974:109). Blackwell (l963:9) quite adequately sums
up when she states,

Clearly, setting a criterion for patient delay in terms
of time does not permit a realistic analysis of the
nature and meaning of delay . . . neither an attempt to
fix responsibility for delay on physician or patient nor
an effort to show differences in delay across time can
do more than point up the fact that the phenomenon of
delay exists, has existed and will continue to exist,
and that inquiry into the nature of the phenomenon might
be more productive in terms of enabling health prac-
titioners to deal effectively with the problem.

Variables in Symptom Duration

 

As mentioned above, the literature on symptom dura-
tion prior to diagnosis and treatment is almost entirely in
the field of cancer. However, the investigators involved in
this research have approached symptom duration from several
directions. Generally there are five classes of variables
that have been used most frequently in the study of delay.
These classes of variables can occur in any combination and
include: (1) sociodemographic factors such as age, sex,
occupation, income, education, social class, religion and
rural and urban residence patterns; (2) cancer related
factors such as knowledge about the existence of cancer and
its various symptoms, and previous experience with cancer in
either a direct or indirect manner; (3) psychological factors

such as a person's general emotional well being, their

46

intellectual capacity, hypochondria, concepts of body image,
and coping behaviors related to fear of cancer; (4) concepts
of physicians and the health care system such as attitudinal
and behavior relationships of an individual to physicians
and the existent health care system; (5) sociocultural
factors like ethnicity and its concomitant aspects such as
culturally learned behavioral dispositions toward medical
care and preventive health behavior, and the effects of
attitudes and behavior of others on individuals who have

noticed the presence of unusual signs or symptoms.

Sociodemographic Factors

 

Although patient age has been a frequently tested
variable for causal effects on delay in a large number of
studies, its significance remains unclear. There appear to
be more positive than negative correlations for the effects
of age. The majority of the research reporting an associ-
ation with age and duration have indicated that older
persons seemed to wait the longest before seeing a doctor
(King and Leach 1951, Cobb et a1. 1954, Shapiro et a1. 1967,
Fink et a1. 1968). However, a few studies (Soost and Thomas
1969, Antonovsky 1972) have reported a tendency for younger
people to wait longer than older people and one study
(Breslow and Hochstim 1964) reported that persons under the
age of thirty and over the age of sixty-four tended to wait
the longest. These seemingly conflicting results do not
necessarily detract from the possibility of an age relation-

ship but the number of non-associations reported certainly

47

calls any assessment of a causal nature into question (Guiss
1955, Titchener et a1. 1956, Antonovsky 1972, Hackett et a1.
1973, Worden and Weisman 1975). For example, since elderly
Apeople tend to be more heavily represented among the lower
socioeconomic and poorly educated groups, there is a
possibility that age could covary with economic and educa-
tional status. Such a relationship could present different
results for age effects depending on the nature of the
intervening variables. Also, as Blackwell (1963:14) and
Goldsen (1963) have mentioned, experience (e.g., past
relationships with physicians) can also be a factor of age
and further confound the influence of age on duration.
Another reason why the correlations with duration
and age should be approached with caution concerns the fact
that the morbidity by age of most cancer sites tends to be
skewed to the right with an over representation of the
elderly in the long duration classes. The only way to avoid
this possible error in sampling is to control for the cancer
site so that the rate of incidence for age is known.
Unfortunately, except for two studies that involved breast
cancer screening clinics (Shapiro et a1. 1967, and Fink et
a1. 1968) none of the studies showing a correlation with
older age and length of duration have controlled for site.
Although the screening studies did control for site of
cancer and reported that older people waited longer than
others, there have been reports of younger people waiting

longer in similar screening programs for the same site

48

Antonovsky 1972). Regardless of the information made avail-
able through these studies, we cannot be sure that the
person who delays seeing a physician for some condition or
symptom would be the same type of person who would reject
participation in a screening program or delay seeing a
physician for a different symptom.

Of all of the studies examined, only one showed any
relationship between the sex of the individual and the
length of "delay" (King and Leach 1951). This relationship
was considered by the investigators as being rather minor
and in this case representing some interaction with patterns
of medical care habits.

Although some of the investigations on symptom dura-
tion have checked such variables as levels of education,
income and occupation, most researchers have collapsed these
factors into the general category of socioeconomic status.
This category is generally believed to "reflect the balance
or net effect of social, environmental, situational,
educational, financial, and other forces in the individual's
personal world" (Coburn and Pope 1974:67) and therefore is
a reflection of a person's general life style. Out of the
studies reviewed that have found correlations with socio-
economic status or its components (King and Leach 1951,

Cobb et a1. 1954, Goldsen et al. 1957, Kutner and Gordon
1961, and Hackett et a1. 1973) only one (Antonovsky 1972)
differed from the general concensus that persons of lower

socioeconomic status tend to wait longer than persons of a

49

higher status. Of the investigations reviewed, only a few
reported no relationship between delay and socioeconomic
status (Mossetti et a1. 1970, Worden and Weisman 1975) or
educational level (Henderson et a1. 1958, Fisher 1967) the
remainder of the studies did not provide any information on
these variables at all.

Further investigation into the effects of socio-
economic status on preventive health behaviors is clearly
indicated because of the underlying interrelationships
involved. Levels of education and income may, in addition
to providing frameworks for the development of personal
habits in health assessment and care utilization (Waitzkin
and Stoeckle 1972), provide differential access to social
networks and thus could be influencial in the development
or maintenance of particular health care patterns (Enelow
1976:66). Also, such complex psychosocial attitudes as
powerlessness or apathy so common in lower socioeconomic
classes could have effects on medical care habits. For
example, in commenting on the supernatural component of
urban Black medical beliefs, Snow (l978a:69) has observed
that for people who feel powerless to control their environ-
ment, beliefs like bad luck, evil influences, or magic as
causes for illness or misfortune are to be expected.
Alternative beliefs could in part explain such observations
like "being black--and especially being poor-~appears to
militate against the utilization of certain health services"

(Cockerham 1978:70). What is suggested here is that under

50

certain conditions, ethnically exclusive social relation-
ships may tend to channel help seeking behavior toward the
group rather than toward orthodox medical care.

Two marginal aspects of life style (i.e., urban-
rural residence and religion) have also been given minor
attention in these investigations. In the only report
reviewed concerning urban-rural residence, Goldsen et a1;
(l957:2) in a study of 727 patients living in New York
reported that persons living in the rural areas of upstate
New York tended to wait longer than persons from New York
City. However, since no mention was made in this study
about the ethnic or racial characteristics of the sample,
it would be somewhat difficult to fully interprete Goldsen's
results. Such a finding could represent some regional
relationship with preventative health behavior and for this
reason should be checked by further research sensitive to
such factors as ethnicity and social class membership.

Occasionally, mention has been made about certain
religious creeds being related to delay such as Protestant-
ism (King and Leach 1951, Kutner and Gordon 1961), Catholi-
cism (Fink et a1. 1968) and religion in general (Titchener
et a1. 1956). The effects of religious behavior can be
quite complex and elusive and, as Comstock and Partridge
(1972) point out, a superficial approach of merely listing
a person's religious preference misses the boat in terms of
controlling for its effects on health behavior. They

suggest that by focusing on church attendance, some of this

51
complexity can be eliminated. Since attendance by itself
may also be a gross measure of influence one could take
things one step further and also look at religiosity or the
content of religious domains. For example, a more formal
religion might affect people of a low religious activity
just as much as a less formalized religion would for people
of a higher activity level. Only one study was reported to
have at least used attendance in relation to religious
influence on cancer patients (Monk et al. 1962). This study
involved colon and rectal cancer patients and although no
relationship was found for colon patients, rectal patients
were reported to attend church services less and were more
likely to not be a member of a religious body. A study of
this character has no value at all for the problem at hand
and if it has any utility at all, it would be to show how
the principles of statistics and epidemiology can be grossly
mishandled.

Whether or not the approach suggested by Comstock
and Partridge can be of any value in the understanding of
symptom duration is a moot point at present because no one
has used it in this regard. To be sure, if I wanted to
control for religious factors operating on health behavior
(e.g., the influence of religious dogma or of reference
groups that patronize particular churches), I would not use
just religious preference. The level of effect that these
factors may have on an individual's health behavior can not

be adequately controlled by merely checking a person's

52

nominal religious affiliation.

Cancer Related Factors

 

One of the most popular cancer related variables
used in studies of delay has been patient awareness or know-
ledge about cancer in general and its various symptoma-
tologies in particular. The usage of this variable has to a
large extent been the result of a heavy emphasis on the
efficacy of cancer education. The literature on delay and
cancer detection has been replete with comments on the use
and effects of cancer education and diachronic studies have
often been designed to permit the evaluation of educational
programs (Robbins et a1. 1959, Blackwell 1964, Kelly and
Thiene 1967, Kegeles 1973). As early as 1943 Harms and
associates reported that in a sample of 158 cancer patients
from a New Haven hospital, the major cause of patient delay
was a lack of "proper" information about cancer regarding
knowledge of its existence and the nature of the associated
symptoms (1943:337). A review of the articles focusing on
the patient's knowledge of cancer shows that there is no
simple relationship between cancer knowledge and preventive
health actions. Although some of the studies suggest that
knowledge areas have a relationship to action (Harms et a1.
1943, King and Leach 1950, Aitken-Swan and Paterson 1955,
Kutner and Gordon 1961), the methodologies are not con-
sistent and comparisions of the results are therefore diffi-
cult. For example, Smith (1935) in a study of 95 cancer

patients (several sites) from a hospital in New York

53

reported that over half of the forty-one patients with
breast symptoms delayed because of "ignorance or procrasti-
nation". Here, two separate concepts of behavior which
differ in their contextual relationship with symptom dura-
tion have been treated as one, and in fact are nothing more
than descriptions of the behavior they are intended to
explain (Kegeles 1976:53). In some of the articles it is
not clear as to whether their reference to ignorance is for
cancer in general or for a failure to recognize and identify
specific symptoms or even a characteristic of the person
involved (Leach and Robbins 1947). Some studies in the
other extreme have ruled out ignorance altogether as a
factor in duration length. Titchener et a1. (1956:1192)
reported that,
people resist diagnosis and treatment seldom because
they are ignorant of the significance of a change in
themselves but oftener because the appearance of a sign
or symptom sets off a sequence of maladaptive and
neurotic behavior resulting in irrational procrasti-
nation.
Titchener noticed, as have others (Hackett et a1. 1973,
Worden and Weisman 1975), that various psychological factors
can confound the relationship between the absence of know-
ledge about cancer and delay by interacting with the symp-
toms the patient perceives. Such factors as fear of treat—
ment, hospitals or doctors (Titchener et a1. 1956), belief
in curability (Eardley 1974), shame (Sandifer and Pritchett
1958), and fear of diagnosis (Worden and Weisman 1975), as

well as others were found to mimic the effects of a lack of

knowledge and deserve adequate attention for research. A

54

more complete discussion of these factors as well as others
will be given in the following section on psychological
factors in symptom duration.

The site of cancer or locale of the symptoms from
the cancer has been used as a criteria for comparing delay
patterns in a number of investigations. The most consistent
finding regarding site show a higher proportion of patients
waiting longer for external or more superficial lesions than
for internal lesions (Goldsen 1953, Goldsen et a1. 1957).
However some investigators have reported shorter lengths of
duration for breast cancer (Leach and Robbins 1947, King and
Leach 1951, Hackett et a1. 1973), some have shown less
duration for other sites (Smith 1935), and some have reported
no difference in site location (Simmons and Daland 1920,
Cobb et a1. 1954).

Another cancer related factor that is intimately
tied to cancer site and knowledge about cancer is the nature
and perception of the symptoms. Many investigators have
reported that patients with long delays tended to believe
their symptoms were not serious (Pack and Gallo 1938, Bates
and Ariel 1948, Aitken-Swan and Paterson 1955, Worden and
Weisman 1975) or at least not unusual (King and Leach 1950,
Goldsen et a1. 1957). Paterson (1955:933) noted that when
a longer delay was found it was "related to the painlessness
and apparent triviality of their lesion"3. Also, other

studies, notably Henerson et a1. (1958:33) and Worden and

Weisman (1975:75) reported either no significant difference

55

between delayers and nondelayers in regard to dramatic symp-
toms or found presenting symptoms to have no relationship at
all with symptom duration. The most interesting interpre-
tations of the present data were reported by Goldsen (1953)
and King and Leach (1950). These investigators felt that
their results indicated that length of duration was more
related to the patient's usual behavior toward any symptom
than towards symptoms of cancer. Unlike the previous
reports and many of the later ones, these researchers did
not view disease events as isolated happenings, but saw them
as occurring on a continuum and therefore they looked at the
behavior of the patient as a continuous series of evalua-
tions.

It would seem that the vast majority of investi-
gations into symptom duration regarding site of cancer and
its resulting symptoms have used the same perspective as for
other factors such as age, socioeconomic status, and know-
ledge about cancer. This perspective, which seems to be the
result of the particulate nature of the Western medical
model, involves a focusing of attention on a particular
variable with an apparent tendency to disregard possible
interaction effects with other factors. Whether or not this
approach bias is real or apparent is not clear; however, the
continued insistence on the relevance of certain variables
over others suggests that the bias is real. In the case of
symptom perception the evidence would seem to point to the

fact that people treat cancer symptoms just like any other

56

symptom and that, if in their mind, the symptom seems
serious they will take action. Also, while certain combina-
tions of symptoms may be given minor attention a different
combination of the same symptoms may give rise to consider-
able concern. For example, while Blacks tend to consider

a symptom set like weight loss coupled with poor appetite

as minor or natural (in the domain of orthodox care), they
become quite concerned about the possibility of witchcraft
when they lose weight and have been eating normally (Snow
l978a:81). This latter case is usually grounds for seeing a
lay healer who is capable of combating the evil influences
of witchcraft. Such different responses could lead to a
differential utilization of health care services and there-
fore conceivably affect symptom duration times in cases of
colon and rectal cancer since the above symptoms are common
for such malignancies. The factors behind this judgment
process would seem to be intimately linked to a variety of
relationships which Titchener et a1. (1956:1190) rightly
interprets as operating before, during, and after an
individual's recognition of a symptom of illness.

One major factor that is involved in this judgment
process is a person's past experience with cancer. Attempts
have been made to assess this experience most often in the
form of whether or not a person has had cancer themselves or
known of someone who has had cancer. It was assumed that
through this contact a person develops certain opinions about

such things as the efficacy of seeking medical care and the

57

treatment regimens involved. These opinions in turn deter-
mine that person's willingness to take part in an examina-
tion for early detection. The data presented indicates,
with the exception of McCullough and Gilbertson (1969),
that, contrary to the belief that persons who know people
that have developed cancer will tend to report early,
increased knowledge of people with cancer tends to promote
delay. For example, Cobb et a1. (1954:922) observed that
patients who reported that cancer had occurred in their
families tended to delay longer than patients who reported
no cancer in their families medical histories. Results
presented by Hackett et a1. (1973:17) indicated that while
patient reported family cancer events seemed to have no
effect on delay, patients who reported that cancer ran in
their families were most often delayers than nondelayers.
Considering these results it would seem that what a patient
subjectively feels (i.e., cancer runs in their family) has
more influence over duration time than what the patient

objectively knows (i.e., cancer has occurred in the family).

Psychological Factors

 

It would seem that although site, age, income,
educational level, symptom perception, and knowledge about
cancer through some experience may, to some extent,
influence symptom duration, they may do so by their relation

to more basic psychological and sociocultural factors. This

58

present section will cover those investigations that have
studied symptom duration in light of particular psycho-
logical issues. As Antonovsky and Hartman (1974) have
noted, there seem to be four major issues that have received
most of the attention in such investigations: general
emotional health; intellectual capacity; hypochondria and
body image; and styles of coping with fear and cancer.
Studies about psychological factors in symptom
duration began in the 1940's (Blackwell 1963: 17-18) with
Youngman's (1947) Australian study. In Youngman's
investigation of persons who "should have come earlier for
treatment," many emotional abnormalities, i.e., uphoria,
apathy, nerousness, anxiety, surliness, etc., were reported
for persons having long symptom duration times. However,
the utility of this study is in question since its
research design is flawed by the absence of any control
group and by the impressionistic character of the data.
Henderson et a1. (1958: 36) reported that in 100 cancer
patients (all sites), from two English-speaking hospitals
in Montreal, over fifty-five per cent of the delay group
had a history of psychiatric symptoms whereas only nineteen
per cent of the non-delay group had such histories. A
later study, unfortunately by the same author, confirmed
these findings (Henderson 1966). On the other hand there
have been studies that have reported no relationships

between psychiatric conditions and delay (Titchener et a1.

59

1956, McCullough and Gilbertson 1969). For example, in
Titchener's study, which was part of a larger psychiatric
survey of surgical patients in the United States, there
were no correlations between delayers and non-delayers

in regard to the presence of neuroses, psychoses, character
and behavioral disorders, mental deficiencies or the
absence of psychiatric illness (1956: 1190). The most
common psychological influence for delay in this study
were fear of punishment from surgical treatment, i.e.,
tensions about pain or mutilation, and fear of death in
surgical treatment (1956: 1190).

While these data would at first seem to suggest
that delay is not the result of any particular psychiatric
condition, an alternative explanation is possible when the
concept of cultural variability is added to the discussion.
With several cultural populations being represented in
these studies, it is possible that such a situation of
cultural variability could account for some of the con—
flicting results. One could suggest that these different
results could be due to some interaction between psycho—
logical and cultural factors or a misrepresentation of
cultural differences as psychological variation by the
authors. By looking at these studies from the perspective
of anthropological theory such seemingly conflicting results

are expected and not considered inconsistent.

60

Other studies have focused on intelligence. These
investigations used intelligence not in the sense of know-
ledge about cancer but in the old tradition of I.Q. tests.
Although Youngman's study reported a low average I.Q. for
patients that wait longer, his research suffers from
certain design errors which result in a very unrepresenta—
tive sample of patients. The remainder of the studies that
investigated I.Q., notably Aitken-Swan and Paterson (1955)
and McCullough and Gilbertson (1969) found no relationship
between intelligence and delay.

The only other two aspects of personality that have
been reported with any frequency have been hypochondria and
body image. Reznikoff (1955: 456) reported a high per-
centage of hypochondriacs among persons who showed up for
screening clinics. Studies by Henderson et a1. (1958),
Hammerschlag (1964), Henderson (1966), and Fisher (1967)
have also offered support to Reznikoff's conclusions. Both
Hammerschlaf and Fisher used the Fisher-Cleveland Barrier
test and found that persons with long symptom durations more
often than not scored high on the test. A high scorer was
an indepentent person with defined standards and goals and
was persistant in achieving their goals without being
diverted by stressful situations. Fisher has suggested
(1967: 677-678) that these persons are often characterized

by a false sense of body security and a need to remain

61

independent and therefore in opposition to the actions of a
hypochondriac will reject a placement of dependence on
therapeutic measures. Other studies such as Sugar and
Watkins (1961) have also reported this tendency of denial
of body changes for persons with longer symptom durations.
To be sure, there are probably many psychological
factors involved in symptom duration but they can not be
easily classified into some particular diagnostic category.
Worden and Weisman's (1975) study reflects this impression
and by using several psychological instruments such as the
profile of mood states (POMS), the Minnesota multiphasic
personality inventory (MMPI), and the index of predominant
concerns (IPC), has approached the phenomenon of symptom
duration from a more person oriented perspective rather than
an impersonal and more pathogenically oriented perspective.
By focusing on the present psychological state of the
patients they found high levels of tension, fatigue, con—
fusion, and total mood disturbance to be significantly
correlated with symptom duration such that delay was
increased. Also, they and others have found relationships
between symptom duration and past situations of psychological
disturbance such as poor relations with one's spouse, multi-
problem family or origin, and marital problems in general
(Worden and Weisman 1975:73-74, Aitken-Swan and Paterson
1955:625). These results and the ones discussed above on
hypochondria and body image would seem to indicate that

psychological factors are important during the decision

62

making process of whether or not to seek care. However, as
yet it is not understood what psychological mechanisms are
in operation except for reactions related to hypochondria,
body image, and fear or apprehensions about the diagnostic,
surgical and social social outcome of the detection visit.
Perhaps these issues could be better examined through longi-
tudinally designed studies rather than the retrospective
ones that have been used in the past.

Probably one of the most widely discussed psycho-
logical factors thought to influence symptom duration has
been cancer phobia (cf. Kutner et a1. 1958, Blackwell 1963,
Antonovsky and Hartman 1974). Although a lot of effort has
been devoted to finding the causes for fear and anxiety in
cancer, the goal of most studies has been to examine the
styles of coping strategies used to handle such fears. In
general, the conclusions of most studies dealing with these
psychological factors have been that there are two basic
patterns of response to fear in regard to cancer. The first
type of response characterizes the patient with a long
symptom duration and involves an immobilization of action
and results from the occurrence of certain defense mechanisms
such as denial and repression of the person's symptoms or
general condition. The second type of fear response
characterizes the prompt patient who tends to mobilize
action in an active fashion to seek medical care or advice.
While a fear of impending death certainly has some important

relationship with cancer phobia, Titchener et a1. (1956:1190)

63

and Aitken-Swan and Paterson (1955:625) have noted that
expectations of pain, mutilation and general fear of doctors
or hospitals also tend to heighten the level of cancer
phobia. Aitken-Swan and Paterson reported that over twenty-
five per cent (78) of all patients they interviewed spoke of
some fear about operations and hospitals in connection with
their delay and Cobb et a1. (1954:923) reported that ninety
per cent of their sample (729) admitted to a general fear
that affected their actions in some way. In Cobb's study,
as with Titchener and Aitken-Swan and Paterson, fear was not
seen as some diffuse entity but as a tangible, lived-with
fear of long-drawn out suffering of some nature. The
pragmatic observations noted above can also be viewed in a
cultural context. For example, Zborowski observed that
Irish patient fears about the crippling effects of surgery
seemed related to a concern over wholeness of body and
strength (1969:218). Also, as Cobb has indicated, this fear
can go beyond the personal level of coping with pain or
mutilation and can range from concern over creating a
financial burden on the family to a fear of how the family
and others will behave toward them in the event of long

term separation (therapy) and potential death (1954:923-924).
MacDonald (1947) and Harms et a1. (1943) have presented
inconsistent results in that they reported that fear was not
very significant as a factor for delay except in a few cases.
Fear is a complex issue composed of a number of factors,

many of which can be anchored into several cultural domains.

64

As with previously discussed factors in delay, the observed
inconsistent results obtained for fear might be better
explained if it were viewed as dependent (in part) upon a
cultural substrate. Fear certainly warrants further investi-
gation, but before new research is conducted it should be
more clearly defined to allow for a more complete analysis
of its effects on symptom duration.

With only a few minor exceptions (Kutner et a1. 1958,
Worden and Weisman 1975) most studies have agreed that fear
in someway determines how a person will act depending on
certain previous conditions. It can result from worries
about physical discomfort, forced dependency, financial
burden, social unacceptability and a variety of other
reasons. All of this points to the fact that although fear
may be a psychological response, it is the result of a
multiplicity of factors which extend across categorical
boundaries and whose effects upon symptom duration are not
linear. For example, Cobb et al. (1954) and Goldsen et a1.
(1957) have shown that a knowledge about the significance of
cancer symptoms tends to increase delay when level of fear
or anxiety are high, but reduces delay when fear levels are
low. Some investigators feel that this low level of fear
reported by Cobb and Goldsen's research should actually be
classified as an intermediate level of fear. Antonovsky and
Hartman (1974:123) support this view and have suggested that
"a moderated degree of fear of the consequences of cancer is

conducive to nondelay, whereas both little and great fear

65

encourage delay".

