”MW/ll

W

J
I

133
809

THS

 

7.51;st

“HIllllllilllllllll‘lllllllllHilllllllllllllllllllllllllllll

3 1293 01771

 

ILHBFUAFU!

i Michigan State
. University

 

 

 

This is to certify that the
thesis entitled
A Case-Control Study of Colon Cancer
and Occupation Among African American

Men and Women - A Population-Based Examination

presented by

Erik Helms

has been accepted towards fulfillment
of the requirements for

Master's aggreeinEpidemiology

6%”

M! or professor

Daeuygij9¥

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

:19; 35- 2’ 9‘ Eol

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ma ciClRCanflS-p.“

A CASE-CONTROL STUDY OF COLON CANCER AND OCCUPATION
AMONG AFRICAN AMERICAN MEN AND WOMEN - A
POPULATION-BASED EXAMINATION

BY
Erik D. Helms

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

MASTER OF SCIENCE

Department of Epidemiology

1998

ABSTRACT
A CASE-CONTROL STUDY OF COLON CANCER AND OCCUPATION

AMONG AFRICAN AMERICAN MEN AND WOMEN - A
POPULATION-BASED EXAMINATION

BY
Erik D. Helms

Information obtained on Blacks and Whites from the
Occupational Cancer Incidence Surveillance Study (OCISS), a
case-referent study investigating workplace risk factors,
was analyzed to identify explanatory factors for racial
differences in colon cancer risk. A case-control analysis
was utilized to test the hypothesis of an increased risk of
colon cancer among Blacks. All analyses were adjusted for
the potential confounding effects of age, gender, occupation
and tobacco use history. No conclusion could be drawn from
the data presented that the risk of colon cancer is
consistently greater among a majority of occupational
groupings for Blacks than for Whites. Sociodemographic or
smoking-related characteristics do not appear to be
associated with an increased risk of colon cancer, which was
consistent among men and women across races. However, the
findings of an age-related increase in risk of colon cancer
among both races are consistent with SEER incidence data.
Although these data cannot address all underlying reasons
for observed differences in occupational risks among Blacks
and Whites, it underscores the need for further study of the

relationship between colon cancer, occupation, and race.

ACKNOWLEDGMENTS

This study was supported by NIOSH grant #: ROl-OH02067
and SEER grant #: NOl-CNOSZZS. I am most grateful to Dr. G.
Marie Swanson for her continued support and guidance
throughout my Master's program. I would also like to thank
my wife, Michelle, for her patience and encouragement, and

my parents and sister for their dedication and support.

iii

LIST OF TABLES
INTRODUCTION
BACKGROUND .
STAGE/SITE .
DIET . . . .
SMOKING . . .
OCCUPATION .
MATERIALS AND
RESULTS . . .

DISCUSSION .

LIST OF REFERENCES

METHODS

TABLE

OF CONTENTS

iv

LIST OF TABLES

Page

Table 1: Characteristics of Colon Cancer Cases
and Population Controls . . . . . . . . . . . . 23

Table 2: Smoking History, Age, and Work Status of
Colon Cancer Cases and Population Controls
Among Blacks and Whites . . . . . . . . . . . . 26

Table 3: Colon Cancer Risk Among White and Black
Subjects for Selected Usual Industries and
Occupations . . . . . . . . . . . . . . . . . . 27

Table 4: Colon Cancer Risk Among Male and Female
Subjects for Selected Usual Industries and
Occupations . . . . . . . . . . . . . . . . . . 28

Table 5: Logistic Regression Results: Risk of Colon
Cancer Among Study Subjects by Age, Race,
Gender, Tobacco Use History, and Usual
Occupation/Industry . . . . . . . . . . . . . . 31

INTRODUCTION

Cancer of the large bowel is second only to lung cancer
as a cause of death among men and women due to malignancies
in the United States (Potter et al., 1993), and is the
fourth most common incident cancer (after breast, lung, and
prostate). It has been estimated by the American Cancer
Society that there were approximately 94,500 incident cases
of colon cancer diagnosed in 1997, and an estimated 46,400
deaths in 1997. The incidence rates show a wide divergence
by racial/ethnic group, while within each racial/ethnic
group, incidence rates for cancer of the colon among women
are slightly lower than those among men (Ries et al., 1994).
While incidence rates have seen a general upward trend since
1973, the age-adjusted mortality rates for all races
combined have actually fallen 31% for women and 9% for men
over the past 30 years (American Cancer Society, 1996).

From 1950 to 1991, the five-year relative survival rates
have increased from 41% to 60.9%. The overall median age of
diagnosis for colon cancer is 72, with the median age of
death being 74 years of age (Ries et al., 1994).