Physigian and Medical Care Factors

 

There have not been very many studies that have
focused on the effects of patient-physician or patient-
medical system relationships and symptom duration in cancer.
However, the nature of these relationships as potential
barriers to prompt treatment makes it necessary to devote
some discussion to them no matter how slight. Of the
studies that have focused on these factors, their emphases
have been divided into two general concerns: availability of
detection and treatment facilities; and doctor-patient
relationships. Several studies have reported on the
relevance of access to treatment and screening centers as
important factors in symptom duration (Braund and Binkley
1942, Henderson 1966). In these studies access was defined
not only in terms of availability of the centers but in
terms of the patients ability to pay for services and their
ability to take time out from normal responsibilities to go
to the center.

Some attention has been given to doctor-patient
relationships. Here as before, fear of doctors or hospitals
can be cited as contributing factors in unsatisfactory
doctor-patient relationships (Aitken-Swan and Paterson 1955,
Titchener et a1. 1956). In addition to apprehension about
doctors and the medical system, others such as Hollingshead
and Redlich (1958) have mentioned communication problems

between doctors or middle or high social class and patients

66
of lower social status. Also, the belief that professional
diagnosis is no better than self-diagnosis (Kegeles gt a1;
1965, Kegeles 1969) or little confidence in the effects of
treatment (Aitken-Swan and Paterson 1955) seem to contribute
to problems of communication between doctors and patients.
In general, any factors that tend to contribute to an
uncomfortable relationship are seen as causal factors in
delay (Cobb et a1. 1954, Henderson et a1. 1958). Another
factor which has not been mentioned in the above investi—
gations centers around the problem causing potential of
cultural differences between the physician and the patient.
For example, Gans (1962:136-138) in his study on Italian-
American patients observed that even subtle ethnic difference
such as social distance could create extreme problems in
physician-patient relationships. In the care of Blacks
Snow (1978b) has noted that they tend to see orthodox
medicine as very strong and expect rapid results irrespec—
tive of whether the condition was clinically diagnosed as
chronic or acute in nature. Also, in situations where
alternative treatment is available "it is not unusual for
an individual to go to the doctor and use home remedies at
the same time" (Snow 1977:79) or discontinue the prescribed
medication in favor of home remedies (Snow 1974:92). Here
it is important to note that in order for such problems to
arise the physician and the patient need not be members of

two very distinct cultures.

67

Sociocultural Factors

 

Preventive health behaviors can be seen as the
result of an interaction between environment (i.e., both
social and material), and a more ideological level involving
values and norms of a person's particular sociocultural
background. The environment presents a set of situations
to which a person applies their knowledge from a particular
cultural milieu and arrives at some conclusion for appro-
priate behavioral action. In other words, an individual
acquires from their culture certain kinds of knowledge, an
evaluation of health in terms of particular values, and a
general understanding of what is considered appropriate and
inappropriate action. These factors, when interfaced with
particular environmental conditions, provide a person with
a choice of acceptable alternative actions or subsets of
cultural norms.

Few investigations have been made into the relation-
ship between sociocultural factors and symptom duration and
those that do treat the impact of such factors on only the
most general level. This general treatment of the data can,
by missing hidden relationships, at worst lead to conclu-
sions totally inconsistent with reality and at least result
in inconsistencies between similar investigative conclusions.
For example, the few reports in relation to a patient's
ethnicity have been quite inconsistent. Whereas King and
Leach (1954:225) found that first generation immigrants

waited less and had better health care habits than second

68

generation Americans, Breslow and Hochstim (1964: 110)
reported that immigrants (foreign-born Whites) waited

longer before seeking care than American-born Whites.
Although it should be noted that the first study was on
cancer patients while the second was on cervical cancer
screening participants, one is not sure as to the cultural
background or even the nationality of these immigrants. It
is very possible that several different cultural populations
have been represented and compared. The few Israeli studies
performed are also clouded by inconsistent results. Grushka
and Steinitz (1955) reported that non-European Israelis
tended to have longer durations, while Bar-Meir and Davies
(1960) and Moses and Cividali (1966) found no correlation,
and Antonovsky (1972) found no relationship in Haifa but in
Jerusalem reported that non-European Israelis wait the long-
est before seeking care. Unfortunately, as in the above
reports on immigrants, information on the cultural criteria
for defining the subject's ethnicity is lacking in these
studies. One wonders about the cultural homogeneity of
these groups. For example, are non-European Israelis from
Israel, North Africa, or other parts of the Middle East;

and are European Israelis from only one country, one region
(e.g., Eastern Europe) or are they from all over Europe.

The lack of sensitivity to such slightly more fine grained
cultural differences in these studies makes interpretation
of these results quite difficult.

As noted above, members of a particular culture have

69

been instilled with some notion about what constitutes
acceptable or unacceptable health behavior. Several of the
present studies on symptom duration have selected this
variable for analysis; however, as could be expected, the
investigators used their concepts of what is good (prompt
action) and what is bad (delayed action) health behavior to
ask questions of these data. With this in mind they focused
on the question of whether good health behavior character—
izes or follows the same pattern as the general medical
habits of an individual. Here it is interesting to note the
many times these investigators, in their discussions of
medical habits and delay, skirted the issue of culture with-
out seeming to grasp its significance for explanation. King
and Leach (1951:225) noted that "the value individuals place
on medical care and health are influenced by community
attitudes toward illness, family conditionings, contacts
with illness, and financial withdrawal." All of these
factors are intimately related with cultural systems of
value and norm but no mention was made about the possibile
implication of such ethnic influences.

Although the studies by King and Leach (1950, 1951),
Goldsen et a1. (1957), and Goldsen (1963) presented evidence
that was contrary to the common belief that patients react
to cancer symptoms in a way specific to the disease (imply-
ing some continuity in illness coping measures), only
Goldsen followed through with further discussion on what she

called socialization and cancer diagnosis. In so many words,

70

she defined this process of evaluation in a cultural context
when she stated that,
it seems to me that the data not only of cancer delay
studies but others . . . indicate that the model for the
approved social behavior we label as rational action on
the part of the laymen is not to expect the individual
to react issue by issue, problem by problem, or symptom
by symptom (like a diagnostician) but rather to be able
to relate each newly arising issue and each newly
appearing symptom to a total context of habits,
attitudes and values . . . almost a way of life
(1963:433).
Here, she was keying in not just on cognition but a special
category of cognition that had as its context a particular
type of socialization background. Aitken-Swan and Paterson
(1955:626) recognized the value of cues given by day—to-day
contacts with people in that they either reinforced the
person's desire to seek help or reaffirmed other attitudes
that resulted in longer durations, thus establishing a
relationship between a person's social environment and the
behavioral actions taken. Hackett et al. (1973:19) in
summing-up his impressions about "delayers" indicated that
he felt people pick up this "trait" from parents or siblings
who had handled similar situations in such a fashion, thus
focusing on a person's immediate social environment where
the ground work is laid for illness behavior.

I could continue to pick out various impressionistic
statements in these studies which imply that the investi-
gators had some peripheral grasp of the cultural influences
on symptom duration but it would be begging the issue. The

fact remains that only one of these investigations really

addresses the issue that cultural training has anything but

71

a minor influence on how long it takes for a person to seek
medical care. In fact there are no other studies reviewed
that used such a culturally oriented perspective to under-
stand behaviors related to cancer in general or symptom
duration in particular, and this observation has been
supported by others, notably Antonovsky and Hartman (1974:
120) and Kegeles (1976:104).

The only study that overtly confronted the relevance
of culture as an influence in symptom duration (albeit
slightly) was Titchener et a1. (1956). This study defined
delay as "the procrastination of a person with an out-
standing, noticable sign or symptom signifying, in his
culture, a deletrious somatic change" (1956:1187).

Titchener was aware that a symptom that could cause consid-
erable alarm for a person in one culture might mean some-
thing minor or even nothing at all in another culture. In a
later comment on the inherent difficulties of obtaining data
through patient interviews he made the observation that
delay seemed to be a complex form of behavior that was not
completely understood or realized by the patient. This
observation, in so many words, is an operational effect of
culture. People who are not in conflict about their value
'or norm orientations are seldom if ever consciously aware of
their cultural world view since such a framework puts logi-
cal boundaries on possible evaluations and subsequent
actions. Unfortunately, outside of supporting his central

hypothesis that delay "is a form of behavior multiply

72

determined by conscious and unconscious psychological factors
that are in operation before, during and after the patients
recognition of a sign or symptom" (1956:1193) he did not
develop the cultural factor further. Culture as an influence
on symptom duration received only minor attention along with
"other causes of delay". Whether or not this cursory

mention reflects a valid minor correlation or one resulting
from design errors in differentiating various cultural

groups can not be determined since the authors did not
provide any research findings or definable criteria for

"culture".

Comments on Past Studies of Delay

 

In this section I will not attempt to summarize the
data and ideas which have been presented in some detail, but
rather concentrate on the nature of these studies in regard
to their design and methodology and their relevance for
further study. First of all, several methodological weak-
nesses have become obvious in these investigations.

Although many of these studies were based on retrospective
data, when presenting their conclusions most did not mention
the inherent errors in such research. Perhaps the most
common weakness was that a large number of investigators
failed to provide explicit information on the measures

used (e.g., methods for establishing ethnicity or even
defining the nature of the sample population). Together
with an apparent lack of continuity and comparability

between studies, several investigations also failed to

73

provide information necessary to maintain any internal
continuity. Such a lack of internal integration seriously
impairs the relevance of any conclusions which are put forth
by such studies.

A second area of methodological weakness concerns
the basic research design in studies attempting to under-
stand symptom duration. While no single study should be
expected to deal with all of the possible relevant variables,
it should be kept in mind that any motivational factor for
symptom duration does not exist in an experimental vacuum.
Most of the studies reviewed would appear to be guilty of
focusing their research emphases in such a way as to present
their case at the exclusion of other possible interpreta-
tions. In other words, the concept of multiple causation or
influence is not often applied in these investigations.

The third methodological point I would like to make
relates to the tendency of these researchers to assume a
causal relationship (almost always linear) between a given
variable and symptom duration. Several of the studies
reviewed appear to be fraught with degrees of logical
positivism in that they tend to assume that by merely
collecting "X" amount of evidence to support an idea is
sufficient to prove that their idea or hypothesis is valid.
A common characteristic of many of these studies is to amass
a number of statistical results (often intertwined with
impressionistic statements) supportive of their hypotheses

and then to apply them in a cause and effect relationship to

74

validate certain impressions such as the effectiveness of
cancer education programs.

Although several of the studies that have been
reviewed provide some very interesting observations and
insights on symptom duration, most of them have fallen short
of providing an adequate explanation of why people behave
the way they do. I fully realize that such variables as a
patient's personality or emotional life experiences are of
utmost importance in influencing attitudes and behavior
towards cancer. Also, I recognize the importance of various
sociological and sociodemographic influences on symptom
duration. However, there are gaps present in these investi-
gations which confound any attempt at arriving at a complete
understanding of the various influences that can affect this
phenomenon. With the underlying assumption that behavioral
responses to cancer present identifiable regularities in
response along cultural models common to groups of similar
ethnic origin, some of these gaps may be filled. Unlike the
approaches used in the past studies on symptom duration,
this present study will be based on concepts drawn from
anthropology and will utilize a research orientation appro-
priate for the study of people as social beings and not as

discrete psychological or pathological categories.

CHAPTER III

COLORECTAL CANCER AND SYMPTOM DURATION

 

In the preceding review of the literature, several
investigations on symptom duration and cancer were discussed
in light of particular drawbacks in research and methodo-
logical design. These drawbacks were seen as seriously
undermining the weight of many of the researchers' conclu-
sions. The present study, to be discussed below, represents
an effort to circumvent some of these problems in order to
obtain a more complete picture of the phenomenon under
investigation.

This present study examines the reported time inter-
val between initial observation of symptoms and the seeking
of medical treatment using a retrospective hospital record
review of deceased cancer patients. The patients are all
males and had received their diagnosis and treatment from
hospitals under the Veterans Administration system. The
main focus of this investigation centers around the role of
ethnic differences in affecting the identification of and
response to a patient's illness. A major criterion of this
study is symptom duration, defined as that time period
between the first appearance or recognition of symptoms and
medical intervention as initiated by the patient. In sum,

this research is concerned with the differential effects of
75

76

ethnicity on secondary preventive health behaviors pertain-
ing to symptoms for colorectal cancer through an analysis of
reported symptom duration prior to diagnosis as a dependent
variable. This analysis was generated to explore the
possible role of sub-cultural differences in patient
response to colorectal cancer. The assumption here is that
these responses will present identifiable regularities
common to groups of similar ethnic origin. These behavioral
responses are therefore in part a reflection of an individ-
ual's cultural milieu.

Four hypotheses are presented below to test for the
general existence of these behavioral differences. 1) The
mean reported symptom duration times for individuals repre-
senting different ethnic backgrounds will be significantly
different. 2) Persons with a positive family medical
history for cancer (i.e., some suspected occurrence on the
part of the patient) will exhibit significantly different
between ethnic group mean reported symptom duration times.
It is also hypothesized that the within ethnic group mean
reporting times will be significantly different for persons
with a positive family medical history when compared to
those with a negative family medical history. 3) Individ-
uals with prior experience of nosologically similar (i.e.,
cancer related) conditions to the one under study will
exhibit significantly different between ethnic group mean
reporting times. 4) Of the symptoms found to be signifi-

cantly related to symptom duration for the sample as a whole,

77

those symptoms reported most often for the present cancer
episode will differ in content between ethnic sub-groups.
These hypotheses are somewhat general in their predictions
of the behavioral responses surrounding the phenomenon of
symptom duration due to the fact that there have been no
similar studies performed.

Research of this nature has the potential of pro-
viding useful insights for both the theoretical and applied
domains of the behavioral sciences in general and anthro-
pology in particular. Within the context of an anthropolog-
ical perspective, such ethnomedical research not only has a
potential for testing certain questions about human behavior
but also for providing insights into the application of
anthropological knowledge to particular medical problems.
Unfortunately, by the time many people have become sympto-
matic for colorectal cancer, the disease has usually trans-
cended the curative benefits of available treatment
regimens4. Thus, although cures have been reported for such
patients, most can be offered little more than pallative
care and survival is usually under five years. Because of
these developmental characteristics of colonic and rectal
carcinoma, this study remains basically a theoretical exer-
cise and can offer little in the way of beneficial insights
for the problems of early detection for colorectal cancer
per se. However, just as cross-cultural research has
implied that behavioral similarities exist for ongoing life-

ways and health matters in differing societies, behavioral

78

relationships found to exist for one form of cancer do not
automatically exclude an application to other cancers when
approached in similar contexts. It is also important to
note that since little behavioral research has been
conducted on colorectal cancer patients, any insights that
come to light may in some way prove beneficial for health
planning concerns, especially when the magnitude of the
colorectal cancer problem is considered.

Although the issue is not free from debate, there
appear to be indications that the period of symptom duration
has some relationship with a lowered survival potential
(poor prognosis) for colorectal cancer patients (Blackwell
1963, Scudamore 1969, Gerard 1975, Potchen 1975, Martin et
a1 1976).5 The longer a person takes to seek intervention
for symptoms resulting from this form of cancer, the greater
is his chance of having a poorer prognosis (Rogers 1974,
Copeland 1976, 1977). The estimated number of new cases of
colorectal cancer for 1978 has been reported at 102,000 men
and women (incidence is roughly equal for sex) with an esti-
mated mortality of 52,000 people (American Cancer Society
1978:9). Approximately 4,000 new cases and 2,000 deaths are
expected fbr the state of Michigan alone in 1978.

Except for skin cancer, colorectal cancer is the
most common single malignancy encountered in males and
females alike, and is second in males after lung cancer and
second in females after breast cancer (Seidman et a1 1976,

Silverberg 1977). These figures have remained quite stable

79

for some time (Rogers 1974:63) and the estimates have been
fairly accurate (e.g., for the 1974 figures the error was

6 per cent). Even with the major emphasis of the national
cancer program aimed at enhancing early diagnosis through
education and screening programs, the mean reported patient
symptom duration for colonic or rectal cancer has been
between four and six months, and the precentages of people
diagnosed at various levels of disease involvement have
remained unchanged for the past thirty to forty years.
Since the application of current means of treatment such as
surgery and chemotherapy have only a limited ability to
improve on colorectal cancer patient prognosis (Muldoon 1977)
every efort should be made to bring people to medical care
as soon as possible.

Thus, with the limited effectiveness of current
treatment and an apparent inability to implement massive
screening programs, the behavior of the patient remains one
of the few accessable components to the problem of early
detection and diagnosis. Although it is somewhat beyond the
scope of this present study to confront the various issues
that exist between cancer detection and public education
about cancer, any additional insights that can be gained
through such research are potentially useful in the
establishment of a more cost effective cancer control

program.

80

The Sample

 

As mentioned above, this study involves a retro-
spective hospital record review of deceased colon and rectum
cancer patients. Although this type of study (when compared
to a contemporary one with live patients) presents certain
problems in regard to an interpretation of the nature of a
person's ethnicity and his response to illness, it is the
most appropriate approach for a pilot study. The topic of
cancer and preventive health behaviors has not been explored
from an ethnomedical perspective and as a result there are
no tested guidelines from which to base a necessary research
design. While this situation does not preclude the develop-
ment of a useful investigative approach, it does present
serious problems when considering the risk to the rights of
living subjects. Since recall of past medical events
related to the patient's diagnosis is required in addition
to standard sociodemographic information, inquiry would have
to take place as soon after diagnosis as possible. In the
initial phases, cancer as an illness experience seems to
entail both a pervading atmosphere of finality and a result-
ing set of socially and emotionally charged issues for many
patients. Considering the potential risk of further dis—
rupting the patient's lives, using past research on acute or
some of the less morbid chronic diseases to guide the type
of multidirected inquiry needed is simply not adequate.
Before living patients are to be subjected to active

investigation, the most fruitful avenues of research on this

81
topic should be outlined by a less risky study of patient
records. However, it is important to note that this type of
pilot study can not serve as a substitute for a more in
depth investigation but only as a necessary prerequisite.

It is essential when investigating illness behavior
and cancer to control for the type of cancer studied in
order to eliminate as much random variation as possible in
regard to differences in symptoms, pathological development,
and age at onset of the disease. The selection of colonic
and rectal cancer meets these requirements and in addition,
its extremely high incidence provides a large population of
patients from which to sample.

The original population for sampling was to be
patients from several private hospitals. Such hospitals
would provide a cross section of the general population and
thus provide several ethnically definable groups for
analysis, in addition to providing data capable of being
generalized to the U.S. population as a whole. However,
such a data base was not available, and the Veterans
Administration Hospitals in Ann Arbor, Allen Park (Detroit),
North Chicago, and Hines (Chicago) were used instead6.

The sample drawn from these four hospitals consists
of data based on 135 cases of deceased colon or rectum
cancer patients. These cases (all male) were selected
according to the following criteria: (1) confirmed diagnosis
of colon or rectum cancer; (2) information on the diagnostic

visit present in the hospital chart; (3) no delay had

82

occurred between the time the patient presented to a private
physician and the time a medical examination was made at one
of the hospitals; (4) the patient had not been referred from
an institutional facilty (e.g., retirement home or mental
institution) since in such a case it could not be certain
that the decision to seek care was arrived at solely by the
patient; (5) patients had been treated since 19647; (6) data
on length of symptoms and symptom duration were present;

(7) information on family medical history and personal
medical history were desirable, but missing data on these
did not preclude the use of the case in the sample; and

(8) a form for request of domiciliary care was necessary to

report relevant sociodemographic data such as ethnicity,

place of birth, and occupation.

Methodology

 

As previously mentioned, the use of patient records
presents several problems in regard to data collection and
levels of precision for data interpretation. The research
design, in being operationally restricted to data available
on the charts, limited the types of variables available for
analysis. These contraints on sampling criteria not only
made certain assumptions about the data obligatory but also
served to limit the sample size. The use of a chart review
also raises problems in regard to the precision of the
information since the data has passed through several

filters during the reporting process. These problems

83

stemming from chart review will be more fully presented
below in light of data collection and variable definition.

Each case was reviewed starting with the most recent
patient record available back to 1964, the date when most
V.A. hospitals up—dated their files by sending all records
prior to it to a centralized records warehouse. However,
many of the charts between 1964 and 1972 were somewhat terse
due to the V.A. policy of reducing patient records to
perpetual files (i.e., a skeleton file with little or no
sociodemographic information) when five years have elapsed
since last contact with the patient or a known date of
death. These conditions and the sampling criteria sub-
stantially reduced the number of usable cases. Thus,
although the original research design called for a random
sample, it was decided that in order to obtain an adequate
number of cases it would be necessary to use all cases that
met the sampling criteria. Even with this compromise, a
review of over 500 medical records produced only 135 usable
cases from which to base an analysis.

Data on these cases reviewed were recorded on a pre-
pared collection sheet (see Appendix A) that was used to
categorize patient related information. These data served
as a basis for the final thirty-three variables used in the
analysis (see Table I). Information drawn from the form for
domiciliary care was the primary source for such socio-
demographic information as the patient's ethnicity, occu-

pation, marital status and population size of birth place at

10.
11.
12.
13.
14.
15.
16.

84

TABLE I.
VARIABLE

Primary site of cancer

Ethnicity of patient

. Occupation of patient

Marital status

Stage of cancer at diagnosis

. Population size of birth place

. Family medical history

. Personal history, occasional

medical problems

. Personal history, occasional

similar medical problems

Personal history, persistant
medical problems

Personal history, persistant
similar medical problems

Personal history, no unusual
medical problems

Personal medical problems
frequency of problems

Personal medical problems
anatomically similar to cancer

Duration of symptoms

Age of patient at diagnosis

la.
lb.

2a.
2b.
2c.
2d.

2e.

3a.
3b.

4a.
4b.

5a.
5b.
5c.
5d.

6a.

6b.

7a.
7b.

8a.
8b.

9a.
9b.

10a.
10b.

11a.
11b.

12a.
12b.

13a.
13b.

14a.
14b.

15.
16.

VARIABLE VALUE

Colon
Rectum

U.S. Black

U.S. Northern White
U.S. Southern White
U.S. White with West
European ancestory
U.S. White with East
European ancestory

Blue collar
White collar

Single or divorced
Married or Widowed

Dukes' class A lesion
Dukes' class B lesion
Dukes' class C lesion
Dukes' class D lesion

Rural

1. Under 1,000

2. 1,000 to 9,999
Urban

1. 10,000 to 99,999
2. 100,000 or more

No cancer reported
Cancer reported

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

None/occasional problems
Persistant problems

Non-ca similarity
Ca similarity

In reported months

In months

85

TABLE I.

VARIABLE

(continued)

VARIABLE VALUE

 

l7. VAR-001, Rectal bleeding

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

VAR-002,

Blood in stools

Abdominal pain at

defecation

VARFOO4,

VAR-005,

VARrOO6,

VARrOO7,

VAR-008,

VAR-009,

VAR-010,

VAR-011,

VAR-012,

VAR-013,

VAR-014,

VAR-015,

VAR-016,

VARr017,
stools

Dull persistant pain

Crampy abdominal pain

General body pain

Diarrhea

Constipation

Weight loss

Shortness of breath

Weakness

Palpable abdominal mass

Distention

Nausea

Poor appetite

Malaize

Decreased caliber of

17a.
17b.

18a.
18b.

19a.
19b.

20a.
20b.

21a.
21b.

22a.
22b.

23a.
23b.

24a.
24b.

25a.
25b.

26a.
26b.

27a.
27b.

28a.
28b.

29a.
29b.

30a.
30b.

31a.
31b.

32a.
32b.

33a.
33b.

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

Absent
Present

86

time of birth.

The patient's ethnic background was determined by
the following criteria: the reported race of the patient,
which in addition to standard Black-White distinctions often
listed the patient's nationality; the place of birth for the
patient; and, if reported, the birth place of his parents.
These criteria resulted in the identification of five
"ethnic" groups; (1) U.S. Northern White, (2) U.S. Southern
White, (3) U.S. Black (almost all southern born), (4) U.S.
White with Western European ancestory (i.e., first or second
generation immigrant), and (5) U.S. White with Eastern
European ancestory. It should be noted that there were not
enough clearly defined origins for the creation of Northern
or Southern European categories. The remaining demographic
variables of occupation, marital status and population size
of birth place were originally divided into four values for
each. This was done in order to allow as much flexibility
as possible for the generation of new variables should a
more in depth analysis be indicated.

These values would be used later to define new
variables depending on both their logical and statistical
relationships with one another. The reported occupation of
the patient was used to classify him as either a blue collar
worker (i.e., a skilled or unskilled laborer) or a white
collar worker (i.e., self-employed or managerial position).
The reduced category for marital status labeled "alone"

included those subjects who were reported as single,

87

divorced, or widowed. The second marital category labeled
"together" represented those patients who were reported as
being married. Once the population size of the subject's
birth place at time of birth was established through a
review of census reports, the patient was placed into one
of four size value categories. A rural population was
identified when the size was below 10,000 people and urban
when the figure was above 10,000.

Information that pertained to the patient's personal
and family medical history was extracted in part from the
reports based on his initial diagnostic visit for cancer and
from hospital records based on past medical intervention.
Family medical history was based on the patient's reported
knowledge of the presence or absence of cancer in his
family. Patient knowledge about whether or not cancer has
occurred in his family was considered more important for
analysis than a historically accurate report; therefore, no
checks were made on the reliability of his account. In the
case where no family history was reported in the records
(either negative or positive) it was assumed that the
question had not been asked by the interviewing physician
and the data were treated as missing.

The variable of personal medical history was divided
into four values which resulted from a matrix of the two
categories of medical similarity and frequency of medical
problems (similarity and frequency of medical problems were

also used in the analysisL Medical similarity refers to the

88

occurrence of past conditions with anatomically similar
sites (e.g., intestines, stomach, or rectum) and/or
experientially similar symptoms to colon or rectum cancer
such as blood in stools, diarrhea, or constipation. In
terms of medical similarity, a patient's history was scored
as either similar or non-similar. The frequency of medical
problems refers to the relative occurrence of medical inter—
vention and whether or not the noted medical conditions
required any monitoring over time. This category was
divided into the two values of occasional problems and
persistent problems. The occasional problems were inciden-
tal events like heart attacks or episodes of pneumonia and
persistent problems were ongoing conditions of a more
chronic nature like diabetes mellitus or rheumatoid
arthritis. These two categories of medical similarity and
frequency were combined to create the four personal history
values of: (l) occasional problems not similar; (2) occa-
sional similar problems; (3) persistent problems not
similar; and (4) persistent similar problems to colorectal
cancer. A fifth personal history value (no problems) was
also included to represent those patients who had experi-
enced only the usual child or adult medical problems like
appendicitis, accidental injury or various infectious
conditions.

Information reported on the initial diagnostic
visit records and the pathological reports that followed

were used to obtain data on the patient's age at diagnosis,

89

duration of symptoms, the content of those symptoms at
presentation, and the stage of cancer involvement at diag-
nosis. Age at diagnosis was computed by subtracting the
date of the patient's birth from the date of his diagnosis
and for ease of analysis was recorded in months. The
patient's symptom duration time was recorded in months just
as reported on the initial visit form. The content of the
symptoms presented (see list in Table I) were also recorded
just as reported on the initial diagnostic visit charts.
The stage of cancer involvement at diagnosis was based on
the patient's pathological reports and was divided into the
four values of Dukes' class "A", "B", "C", and "D" lesions.
The Dukes' system of classification was specifically
developed for the ordering of cancer involvement for colonic
and rectal tumors. This involvement represents a pattern of
growth from a superficial lesion to distant organ involve-
ment. A class "A" lesion is where a cancerous growth is
limited in its invasive process to the mucosa and submucosa
of the bowel without penetrating the muscular layers of the
bowel wall. Class "B" lesions have penetrated the bowel
wall, the serosa, and/or the pericolic fat which surrounds
the intestine. Class "C" lesions not only have penetrated
all of the bowel wall layers, including the serosa and
pericolic fat, but have invaded the regional lymph nodes
that are either proximal or distal to the tumor-bearing
bowel segment. Sometimes a class "C" lesion can be com-

pounded with additional involvement of an adjacent organ

90
such as the urinary bladder, small bowel segment, prostate
or abdominal wall structures. To avoid any confusion,
patients who were reported as having regional lymph node
involvement only were recorded as class "C" lesion patients.
Any individual reported as having an involvement beyond the
regional lymph nodes was recorded as a class "D" patient.
The Dukes' class "D" lesion represents those patients who
have widespread metastases involving such organs as the
liver, brain, or lungs.

As can be seen by a comparison of the data sheet
(Appendix A) with the list of variables used in the analysis
(Table I), there are several variables on which data were
gathered but not analyzed. These variables included data on
both sociodemographic information and several medically
related categories. The sociodemographic variables of
education, family size, religion, and economic status were
originally considered for analysis but once data collection
began it was decided that their frequency and/or reliability
or reporting made them inadequate for testing. Information
on education and family size were reported in too low a
frequency to warrant further consideration. For reasons
stated at the end of chapter two, the religious preference
of the patient was not used in the analysis. Economic
status presented a different situation in that information
was present inthe form of patient statements about ability
to pay for medical care, transportation to and from the

hospital, and pension allotments. It was decided that since

91

all of the patients in the sample appeared to be in the same
low income level, economic status would not prove to be
fruitful for analysis.

The medically related variables of health at diag-
nosis, date of treatment and stage at treatment, differen-
tiation at diagnosis, and primary tumor type were also found
to be either useless or unnecessary for data analysis. The
information on date of treatment and stage at treatment were
found to be meaningless, since the majority of the cases had
been treated within a week of the time of diagnosis by a
private physiciana. Also, it should be added that, of the
patients who were listed as seeing a private physician
before going to the V.A. Hospital, all received diagnostic
consultation and evaluation by hospital based physicians
within a matter of days from the date of referral. The
variable of differentiation at diagnosis was not useful
because over eighty per cent of the sample was reported to
have either well or moderately-well differentiated tumors.
At present there seems to be some debate between pathologists
about what criteria separate these two classifications;
therefore, the reliability of reporting was not considered
high enough to warrant analysis. The variable of primary
tumor type was not used in the analysis sample because all
but two patients, who were dropped from the study, were
classified under the single type of adenocarcinoma.

A retrospective investigation involves an inductive

approach that is particularly prone to problems of precision

92

concerning the implications of data interpretation. This
stems from the fact that the researcher has no control over
the conditions under which the data were initially recorded.
The only way around this handicap is to acknowledge the
varied sources of interpretational error and to take them
into account when viewing any results from analyses based
on such data.

There are several sources of error which could stem
from particular sociodemographic variables. For example,
although the economic information on the patients indicated
that all of them were essentially on the same low income
level, there is no way to determine if one patient had
access to resources which were unavailable to others. Such
differences could affect decisions about obtaining medical
intervention. Alternative resources need not always mean
material sources but can also represent such resources as
access to various social networks. These networks could
provide access to alternative consultants who have medical
knowledge but who are not formally members of the profes-
sional medical community. If individuals used these
consultation sources before seeing a physician, their
behavior as based on the chart review, would show them
taking a longer time to come in than others who did not use
the resources. Thus, an error would result such that they
would be seen as not seeking care as soon as they actually
did.

Another variable which could have a bearing on the

93

decision to seek medical intervention is occupation. Since
the reported occupation of the patient does not include any
details of what the person's responsibilities were or how
long he held the position, considerable variability could
be hidden. This hidden variability could, if known, imply
certain conditions such as an inability to get along with
others or a very strong orientation towards the self rather
than other. These conditions could affect the patient's
decision to seek care since this act is just as much a
social process as it is an individual one. In addition to
occupation, marital status could also point to problems in
getting along with people (e.g., repeated divorces).
However, since there was no available information on fre-
quency of marital status change or marital problems, such
situations are not identifiable.

One of the more important sources of imprecision in
this study involves the medical categories of family medical
history, personal medical history, stage at diagnosis,
symptom duration, and the symptoms presented. One factor
that can affect these variables is the nature of the
doctor-patient relationship as discussed in an earlier
chapter. For all the researcher knows, there could have
been major conflicts between the personalities of the
patient and the physician which could seriously affect the
nature of the information reported in the hospital records.
These effects could take a variety of forms from intentional

misinformation or reticence by the patient to only a sketchy

94

interview by a physician who wants to be rid of a trouble-
some patient. The physician may or may not notice the
situation and, in the event it was obvious, it can not be
assumed that comments will be made to demonstrate the
situations existence. Therefore, such misinformation could
not be controlled for by checking the records for comments
about relationship and/or memory problems.

Another source of error ensuing from the doctor-
patient encounter could result from certain filtering
mechanisms on the part of the physician. A standard filter
for all physicians which stems from their contact with
general biomedical theory involves certain beliefs and
expectations about what information is necessary to record.
Such a filter could result in a failure to record data
important for behavioral research but considered by the
physician to be unimportant (e.g., the other symptoms
presented at the diagnostic visit but not recorded).
Another more idiosyncratic filter to the physician relates
to problems of personality such as discussed above, and also
what the doctor thinks of the disease situation and its
relevant aspects. This filter can take the form of
expectations of how a sick person should behave and what
his problems should be. Deviations from these expectations
could result in misinformation on the behavior of the
patient. For example, symptom duration times might be
considered only a very minor preliminary to treatment for

the patient and therefore receive only a terse comment in

95

the hospital records.

The last major source of misinformation to be con-
sidered for this study is subject recall. Considering the
levels and varieties of stress these patients are subject to
and their ages, recall becomes a potential source of error
that needs to be recognized. Although reports were made
about poor memories, it can not be assumed that all patients
with poor memories were properly identified. In cases where
the patient was so identified the case was dropped from the
sample. It was assumed that a subjective belief by the
patient about family history or symptoms and their duration
would be sufficient for analysis of his behavioral response.
However, if the patient fails to remember facts that lead to
his health seeking behavior, several problems in interpre-
tation can arise, especially if the patient has not been
identified as having a poor memory. Problems in patient
recall can also generate errors in the reported symptom
duration intervals. While symptom duration reporting may be
less of a problem for error than family history because of
the nature of the questions asked by a physician, it can
still lead to errors in interpretations about the patient's

behavioral response.

96
Analysis

Once the data collection was completed, the thirty-
three variables to be tested (Table I) were prepared for
computerized data analysis. One form of this analysis was
a program that produced various descriptive statistics such
as frequencies and group means, this was used to obtain a
general profile of the sample distribution for each variable
separately.

In order to determine if the ethnic groups identi-
fied represented statistically separate groups in terms of
mean duration time, an analysis of variance was performed.
The results of this test were considered to be significant
at the five per cent level. Although t-tests are normally
performed when the question concerns a comparison of means
for the sample groups, they are statistically useful only
when testing the means of two groups. Since the variable
of ethnicity was composed of five groups the t-test could
not be reliably employedg. An analysis of variance is based
on the condition that if the means of subgroups are greatly
different, the variance of the combined groups is much
larger than the variances of the separate groups. This type
of analysis rests on a separation of the variance of all the
observations into parts (ethnic groups), each of which
ineasures variability attributable to some specific source
(mean symptom duration time).

Migration tendencies in between group relationships

for the five ethnic groups were checked by using a oneway

97

analysis of variance in the format of a posteriori contrast

 

tests. These contrasts provided a multiple range test for
the mean duration times of the ethnic groups and as a result,
group them into homogeneous subsets. A subset was con—
sidered homogeneous when the means of the first and last
group differed by less than the critical value for a subset
of that size. Although a five per cent level was necessary
for significance, the program used provided several tests

of differing degrees of conservativeness at both the .05 and
.10 levels. This flexibility made it possible to observe
segregation characteristics for the groups should their case
sizes be increased.

An analysis of variance for between group mean dura-
tion times was also performed on the variables of age at
diagnosis and stage at diagnosis to determine if either of
these variables demonstrated any significant relationship
with symptom duration. Before the variances of the within
variable groups were considered to be indicative of any
heterogeneity, the probability of the F-ratio needed to
meet or exceed an alpha level of .05. Once this analysis

was performed, a posteriori contrast tests were computed to

 

check for any trends in heterogeneity. Age at diagnosis

was seen as a possible confounding variable which could mask
possible relationships between the other tested variables
and symptom duration time. Stage at diagnosis was checked
for similar reasons and to see if length of duration had

any correlation with disease involvement.

98

In addition to the analyses of variance performed
above, similar procedures were used on the variables of
occupation, marital status, and population size of birth
place. These analyses were necessary to determine the most
fruitful value combinations to be used in the formation of
the dichotomous categories of each variable. As was the
case for ethnicity, age and stage at diagnosis, a ninety-
five per cent confidence level was needed before any hetero-
geneous relationships were considered significant. A

posteriori contrasts were then performed to provide a more

 

descriptive representation of any trends in heterogeneity.
Once these above analyses were computed, t-tests
were performed on the dichotomous values of all relevant
variables to see if any showed a significant relationship
in terms of reported mean duration times. The results were
not considered significant unless the t-values fell at the
five per cent level or less. All variables which showed a
significant relationship with duration and those included
in the stated hypotheses were then subjected to a second
series of t-tests for means when controlling for ethnic
background. These tests were performed to determine if the
previous significant relationship held when applied to with—
in ethnic group classifications. Any variables that main—
tained a significant relationship with duration as per
ethnic group were then to be subjected to another t-test of
mean time for each of the matching values of the variables

between the ethnic groups.

99

These first t-tests on within variable group means
were performed on the sociodemographic variables of occu-
pation, marital status, and population size of birth place.
The two values for occupation (i.e., blue collar and white
collar) were seen as representing differences in terms of
income, education, and ability of opportunity to accept
various levels of responsibility, etc., and the test was
performed to check for any effect of labor classification
on reporting time. Marital status was tested to see if it
made any difference whether a person was married or not
married (i.e., single, divorced, or widowed). These within
variable groups were seen as possible indicators of different
sets of social, emotional, and economic responsibilities
and considerations. The values for population size of birth
place (i.e., rural and urban) were considered valuable for
testing because of their use in past health related research.
They were also seen as representing basic differences in
access to medical information and institutionalized medical
care; and differences in the level of assimilation of bio-
medical knowledge and the maintenance of more traditional
medical beliefs.

In addition to the above mentioned sociodemographic
variables, the first round of t-tests were also performed
on the several medical variables of family medical history,
personal medical history, and the symptoms presented at the
diagnostic visit. The two values for family history (i.e.,

cancer, no cancer) were seen as respectively representing

100

the presence or lack of particular concerns regarding cancer.
The reported presence of cancer was seen as an indicator of
both an awareness by the patient of the possibility of
having cancer, and the possible knowledge of what the
disease may entail in terms of medical, social, and economic
costs. A comparison between the presence and absence of
these concerns was hypothesized to have a significant
relationship with reported symptom duration such that the
means for these two values would be different.

The values for personal medical history (i.e.,
frequency of problems and similarity of problems) were also
hypothesized to result in significantly different reported
mean duration times.» The frequency of problems was seen as
an indicator of an individual's exposure to the medical
system, and their inclination towards medical care. Medical
similarity was seen as an indicator of an individual's
exposure to similar symptoms as those for the present illness
episode. This exposure was considered as a possible factor
in either heightening or lowering the perceptual threshold
of the patient for such symptoms. The following group mean
comparisons were therefore expected to demonstrate signif-
icant relationships: (1) presence of medical similarity
versus absence of similarity; (2) occasional problems versus
persistent problems; and (3) occasional similar problems
versus persistent similar problems.

The within variable means for each symptom presented

(i.e., when present versus when not reported) were compared

101

to check for any relationship with duration time. As was
the case for the above mentioned variables, this first bat-
tery of t-tests was considered preliminary to a test of
between ethnic group differences for each variable.

Although the within variable test focused upon the presence
and absence of the symptom, the primary concern here is with
the presence of the symptom. It was suspected that the
presence of particular symptoms by their interrelationship
with past medical experiences and social orientations (i.e.,
an individual's ethnic background in conjunction with their
family medical history and/or personal medical history) will
result in significantly different mean duration times when
compared with their mutually exclusive counterparts. This
first set of tests was simply a means of establishing which
symptoms should be analyzed at a deeper level of relation-
ship with reported duration time.

When sample sizes are small as in the case of this
investigation, the depth of analysis becomes somewhat
limited because of reduced cell sizes. In effect, these
small cell sizes when analyzed have considerable error in
data interpretation. In order to circumvent this handicap,
a log linear model was employed. This procedure is an
analog to analysis of variance and involves the use of
cross-product ratios of particular dichotomous variables
contained within a 2 X 2 table format (Goodman 1965, 1972,
and Marascuilo and McSweeney 1977). In order for this

analysis to be performed it was necessary to transform the

102

continuous dependent variable of reported symptom duration
into a discrete variable with only two values. This trans-
formation was accomplished by dividing the reported symptom
duration times into categories based on the median duration
time. The median duration was used instead of the mean
duration time because the distribution of this variable was
skewed to the left of a normal curve. By using the median
as a point of comparison, the sample was broken into two
evenly weighted groups. Unlike earlier delay studies, there
was no predetermined time marker for analysis. The two
categories of median duration are: (1) less than or equal
to the median; and (2) greater than the median. The median
of the sample was three months (the correspondence to Pack
and Gallo's 1938 time marker is only coincidental).

The log linear model used here computes statistical
values which indicate the nature of the relationships be-
tween four main elements; symptom duration, ethnicity, the
frequency of variable reporting, and the reported values for
that variable. These relationships, which are listed in a
ratio format, are between: (1) reported symptom duration and
ethnic background; (2) frequency of reported variable and
background; (3) the reported variable values and symptom
duration; and (4) the interaction between ethnicity, report-
ed variable value, and symptom duration. A fifth permuta—
tion includes the relationship between ethnicity and the
reported variable values. Although the ratio values for

this relationship will be given in the tables, they are

103
useful at only the most descriptive level of interpretation
about the sample. This stems from the fact that these
ratios for ethnicity and variable value can not be
statistically tested separately. They merely result from
the computation of the ratio values for the interaction
effects.

The first relationship between reported symptom
duration and ethnicity concerns the overall main effects of
ethnicity on reporting time. This relationship is of
particular interest because it tests the main hypothesis of
the study. The second relationship on frequency of reported
variable and ethnicity provides information about whether or
not the first listed (i.e., in a test format) value of the
variable occurs equally between ethnic groups. This listed
value is arbitrary and is dependent only on the convention
used by the researcher. For example, either the presence
or the absence of a symptom could be used since they are
mutually exclusive categories. This test is essential
before proceeding with any other tests on the variable in
question. If the variable was unevenly distributed among
the sample, here considered to be a ratio of 2:1 or greater,
it would be meaningless to compare between ethnic group
relationships. To do this would be like comparing the
relationships between two separate populations. The third
relationship indicates the association between each of the
listed variable values and reporting time. For example,

this test would be appropriate for such questions as: do

104

people who have reported rectal bleeding tend to report
before the median time or after, and how do they compare to
people who have not reported such bleeding? The last
relationship on the interaction between ethnicity, reported
variable value, and duration time will be the keystone for
discussing the relationships for each variable. This test
provides ratio values representing the interaction between
a listed variable value with a given ethnic group in terms
of reporting before or after the median.

The ratio values are computed by taking the cross-
product ratio from the 2 X 2 table for the variable in
question. This odds ratio which is labeled (9) will in all
references involve the ratio of one analytical category to
another where the second category is equal to one. The
statistical significance of this odds ratio is computed by
taking Goodman's gamma (9), which is the natural log of the
odds ratio, and dividing it by the standard deviation for
the table under consideration. This procedure results in a
standard score which, when looked up in a Z score distri-
bution table, indicates the alpha level. Thus the overall
procedure involves an odds ratio which indicates how the
sample relates to particular variables and then computes
a statistical test for this ratio in terms of a standard
score. The interaction effect, although more involved than
the first three tests because it is an odds ratio of an odds
ratio, is tested statistically by a similar procedure which

takes into account the increased complexity of the

105

relationship. For examples of this procedure, see

Appendix B.

Results

In discussing the results of the preceding analyses,
various figures and tables will be used in addition to the
text. These figures and tables will be found throughout
this chapter. The reporting of these results will be
presented in the following sequence: (1) the sample
distribution; (2) the reulsts from tests involving the
continuous dependent variable of reported symptom duration;
and (3) the results from tests on reported sypmtom duration

as a discrete variable.

The Sample Distribution

 

The sample distribution for patients reported to
have either colonic or rectal carcinoma was fairly even with
only a few more colon cases than rectum cases (Figure 1.,
Table II). Seventy cases or 52 per cent of the sample were
reported as having colon cancer, while sixty-five cases or
forty-eight per cent were reported to have rectal cancer.

The distribution of the sample by the five categories
for ethnicity indicated that by far the majority of patients
were Northern U.S. White (60 cases) while the smallest group
was Southern U.S. White (13 cases)(Figure 2., Table IIIa).
The second largest group reported was U.S. Black (28 cases)

followed by people with Eastern European ancestory (19 cases)

Frequency of reported site

70

50

30

106

—

 

 

COLON RECTUM

Figure 1
Distribution of cancer sites (N-l35)

107

 

 

 

 

 

TABLE II.
PRIMARY SITE OF CANCER
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
1. COLON 70 51.9 51.9
2. RECTUM 65 48.1 100.0
TOTAL 135 100.0
TABLE IIIa.
ETHNICITY OF PATIENT
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
1. NORTH US
WHITE 60 44.4 44.4
2. SOUTH US
WHITE 13 9.6 54.0
3. US BLACK 28 20.8 74.8
4. US WEST
EUROPE 15 11.1 85.9
5. US EAST
EUROPE 19 14.1 100.0
TOTAL 135 100.0
TABLE IIIb.
ETHNICITY OF PATIENT
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
1. US WHITE 107 79.2 79.2
2. US BLACK 28 20.8 100.0
135 100.0

70

Frequency of cases
no u:
C O

0

Frequency of reported race

5
o
I

60-

to
O
l

108

 

 

NW SW 8 WE EE

Nw-North US White WE'US West Europe
Sow-South US White EE-US East Europe
8 - US Block

Figure 2
Distribution of ethnic groups (N=l35)

 

 

US WHITE US BLACK

Figure 3
Distribution of subiects by race (N‘IBS)

109

and then people with Western European ancestory (15 cases).
When simply considering the Black-White racial distribution
(Figure 3., Table IIIb) the frequency of Whites to Blacks
was almost 4:1. The distribution by race of site (Figure
4., Table IIIc) showed that while rectal cancer represented
the majority of White cases (57 per cent), it comprised the
minority of cases for Blacks (36 per cent).

The distribution for occupational status indicated
that the largest number of patients (over 50 per cent) were
classed as unskilled laborers (63 cases)(Figure 5., Table
IV). The second largest group was people who were self-
employed (31 cases) followed by skilled laborers (25 cases)
and people occupying managerial positions (8 cases). When
occupation was divided into the categories of blue collar
worker (BC) and white collar worker (WC) the (BC) workers
out numbered the (WC) workers by more than two to one with
88 and 39 cases respectively. Both of these distributions
had eight cases missing because of inadequate data.

The distribution of the variable marital status
(Figure 6., Table V) showed that a majority of the patients
were married at diagnosis (78 cases). Following the married
category were the groups representing: divorced (23 cases),
single/never married (21 cases), and widowed patients (13
cases). When marital status was divided into the categories
of alone (A) and together (T), the (T) group was only
slightly larger than the (A) group at fifty-eight and

forty-two per cent respectively.