Colon cancer is essentially the only cancer that occurs
with approximately equal frequency in men and women. The
lifetime risk of being diagnosed with colon cancer is
approximately 6.14% for males and 5.92% for females.
Lifetime risk of dying of cancer of the colon is 2.60% for
males and 2.65% for females (Ries et al., 1994). Although
cancers of the colon are currently the second and third most

1

frequently diagnosed malignancy among men and women
respectively, and have increased in incidence during the
past two decades, the all-race and all-age incidence had
decreased slightly (by 1.3%) between 1988 and 1992 (Ries et
al., 1994). However, when this decrease was stratified by
race, there was actually an increased incidence among Blacks
(.9%).

Although incidence rates of colon cancer are not as
divergent for Blacks and Whites as are other common
malignancies (lung and prostate), the difference is still
important to examine due to the higher levels of mortality
experienced among the Black population from colon cancer.
From the period between 1988 through 1992, the age-adjusted
incidence rate of colon cancer among Blacks was 40.4 per
100,000, and 35.5 per 100,000 among Non-Hispanic Whites
(Miller et al., 1996). Data indicate that greater numbers
of Black males and females are diagnosed with colon cancer
at an earlier age than their age-adjusted White
counterparts. For Black males, rates per 100,000 begin to
differ drastically from Whites in the 35-39 year age group
(18.1 vs. 11.2 per 100,000) (Miller et al., 1996), and
continue to be diagnosed at higher rates until age 74. The
numbers and the extent of disparity between Blacks and
Whites are similar for women. In addition, mortality rates
from cancer of the colon and rectum combined were generally
higher for Blacks (23.4 per 100,000) compared to Non-

Hispanic Whites (18.4 per 100,000) (Miller et al., 1996).

Cancer of the colon is a highly treatable and often
curable disease when localized to the bowel. It is also the
second most common cause of cancer death” inore than 90% of
cancers of the colon are adenocarcinomas. Very little is
known about the etiology of colonic lymphomas and carcinoids
which compromise almost all of the remaining histologic
types. Little is also known about the relationship between
specific histologic subtypes of adenocarcinoma to the
environmental risk factors, physiologic processes, and
molecular changes? (Potter et al., 1993). Therefore,
because of the frequency of this disease, lack of knowledge
regarding its specific etiology, and its substantial public
health impact in terms of both mortality and morbidity
(recurrence is commonly the ultimate cause of death and is
associated with the greatest cost), the identification and
follow-up of high-risk groups for early stage diagnosis
should be a priority for certain subsets of the population.

Indeed, the impact of stage severity and comorbidity
increases the costs of colon cancer, both in terms of
resource use and patient suffering. Taplin et al. (1995)
found that the age-adjusted costs of initial care for a
patient diagnosed with carcinoma in situ of the colon was
$13,848 for local, $15,398 for regional, and $17,223 for
distant sites. Additionally, stage and comorbidity were
found to be more predictive of cost than age, indicating the
need for increased early detection among at-risk
populations. Based upon the statistics presented by SEER
since 1971, it appears that Blacks fall into this high-risk

3

category. The improvements in survival rates (49% in 1973
to 62% in 1991) and decreased mortality that have been seen
in the White population have fallen short of those levels in
the Black population (52% in 1991). The all race and age
five year relative survival rate for the total population
from 1986-1991 was 61%. The rate for Whites over the same
period of time was 61.8%, while for Blacks, five-year
relative survival rates were 52.4%. Blacks also have a
greater likelihood of dying at a younger age than Whites
(115/100,000 vs. 81/100,000) between 55 and 64 years of age
(Ries et al., 1994).

Because of the documented differences in colon cancer
incidence and mortality between Blacks and Whites, as well
as the overall impact of this disease upon public health, it
is important to obtain a better understanding of factors
which may contribute to differential rates. The objective
of this study is to determine the possible explanatory
factors for racial differences in colon cancer risk.
Further, this study will examine the association between the
occurrence of colon cancer among Black and White males and
females and the usual occupations and industries in which
they were ever employed, as well as the tobacco use
histories of the study subjects. This research utilizes one
of the largest population-based samples available to conduct
an examination of race, colon cancer, and occupation.

BACKGROUND

The need to evaluate race and the differential rates in

cancer incidence, morbidity, mortality, and survival has

4

been recognized for several decades. However, the
difficulty in establishing the determinants of these
differences has also been evident. Current state-of-the-art
literature concerned with colon cancer has focused mainly
upon dietary patterns, exercise habits, family history, and
to some extent, stage at diagnosis or tumor site. Much of
this research has been limited to samples containing only
White males or very small sample sizes. Additionally, very
limited smoking histories have been available from
population-based studies, and occupational exposures were
traditionally obtained from death certificate data, which
have been shown to misclassify usual occupation and industry
by as much as 30 to 50 percent (Swanson Final Performance
Report, National Institute of Occupational Safety and
Health, 1995). To the large extent that cancer of the colon
is described by heredity and diet, these factors do not
fully ascertain causation, nor do they completely describe
the disproportionate levels of morbidity and mortality among
the Black population.
STAGE/SITE