110

70-

Frequency of reported cases

 

 

COLON RECTUM COLON RECTUM
WHI'I’E BLACK

Figure 4
Distribution of cancer site by race

O
O
L 1

Frequency of cases
1s
0
l

 

 

20 q
-i
USK S K S E M B C W C
USK "' Unskilled M - Managerial
SK - Skilled IC- Blue Collar
SE - Sell - Employed WC- White Collar
Figure 5

Distribution by occupation (NI-127)

111

 

 

 

TABLE IIIC.
ETHNICITY BY SITE
ABSOLUTE RELATIVE FREQ. RELATIVE FREQ.
VALUE FREQ. (%) BY ETHNIC (%) BY SAMPLE
1. US WHITE
a. COLON 46 43.0 34.1
b. RECTUM 61 57.0 45.2
2. US BLACK
a. COLON 18 64.3 13.0
b. RECTUM 10 35.7 7.4
TOTAL 135
TABLE IV.
OCCUPATION OF PATIENT
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQUENCY
1. BLUE COLLAR 88 65.2 69.3
a. UNSKILLED (63) (46.7) (49.6) 49.6
b. SKILLED (25) (18.5) (19.7) 69.3
2. WHITE COLLAR 39 28.9 30.7
a. SELF-EMPLYD (31) (23.0) (24.4) 93.7
b. MANAGERIAL ( 8) ( 5.9) ( 6.3) 100.0
3. OUT OF RANGE 8 5.9 MISSING
TOTAL 13 100.0 100.0

 

 

 

so...
oeusueeeeeeeeee.

0900900000 0

 

 

112

 

 

801

_ 4 _ _ _

o o O
6 4 2

«once v2.8.2 .0 3:252“.

1- together

W-Widowed
A- Alone

S-Single
D-Divorced

M—Married

6
Distribution by marital status (N=l35)

Figure

113

 

 

 

TABLE V.
MARITAL STATUS OF PATIENT
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
1. ALONE 57 42.2
a. SINGLE (21) (15.6) 15.6
b. DIVORCED (23) (17.0) 32.6
C. WIDOWED (13) ( 9.6) 42.2
2. TOGETHER
a. MARRIED (78) (57.8) 100.0
TOTAL 35 100.0
TABLE VI.
POPULATION SIZE OF BIRTH PLACE
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. RURAL 45 33.3 51.7
a. UNDER 1000 (21) (15.6) (24.1) 24.1
b. 1000-9999 (24) (17.8) (27.6) 51.7
2. URBAN 42 31.1 48.3
C. 10000-99999 (16) (11.8) (18.4) 70.1
d. OVER 100000 (26) (19.3) (29.9) 100.0
3. OUT OF RANGE 48 35.6 MISSING
TOTAL 135 100.0 100.0
TABLE VII.
AGE AT DIAGNOSIS
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
1. 30-35 1 .7 .7
2. 25-40 1 .7 1.5
3. 40-45 6 4.4 5.9
4. 45-50 12 8.9 14.8
5. 50-55 6 4.4 19.3
6. 55-60 23 17.0 36.3
7. 60-65 20 14.8 51.1
8. 65-70 18 13.3 64.4
9. 70-75 18 13.3 77.8
10. 75-80 19 14.1 91.9
11. 80-85 9 6.7 98.5
12. 85-90 2 1.5 100.0
TOTAL 35 100.0

 

Frequency of cases

Frequency of cases

114

 

 

A B C D R U
A I Under 1,000 D - Over l,000,000
B ' 1,000 - 9,999 I ' Rural
C " IO, 000 - 999,999 U ' Urban
Figure 7

Distribution of population by birth place (N=87)

 

 

l 2 3 4 5 6 7 8 9 10 ll 12

1'30‘35 5'50'55 9'70'75

2‘35‘40 6'55'60 '0‘75'00

3'40'45 7'60'65 "‘00'05

4' 45'50 0'65 ' 70 12".5 ‘90
Figure 8

Distribution of age at diagnosis (N=l35)

115

Population size of birth place at birth (Figure 7.,
Table VI) was rather evenly distributed for both the
explicit category of size and the more general category of
rural-urban population type. However, it should be noted
that forty-eight cases had to be excluded from these
distributions because of inadequate data.

Age at diagnosis (Figure 8., Table VII) was found
to have a mean value of 65 years and a median value of 64
years with a minimum and maximum age of 32 and 86 years
respectively. The distribution indicates a minor peak at
the 45-50 year age range with the major peak at the 55-60
year age range. Following this main frequency peak there
is a plateau until the 80-85 year age range.

The distribution for the stage at diagnosis (Figure
9., Table VIII) indicates that, with the exception of the
reporting of Dukes' class "A" lesions, the variable is
evenly reported.

Differentiation at diagnosis (Figure 10., Table IX)
had a distribution in which almost half of the cases were
classified as well differentiated (67 cases) and another
thirty per cent were moderately-well differentiated (41
cases). The remaining three classes of moderately differ-
entiated (8 cases), poorly differentiated (14 cases) and
undifferentiated (2 cases) made up only twenty-four per cent
of the sample with three cases missing.

Family medical history (Figure 11., Table X)

indicated that the reported absence of cancer in an

Frequency ol cases

Frequency ol cases

116

 

 

50 '1
30 —
IO ‘
A
Figure 9

Distribution ol Dukes’ stage at diagnosis (N=I3S)

 

 

W MW M P U
W ' Well Dill. P "' Poorly Dill.
mw- Mod. Well Dill. U- Undill.
M- Mod. Dill.
Figure 10

Distribution of dillerentiation at diagnosis (N432)

117

 

 

 

 

 

TABLE VIII.
STAGE AT DIAGNOSIS
ABSOLUTE RELATIVE CUMULATIVE
VALUE FREQ. FREQ.(%) RELATIVE FREQ.(%)
l. DUKES' A 2 1.5 1.5
2. DUKES' B 46 34.1 35.6
3. DUKES' C 41 30.4 65.9
4. DUKES' D 46 34.1 100.0
TOTAL 135 100.0
TABLE IX.
DIFFERENTIATION AT DIAGNOSIS
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. WELL DIFF. 67 49.5 50.8 50.8
2. MOD-WELL DIFF. 41 30.4 31.1 81.8
3. MOD DIFF. 8 5.9 6.1 87.9
4. POORLY DIFF. 14 10.4 10.5 98.5
5. UNDIFF. 2 1.5 1.5 100.0
6. OUT OF RANGE 3 2.2 MISSING
TOTAL 135 100.0 100.0
TABLE X.
FAMILY MEDICAL HISTORY
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. NO CANCER 81 60.0 74.3 74.3
2. CANCER 28 20.7 25.7 100.0
3. OUT OF RANGE 26 19.3 MISSING

 

TOTAL 135 100.0 100.0

Frequency ol cases

Frequency ol cases

V
(II

M
(It

118

 

 

A P

A- Absent P - Present

Figure ll
Distribution ol reported cancer in lamily (N=u09)

 

 

N/O PST NC A CA5
Problem occurrence
N /O’ None or occasional problems NCA- Non -cancer similarity
PS‘I’ - Persistent problems CAS - Cancer similarity
Figure l 2

Distribution ol personal medical problems (N=130)

119

individual's family was almost three times as frequent as
its reported presence in the rest of the sample with 81 and
28 cases respectively. Information on this variable was not
always present and twenty-six cases were recorded as missing.

Personal medical history was separated into the four
main elements of: occasional problems, persistent problems,
no problems/usual adult and child conditions, and personal
medical problems (five cases missing). The distribution for
occasional medical problems (Tables XIa and XIb) indicated
that individual's with no similar occasional problems (51
cases) were half again as frequent as those patients who
were reported as having similar problems (35 cases). The
distribution for patients with persistent problems (Tables
XIc and XId) indicated that non-similar persistent problems
(37 cases) were reported most frequently for the sample with
an increase over similar problems (21 cases) of 76 per cent.
The distribution for no reported unusual problems indicated
that only 22 per cent of the sample or 29 cases had gone
through life without any unusual medical problems (Table
XIe).

Personal medical problems (Figure 12., Table XIf
and XIg) includes problem occurrence and medical similarity.
The distribution of the problem occurrence showed that by
far the majority of patients had only occasional problems or
nothing unusual at all. The distribution for similar
medical problems to cancer indicated that 50 per cent more

patients were recorded to have non-similar problems than

120

TABLE XI.

PERSONAL MEDICAL HISTORY:

 

 

 

 

 

 

 

XIa. OCCASIONAL MEDICAL PROBLEMS NON-SIMILAR
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. ABSENT 79 58.5 60.8 60.8
2. PRESENT 51 37.8 39.2 100.0
3. OUT OF RANGE __5 3.7 MISSING
TOTAL 35 100.0 100.0
XIb. OCCASIONAL MEDICAL PROBLEMS SIMILAR
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. ABSENT 95 70.4 73.1 73.1
2. PRESENT 35 25.9 26.9 100.0
3. OUT OF RANGE __§ 3.7 MISSING
TOTAL 35 100.0 100.0
XIC. PERSISTANT MEDICAL PROBLEMS NON-SIMILAR
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. ABSENT 93 68.9 71.5 71.5
2. PRESENT 37 27.4 28.5 100.0
3. OUT OF RANGE __§ 3.7 MISSING
TOTAL 135 100.0 100.0
XId. PERSISTANT MEDICAL PROBLEMS SIMILAR
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. ABSENT 109 80.7 83.8 83.8
2. PRESENT 21 15.6 16.2 100.0
3. OUT OF RANGE __§ 3.7 MISSING
TOTAL 35 100.0 100.0

121

TABLE XI. (continued)

XIe. NO REPORTED UNUSUAL PROBLEMS

 

 

 

CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (8)
1. ABSENT 101 74.8 77.7 77.7
2. PRESENT 29 21.5 22.3 100.0
3. OUT OF RANGE __§ 3.7 MISSING
TOTAL 35 100.0 100.0
XIf. PERSONAL MEDICAL PROBLEMS
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREQ.(%) FREQ. (%)
1. ABSENT 93 68.9 71.5 71.5
2. PRESENT 37 27.4 28.5 100.0
3. OUT OF RANGE __§ 3.7 MISSING
TOTAL 13 100.0 100.0
XIg. PERSONAL MEDICAL SIMILARITY
CUMULATIVE
ABSOLUTE RELATIVE ADJUSTED ADJUSTED
VALUE FREQ. FREQ.(%) FREg;(%) FREQ. (%)
1. NON-CA RELATED 77 57.0 59.2 59.2
2. CA RELATED 53 39.3 40.8 100.0
3. OUT OF RANGE 5 3.7 MISSING

TOTAL 35 100.0 100.0

 

122

similar ones with 77 and 53 cases respectively.

The distribution of reported symptoms (Figure 13.,
Table XII) indicated three symptom groupings that could be
isolated by their frequency of reporting from the remaining
nine symptoms. The first group consisted of the two
symptoms; weight loss (1) with 71 cases, and rectal bleeding
(a) with 70 cases. The second group was more than twenty-
three per cent lower in frequency than rectal bleeding and
included the symptoms: dull persistent pain (d) with 57
cases, constipation (h) with 54 cases, and diarrhea (g) with
53 cases. The third group was only fifty per cent of group
one and only 75 per cent of group two and included the
symptoms: weakness (k) with 34 cases, nausea (n) with 28
cases, and poor appetite (o) with 28 cases. Of the
remaining nine symptoms, the highest frequency was 16 cases
and the lowest was 8 cases (for a symptom distribution by
race see Table XIII).

A X2 was used here to determine if the symptoms
within each observed cluster covaried or were merely an
artifact of reporting and therefore random in their occur-
rence. In this case with one degree of freedom, the
critical value for X2 for an alpha of .05 was 3.84 and for
an alpha of .01 it was 6.63. When the first group (rectal
bleeding and weight loss) was tested the results indicated

2

no covariance with a X =.20. In the second group, although

diarrhea was not found to covary with either constipation

(X2=.60) or dull persistent pain (X2=.70), constipation and

Frequency of reported symptoms

80-

60-

404

20-

123

 

abcdelghiiklmnop

Figure 13
Distribution ol reported symptoms (See table XII lor delinitions) (N=l35)

q

124

TABLE XII.

SYMPTOMS PRESENTED AT DIAGNOSIS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ABSOLUTE RELATIVE
VALUE FREQ. FREQ.(%)
XIIa. RECTAL BLEEDING
1. ABSENT 65 48.1
2. PRESENT 70 51.9
TOTAL 135 100.0
XIIb. BLOOD IN STOOLS
1. ABSENT 122 90.4
2. PRESENT 13 9.6
TOTAL 135 100.0
XIIC. ABDM. PAIN AT DEF.
1. ABSENT 120 88.9
2. PRESENT 15 11.1
TOTAL 135 100.0
XIId. DULL PERSIST. PAIN
1. ABSENT 78 57.8
2. PRESENT 57 42.2
TOTAL 135 100.0
XIIe. CRAMPY ABDM. PAIN
1. ABSENT 125 92.6
2. PRESENT 10 7.4
TOTAL 135 100.0
XIIf. GENERAL BODY PAIN
1. ABSENT 119 88.1
2. PRESENT 16 11.9
TOTAL 135 100.0
XIIg. DIARRHEA
1. ABSENT 82 60.7
2. PRESENT 53 39.3
TOTAL 13 100.0
XIIh. CONSTIPATION
1. ABSENT 81 60.0
2. PRESENT 54 40.0
TOTAL 135 100.0
XIIi. WEIGHT LOSS
1. ABSENT 64 47.4
2. PRESENT 71 52.6

 

TOTAL 135 100.0

125
TABLE XII. (continued)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ABSOLUTE RELATIVE
VALUE FREQ. FREQ.(%)
XIIj. SHORTNESS OF BREATH
1. ABSENT 127 94.1
2. PRESENT 8 5.9
TOTAL 35 100.0
XIIk. WEAKNESS
1. ABSENT 101 74.8
2. PRESENT 34 25.2
TOTAL 135 100.0
XIII. PALPABLE MASS
1. ABSENT 124 91.9
2. PRESENT 11 8.1
TOTAL 35 100.0
XIIm. DISTENTION
1. ABSENT 119 88.1
2. PRESENT 16 11.9
135 100.0
XXIn. NAUSEA
1. ABSENT 107 79.3
2. PRESENT 28 20.7
TOTAL 35 100.0
XIIO. POOR APPETITE
1. ABSENT 107 79.3
2. PRESENT 28 20.7
TOTAL 135 100.0
XIIp. MALAIZE
1. ABSENT 127 94.1
2. PRESENT 8 5.9
TOTAL 35 100.0
XIIq. DECRSD. CALIBER STOOLS
1. ABSENT 124 91.9
2. PRESENT 11 8.1

 

TOTAL 135 100.0

 

126
TABLE XIII.

SYMPTOM DISTRIBUTION BY RACE

 

RELATIVE
ABSOLUTE FREQ.(%) CUMULATIVE
SYMPTOM RACE FREQ. BY RACE FREQ. (%)

1. RECTAL W 55 13.8
BLEEDING B 16 14.9 14.1 14.1

2. BLOOD IN W 11 3.0
STOOL B 3 2.8 2.8 16.9

3. ABDOMINAL PAIN W 13 3.3
AT DEFECATION B l .9 2.8 19.7

4. DULL PERSISTANT W 46 11.6
PAIN B 11 10.3 11.3 31.0

5. CRAMPY ABDOMINAL W 6 1.5
PAIN B 4 3.7 2.0 33.0

6. GENERAL BODY W 12 3.0
PAIN B 4 3.7 3.2 36.2

7. DIARRHEA W 44 11.1
B 9 8.4 10.5 46.7

8. CONSTIPATION W 46 11.6
B 8 7.5 10.7 57.4

9. WEIGHT LOSS W 59 14.9
B 12 11.2 14.1 71.5

10. SHORTNESS OF W 5 1.2
BREATH B 3 2.8 1.6 73.1

11. WEAKNESS W 26 6.5
B 8 7.5 6.7 79.8

12. PALPABLE MASS W 9 2.3
B 2 1.9 2.2 82.0

13. DISTENTION W 15 3.8
B 2 1.9 3.4 85.4

14. NAUSEA W 17 4.3
B 10 9.3 5.3 90.7

15. POOR APPETITE W 21 5.3
B 7 6.5 5.5 96.2

16. MALAIZE W 4 1.0
B 4 3.7 1.6 97.8

17 DECREASED CALIBER W 8 2.0
OF STOOLS B 3 2.8 2.2 100.0

127

dull persistent pain were found to covary (X2=4.38) with an
alpha of .05. Tests on the third group indicated that
nausea and poor appetite covaried with a (corrected)
X2=l6.27 that was highly significant, whereas weakness was
found to occur at random and independent of poor appetite

(x2

=.50) and, by statistical inference, nausea.

Reported symptom duration (Figure 14., Table XIV)
had a mean of five months, a median of a little over three
months and a standard deviation of five months. This
distribution was heavily weighted to the left with a mini-
mum duration of one week and a maximum duration of thirty
months. Although it may be the result of the way people in
this culture estimate and round time, there appear to be
four frequency peaks: at one, three, six, and tweleve
months. The most frequently reported duration time was one
month or less with 33 cases; followed by three and six

months with 19 cases each; two months with 15 cases; four

months with 13 cases; and twelve months with 10 cases.

128

Ann—nzv «5:2: am «E038: .o cot—top .o cot-5:35
I 256:

omommeNONnNVNMNNN—NONQ— 950— n— 3 2 N— 2 o. o m A 0 m. v n N —

+_._p

 

 

o—

m—

cm

mm

on

mm

sesos POHOdOJ go Asuenbeq

129
TABLE XIV.

REPORTED SYMPTOM DURATION PRIOR TO DIAGNOSIS

 

 

VALUE ABSOLUTE RELATIVE CUMULATIVE
IN MONTHS FREQ. FREQ.(%) RELATIVE FREQ.(%)
1.00 33 24.4 24.4
2.00 15 11.1 35.5
3.00 22 16.3 51.8
4.00 13 9.6 61.4
5.00 4 3.0 64.4
6.00 19 14.1 78.5
7.00 1.5 80.0
8.00 3.0 83.0
9.00 1.5 84.5
10.00 1.5 86.0
11.00 ** **** 86.0
12.00 10 7.4 93.4
13.00 ** **** 93.4
14.00 1 .7 94.1
15.00 ** **** 94.1
16.00 ** **** 94.1
17.00 .7 94.8
18.00 3 2.2 97.0
19.00 ** **** 97.0
20.00 1 .7 97.7
24.00 2 1.5 99.2
30.00 __1 .7 99.9
TOTAL 135 100.0
MEAN 5.07 MONTHS

3.29 MONTHS
5.09 MONTHS

MEDIAN
STD DEV

130

Symptom Duration As A Continuous Variable

 

The results from a oneway analysis of variance on
symptom duration time by ethnic background (Table XV)
indicated that the groups under consideration had signifi—
cantly different variances and mean duration times. The F
statistic has a .03 significance level with an F-ratio of
2.76.

Two a posteriori contrast tests were also performed

 

on the ethnic groups and symptom duration to provide a better
descriptive picture of how they were distributed. The first
of these tests was run at a significance of .05 and the
second has run at a significance level of .10 to discover
any possible trends or shifting relationships. The results
of the first contrast test with an alpha of .05 (Table XVa)
indicated that while the Black and Northern White groups
were not homogeneous, the Southern White, West European and
East European groups could not be differentiated from either
the Black or the Northern White groups. The results of the
second test run at an alpha of .10 (Table XVb) indicated a
trend for a heterogeneous relationship between the Black
group on one side and the Northern and Southern White groups
on the other; however, the West and East European groups
still could not be differentiated from the previously
defined Black and White groups.

Considering the above results it was decided to
collapse the groups into the categories of U.S. Black,

U.S. White (NW and SW), and U.S. European (WE and EE). The

131
TABLE XV.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON
OF SYMPTOMS BY ETHNIC BACKGROUND.

REPORTED DURATION

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 4 300.34 75.09 2 76 03
WITHIN GROUPS 130 3532.72 27.17
TOTAL 134 3833.06
_ 95% CONFIDENCE
GROUP N X S S.E. MIN MAX LIMITS
1. (NW) 60 6.36 6.54 .84 .25 30.00 4.67 -- 8.05
2. (SW) 13 5.71 5.07 1.36 1.00 18.00 2.78 -- 8.64
3. (B) 28 2.60 3.34 .63 .25 17.00 1.30 -- 3.89
4. (WE) 15 3.88 3.18 .82 .25 12.00 2.12 -— 5.64
5. (BE) 19 4.80 3.60 .83 .25 12.00 3.07 -- 6.54
TABLE XVa.
RESULTS OF MULTIPLE RANGE TEST, TUKEY-B AT AN ALPHA OF .05.
SUBSET 1
GROUP GRP (B) GRP (WE) GRP (EE) GRP (SW)
MEAN 2.60 3.88 4.80 5.71
SUBSET 2
GROUP GRP (WE) GRP (EE) GRP (SW) GRP (NW)
MEAN 3.88 4.80 5.71 6.36
TABLE XVb.
RESULTS OF MULTIPLE RANGE TEST, DUNCAN AT AN ALPHA OF .10.
SUBSET 1
GROUP GRP (B) GRP (WE) GRP (EE)
MEAN 2.60 3.88 4.80
GROUP GRP (WE) GRP (EB) GRP (SW) GRP (NW)
MEAN 3.88 4.80 5.71 6.36

132

results of an analysis of variance on these three groups
(Table XVI) Showed an F-ratio of 5.12 at an alpha value of
.007. When the multiple range test for the five per cent
level was performed as above, the results (Table XVa)
indicated that the Europeans still overlapped the White

and Black groups. A more liberal contrast test was run at
an alpha of .10 to see if the European group would shift
towards the White group and the opposite was observed. The
European group appears more homogeneous to the Black rather
than the White group, albeit with very little statistical
significance.

It was decided to collapse the European group in
with the U.S. White group. Here the meaningful Significance
was considered more important than the statistical signifi-
cance. Not only did the groups logically break down this
way but with a further inspection of their means, they
separated in a way that was expected. In fact when the
standard deviations for the groups are considered, it could
be hypothesized that with a larger sample size, the Europeans
would segregate into a subset separate from both Blacks and
Whites. With the only logical heterogeneous groups being
White and Black the remaining tests were performed in light
of these racial groupings. A oneway analysis of variance on
these two groups indicated that they were heterogeneous for
reported symptom duration with an alpha of .008 (Table XVII).
Even with the tendency for the European group to pull the

White values closer to the Black group values, their

133

TABLE XVI

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON REPORTED DURATION
OF SYMPTOMS BY ETHNIC BACKGROUND

 

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 2 272.92 136.46 5.12 .007
WITHIN GROUPS 132 3520.14 26.67
TOTAL 134 3793.06
_ 95% CONFIDENCE
GROUP N X S S.E. MIN MAX LIMITS
1. (B) 28 2.60 3.34 .63 .25 17.00 1.30 —- 3.89
2. (W) 73 6.15 6.28 .73 .25 30.00 4.68 -- 7.61
3. (E) 34 4.36 3.36 .57 .25 12.00 3.20 -- 5.51
TOTAL 135 4.96 .25 30.00

TABLE XVIa.

RESULTS OF MULTIPLE RANGE TEST, TUKEY-B AT AN ALPHA OF .05.

SUBSET 1

GROUP GRP (B) GRP (E)
MEAN 2.60 4.36
SUBSET 2

GROUP GRP (E) GRP (W)
MEAN 4.36 6.15

TABLE XVIb.

RESULTS OF MULTIPLE RANGE TEST, DUNCAN AT AN ALPHA OF .10.

SUBSET 1

GROUP GRP (B) GRP (E)
MEAN 2.60 4.36
GROUP GRP (W)

MEAN 6.15

134

TABLE XVII

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON REPORTED DURATION
OF SYMPTOMS BY ETHNIC BACKGROUND.

 

SUM OF MEAN F F
SOURCE D.F SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 1 197.55 197.55 7.31 .008
WITHIN GROUPS 33 3595.51 27.03
TOTAL 134 3793.06

_ 95% CONFIDENCE
GROUP N x S S.E. MIN MAX LIMITS

1. US BLACK 28 2.60 3.34 .63 .25 17.00 1.30 -- 3.89
2. US WHITE 107 5.58 5.58 .54 .25 30.00 4.51 -- 6.65
TOTAL 135 .25 30.00

 

 

TABLE XVIIIa.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON OCCUPATION BY
REPORTED SYMPTOM DURATION.

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 299.13 99.71 3.38 .021
WITHIN GROUPS 112 3299.40
TOTAL 115 3598.53

_ 95% CONFICENCE
GROUP N x S S.E. MIN MAX LIMITS

SKILLED 60 6.36 6.54 .84 .25 30.00 4.67 -- 8.05
UNSKILLED 14 5.71 5.07 1.36 1.00 18.00 2.78 -- 8.64
S-EMPLD 27 2.55 3.39 .65 .25 17.00 1.20 -- 3.89
. W-COLLAR 15 3.88 3.18 .82 .25 12.00 2.12 -- 5.64
TOTAL 116 5.07 .25 30.00

 

b 041v P1
0 0 O

 

 

135

placement within the White group did not inhibit the observ-
ance of a strong racial difference in duration times. The
distribution of reported symptom duration times as a
function of the per cent reporting per racial group is
presented in Figure 15.

Analysis of variance for occupation by duration
(Table XVIIIa) indicated that the variances were hetero—
geneous (p=.021). The labor classification with the longest
duration was skilled workers and the shortest mean duration
was represented by patients who were self-employed. An a

posteriori contrast test on these four groups (Table XVIIIb)

 

showed a tendency for self-employed and white collar workers
to group together and for skilled and unskilled workers to
group together. The average difference between these groups
means was three months.

A oneway analysis of variance on marital status
(Table XIXa) Showed a tendency toward heterogeneity of value
means with an alpha of .06. However, the results of an

a posteriori contrast test indicated that all marital values

 

belonged in the same subset (Table XIXb) at the five per

cent level with married patients having the longest mean

duration and widowed patients having the shortest duration.
The results of a oneway analysis of variance on age

at diagnosis by duration (Table XXa) indicated that the age

groups were homogeneous in regard to duration with an F-ratio

of 1.15 and an alpha of .327. In addition to these results,

an a posteriori test performed on these data indicated that

 

136

96.5 35.... .0 E032. >3 Stoop. E «529:: .o c0205? .0 octantfio
9 about

.7-.. 183.3 I 2231...:
0—0 0pm JNKW O-N MN VFW MN N—N —.N O_N 0.— Q... kw 0——

‘

3 N. o. m c 4 a
Wp— — mp — —.——@ P

 

I

In
('9

l

l

to
V

 

l—l

II)
V)

dnoJB sguqie Jed seso: peuode: go weaned

If)

II)

In
N

137

TABLE XVIIIb.

RESULTS OF MULTIPLE RANGE TEST,

TUKEY-B AT AN ALPHA OF

GRP (W-C)

3.88

GRP (USK)

5.71

SUBSET 1

GROUP GRP (S-E)
MEAN 2.55
SUBSET 2

GROUP GRP (W-C)
MEAN 3.88

GRP (USK)

5.71

TABLE

XIXa.

GRP (SKD)

6.36

.05.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON MARITAL STATUS BY
REPORTED SYMPTOM DURATION.

 

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB
BETWEEN GROUPS 3 206.97 68.99 2.49 .063
WITHIN GROUPS 131 3626.09 27.68
TOTAL 134 3833.06
_ 95% CONFIDENCE
GROUP N x S S.E. MIN MAX LIMITS
1. SINGLE 21 3.00 3.56 .78 .25 12.00 1.37 -- 4.61
2. MARRIED 78 5.87 5.44 .62 .25 24.00 4.65 -- 7.10
DIVORCED 23 5.30 6.86 1.43 .25 30.00 2.34 -- 8.27
4. WIDOWED 13 2.85 2.14 .59 .25 6.00 1.55 -- 4.14
TOTAL 135 5.04 .25 30.00

TABLE XIXb.

RESULTS OF MULTIPLE RANGE TEST, TUKEY-B AT AN ALPHA OF

SUBSET 1

GROUP
MEAN

GRP (W)
2.85

GRP (S)
3.00

GRP (D)

5.30

GRP (M)
5.

87

.05.

SUBSET 2

GROUP
MEAN

*******

*******

*******

*******

138
TABLE XXa.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON REPORTED DURATION
OF SYMPTOMS BY AGE AT DIAGNOSIS.

 

 

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 11 354.63 32.24 1.15 .327
WITHIN GROUPS 123 3438.44
TOTAL 135 3793.06
_ 95% CONFIDENCE
GROUP N X S S.E. MIN MAX LIMITS
(30-35) 1 6.00 0.00 0.00 6.00 6.00 6.00 - 6.00
2. (35-40) 1 3.50 0.00 0.00 3.50 3.50 3.50 - 3.50
3. (40-45) 6 2.08 .92 .37 1.00 3.00 1.12 - 3.04
4. (45-50) 12 5.27 7.97 2.30 .25 30.00 .20 — 10.34
5. (50-55) 6 8.21 7.78 3.18 .25 18.00 .04 - 16.28
6. (55-60) 23 4.17 3.64 .76 .25 12.00 2.60 - 5.75
7. (60-65) 20 4.85 4.29 .96 .25 12.00 2.84 - 6.86
8. (65-70) 18 6.83 7.70 1.81 .25 24.00 3.00 - 10.66
. (70-75) 18 3.82 3.23 .76 .25 12.00 2.21 - 5.42
10. (75-80) 19 6.64 5.61 1.29 .25 20.00 3.94 - 9.35
11. (80-85) 9 2.17 2.20 .73 .25 6.00 .47 - 3.86
12. (85-90) 2 2.50 .71 .50 .25 3.00 -3.85 - 8.85
TOTAL 135 4.96 .25 30.00
TABLE XXb.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON AGE AT DIAGNOSIS
BY ETHNIC BACKGROUND.

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB
BETWEEN GROUPS 1 291.49 291.49 17 681
WITHIN GROUPS 133 227984.91 1714.17
TOTAL 134 228276.40
_ 95% CONFIDENCE
GROUP N x S S.E. MIN MAX LIMITS

 

1. US BLACK 28 63.64 15.05 2.85
2. US WHITE 107 64.50 11.02 1.06
TOTAL 135 64.28

36.08 85.50 57.62 -- 69.29
31.58 86.33 62.39 -- 66.61
31.58 86.33

 

 

139

all age groups occupied the same subset at the five per cent
level. A second analysis of variance was performed on age
at diagnosis by ethnic group (Table XXb) and indicated that
the age groups were homogeneous in terms of their variances
and means when ethnicity was controlled. The mean age for
Black cancer patients was 63.5 years and for White patients
it was 64.5 years.

A oneway analysis of variance on stage at diagnosis
by duration (Table XXI) showed that the variance between
stage classifications were equal. A similar result was also

obtained through an a posteriori test whiCh groups all stage

 

classifications into a single subset at a ninety-five per
cent confidence level.

A test of variance for population size of birth
place at time of birth (Table XXIIa) showed the variance to

be homogeneous (p=.95). And, an a posteriori test (Table

 

XXIIb) supported this finding by placing all values into the
same subset at the five per cent level.

Comparisons of within variable group mean duration
times based on all cases in the sample were computed by
t-test procedures (Tables XXIII and XXIV) and the results
are presented below. The variables which demonstrated a
significant relationship with reported symptom duration time
at a five per cent level or less are: poor appetite (p=.000),
ethnic background (p=.001), nausea (p=.001), Shortness of
breath (p=.003), primary Site (p=.008), marital status

(p=.03), occasional similar medical problems (p=.04),

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON

140

TABLE XXI.

OF SYMPTOMS BY STAGE AT DIAGNOSIS.

REPORTED DURATION

 

 

 

SUM OF MEAN F F

SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 97.99 32.66 1.15 .332
WITHIN GROUPS 130 3694.14 28.42
TOTAL 133 3792.13

_ 95% CONFIDENCE

GROUP N x S S.E. MIN MAX LIMITS
1. (A) 2 2.50 2.12 1.50 1.00 4.00 -l6.56 -- 21.56
2. (B) 46 5.55 6.23 .92 .25 30.00 3.70 -— 7.40
3. (C) 40 5.64 6.14 .97 .25 24.00 3.68 -- 7.61
4. (D) 46 3.91 3.24 .48 .2 12.00 2.94 -- 4.87

TOTAL 134 4.97 .25 30.00

TABLE XXIIa.

RESULTS OF ONEWAY ANALYSIS OF VARIANCE ON POPULATION SIZE OF
BIRTH PLACE BY REPORTED SYMPTOM DURATION.

 

 

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB
BETWEEN GROUPS 3 12.05 4.02 .11 .955
WITHIN GROUPS 8; 3061.61 36.89
TOTAL 86 3073.66
_ 95% CONFIDENCE
GROUP N x S S MIN MAX LIMITS
1. UNDER
1,000 21 5.86 6 05 l 32 .25 24.00 3.10 -- 8.61
2. 1,000-
99,999 24 5. 7.69 1.57 .25 30.00 2.73 —— 9.22
3. 10,000-
999,999 16 5. 5.10 1.27 .25 18.00 2.31 -- 7.75
4 O OVER
1,000,000 26 5.32 4.81 .94 .25 20.00 3.37 -- 7.26
TOTAL 87 5. .25 30.00

141
TABLE XXIIb.

RESULTS OF MULTIPLE RANGE TEST, TUKEY-B AT AN ALPHA OF .05.

SUBSET 1
GROUP GRP (3) GRP (4) GRP (1) GRP (2)
MEAN 5.03 5.32 5.86 5.98
SUBSET 2
GROUP ******* ******* ******* *******

MEAN

142

TABLE XXIII

BETWEEN GROUP VARIANCES OF ALL VARIABLES SELECTED FOR STUDY
AS GENERATED BY T-TEST PROCEDURES.

 

GRP GRP
VARIABLE GRP SIZE F p/F VAR
E3h33333§ 33 33353 133 2.79 .003 ¢
38.3333333 33333333 33 1-09 .789 =
giaffiilTAL QSEESHER 3; 1.35 .238 =
3ir33p3i33 33 33333 33 1-99 .029 ¢
35m333HiiT E2 CA 3; 1.75 .102 =
3§.P3333§3 333333335 33 1-01 1 000 =
33 3:33:33 EANREETD 3; 1-37 -234 =
33a3303333 I 23 1-54 .087 =
33a3303333. I 33 2-28 .008 H
33L533P333. : :3 1-59 -080 =
333333P§333. I 133 1-95 .080 =
12. NO PROB ; 13; 1.71 .105 =
fiiénAggglAG 33/37371 2% 1.07 .777 =
333.V33e331 I 33 1-99 .006 #
3336333t333 I Iii 1-49 .274 =
iEAmYA§a333 I 1:2 1.43 .302 =
3311V333303 ; 23 1.53 .082 =
18. VAR 005 - 125 2.26 .181 =

+

Crampy P 10

143

TABLE XXIII (continued)

 

GRP GRP

VARIABLE GRP SIZE F p/F VAR
éigrYAfidgog I 1:2 1.60 .298 =
Sggr¥§§a°°7 I g; 1.58 .064 =
33353331338 I 33 1.73 .026 ¢
333h33RL333 I 33 2.42 .000 #
330.33R 010 l 12; 7.54 .009 g
3:3k3335011 I 13) 1.74 .072 =
3335333 333s I 133 2-40 .127 =
333t3331333 I Ii: 1-52 .219 =
333533R 014 I 13; 3-83 .000 ¢
333rv33p333 I 13; 5~07 .000 #
3:93: 016 I 12; 1.72 .458 =
333.V3:1333. ; Iii 2.90 .006 #
3.1.4.2883: 33333 333 33 1. 56 -072 =
3§im§§§M333§ 33333M 22 2.87 .ooo #

144

TABLE XXIV.

STATISTICAL RELATIONSHIP OF VARIABLE MEANS FOR THE CONTINUOUS DEPENDENT
VARIABLE OF SYMPTOM DURATION, GENERATED BY T-TEST PROCEDURES.

 

GRP GRP GRP GRP
VARIABLE GROUP SIZE MEAN S.D. T df. p/T VAR
23222222 2222222 132 2:22 2:22 71.14 2
32.333333: 33333333 33 3233 3233 ~60 125-00 -955 =
géaifingAL TOQETHER 33 3:33 3:33 "2'20 133'00 '030 =
2322222: 2228888 22 2:22 2:23 2
33miigfléiT NSACA 3; 2:2: 3:;: - .23 107.00 .822 =
25222222 222222225 23 2:22 2:33 - 128-00 =
2.222222222122213 23 2:22 2:22 1.14 =
33aZ3OSSE3 - :3 2:23 3:23 -1.39 128.00 .167 =
33a3303333 I 3: 3:2: 2:2; 2.08 91.71 .040 y
3:;si3P:::. I :3 2::; 3:3: - .97 128.00 .333 =
3335:3P3353 I 123 2:3: 3:2: .56 128.00 .574 S
12' NO PROB : 132 :23: 3:3: .02 128.00 .985 =
221.222?“ 222332 22 2:22 2:32 .721 =
3:3,V238231 I 33 ::23 2:32 -1.95 124.53 .054 ¢
3330333t223 I 13: 2:33 2:3: - .46 133.00 .644 =
:Eém7335233 : Ii: 2:3: 2::: -1.40 133.00 .165 =
:ZiIVEEgEO: ; :3 2:3: :23: - .30 133.00 .767 =
334333333 I 133 3233 3233 ~24 133-00 m =

145

TABLE XXIV (continued)

 

GRP GRP GRP GRP
VARIABLE GROUP SIZE MEAN S.D. T df. AP/T VAR
222.2213": ; 1:: :33 33;: - .03 133.00 .976 =
333x333a007 I 33 2:3: 2:22 -1.94 133.00 .055 =
32253251223 1 E: 2:12 2:8: -2.05 92.57 .044 g
333HZ3RL233 3 3: 2:3: 2:33 .38 105.55 .702 g
336,:3R 010 I 12; 2:1: 3:33 3.56 15.04 .003 ¢
333kZZSSOll I 13: 2:3: 2:3: — .12 133.00 .904 =
33§b323 33:5 I If? 3:23 3::: 1.00 133.00 .321 =
332t3231223 ; 1:: 2:3: 22:: .01 133.00 .994 =
:3fi5:2R 014 I 1:; 3:3: 3:28 3.60 81.16 .001 g
332rvifiéiii I 13; 3:2: 3:23 4.10 100.95 .000 y
3:1a333 016 I 12; 3:3: ::ii .75 133.00 .455 =
33;,V231323, I 133 3:2: 2:32 -1.46 10.62 .172 g
332.3333: 33333 333 33 3233 3233 - .7?- 133-00 .475 =
332miiss§fie EEEEEM 2? 2:2: 2:23 -2.69 102.53 .008 g

146

constipation (p=.044). In addition to these above listed
variables there were two marginally significant variables:
rectal bleeding (p=.054), and diarrhea (p=.055). These

variables were grouped into either symptom or non-symptom
classes and were categorized as to whether they speeded or

slowed detection (see Figures 16 and 17 below).

 

Symptom Speeds Detection Slows Detection
Rectal Bleeding *****
Diarrhea *****
Constipation *****
Shortness of Breath *****

Nausea *****
Poor Appetite *****

 

Figure 16. Symptom profile for detection

 

Non-Symptom Variable Speeds Detection Slows Detection
Marital Status (Married) *****
Primary Site (Rectum) *****
Occasional Similar Prbms. *****

Ethnic Group (Black) *****

 

Figure 17. Non-symptom profile for detection

These profiles would suggest that for this sample a
patient with a slow detection rate would tend to be White,
married, with a medical history of occasional similar

problems, diagnosed with rectal cancer, and having the

147

symptoms of rectal bleeding, diarrhea, and/or constipation.

A patient with a speedy detection rate would tend to be
Black, unmarried, without a personal medical history of
occasional similar problems, diagnosed with colonic cancer,
and having any of the following symptoms of shortness of
breath, nausea, and poor appetite. It is interesting to

note that when the ethnic groups were compared, the relation-
ship for symptom and non-symptom profiles were the same.

A X2 test of independence was performed here to determine if
any of the variables in these profiles covaried with ethnic
background. As in the tests of independence for the pre-
viously discussed symptom clusters, the critical values for
chi square in alphas of .05 and .01 are 3.84 and 6.63
respectively. Rectal bleeding was found to occur independent
of ethnic group (X2=.04) as was diarrhea (X2=.75), constipa-
tion (X2=2.l7), shortness of breath (X2=1.45), and primary
site (X2=2.l7). However, the symptom of nausea (x:=6.76)
highly significant with an alpha of .01, and poor appetite

by inference were found to occur more often among Whites

than Blacks.

The remaining twenty-two variables tested did not
show a significant relationship with reported duration time
at the five per cent critical level. This group of variables
included both sociodemographic and medically related
variables. The sociodemographic variables included: popula-
tion size of birth place (p=.699), age at diagnosis (p=.721),

and patient occupation (p=.955). The medical history

148

variables were: occasional similar problems (p=.l67),
medical similarity (p=-256), persistent medical problems
(p=.333), persistent similar medical problems (p=.574),
frequency of medical problems (p=.86), and no unusual
problems (p=.985). Family medical history was also not
significantly related to duration (p=.822). Stage at
diagnosis was tested with one group representing stages
"A" and "B" and the other being stages "C" and "D", and was
not significant (p=.475). The remaining eleven variables
that showed no relationship with duration were: abdominal
pain at defecation (p=.644); decreased caliber of stools
(p=.455); blood in stools (p=.644); weight loss (p=.702);
dull persistent pain (p=.767); crampy abdominal pain
(p=.812); weakness (p=.904); general body pain (p=.976);
and distention (p=.994).

A second series of group mean comparisons was made
on those variables relevant to the previously stated hypo-
theses and those that showed a significant relationship with
duration time from the first series of t-tests (see Tables
XXVa, XXVb, XXVIa, and XXVIb). Out of the twelve tested
variables for Blacks and Whites, there are four that were
relevant to the stated hypotheses. These four variables are
family medical history, medical similarity, persistent
similar medical problems, and occasional similar medical
problems. Of these four variables only occasional similar
medical problems was previously found to have any significant

relationship with duration time. The results of this second

149

TABLE XXVa.

RESULTS OF T-TEST PROCEDURES ON VARIABLES DEMONSTRATING A
SIGNIFICANT RELATIONSHIP WITH "CONTINUOUS" DURATION, AND
ALL VARIABLES INCLUDED IN HYPOTHESES (*). BETWEEN GROUP
VARIANCES FOR WITHIN ETHNIC GROUP MEANS BY SYMPTOM DURATION.

A. U.S. BLACK

 

 

 

GRP GRP
VARIABLE GRP SIZE F IE/F VAR
3£a333ITAL TOGETHER ii 4.41 .017 7e
Priigfiglggte Eggggm 33 5.68 .003 g
*33m333H33T NOCEA 1: 9.75 .040 g
*3; 3333333 EgNEgén 3: 8.21 .000 g
*géa3303333 I 13 9.22 .000 g
33.333333. I 23 0.00 1.000 =
336331333 1 1; 3.30 .045 ,.
gia¥:§6207 I 13 2.48 .192 =
33n333t333. I 13 8.32 .000 g
13:0333 010 I 2; 1.20 1.000 =
135u33§ 014 I i3 4.00 .039 g
lgoozAApgiie. ; 2% 2-90 -l90 =
B. U.S. WHITE
ééafifingAL TOGETHER 23 1.17 .596 =
3&13333133te 33333». 33 2 - 61 . 001 ,.
*33m333fl33T Noe? Si 1. 87 . 106 =
*3; 3:32:33 EQNQEQD 23 1.66 .088 =

150
TABLE XXVa. (continued)

B. U.S. WHITE

 

GRP GRP

VARIABLE GRP SIZE F p/F VAR
*3;.:?°§EE? : g: 4.10 .000 I
*IIrIIIRIII. I 33 2.24 .050 =
EécYA§12gé I 3: 1.89 .024 #
giaZ§fie207 I 22 1.66 .066 =
300323133? I 2: 1.54 .119 =
lg:oY§§ 010 I 10: 17.46 .012 g
1&50222 014 I 33 3.40 .000 #
2300¥A§02I30. I Si 5.40 .000 ¢

TABLE XXVb.

RESULTS OF T-TEST PROCEDURES ON VARIABLES CONTAINED WITHIN
HYPOTHESES. BETWEEN GROUP VARIANCES FOR BETWEEN ETHNIC
GROUP MEANS BY SYMPTOM DURATION.

 

GRP GRP

VARIABLE GRP SIZE F p/B VAR
IImIIIHESI I IIIIE 23 10-75 -°33 #
2;.PEEI2520 I SIIIE II 1-34 .402 =
36.2?Oiii? I SIIEE 22 3.07 -°°9 I
IIrZIIRIII. I SIIII 15 0-00 1-000 =

151

TABLE XXVIa.

RESULTS OF T-TEST PROCEDURES ON VARIABLES DEMONSTRATING A SIGNIFICANT
RELATIONSHIP WITH "CONTINUOUS" DURATION, AND ALL VARIABLES INCLUDED IN

HYPOTHESES (*).

A. U.S. BLACK

WITHIN ETHNIC GROUP MEANS BY SYMPTOM DURATION.

 

 

 

GRP GRP GRP 2 GRP

VARIABLE GROUP SIZE MEAN S.D. T df p/T VAR
:Ea:::ITAL TOGETHER I: 3:2: 1:3: ' '49 22'62 '632 ¢
EAIZISIIEI. £32330 :3 3:23 2:33 - 10.90 3
*gémiIgHéi? NOCEA 1: i::: iIE: 1.10 20.65 .282 ¢
*3; £25222: ggNEgiTD I: §:§g 22;: ' ’52 11'59 '612 #
*Sgaif‘o‘éii‘: ;; 12 IS: :32 - 91 8-83 9
*gér:E?R:§:. ; 2: 2:2: 3:38 1.03 26.00 .311 =
ISIS: 1 :2 3:3; 2:3 - I
gia:::e207 I 1: §Iii 3:3: - .55 26.00 .585 =
EInZIEtggi, ; 1: 1:33 ::38 -1.16 8.92 .276 g
12207:? 010 I 2: :22: 3::3 .10 26.00 .923 =
liéu::: 014 I 13 3::2 i:§i 1.53 25.85 .137 ¢
11350333.. I "I i133 iii? 1.21 26-00 .237 =

B. U.S. WHITE

:éaifiilTAL TSEEESER 2: ::§: 2:22 —2.31 105.00 .030 =
IIIIEI31§§te 22:33“ 2: 2:3: ::g: -2.31 90.84 .023 g
$033“? 33 2:22 2:33 - =
*3;.P§§I‘§Ii‘d EINIEfI. 23 2:33 32:33 1.50 100.00 .137 =

152

 

TABLE XXVIa. (continued)

GRP GRP GRP 2 GRP

VARIABLE GROUP SIZE MEAN S.D. T df, p/T VAR
*igaZIOEEE? I :2 3:3: 3:32 3.02 87.76 .003 ' g
*SngIIRifi I 3: 2:2: 3:23 .68 100.00 .496 =
lgcYAfilggé ; 2: 2:28 2:22 -1.99 98.65 .050 g
SiaZIEeSO7 : :2 2:2: 2:33 -1.68 105.00 .095 =
Saniiitfiii. : 2: 2:33 2:38 =
1210727 010 I 10: I:;: EZSE 4.63 13.39 .000 g
IIIUZ2: 014 I i: EIS: 3:2; 2.65 44.84 .011 ¢
IIAOZAIISIie. 1 3i §:§2 2:22 3-74 76-76 .000 3

TABLE XXVIb.

RESULTS OF T-TEST PROCEDURES ON VARIABLES CONTAINED WITHIN HYPOTHESES.
BETWEEN ETHNIC GROUP MEANS BY SYMPTOM DURATION.

 

GRP GRP GRP GRP

VARIABLE GROUP SIZE MEAN S.D. T df. p/T VAR
IAmIIIHIET I IIIII 23 I:§Z 1:33 ~3-71 22.35 .001 3
2;,Piifiiid : 332;: 1; 3;33 3;33 - .08 51.00 .302 .
38.:IOIEI? : IIIII 22 3:22 3:33 -15 9-47 .884 3
2.4253532. 13:23; 13 II: 3:28 ~1-67 =

153

series of t-tests will first be presented for U.S. Blacks
and then for U.S. Whites.

None of the twelve variables tested within the
Black group of patients (N=28) indicated any significant
relationship with symptom duration time. The significance
levels for the four hypothesis related variables were:
family medical history (p=.282); medical similarity (p=.612);
persistent similar problems (p=.311); and occasional similar
problems (p=.388). The remaining eight variables were:
nausea (p=.137); poor appetite (p=.237); constipation
(p=.624); marital status (p=.632); and shortness of breath
(p=.923).

For the White group of patients (N=107), there were
a number of significant relationships noted. The test
results for the hypothesis related variables were: family
medical history (p=.753); medical similarity (p=.137);
persistent similar problems (p=.496). The fourth related
variable of occasional similar problems was found to be
highly significant (p=.003) with the mean time for its
presence being two and one half months shorter than for the
conditions absence.

Of the eight remaining variables tested for the
White group, six were found to be significantly related to
duration time. The symptom of poor appetite was highly
significant (p=.000) with the mean reported duration for
its presence being three months shorter than for its

absence. Shortness of breath was also found to be highly

154

significant (p=.000) with the mean reported duration for its
presence being almost four months shorter than for its
absence. Nausea was significant (p=.Ol) with the mean
reported time for its presence being almost four months
shorter than for its absence. Nausea was significant
(p=.Oll) with the mean reported time for its presence being
almost three months shorter than for its absence. The
variable of primary site (p=.023) indicated that the mean
reported time for White colon patients was almost two and a
half months shorter than for White rectal patients. Marital
status was shown to be significant (p=.03) with single,
divorced, and widowed patients having a mean that was two
months shorter than for patients reported as being married.
The symptom of rectal bleeding (p=.05) showed a mean time
for its presence as being two months longer than for its
absence. The final two symptoms of diarrhea and constipa—
tion were both non-significant with alpha levels of (p=.095)
and (p=.112) respectively.

Following up on observations noted by Snow (l978a:
81), the compound symptom of nausea/poor appetite (used
together because of indicated covariance) and the symptom
of weight loss were used to test for possible cultural
differences in reporting times within and between racial
groups. As mentioned earlier, low income Blacks tend to
classify symptoms as natural or unnatural relative to their
medical belief system. In order to test whether this

natural-unnatural classification had any relationship with

155

duration time two groups of Blacks and Whites were compared

(Table XXVII). The first group reported a weight loss and

TABLE XXVII--Effects of normal appetite and loss of appetite
with weight loss on symptom duration time.

POOR APPETITE

 

RACE N NAUSEA X S.D. RANGE SIGNF.
BLACK 5 + 1.1 1.13 .25 - 3.00
7 - 5.7 5.20 2.50 - 17.00 p=.45
WHITE 20 + 4.0 2.65 .25 - 12.00
37 - 5.9 4.73 .50 - 24.00 p=.90

 

poor appetite or nausea. In the second group the patients
reported a weight loss but no nausea or poor appetite. A
comparison of the two Black groups showed that the natural
condition of weight loss and poor appetite/nausea had a mean
duration time that was considerable less than the unnatural
condition of weight loss with normal appetite. The same
directional relationship existed for the two White groups
but the difference was not as large. Also, while the ranges
of duration times for the Black groups were almost mutually
exclusive this was not the case for the White groups.
Although the tests were not statistically significant a chi
square test of independence for the Blacks indicated that
the relationship between the occurrence of poor appetite and
duration time (before or after the median) was not random
(x2=3.7 at a .05 level).

A final series of t-tests were performed on the
between ethnic group relationships for the four hypotheti-

cally related variables of: 1) positive family medical

156

history; 2) presence of similar medical problems; 3)
presence of occasional similar medical problems; and 4)
presence or persistent similar medical problems. Family
medical history was found to be highly significant (p=.001)
with the Blacks tending to report in almost four months
sooner than the Whites (Black N=5, White N=23). The
remaining three variables were not found to be significantly
related to reported duration and their test results are:
presence of similar medical problems (p=.382); presence of
occasional similar medical problems (p=.884); and presence

of persistent similar medical problems (p=.112).

Symptom Duration as a Discrete Variable

 

In presenting these results, any mention of symptom
duration will be in the context of before or equal to, or
after the median. That is before or after a three month
time marker.

The results from this log linear procedure include
an analysis of the relationship between reported symptom
duration and ethnic background; variable occurrence by
ethnic background; the variable under consideration by
reported symptom duration; and the interaction between
ethnic background, reported variable value, and reported
symptom duration.

The 109 linear analysis of reported symptom duration
by ethnic background indicated that in this sample Blacks

appear to be four times more likely to report before the

157

median than Whites. This relationship was highly signif-
icant with an alpha of .002 (Table XXVIII).

The analysis of the relationship between variable
occurrence and ethnic background involved the nine variables
shown to be significant by the first battery of t-tests and
four other variables that had approached significance (i.e.,
an alpha less than .30). In order for a variable to be
considered to have occurred more often for one group than
the other, a 9 value of less than .49 or greater than 1.9
was necessary (for a complete listing of the computed values
see Table XXIX). Out of the fourteen variables tested by
this analysis, four were shown to be significant at the five
per cent level. These four variables were: nausea,
decreased caliber of stools, persistant similar problems,
and occasional medical problems.

With an alpha of .016, the symptom of nausea was
indicated to be three times more likely to be reported for
Whites than for Blacks. The symptom of decreased caliber
of stools was shown to be significant (p=.027) and the
results indicated that this symptom was four times more
likely to be reported by Whites than Blacks. The condition
of persistant similar problems was shown to be occurring
evenly among Blacks and Whites and was significant (p=.045).
Occasional problems were significant with an alpha of .049
and were shown to be twice as likely to be reported by
Blacks than Whites. The remaining ten variables were not

found to be significant and include: primary site (p=.l7l);

158

TABLE XXVIII.

RESULTS OF CROSS-PRODUCT RATIO ANALYSIS: DURATION OF SYMPTOMS
(ABOVE/BELOW MEDIAN) BY ETHNIC GROUP.

 

ETHNICITY (9) (S'D-’ (Z) (p/Z) (STAT. DECSN.)
1. U.S. BLACK 4.18
2. U.S. WHITE .24 4'99 2-87 -002 s.

(9): Gamma or cross-product ratio based on a Black to White
comparison (i.e., the Black gamma is the reciprocal of
the White gamma).

(S.D.)= Standard deviation for the 2 X 2 table values.

(z)= Standard score for reported gamma.

(p/Z)= Probability for A score.
(STAT.
DECSN.)= Statistical decision.
S= Significant relationship
NS= Nonsignificant relationship

MAIN EFFECTS FOR VARIABLE OCCURANCE BY ETHNICITY.

159

TABLE XXIX

HERE GAMMA REPRESENTS
THE RATIO OF OCCURANCE FOR BLACKS GIVEN THE FIRST LISTED VARIABLE VALUE.

 

VAR
VARIABLE VALUE (3) (S.D.) (Z) (p/z) (STAT. DECSN.)
*1. PRMSITE COLON
.9 . , . _ .
Primary Site RECTUM l O 439 l 47 071 N S
*2. MARITAL ALONE
. 7 . -, . . .
Status TOGETHER 9 474 07 472 N S
3. PERMDRL NON-CA
1. . , , .
Ca. Related CA RLTD 08 436 18 429 N S
4' PHOCCPR ' 2.22 .480 1.66 .049 s,
Ocas. Prb. +
5. PHPRSRL -
Perst. th. + ~78 ~465 1.69 .045 s.
*6. VAR 001 -
Rec. Blood + ’92 '425 '20 ~421 N.S
7. VAR 003 -
Abdm. Pain 1.80 .791 .74 .230 n.5,
* -
8i VAR 007 1.47 .450 .86 .195 n.5,
Diarrhea +
* -
9' VAR 908 1.53 .450 .95 .171 N,s
Constption. +
*10. VAR 010 -
.40 .7 - . 7 . , ,
s.0.B. + 64 1 1 121 N s
* -
11’ VAR 014 .36 .471 -2.14 .016 s,
Nausea +
* —
12. VAR 015 .73 .500 _.62 .268 N's
Poor Appite. +
13' VAR 017 7 .25 .713 -1.93 .027 5,
Dec. Calibr. +
*14. PHOCCRL -
Ocas. thd. + ’84 ’480 '35 -360 N.S.

(*) Variables that showed a significant relationship by t—test.
All other variables were significant at least at the .25 level.

160

shortness of breath (p=.121); constipation (p=.l7l);

diarrhea (p=.l95); abdominal pain at defecation (p=.230);
poor appetite (p=.268); occasional similar problems (p=.360);
rectal bleeding (p=.421); similar medical problems (p=.429);
and marital status (p=.472).

In an analysis of the relationship between a partic-
ular value of the variable and reported symptom duration
(Table XXIX), eight out of fourteen variables were found to
show a significant relationship. These variables were:
constipation, persistent similar problems, occasional
similar problems, diarrhea, marital status, nausea, decreased
caliber of stools, and poor appetite. Of these eight
variables, four seemed to be correlated with a slowness in
detection: constipation, diarrhea, marital status (married),
and decreased caliber of stools. The remaining four
variables of: persistent similar medical problems, occasional
similar problems, nausea, and poor appetite seemed to

correspond with a speedy detection (Figure 18).

 

Variable Speedy Detection Slow Detection
Constipation ******
Diarrhea ******
Marital Status (married) 1 ******
Decreased Caliber Stools ******
Persistent Similar Prblms. ******

Occasional Similar Prblms. ******
Nausea ******
Poor Appetite ******

 

Figure 18. Discrete variable profile for detection

161

TABLE XXX,

MAIN EFFECTS FOR VARIABLE VALUE BY THE DISCRETE VARIABLE SYMPTOM

DURATION.

HERE,

GAMMA REPRESENTS THE RATIO FOR BEFORE MEDIAN
REPORTING GIVEN THE FIRST LISTED VARIABLE VALUE.

 

VARIABLE VALUE (9) (S.D.) (z) (p/Z) (STAT. DESCN.)
*1. PRMSITE COLON
Primary site RECTUM 1.46 .346 1.10 .136 N.S.
*2 . MARITAL ALONE
1.95 . 77 . . .
Status TOGETHER 3 1 78 038 S
3. PERMDRL NON-CA
.77 . —. . . .
Ca. Related CA RLTD 360 74 230 N S
4‘ PHOCCPR ' 1.27 .361 .67 .250 N.S.
Ocas. Prb. +
5. PHPRSRL -
. 7 . - . . .
Perst. th. + 3 421 2 38 009 S
*6. VAR 001 -
Rec. Bleed_ + 1.49 .345 1.16 .123 N.S.
7. VAR 003 -
Abdm. Pain + 1.83 .558 1.09 .138 N.S.
* .-
8? VAR 007 2.20 .359 2.20 .014 S.
Diarrhea +
* ..
9' VAR 908 2.67 .362 2.72 .003 S.
Constption. +
* ..
10' VAR 010 .36 .836 -1.22 .111 N.S.
S.O.B. +
* _
11‘ VAR 014 .46 .449 -1.72 .043 S.
Nau sea +
*12' VAR 015 ' .46 .449 -1.71 .044 S.
Poor Appite. +
13' VAR °i7 ' 3.34 .701 1.72 .043 8.
Dec. Calibr. +
*14. PHOCCRL -
.45 .42 -1.92 .030 S.
Ocas. thd. +

162

The symptom of constipation was significant with an
alpha of .003 and the test indication here was that in the
presence of this symptom people were almost three times as
likely to report in after the median time than when it was
absent. Diarrhea was significant (p=.Ol4) and in its
presence people were shown to be twice as likely to report
in after the median time than when it was absent. Marital
status was significant at an alpha of .038 and the results
for the variable indicated that married people were twice
as likely as single, widowed or divorced patients to report
in after the median time. The symptom of decreased caliber
of stools with an alpha of .043 was indicated for people to
be three times more likely to report in after the median
time when it was present than when it was absent.

Persistent similar problems was found to be signfi-
cant (p=.009) and when present people were almost three
times as likely to report in before the median than in its
absence. Occasional similar medical problems seemed to
create a situation where people were twice as likely to
report in before the median in its presence than in its
absence. Both of the symptoms of nausea (p=.043) and poor
appetite (p=.044) when present caused people to be twice as
likely to report in before the median than if either of
them were absent.

The six remaining variables were not found to have
a significant relationship with median duration reporting.

These six variables were: shortness of breath (p=.lll);

163

rectal bleeding (p=.l23); primary site (p=.136); abdominal
pain at defecation (p=.l38); medical similarity (p=.230);
and occasional medical problems (p=.250).

The final series of log linear procedures concerns
the interaction between ethnic background, variable value,
and reported symptom duration (Table XXX). Three gévalues
were used (g—l, g-2, g-3). The g-l values represents the
main effects for ethnic background given the first listed
value of the variable. The g-Z value refers to the main
effects of ethnicity given the second listed value of the
variable. The g-3 value represents the interaction effects
between these variables and reported symptom duration.
Thirteen out of the fourteen variables tested were found to
not be significant for the interaction relationships.

The only variable that was demonstrated to have a
significant interaction relationship was the personal
medical history variable of persistent similar problems.
This variable was significant (p=.Oll) and the test results
implied that Blacks (in the absence of persistent similar
problems) are considerably more likely to report in before
the median time than Whites.

The thirteen variables shown to not have a signifi-
cant relationship in terms of the interaction effects are:
shortness of breath (p=.095); occasional similar problems
(p=.lOl); medical similarity (p=.156); rectal bleeding
(p=.230); decreased caliber of stools (p=.236); marital

status (p=.255); nausea (p=.264); occasional medical

164
TABLE xxxI.

INTERACTION EFFECTS BETWEEN ETHNICITY AND LISTED VARIABLES BY THE
DISCRETE VARIABLE OF SYMPTOM DURATION. HERE, THERE ARE THREE GAMMAS
PRESENTED: (1) MAIN EFFECTS FOR ETHNICITY GIVEN THE FIRST VARIABLE
VALUE; (2) MAIN EFFECTS FOR ETHNICITY GIVEN THE SECOND VARIABLE VALUE;
AND (3) INTERACTION EFFECTS. THE REMAINING STATISTICS REFER TO THE
INTERACTION GAMMA.

VAR 1 2 3
VARIABLE VALUE (9) (g) (9) (SD) (Z) (p/Z) (S.DECS.)

1. PRMSITE COLON
3.50 5.57 .63 1.05 . , 0 . .
Primary Site RECTUM 44 33 N S

 

2. MARITAL ALONE
Status TOGETHER 5.00 12.32 .41 1.37 .66 .255 N.S.

3. PERMDRL NON-CA
Ca. Related CA RTLD 6.10 2.20 2.77 1.01 1.01 .156 N.S.

4- PHOCCPR ' 4.55 2.62 1.74 1.08 .51 .305 N.S.
Ocas. Prb. +

5. PHPRSRL

Perst. th. + 10.22 .88 11.61 1.07 2.29 .011 S.
6. VAR 001 -

Rec. Bleed + 2.86 6.00 .48 1.00 - .74 .230 N.S.
7° VAR 093 - 3.48 **** **** **** **** **** ****
Abdm. Pain +

8? VAR 007 ' 7.73 1.99 3.89 1.08 1.26 .104 N.S.
Diarrhea +

9- VAR 908 ’ 4.11 4.00 1.03 1.03 .03 .488 N.S.
Constption. +

10- VAR 01° ’ 4.87 .50 9.73 1.74 -1.31 .095 N.S.
5.0.8. +
11- VAR 014 ’ 3.19 7.20 .44 1.29 - .63 .264 N.S.
Nausea +
12- VAR °}5 ‘ 4.24 3.69 1.15 1.30 .11 .456 N.S.
Poor Appite. +

13- VAR 0}7 ’ 4.08 14.00 .29 1.71 - .72 .236 N.S.
Dec. Calibr. +

14- PHOCCRL ‘ 7.74 1.50 5.16 1.28 1.28 .101 N.S.
Ocas. thd. +

(****)= Statistic unavailable due to insufficient data.

165
problems (p=.305); primary site (p=.330); poor appetite

(p=.456); and constipation (p=.488).

Discussion

 

The most important finding in the preceding analysis
was that Black veterans with colorectal cancer used health
care facilities sooner than White veterans with colorectal
cancerlo. Thus, an important question that arises from this
analysis is: why are the Black patients coming in sooner for
medical care than White patients? Although a difference in
reporting between these two groups was expected, it was
expected that White veterans would have utilized the health
facilities more readily than the Blacks. This was because
of the commonly held notion that Blacks utilize care
facilities less than do Whitesll-lz. With this unexpected
shift in utilization, two additional questions also arise:
is this pattern of utilization limited to V.A. hospitals;
and, is this pattern limited to colonic and rectal cancer
or to gastro-intestional conditions in general? Unfortu—
nately, questions such as these are not resolvable by the
analytic tools available in a pilot study such as the one
presented here. The limited nature of available data for
this study permitted only a general illumination of such
differences, not the rationale behind them. However, one
can generate hypotheses for testing in future research which
may provide some explanations for the nature of such

differences.

166

The first step would be to check if these differ-
ences in care patterns existed only within V.A. hospitals or
among all types of hospitals for various medical conditions
and ethnic groups. If future research indicated that this
pattern of care use was different for non-V.A. hospitals,
questions should be directed toward such factors as socio-
economic status or any of the other characteristics which
differentiate veterans from the rest of society (e.g.,
military medical care and education). If it is indicated
that such patterns are limited to colorectal cancer or
general gastro-intestinal conditions, questions should focus
on symptomatologies and ethnicity.

Two main hypotheses can be generated to test some of
these possible contributing factors in the Black-White
utilization gap. 1) Persons who have used low cost health
services in the past should have shorter mean duration times
when using similar services than those who have not had this
prior experience. Here the context in question is the use
of services for such low cost facilities as V.A. hospitals
by persons who at this time are all in the same economic
situation. Proportionally, Blacks are more often repre-
sented in the lower income classes than Whites in the United
States. Therefore, it would be expected that Black veterans
are more likely to have had a history (prior to entering the
service) of low income status than White veterans and thus
more readily use such low cost facilities. 2) The direction

of the relationship between Black and White mean duration

167

times for colorectal cancer (or gastro-intestinal conditions
in general) differs from the direction of the relationship
between Black and White mean durations for other medical
conditions. In other words, the symptomatologies existing
for colorectal cancer (or other G-I conditions) have
received a different emphasis in concern with low income
Blacks versus Whites. A logical extension of this would be
that these concerns would not be equal for other conditions
with different symptoms or anatomical locations (e.g.,
general cold symptoms like fever or sore throat, and
skeletal conditions with the symptoms of joint pain or
swelling).

Although there are a number of factors that could
not be checked for effects on the observed relationships
between ethnicity, symptom duration and the other variables
in the study, it was possible to identify and test three
potentially confounding factors. These factors were repre-
sented by the variables of site of cancer, age at diagnosis,
and stage at diagnosis.

The variable of primary site was found to be related
to symptom duration for the sample in general and for the
White but not the Black group. In all relationships, colon
patients took less time to seek care than rectal cancer
patients. For the White patients, those with colon cancer
had a mean duration that was two and one half months
shorter than rectal cancer patients. While not a signifi-

cant variable among Blacks, the Black colon cancer patients

168

did have a mean duration that was shorter than the Black
rectal cancer patients. Since other studies have also noted
this trend, it was decided to test for the existence of any
relationship between site and the other variables in the
study. These tests had a format very similar to those
conducted for the listed variables in the study and eth-
nicity (i.e., t-test comparison of means). The results
indicated that cancer site was not a significant contributor
to the observed differences between the ethnic groups.

Also, it is important to note that in all tests involving

a comparison between the two sites, the variances were shown
to be heterogeneous. Since this significant relationship
was present for duration as a continuous variable and not as
a discrete one it is suggested that a likely source of this
difference is due to extreme duration times for rectal
cancer patients in general. However, since only one Black
patient was observed to take longer than seven months to
seek care, this suggestion is made with a degree of caution.
Whether this relationship represents factors such as
differential symptom perception, concern over the anatomical
location of the symptoms, or the physiological function of
the symptom site is uncertain at this time. In order to
reduce some of this confusion it is suggested that future
studies should not lump colonic and rectal cancers together
as colorectal cancer has been handled so often in the past.
In future investigations that include colon or rectal cancer

patients, care should be taken to get adequate numbers of

169

patients for each site so that both cancers can be analyzed
separately.

The results of analyses of variance on age at diag-
nosis were significant in that they showed age to be un-
related to symptom duration or ethnic background. Age had
been seen as a possible confounding variable in that it
would have been a factor in such relationships as symptom
onset, symptom tolerance, concern over health status, or any
of several other relationships which could effect duration
time.

With stage at diagnosis representing a measure of
cancer involvement, there was a possibility that it could
cloud the interpretation of the studies results by affecting
such factors as symptom severity, onset, or content. The
tests performed on this variable focused on two levels: by
comparing between site data (i.e., colon versus rectum), and
within site data (e.g., ascending, descending, and sigmoid
colon). The results from all testing indicated that stage
at diagnosis had no correlation with symptom duration, thus
discounting it as a potential problem for interpretation.

To be sure, there are many aspects of this situation that
could introduce errors into the above interpretation, for
example, small sample sizes for some colon sites, differ-
ential growth rates for tumors, the effects of patient recall
and the various filtering mechanisms previously discussed.
Should the effects of any of these sources of random vari-

ation be reduced by modifications in research design, it is

170

quite possible that a correlation between duration and stage
at diagnosis could be uncovered. For example, the use of a
prospective study could reduce some of the noise created by
problems in patient recall and physician filtering of data.
However, until such information comes to light the present
finding of no correlation between stage at diagnosis and
reported symptom duration would seem to have serious impli-
cations for the planning of early detection programs. These
data indicate that a person has an equal chance of being
diagnosed with stage "B", "C", or "D" cancers regardless

of whether he reported within one month or twelve. If this
is indeed the case, it would seem that funding for cancer
education to bring people in as soon as possible for symptoms
of colonic or rectal carcinoma would be better put to use in
the development of various cancer screening programs and thus
increase access to asymptomatic individuals.

The observed relationships between duration and the
variables in the study for the entire sample appear to have
been strangely determined by the relationships between each
variable and the duration among White patients. There are
two possible explanations for this which are not mutually
exclusive. First, it could result from the small sample
size of the Black population. Secondly, with so many Blacks
reporting so early, the within variable mean comparisons for
this group could have been significantly affected by a
restriction on the total range of possible variability. In

order to separate these effects a larger sample of Blacks is

171

clearly indicated.

From the analysis on the variable of marital status
a trend was noted such that single, divorced, or widowed
patients tended to report earlier than patients who were
married. The differences between these two categories of
married and unmarried may have stemmed from the effects of
factors such as multiple member households, the economic
costs of lost work time, or even occupational status (e.g.,
different perceptions of job responsibility). Although
tests on the effects of multiple member households were not
possible, patients who were reported as married may have
more factors operating on them to retard seeking care than
patients who may have had only themselves to look after
(i.e., single, divorced, or widowed). Tests for the effects
of economic cost of work loss were checked by comparing
retired married patients with non-retired married patients
(determined by age at diagnosis being below 65 years or 65
and above). Although the results indicated that non-retired
patients had a mean duration of about a month less than
retired patients the difference was not statistically
significant (p=.4l). The results on tests on occupational
status were also nonsignificant (p=.5), albeit the mean for
blue collar workers was one month less than the mean for
white collar workers. Considering the fact that a signif—
icant relationship for marital status and duration was
indicated, this variable should be further investigated in

future studies. Such a study should be conducted on

172

subjects for which there are more detailed data on such
issues as household composition, employment, and income.
Several hypotheses could be suggested to further
test for the possible effects of these issues in future
studies. 1) The mean symptom duration time for married
patients of households including members other than the
married couple will be significantly longer than for married
patients with only two member households. This hypothesis
would emphasize increased economic and social responsibil-
ities on the male in large households versus small house-
holds. 2) Individuals whose spouse is employed will have
significantly shorter mean duration times than persons
whose spouse is not working. In this hypothesis the
emphasis on economic costs would be further refined.
3) Blue collar patient mean duration times will be signifi-
cantly shorter than white collar patient mean duration times.
The results of the preceding analysis indicate that
a person's past personal medical history may prove to be a
valuable key to understanding why he responds to a given
medical condition in a particular way. In this study, the
presence of occasional similar problems was found to enhance
one's likelihood of a speedy detection. This finding was
also found for persistent similar problems, although the
relationship is considered less significant than for
occasional similar problems because only two Black patients
were reported to have a presence of a persistent similar

problem history. In other words, this correlation of

173

persistent similar problems and a speedy detection is
suggested to be an artifact of sample size, especially
since such a finding was not corroborated by the non-
parametric testing. Although a further investigation into
the influences of personal medical history is clearly
indicated, a more refined method of listing past medical
experiences will be required. One of the problems with the
present study was that the only information available was
from hospital charts, which did not include any detailed
data on private practice care for the patient. Obviously,
people do not report to a hospital for every medical
problem and by restricting one's information base to these
charts a considerable amount of experiencial data could be
beyond reach. It is suggested that a more indepth study
involving illness episodes obtained through interviewing
methods would be the most efficient and meaningful way of
obtaining the necessary information for analysis.

The finding of a significant correlation between
seven symptoms and duration time is most interesting
considering that the studies reviewed reported no such
relationships. However, of these seven correlations, the
relationships for two symptoms (shortness of breath and
decreased caliber of stools) may be spurious because of a
small sample size. Only six per cent of the sample reported
having a shortness of breath and only eight per cent
reported having decreased caliber of stools. Also, the two

symptoms of nausea and poor appetite (which have similar

174

correlations with duration time) covary and thus effectively
represent one symptom type and not two as previously
defined.

When analyses were performed on the combined symptom
of nausea/poor appetite and the symptom of weight loss (see
pages 164-65) some very illuminating trends were observed.
As previously mentioned, low income Blacks tend to classify
medical conditions into one of two categories: natural and,
unnatural. When weight loss occurs in conjunction with
eating less they tend to view such a situation as a natural
event; however, when weight loss occurs when eating has been
normal, they consider it quite unnatural and possible the
result of root work or evil influences. It was observed
that for the Black patients, the natural condition corre-
lated with a speedy detection whereas the unnatural condi-
tion correlated with a slower detection. Although a similar
trend in mean duration differences could be noted for the
White patients, a comparison of the ranges for duration per
condition and racial group revealed that the White patients
had considerable overlap while the Black patient ranges were
almost mutually exclusive. Even though the differences in
group means for the Black groups were not statistically
significant, a chi square test of independence indicated
that the trend was not random. Considering the literature
on low income Black medical beliefs and these symptoms it
could be suggested that natural conditions are more readily

brought to the attention of orthodox practioners than

175

unnatural conditions because of perceived domains of control
for healers. In other words, alternative treatment may be
sought before going to Western physicians because the more
traditional practioners are thought to be more effective in
cases of unnatural etiologies. While these observed
differences in reporting certainly suffer from problems of
sample size and could result from other non-ethnically
related factors; such a phenomena readily lends itself to
cultural explanation.

Another aspect of symptom reporting behavior which
might result from the influence of various cultural differ-
ences is the apparent grouping of reporting times about
particular time markers (i.e., three, six and twelve
months). It is quite possible that some unconscious
selection of culturally emphasized time intervals may
result from physician filtering mechanisms, patient recall,
or problems occuring in the doctor-patient relationship.
For example, if a patient reported some vague time interval
a physician might record the duration time in a time frame
familiar to the physician. Conversely, in order not to
appear uninformed, unconcerned, or just to have a response
to questions, the patient might conceivably fabricate some
time interval familiar to him if he is not sure of a
definite duration time for his symptoms. In either
situation the wrong duration time is reported and the
chances of obtaining an erroneous result from analyses of

symptom duration are increased. As mentioned earlier, the

176

use of a prospective study might reduce the effects of such
variables by obtaining information directly from the patient
and the physician without relying on chart sources. As the
above discussions would seem to indicate, further research
emphasizing such sub-cultural issues is clearly warranted in
further detail.

The symptoms of diarrhea and constipation could not
be anchored to any specific sub-cultural medical beliefs and
seem to indicate a broader more socially oriented trend for
American veterans in general. The relationship between
these symptoms and a slower detection rate was found to be
significant only for the sample in general and not for
either of the racial groups in particular. People with
diarrhea were only one half as likely to report in before
the three month median time, and those with constipation
were only one third as likely to report in before the median
when compared to those without the respective symptoms. It
is possible that patients with these symptoms (perhaps more
so than for any of the other symptoms) are quite susceptable
to commercially advertised (over-the-counter) treatments
(e.g., Exlax, enemas, Kaopectate). Thus, such a trend could
be a reflection of some broader social influences involving
concepts of health and self-medication which tend to be
supported through the advertising campaigns of various
American pharmaceutical companies. As Kutner et al (1958:
97) have noted, "often the symptoms of cancer are indefinite,

commonly mimicking the symptoms of a variety of other

177
insignificant ailments." The vague etiologies of symptoms
like diarrhea and constipation may indeed readily lend them-
selves to the application of health care practices which
have become firmly entrenched in the general American
society. For example, there are ample data on the over-
dependence on laxatives for Whites in general (National

Analysts 1972) and for Blacks in particular (Snow l978b).

Conclusions

 

Of the four originally proposed hypotheses, only one
was retained with any strong statistical significance. The
main hypothesis concerning ethnic group difference by mean
reported symptom duration time was retained. This may be
the first report of any strong correlation between duration
and ethnicity. It is quite possible that this relationship
was more amenable to analysis here because only one general
cancer type was present in the sample. With more refined
instruments for establishing ethnicity within White patient
populations, further subtle ethnic differences in illness
response may become evident.

The second hypothesis involving significant differ-
ences between group mean duration times for a positive
family medical history was retained with some reservations.
Although the means were shown to be significantly different
in terms of reported duration time, the extremely small
group sizes coupled with the overall ethnic influence toward

shorter duration times made it difficult to put much weight

178

behind the results. The corollary to this hypothesis re-
garding within ethnic group mean differences for positive
versus negative family histories was rejected. Also, since
no relationship was observed in regard to this hypothesis
and the sample in general, the finding of Cobb gt a1 (1954:
17) of longer duration for positive histories was not sub-
stantiated. However, it is quite possible that with a
larger sample size and better ethnic identification that
such differences may become more distinct.

The third hypothesis concerning between ethnic group
mean duration differences for prior experience of nosologi-
cally similar conditions and the corollary to this were
rejected. However, for the sample in general this variable
of personal medical history was indicated as a potentially
useful instrument for analyzing duration time (i.e., the
correlations for occasional and persistent similar problems).
Considering these implications and the observations of
Goldsen (1953) and King and Leach (1950) on illness response,
personal medical history should be further tested under more
controlled conditions before any decision is made about its
potential usefulness.

The fourth hypothesis on symptom content relative to
each ethnic group could not be tested due to inadequate data
stemming from problems in sample size. However, it is
interesting to note that when comparing the racial groups in
terms of a general symptom index (Table XIII) they appeared

to be roughly similar in content and order of occurrence.

179

The purpose of this pilot study was to determine if
there existed any ethnically related differences in medical
care seeking behavior (as measured by reported symptom
duration time). In addition to the ethnic differences,
several potentially fruitful avenues for further research
have been identified, including family medical history,
personal medical history, and such relationships as family
and work related dynamics. The potential usefulness for
more indepth culturally oriented research is further
indicated by the fact that these differences were observed
in sketchy hospital records. If such behaviorally related
indications can be drawn from terse data sources like
medical charts it could be suggested that studies involving
living patients would provide an even better picture of
how culture may influence illness responses in the case of
cancer. Such future research would be designed to study the
ways in which people cope in terms of their feelings with
cancer. The sample criteria for such a study could involve
people who are still active in everyday affairs, whose
illness is not immediately terminal, and whose diagnoses are
limited to only a few types of cancers. The major thrust of
this study would involve the use of various psychological
instruments for measuring such concepts as self-esteem,
social avoidance and distress, locus of control and
definitions of self. In addition to these and other
psychological measures would be included instruments for

measuring such sociodemographic variables as ethnic

180

background, household compositions, marital relationships,
socioeconomic status, and various aspects of past personal
and family medical history. With more refined categories
for ethnic background, and detailed information on medical
histories (including symptom presented at diagnosis from
the patients point of view) coupled with additional socio-
demographic and psychological data, the issues presented
in the delay literature could be more meaningfully
approached.

This pilot study has supported the premise that
health and illness do not result simply from an individual's
state of being. It has been pointed out that health and
illness are intimately connected with the way in which
people construct reality and with the way they interact with
their social environment. Additional studies on living
people should bring these issues into a clearer focus than
was possible in the symptom duration pilot study. Although
this pilot study was somewhat removed from the ground of
social interaction, it has served as a useful data base for
the generation of additional potentially more fruitful

research.

APPENDICES

11.

APPENDIX A.

DATA RECORD SHEET

 

 

 

 

 

 

 

Case #
Primary site ( ) Colon
( ) Rectum
Ethnicity ( ) White
( ) Black
( ) Spanish surname
( ) Other
Patient sex ( ) Male
( ) Female
Occupation
Education
Marital status ( ) Single
( ) Married
( ) Divorced
( ) Widowed
Children
Religion

 

Date of birth

 

Place of birth (parents)

Place of birth (patient)

(month/yr.)

 

 

 

181

182

 

 

12. Date of death (month/yr.)
13. Date of diag. (month/yr.)
14. Date of Rx (month/yr.)

 

15. Stage at Rx

 

16. Stage at diagnosis ( ) Dukes A
( ) Dukes B
( ) Dukes C
( ) Dukes D
17. Differentiation at diagnosis
( ) Well differentiated
( ) Moderately-well differentiated
( ) Moderately differentiated

( ) Poorly differentiated

 

 

 

( ) Undifferentiated
18. Primary tumor type
( ) Adenocarcinoma ( ) Squamous cell Carcinoma
( ) Mucinous ( ) Adenosquamous Carcinoma
Adenocarcinoma
( ) Basaloid Carcinoma ( ) Undifferentiated Carcinoma
( ) Muco-Epidermoid ( ) Unclassified Carcinoma
Carcinoma
( ) Signet-Ring Cell ( ) Other
Carcinoma
19. Diagnostic comments
20. Secondary involvement ( ) At diagnosis

( ) After primary Rx

 

 

183

21. Relative health at diagnosis

( ) Good (no complications)

(
(

(

22. Personal medical history

(

23. Family medical history

24. Symptoms presented
( ) Rectal bleeding
( ) Constipation
( ) Weakness

( ) Abdominal or rectal
pain at defecation

( ) Other

)
)

)

)

)
)

 

Fair (minor complications)

Poor (complications requiring
monitoring)

Other

 

Usual childhood and adult
medical problems

Occasional crises requiring
intervention

Occasional crises anatomically
similar to present cancer

Persistant problems requiring
monitoring

Persistant problems
anatomically related to cancer

Other

 

No occurrence

Occurrence of cancer

 

 

( ) Diarrhea
( ) Weight loss
( ) Palpable abdominal mass

( ) Dull and persistant
abdominal pain

25. Duration of symptoms (in months)

 

184

APPENDIX B.

EXAMPLE OF GOODMAN'S PROCEDURE, USING THE SYMPTOM
OF RECTAL BLEEDING FOR ILLUSTRATION

RECTAL BLEEDING (VAR 001)

 

MAIN EFFECTS 1.
Symptom by
reporting time.

VALUE 2. ETHNIC
Symptom presence
by reporting time.

VALUE 1. ETHNIC
Symptom absence by
reporting time.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- B W + B W - +
I, 10 28 38 q, 12 22 34 B 38 34 72
1‘ 3 24 27 Ar 3 33 36 W 27 36 63
13 52 65 15 55 70 65 70 135
MAIN EFFECTS 2. MAIN EFFECTS 3.
Symptom occurrence Ethnic by reporting
by ETHNICITY time.
- + B W
B 13 15 28 .1 22 50 72
W 52 55 107 4. 6 57 63
65 70 135 28 107 135
Key:
. . X X X
B = Black Patients reporting. 11 21 11+21
W = White Patients reporting. X X x
4 = Before or on the Median time. X12 X22 X12+22
T = After the Median time. 11 21 11+12
- = Absence of symptom. + + +21+22
+ = Presence of symptom. X12 X22

 

 

 

1.

2.

185

APPENDIX B (continued)

 

MAIN EFFECTS FOR RECTAL BLEEDING:
g: X11 - x22 = 38 x 36 = 1 49
X12 . X21 27 X 34 =::::

People are 1% times more likely to report in before the median given
the symptom is absent rather than present. (The inverse g for
presence would be .67 indicating that the presence of the symptom
causes people to delay although it is a small proportion).

§= 1.49 log12 = .399

 

 

 

l 1 1 1 1 1

variance (8) = _ + —— + —— = + + + = .119
X11 X12 X21 38 36 27 36

Standard Dev. = V variance = .345

Standard score (2) = _§_269__ = ;§2§Z%_Q__ = 1.16 Not Sig.at P .05

MAIN EFFECTS FOR SYMPTOM OCCURRENCE:

13 x 55
‘ 52 x 15 4311-
The symptom of rectal bleeding occurs, in the sample, a little more
frequently in Whites than in Blacks but the difference is not
remarkable.

g= .917 log12 = -.O87

. 1 l l l
Variance — 13 + 15 + 52 + 55 — .181

Standard Dev. = .425

-.0 7 - 0
Z score= 8425 = -.205 Not Sig. at P .05

 

 

3. MAIN EFFECTS FOR ETHNICITY:

22 x 57 _
9: 6 x 50 ‘ 54kg

Given that the patient is Black, he is a little over 4 times more likely
to report in before the median than if he was White.
g= 4.18 log12 = 1.43

1 1
22 + 60

Standard Dev. = .500

 

.3_

Variance = 57

1
+ ‘g—'+ — .250

186

APPENDIX B (continued)

1.4 - . . .
Z score = 35000 = 2.86 Significant at P= .002

 

 

4. MAIN EFFECTS FOR ETHNICITY AND PRESENCE OR ABSENCE OF RECTAL BLEEDING
(USED FOR COMPUTING INTERACTION EFFECTS).
a. Absence of rectal bleeding:

10x24
‘ 3x28 ‘2—‘§§

Given the absence of rectal bleeding, Blacks are almost 3 times more
likely to report in before the median than Whites.

6: 2.86 log12 = 1.05

 

 

. _ 1 l l l _
Variance — 10 + 28 + 3 + 24 — .511
Standard Dev. = .715
Z score = 1'797150 = 1.47 Significant at P: .071

b. Presence of rectal bleeding:

12 x 33 _
9’ W’ ——6-00

Given the presence of rectal bleeding, Blacks are 6 times more likely
to report in before than median than Whites.

§= 6.00 log12 = 1.79

 

. l l 1 l
= —- —— + "— __ = o
Variance 2 + 22 3 + 33 492
Standard Dev. = .702
Z score = 1'797620 = 2.55 Significant at P= .005

 

5. INTERACTION EFFECTS BETWEEN ETHNICITY AND THE SYMPTOM OF RECTAL
BLEEDING

9 (SK -) 2.86
= = ——-—— = . 7
9 9 (SK +) 6.00 —4—7—

 

When controlling for the overall ethnicity effect, Blacks tend to report
later than Whites given the symptom of rectal bleeding is present.
(Whites with rectal bleeding are twice (2.10) as likely to report in
before the median than Blacks with rectal bleeding).

6: .477

Variance

Standard

Z score =

187

APPENDIX B (continued)

log12 = -.741
A A
= U (Symptom absence) + 5 (Symptom presence)
1 l l 1 l l l l
’ 10 + 28 + 3 + 24 + 12 + 22 + 3 + 33 ‘ 1'003
Dev. = 1.001

 

-.741 - 0
1.001

 

= -.74 Not Significant at P .05

 

FOOTNOTES

188

FOOTNOTES

 

The use of the word postponed is not intended to imply

an intentional act of delay on the part of the patient,
but merely that some time has elapsed between the onset
of symptoms and the patient's first diagnostic visit.

There appears to be a considerable degree of debate on
whether or not early diagnosis has a definite impact on
patient prognosis. On the one hand, delay in diagnosis
and treatment of malignancy has been indicated as being
significant in reducing a patient's cure potential for
cancer in general (Blackwell l963:3, Lynch and Krush
1968:204, Antonovsky and Hartman 1974:98, American
Cancer Society 1978) and for colorectal cancer in
particular (Scudamore 1969, Potchen 1975:5, Martin et al
1976:430). On the other hand, studies such as Copeland
et a1 (1968) for colorectal cancer and Haagensen (1971)
for breast cancer found no significant changes in
prognoses with delay. To be sure, one of the problems
which confounds the issue lies in the relationship
between the tumor and the host. Depending on the
activity level of the tumor and the resistance of the
host, a number of results are possible. For example,

a person with a slow growing tumor that waited eight
months before seeking intervention may have the same
prognosis as a person (with a fast growing tumor) who
waited two months. Logically if one considers the known
aspects of the natural history of solid tumors (as is
the case with colorectal cancer) early intervention
would seem to provide more benefits than costs for the
patient (Scudamore 1969). In this case a tumor begins
in situ (the most responsive state for treatment) and
in time proceeds to involve more and more of the host
until distant organs such as the lungs, brain, and
spinal cord are also affected. Once the malignancy

has reached the lymphatic or circulatory system (a
prerequisite for metastasis) current treatment regimens
appear to have little effect on survival.

In this case, triviality does not refer to slow growing
tumors, but to the patient's perception of the lesion.

Martin et al (1976:428-431) in discussing the national
percentages for patients treated at certain stages of
cancer involvement reported that 10% were class "A"
patients, while class "B", "C", and "D" patients
represented 30-40%, 25-30%, and 20-25% respectively
(for a complete discussion of this classification
system see pages 89-90) . The five year survival rates
for "A", "B", "C" lesions was reported at 95%, 65%, 30%,
and for class "D" lesion patients it is virtually nill.

189

Since class "A" patients are rarely symptomatic and
class "C" and "D" patients can expect little in the way
of extended life spans from current treatment, and
benefits that may stem from studies such as this pilot
investigation would be small if any at all. However,
considering the large numbers of people expected to
develop this cancer, class "B" patients, with their

65% five year survival rates, represent a segment of
the population which could indeed benefit from such
studies.

It has been reported (Martin et a1 1976:430) that among
class "D" patients who were originally diagnosed as
class "A", "B", or "C", the survival from time of
metastatic development was longest among original class
"A" patients, intermediate for class "B" and shortest
for class "C" patients. This would seem to imply that
persons diagnosed at less invasive stages tend to have

a better prognosis than those diagnosed at more invasive
stages.

Unfortunately, due to the prevailing opinions of several
hospital administrators, this sampling was not possible.
Whether this refusal to grant access reflects a general
tendency to comply with the privacy act of 1974 or is in
response to such issues as malpractice suits is
uncertain. However, the fact remains that regardless

of the tack taken by the researchers, any request for
access to complete medical records was uncategorically
denied. Fortunately, several administrators in charge
of the V.A. hospitals were more open to the possible
benefits that could be obtained through unconventional
(non—biomedically oriented) research.

According to one of the resident pathologists at Allen
Park in Detroit (Wheatherbee 1978) using diagnostic
information back to 1964 would not impose any problems
in regard to differing techniques and reliability of
diagnosis.

By private physician, I do not mean to imply that the
patient was seeing the doctor on a regular basis, for
this cannot be discerned from the available information.
This useage simply means that the initial diagnosis
normally came from a physician working outside of the
V.A. hospital.

If one used t-tests to investigate the differences of
the five means there would be ten t-values to compute.
This sort of multiple testing would in effect inflate
the alpha error and increase one's chances of rejecting
the hypothesis of equal means when in fact it should
have been retained. If the hypothesis that all five

10.

11.

12.

190

means are equal is rejected when some one of the ten
t-values exceed the selected critical value of 5 per
cent, this hypothesis will be rejected with a chance
which may be many times the level of significance .05.

A certain degree of caution whould be exercised when
considering sucn a result because of possible sampling
error. The chances of such an error in this study have
been inflated because sampling was not random (i.e.,
all Blacks that fit the sample criteria were selected
for analysis). Such an error could present a picture
of Black medical responses which in reality do not
represent responses common to the Black population at
large.

Rivlin in reviewing current health care utilization data
from vital and health statistical surveys observed that
"non-whites were significantly less likely to see a
physician and slightly less likely to enter a hospital
than the white majority" (1977:11). Evidence in support
of this observation in the case of low income urban
Blacks and Whites has also been presented by Hulka,
Kupper and Cassel (1972:307). Also, a review by Haynes
(1975), that makes a finer distinction for the non-white
class, has provided additional support for such observa-
tions on the "health gap" between American White and
Black patients. Haynes noted that "in 1970, 50 per cent
of Blacks and 70 per cent of Whites reported seeing a
physician during the year" (1975:19), and that although
there has been an increase in care utilization for both
groups since the 1963-1964 national survey, the gap has
remained the same. In regard to hospital utilization

he noted that a larger proportion of Whites had reported
care episodes in short stay hospitals (10.2% for Whites
and 8.2% for Blacks) and that these differences were

the greatest between White and Black males and in
persons 45 years and older (1975:24).

While there appears to be a fairly strong case for con-
sidering the medical beliefs of low income Blacks as an
essentially homogeneous system, it seems to be less so
in the case of White low income patients (at least in
this study). Trends noted in the data indicated that if
the sample size were larger for the Southern and
European Whites, their differences would have become
significant in relation to the Northern White group.

The implication here is that the White group would be
less homogeneous in regard to medical response behavior
than the Black group. Such a high level of variation

in response for the White group would seem to lessen any
difference between Blacks and Whites thus increasing the
importance of the observed difference, especially since
it was in the opposite direction expected.

LI ST OF REFERENCES

L I ST OF REFERENCES

American Cancer Society
1978 Cancer statistics 1978. New York, American
Cancer Society Publications.

Adams, R. N. and A. J. Rubel
1967 Sickness and social relations. Handbook of
Middle American Indians 6: 333-56. Austin,
University of Texas Press.

Aitken-Swan, J. and R. Paterson
1955 The cancer patient: delay in seeking advice.
British Medical Journal 1: 623-27.

Alland, A.
1970 The cancer problem from the standpoint of the
practicing physician in the small community.
Minnesota Medicine 32: 65-9.

Antonovsky, A.
1972 Comparison of examinees in breast cancer
detection clinics in Haifa and Jerusalem.
Harefuah 83: 598-603. (Hebrew)

Antonovsky, A. and H. Hartman
1974 Delay in the detection of cancer: a review of
the literature. Health Ed. Mong. 2(2): 98-128.

Atkinson, D.
1962 Magic, myth and medicine. London, Routledge
and Kegan.

Bar—Meir, A. and A. Davies

1960 Delay in the detection of cancer of the
digestive tract. Harefuah 59: 317-23.
(Hebrew)

Bates, F. and I. Ariel
1948 Delay in treatment of cancer. Illinois Medical
Journal (cited in Blackwell 1963. The
literature of delay in seeking medical care for
chronic illness. Health Ed. Mong. 16: 3-31).

191

192

Bateson, G.
1965 Naven: a survey of the problems suggested by a
composite picture of the culture of a New
Guinea tribe drawn from three points of view.
Stanford, Stanford University Press.

Benedict, R.
1959 Patterns of culture. Boston, Houghton-Mifflin.

Blackwell, B.
1963 The literature of delay in seeking medical care
for chronic illness. Health Ed. Mong. 16: 3-31.

1964 Anticipated premedical care activities of upper
middle class adults and their implications for
health education practice. Health Ed. Mong.
17(17): 36.

Bohannon, P.
1973 Rethinking culture: a project for current
anthropologists. CA 14(4): 357-65.

Braund, R. and G. Binkley
1942 A plea for the earlier diagnosis of rectal
cancer. New York State Journal Med. 42: 33-7.

Breslow, L. and J. Hochstim
1964 Sociocultural aspects of cervical cytology in
Alameda County, California. Public Health
Report 79: 107-112.

Burdick, D. and F. Chantry
1954 Central New York Surgical Society Survey on
breast carcinoma, 1920-1952. Cancer 7: 47-53.

Cameron, A. and J. Hinton
1968 Delay in seeking treatment for mammary tumors.
Cancer 21: 1121-1126.

Cameron, V. K.
1930 Folk beliefs pertaining to health of the
southern Negro. Unpublished Masters thesis.
Department of Sociology, Northwestern
University.

Cobb, B., et a1.
1954 Patient-responsible delay of treatment in
cancer. Cancer 7: 920-26.

Coburn, B. and C. Pope
1974 Socioeconomic status and preventative health
behavior. J. Health Soc. Behavior 15: 67-78.

Cockerham, W.

1978

Comstock, G.
1972

193

Medical sociology. Englewood Cliffs, NJ,
Prentice-Hall.

and K. Partridge
Church attendance and health. J. Chronic
Disease 25: 665-72.

Cooper, E. and P. Smith

1975

Copeland, M.
1976

1977
Copeland, M.
1968

Currier, R.
1966

DeLawter, D.
1948

Dubos, R.
1973

Eardley, A.
1974

Bladder cancer: prognostic factors and
assessment of response to treatment.
Cancer therapy: prognostic factors and
criteria of response. M. Staquet, ed.
New York, Raven Press.

Efforts in cancer diagnosis: national task
forces--Large bowel cancer. J. Surg.
Oncology 8: 555-63.

Personal communication.

et a1.
Prognostic factors in carcinoma of the colon
and rectum. Am. J. Surg. 116: 875-81.

The hot-cold syndrome and symbolic balance in
Mexican and Spanish American folk medicine.
Ethnology 5: 251-63.

Culpability for delay in management of cancer.
In Antonovsky and Hartman, 1974. Health Ed.
Mong. 2(2): 98-128.

Man adapting. New Haven, Yale University Press.

Triggers to action: a study of what makes women
seek advice for breast conditions. Intl. J.
Health Ed. 17: 256-65.

Ehrenreich, B. and J. Ehrenreich

1971

Eisenberg, L.

1977

The American health empire: power, profits and
politics. A report from the health policy
advisory center (Health-PAC). New York,
Vintage Books.

Disease and illness. Culture, Med. Psych.
1(1): 9-23.

Engel, G.
1977

Enelow, A.

1976

Fabrega, H.
1971

1974

1975

1977a

1977b

Fabrega, H.
1972

Fabrega, H.
1977

'Farr, C.
1919

Fink, R., et
1968

194

The need for a new medical model: a challenge
for-biomedicine. Science 196(4286): 129-36.

Group influences on health behavior: a social
learning perspective. Cancer: the behavioral
dimensions. J. Cullen, et al., eds, New
York, Raven Press.

Medical anthropology. Biennial review of
anthropology. B. Siegal, ed. Stanford,
Stanford University Press.

Disease and social behavior: an interdisciplin-
ary perspective. Cambridge, M.I.T. Press.

The need for an ethnomedical science. 189:
969-975.

Group differences in the structure of illness.
Culture, iedicine and Psychiatry 1: 379-94.

The scope of ethnomedical science. Culture,
Medicine and Psychiatry 1(2): 201-8.

and P. Manning

Self and body. Paper presented at the Third
Annual Conference in Social Science and
Medicine, Elsinore, Denmark, August, 1972.

and M. Zucker

Comparisons of illness episodes in the
pluralistic setting. J. Psychosom. Med.
39(5): 325-43.

Delay in the treatment of cancer. American
Journal Med. Sci. 157(Jan.): 34-5.

al.
The reluctant participant in a breast cancer
screening program. Public Health Report

83: 479-90.

Firth, R.
1957

Fisher, S.
1967

Health planning and community organization.
Health Education Journal 15: 118-25.

Motivation for patient delay. Arch. Gen.
Psychiat. 16: 676-78.

Foster, G.

1961

Freidson,
1970

Gans, H.
1962

E.

Geertz, C.

1957

1968

1973

Gerard, A
1975

Gibson, C.

1915

Glick, L.
1967

Goldsen,
1953

1963

Goldsen,
1957

R.

R.,

195

Public health and behavioral science: the
problems of teamwork. American Journal of
Public Health 51(9): 1286-93.

Profession of medicine: a study of the sociology
of applied knowledge. New York, Harper and Row.

The urban villagers. New York, The Free Press.

Ritual and social change: a Javanese example.
American Anthropologist 59: 32-54.

The impact of the concept of culture on the
concept of man. Man in adaptation: the
cultural present. Y. Cohen, ed. Chicago,
Aldine.

Religion as a cultural system. The interpreta-
tion of cultures: selected essays. C. Geertz,
ed. New York, Basic Books.

Carcinoma of the colon and rectum: prognostic
factors and criteria of response. Cancer
therapy: prognostic factors and criteria of
response. M Staquet, ed. New York, Raven
Books.

Cancer surgery. Am. J. Surg. 29: 374—77.

Medicine as a enthnographic category: the Gimi
of the New Guinea highlands. Ethnology 6:
31-56.

Factors related to delay in seeking diagnosis
for the danger signals of cancer. In Kutner
et al., eds., 1958. Ithaca, New York, Social
Science Research Center, Cornell University.

Patient delay in seeking cancer diagnosis:
behavioral aspects- J. Chronic Disease
16: 427-36.

P. Gerhardt and V. Handy .
Some factors related to patient delay in seeking
diagnosis for cancer symptoms. Cancer 10(1):l-7

196

Goodman, L. A.
1965 On the multivariate analysis of three dichoto-
mous variables. Am. J. Sociology 71: 290-301.

1972 A general model for the analysis of surveys.
Am. J. Sociology 77(6): 1135-86.

Gray, D. and G. Ward

1952 Delay in diagnosis of carcinoma of the stomach.
Am. J. Surg. 83: 524-26.

Grushka, T. and R. Steintz
1955 Delays in diagnosis and treatment of cancer of
the female breast. In Antonovsky and Hartman,
1974. Harefuah 49: 6-8.

Guiss, L.
1955 Mass roentgenographic screening as a lung-
cancer control measure. Cancer 8: 219-36.

Hackett, T., N. Cassem and J. Raker
1973 Patient delay in cancer. New Eng. J. Med.
289(1): 14-20.

Haggerty, R.
1965 Family diagnosis: research methods and their
reliability for studies of the medical social
unit, the family. Am. J. Public Health 55:
1521-33.

Hallover, D.
1972 Illness behavior--an experimental investigation.
J. Chron. Dis. 25: 599-610.

Hallowell, I.
1953 Culture personality and society. Anthropology
today, A. Kroeber, ed. Chicago, University of
Chicago Press.

1963 Ojibwa world view and diease. Man's image of
medicine and anthropology. I. Galdston, ed.
New York, International Universities Press.

Hammerschlag, C.
1964 Breast symptoms and patient delay: psychological
variables involved. Cancer 17: 1480-85.

Harms, C., J. Plant, and A. Oughterson
1943 Delay in the treatment of cancer. JAMA 121(5):
335-38.

197

Harwood, A.
1971 The hot-cold theory of disease: implications
treatment of Puerto Rican patients. JAMA
216(7) 1153-58.

Haynes, M. A.

1975 The gap in health states between black and
white Americans. Textbook of black related
diseases. A. Williams, ed. New York, McGraw
Hill Book Co.

Henderson, J.

1966 Denial and regression as factors in the delay
of patients with cancer presenting themselves
to the physician. Annals New York Acad.
Science 125: 856-64.

Henderson, J., et a1.
1958 A psychiatric investigation of the delay factor
in patient doctor presentation in cancer.
J. Psychosom. Res. 3:27-41.

Hetherington, R. and C. Hopkins
1969 Symptom sensitivity: its social and cultural
correlates. Health Serv. Res. 4: 63-75.

Hillard, J. and W. Rockwell
1978 Dysesthesia, witchcraft and conversion
reaction: a case successfully treated with
psychotherapy. JAMA 241: 1742-44.

Hochstrausser, D. and J. Tapp
1970 Social anthropology and public health.
Anthropology and the behavioral and health
sciences. 0. von Merring and L. Kasden, eds.
Pittsburgh, University of Pittsburgh Press.

Hollingshead, A. and F. Redlich
1958 Social class and mental illness. New York,
John Wiley and Sons.

Howson, J.
1948 Pelvic cancer delay. Am. J. Obstet. Gynecol.
55: 538-40.

Hulka, B., L. Kupper and J. Cassel
1972 Determinants of physician utilization:
approach to a service-oriented classification
of symptoms. Medical Care 10(4): 300-9.

Hughes, B.
1976 Long-term study of large-bowel cancer. Med. J.
Australia 2:365-68.

Imperato, P.
1969

Ingham, J.

1970

Kegeles, S.

1969

1973

1976

Kegeles, S.,
1965

198

Traditional attitudes towards measles in the
republic of Mali. Trans. Royal Society
Tropical Med. Hygiene 63: 768-80.

On Mexican folk medicine. American
Anthropologist 72: 76-87.

A field experimental attempt to change beliefs
and behavior of women in an urban ghetto.
J. Health Social Behavior 10: 115-25.

Behavioral sciences data and approaches
relevant to the development of education
programs in cancer. Health Ed. Mong. 36: 18-33.

Relationship of sociocultural factors to cancer.
Cancer: the behavioral dimensions. J. Cullen,
B. Fox, and R. Isom, eds. New York, Raven
Press.

et a1.

Survey of beliefs about cancer detection and
taking Papanicolaon tests. Public Health
Report 80: 815-23.

Kelly, J. and C. Thieme

1967

Kiev, A., ed.
1964

King, R. and
1950

Kleinman, A.
1974

Impact of public education about cancer on
detection and treatment of cancer of the
breast by surgery. Cancer 20: 260-62.

Magic, faith and healing. New York, Glencoe.

J. Leach
Factors contributing to delay by patients
seeking medical care. Cancer 3: 571-79.

Cognitive structures of traditional medical
systems: ordering, explaining and interpreting
the human experience of illness. Ethnomedizin
III, 8 (1974/1975): 27-49.

199

1975 Social, cultural, and historical themes in the
study of medicine in Chinese societies:
problems and prospects for the comparative
study of medicine and psychiatry. Medicine in
Chinese cultures: comparative studies of health
care in Chinese and other societies. U.S.
DHEW, DHEW publication no. (NIH) 75-653.
Washington, D.C., U.S. Government Printing
Office.

1977 Editorial: Culture and illness: a question of
models. Culture Med. Psychiatry 1(3): 229-31.

Koos, E.
1954 The health of regionville. New York, Columbia
University Press.

Kroeber, A.
1957 Style and civilizations. Ithaca, Cornell
University Press.

Kroeber, A. and C. Kluckhohn
1952 Culture: a critical review of concepts and
definitions. Vintage Books.

Kutner, B. and G. Gordon
1961 Seeking care for cancer. J. Health Hum. Behav.
2: 128-45.

Kutner, B., H. Makover, and A. Oppenheim
1958 Delay in diagnosis and the treatment of cancer:
a critical analysis of the literature. J.
Chron. Dis. 7: 95-120.

Leach, J. and G. Robbins
1947 Delay in the diagnosis of cancer. JAMA 135:
5-8.

Levin, A.
1947 Certain factors influencing cancer curability.
Southern Med. J. 40: 259-61.

Lieban, R.
1962 The dangerous Ingkantos: illness and social
control in a Philippine community. American
Anthropologist 64: 306-12.

1973 Medical anthropology. Handbook of social and
cultural anthropology. J. Honigmann, ed.
Chicago, Rand McNally and Co.

200

Litman, T.
1974 The family as a basic unit in health and
medical care: a social-behavioral overview.
Soc. Sci. Med. 8: 495-519.

Logan, M.

1978 Humoral medicine in Guatemala and peasant
acceptance of modern medicine. Health and the
human condition: perspectives on medical
anthropology. M. Logan and C. Hunt, eds.
North Scituate, Mass., Duxbary Press.

Lynch, H. and A. Krush
1968 Attitudes and delay in cancer detection. Arch.
Environ. Health 17: 204-09.

Mabry, J.
1964 Medicine and the family. J. Marriage and the
Family 26: 160-65.

MacDonald, E. J.
1947 Changing reasons for cancer patient delay.
Connecticut Health Bull. 61: 99

MacDonald, I.
1951 Biological predeterminism in human cancer.
Surg. Gynecol. Obstet. 92: 443-52.

Marascuilo, L. and M. McSweeney

1977 Nonparametric and distribution-free methods for
the social sciences. Monetery, Brooks and
Cole.

Martin, R. G. et a1.
1976 Gastrointestinal cancer. Cancer patient care
at M. D. Anderson Hospital and Tumor Institute.
R. Clark and C. Howe, eds. Chicago, Yearbook
Medical Pub. Inc.

McCullough, J. and V. Gilbertsen
1969 Motivational factors in persons seeking early
diagnosis of cancer: a preliminary report.
Geriatrics 24: 117-25.

McKinley, J.
1972 Some approaches and problems in the study of
the use of services--an overview. J. Health
Soc. Behav. 13(2): 115-52.

201

Mechanic, D.

1974 Social psychologic factors affecting the pre-
sentation of bodily complaints. Health and
the social environment, P. Insel and R. Moos,
eds. Lexington, Mass., D.C. Heath.

Messing, S.
1976 Emics and etics of health problems in Ethiopia.
Medical anthropology, F. Grallig and H. Haley,
eds. Paris, Mouton.

Miner, H.
1960 Culture change under pressure: a Hausa case.
Human Organization 19(3): 164-67.

Monk, M., A. Lilienfeld, and A. Mendeloff
1962 Preliminary report of a epidemiologic study of
cancers of the colon and rectum. Paper
presented at meeting of epidemiology section,
American Public Health Association (cited in
Comstock and Partridge 1972, Church attendance
and health. J. Chron. Dis. 25: 665-72).

Moses, R. and N. Cividali
1966 Differential levels of awareness of illness:
their relation to some salient features in
cancer patients. Annals N. Y. Acad. Sci.
125: 984-94.

Mossetli, C., et. a1.

1970 Sociocultural aspects and motivational reactions
of populations involved in mass educational
programs for the early diagnosis of tumors of
the female genital organs. Minerva Med.

61: 4973-90. (Italian)

Muldoon, J. P.

1977 The role of surgery in colon and rectal cancer.
Paper presented at a symposium on colon and
rectal cancer, St. Mary's Hospital, November
1977, Grand Rapids, Michigan.

Nash, J.
1967 The logic of behavior: curing in a Mayan Indian
town. Human Organization 26: 132-40.

National Analysts, Inc.
1972 A study of health practices and opinions (final
report). National Technical Information
Service, U.S. Dept. Commerce., Springfield, Va.

202

O'Frake, C.
1961 The diagnosis of disease among the Subanun of
Mindanao. American Anthropologist 63: 113-32.
1977 The diagnosis of disease among the Subanun of
Mindano. Culture, disease and healing: studies
in medical anthropology. D. Landy, ed.
New York, Maxmillan.
Opler, M.
1945 Themes as dynamic forces in culture. Am. J.
Soc. 51: 198-206.
1961 Ethnic differences in behavior and health

practices. The family: a focal point for
health education. I Galdson, ed. New York,
New York Acad. Med.

Pack, G., and J. Gallo
1938 The culpability for delay in the treatment of
cancer. Am. J. Cancer 33(July): 443-62.

Paterson, R.
1955 Why do cancer patients delay? Canadian Med.
Assoc. J. 73(12): 931-39.

Paul, B., ed.
1955 Health, culture and community. New York,
Russell Sage Foundation.

Paul, B.
1963 Anthropological perspectives on medicine and
public health. Annals Am. Acd. Polt. Soc.
Sci. 346: 34-43.

Pellegrino, E. D.
1963 Medicine, history and the idea of man.
Medicine and society. J. Clausen and R.
Strauss, eds. The Annals of the American
Academy of Political and Social Science 346:
9-20.

Pless, I. and B. Satterwhite
1973 A measure of family functioning and its
application. Soc. Sci. Med. 7: 613-21.

Potchen, E. J.
1975 A community based study of colorectal cancer
control. Project sponsored by M.S.U. Dept.
of Radiology (in progress).

Preston, R.
1966

203

Edward Sapier's anthropology: style, structure
and method. American Anthropologist 68: 1105-
28. '

Price-Wiliams, D.

1962

Reznekoff, M.
1955

Rivlin, A. M.

1977

Robbins, G.,

1950

1953

1959

Rogers, A.
1974

Rubel, A.
1960

1964

Sandifer, W.
1958

Scheffey, L.
1953

A case-study of ideas concerning disease among
the Tiv. Africa 32: 123-31.

Motivational factors in persons attending a
cancer detection center. Cancer 8: 454-58.

Health differentials between white and non-
white Americans: a background paper. Congress
of the United States. Washington, DC, U.S.
Government Printing Office.

et a1.
Delay in diagnosis and treatment of cancer.
JAMA 143: 346-48.

Delay in the diagnosis and treatment of
physicians with cancer. Cancer 6:624-26.

The significance of early treatment of breast
cancer: changes correlated with the cancer
education programs of 1940-55. Cancer 12:
688-92.

Cancer of the colon: magnitude of the problem
and clinical aspects. Selected topics of
cancer: current concepts. K. Charyulu and

A. Sudarsanam, eds. New York, Intercontinental
Medical Book Corp.

Concepts of disease in Mexican-American
culture. American Anthropologist 62: 795-814.

The epidemiology of a folk illness: susto in
Hispanic America. Ethnology 3: 268-83.

and B. Pritchett

Psychologic reactions causing delay in treatment
of cancer of the cervix. J. Obstet. Gynecol.
Br. Commonwealth 11:82-88.

The delay period in the diagnosis of pelvic
malignancy. Obstet. Gyn. 1: 554-63.

204

Scotch, N.
1963 Medical anthropology. Biennial review of
anthropology. B. Siegel, ed. Stanford,
Stanford University Press.

Scudamore, B.
1969 Cancer of the colon and rectum: general
aspects, diagnosis, treatment and prognosis;
a review. Dis. Colon Rect. 12: 105-14.

Seidman, H., E. Silverberg, and A. Holleb
1976 Cancer statistics, 1976 a comparison of White
and Black populations. American Cancer
Society, Professional Education Publications.

Seijas, H.
1973 An approach to the study of medical aspects of
culture. Current Anthropology 14(5): 544-45.

Shapiro, S., et a1.

1967 Periodic breast cancer screening in reducing
mortality from breast cancer. JAMA 215: 1777-
85.

Shedden, W.
1939 Carcinoma of the rectum and sigmoid. Am. J.
Surg. 44: 482-511.

Shuval, J.
1970 Social functions of medical practice. San
Francisco, Jossey-Bass Inc.

Silverberg, E.
1977 Cancer statistics, 1977. Cancer 27(1): 26-41.

Simmons, C. and E. Daland
1920 Cancer factors entering into delay in its
surgical treatment. Boston Med. Surg. J.
183(September): 298-306.

1924 Cancer: delay in surgical treatment. Boston
Med. Surg. J. 190(January): 15-19.

Simmons, 0.
1953 Popular and modern medicine in mestizo
communities of coastal Peru and Chile.
J. Am. Folklore 66: 57-71.

Smith, F.
1935 Factors causing delay in operative therapy of
carcinoma. Surg. Gynecol. Obst. 60: 45-53.

205

Snow, L. F.

1974 Folk medical beliefs and their implications for
care of patients: a review based on studies
among black Americans. Annals of Internal
Medicine 81(1): 92-6.

1977 Popular medicine in a black neighborhood.
Ethnic medicine in the Southwest. E. Spicer,
ed. Tucson, University of Arizona Press.

l978a Sorcerers, saints and Charlatans: black folk
healers in urban America. Culture, Medicine
and Psychiatry 2: 69-106.

l978b Personal communication.

Soost, H. and H. Thomas
1969 Delayed treatment of carcinoma of uterine body
in cervix: incidence and causes. Dtsch. Med.
Wes. 94: 633-38.

Staiano, K. V.
1974 The integration of modern medicine into a
system of folk beliefs: study of patients in
a clinic serving a low-income community.
(unpublished manuscript)

Stearns, H.

1950 Discussion of A. Diddle, genital cancer among
women. Am. J. Obstet. and Gynecol. 59: 1381-
82.

Sugar, M. and C. Watkins
1961 Some observations about patients with a breast
mass. Cancer 14: 979-88.

Titchener, J., et a1.
1956 Problem of delay in seeking medical care.
JAMA 160: 1187-93.

Turner, V.
1968 The drums of affliction. London, Oxford
University Press.

Wainwright, J. M.

1911 The reduction of cancer mortality. New York
Journal of Medicine 95(December 9, 1911):
1165-68.

Waitzkin, H. and J. Stolckle
1972 The communication of information about illness.
Adv. Psychosom. Med. 8: 180-215.

206

Wakefield, J.
1969 Social and education factors affecting the
early diagnosis of cancer. In Antonovsky and
Hartman, 1974. Schweizerische Medizinische
Wochenschrift 99: 828-33.

Wax, M.

1970 Sociology. Anthropology and the behavioral and
health sciences. 0. von Mering and L. Kasden,
eds. Pittsburgh, University of Pittsburgh
Press.

Wheatherbee, R.
1978 Personal communication.

Whitten, N.
1962 Contemporary patterns of malign occultism among
Negroes in North Carolina. J. Am. Folklore
75: 311-25.

Williams, R. A., ed.
1975 Textbook of black related diseases. New York,
McGraw Hill Book Co.

Wintrob, R.
1973 The influence of others: witchcraft and root-
work as explanations of behavior disturbances.
J. of Nervous and Mental Disease 150(5): 318-
26.

Wintrob, R., R. Fox and E. O'Brien
1971 Rootwork beliefs and psychiatric disorder among
blacks in a northern United States city. Paper
presented at the Vth World Congress of
Psychiatry, Mexico City.

Wolff, B. and S. Langley
1968 Cultural factors and the response to pain: a
review. American Anthropologist 70: 494-501.

Worden, J. and A. Weisman
1975 Psycho-social components of lagtime in cancer
diagnosis. J. Psychosomatic Res. 19: 69-79.

Vargas, L.
1976 Reflections. Medical anthropology. F. Gralliz
and H. Haley, eds. Paris, Mouton.

Young, A.
1975 Why Amhara get Kuregnga: sickness and posses-
sion in a Ethiopian Zar cult. American
Ethnologist 2: 567-84.

1976

Youngman,
1947

Zborowski,
1952
1969

Zola, I.
1966

N.

M.

207

Some implications of medical beliefs and
practices for social anthropology. American
Anthropologist 78(1): 5-24.

Psychological aspects of the early diagnosis
of cancer. Med. J. Australia 34: 581-87.

Cultural components in response to pain. J.
Soc. Issues 8:16-30.

People in pain. San Francisco, Jossey-Bass.
Culture and symptoms: an analysis of patients

presenting complaints. Am. Soc. Rev. 31:
615-30.