Determinants of Black/White differences are often
postulated to be a direct result of more advanced disease at
diagnosis and socioeconomic differences between Blacks and
Whites. Data show that survival rate differences may be
due, in part, to the different stage distributions at
diagnosis. For White males and females, 38% of all cases
were diagnosed as having localized tumors, 37% as regional,
19% as distant, and 6% unstaged. Only 32% of Blacks were

5

diagnosed as localized, while 35% were regional, 25%
distant, and 8% unstaged. Overall, relative survival rates
vary only slightly by age at diagnosis for both racial
groups, although rates are lower for Blacks throughout all
age groups than survival rates for Whites (Ries et al.,
1994). However, according to Mayberry et a1. (1995), the
few studies that have investigated stage differences have
been limited by the modest number and type of explanatory
factors that were considered. They attempted to determine
what characteristics might contribute to the racial
differences in colon cancer survival and if a survival
disparity remained between Blacks and Whites after
adjustment was made for these characteristics.

Several investigators have studied the possible
relationship between stage and site with respect to
Black/White colon cancer survival differences. Mayberry et
al. (1996), in a prospective NCI Black/White cancer survival
study of 454 Blacks and 521 Whites, found that, after age
and gender adjustment, mortality hazard ratios for Blacks
were 50% higher than for Whites. Further adjustment for
stage reduced the excess mortality to 20% (RR=1.2, CI =1.0-
1.5). Adjustment for poverty and other sociodemographic
conditions did not further explain the survival disadvantage
among Blacks. It was concluded that stage at diagnosis
accounted for more than half of the excess colon cancer
mortality experienced among Blacks. An additional study by
Coates et a1. (1995) explored specific anatomic site
distribution within the colon by race and other colon cancer

6

risk factors to determine whether or not this may contribute
to survival differences. Among 1,045 patients from Atlanta,
New Orleans, and San Francisco/Oakland, it was found that
site was related to stage, grade, and histologic type, and
possibly smoking, however, it was not related to race,
suggesting that site differences are unlikely to contribute
to poorer survival among Blacks/Whites.
DIET

While as many as 25% of patients with colon cancer have
a family history of the disease (Potter et al., 1993), the
vast majority of cases of colon cancer appear to be related
to environmental factors. A substantial body of research
exists on the relationship between mortality from colon
cancer and per capita consumption of calories, with special
emphasis placed upon examining dietary fat, meat proteins,
fiber, cruciferous vegetables, and calcium and other
micronutrients (Potter et al., 1993). In general, it
appears that a large portion of the colon cancer incidence
may be attributed to a diet with high fat and low
fiber/vegetable intake. Migrant data have shown that the
disease is particularly sensitive to changes in the
environment and diet, with incidence rates reaching those of
the host country within one to two generations (Potter et
al., 1993). Internationally, colon cancer rates are the
highest in more developed countries of the Western world,
including Europe, Australia, the U.S., and Canada, with
incidence rates in the range of 25-35/100,000 during the
1980's. These rates are vastly different from some of the

7

lesser developed nations, where in India for example,
incidence rates are 1-3/100,000 and 3.4/100,000 in Nigeria
(Potter et al., 1993). Several ecologic and case-control
studies demonstrate positive associations between the risk
for colon cancer and dietary fat (Willett et al., 1990;
Giovannuci et al., 1992; Grahm et al., 1988). Others have
found positive associations with the consumption of meat
(Willett et al., 1990; Bostick, 1994; Manousos et al.,
1983). Separating the independent effects of dietary fats
and meats proved to be difficult in these studies.

Many etiologic hypotheses related to colon cancer have
evolved from the examination of dietary patterns and alcohol
consumption upon rate differences between Blacks and Whites.
Findings have indicated some geographic differences based
upon food frequency questionnaire methodologies. Blacks
have been found in general to consume higher fat, lower
fiber diets. Finally, as with diet, an association between
alcohol and colon cancer has generally been found to be
inconsistent (Miller et al., 1983; Ferraroni et al., 1989;
Murata et al., 1996; Potter et al., 1993).

The topic of diet and its relationship to colon cancer
has been explored in detail (Swanson et al., 1993; Dales et
al., 1978; Potter et al., 1993; National Academy Press,
1989) and will not be discussed further in this paper. Very
little data have been provided to determine the impact of
additional exposures and risk factors, such as cigarette
smoking and occupational exposures upon colon cancer. Less
emphasis has been placed upon these variables in explaining

8

the disproportionate levels of colon cancer morbidity and
mortality by race. It is important to recognize that the
data on diet remain inconsistent and inconclusive due to
population differences and issues related to the validity of
measurement instruments. The ascertainment of past and
present cigarette smoking habits is much easier to obtain
and is generally more reliable.
SMOKING

There is little debate over the impact and contribution
that cigarettes have had upon cancer among the general U.S.
population. The first official statement on smoking and
health by the U.S. government was contained in the Report of
the Advisory Committee to the Surgeon General of the U.S.
Public Health Service in 1964 (U.S. Department of Health,
Education, and Welfare). Cigarette smoking has been
implicated in cancer of the lung, larynx, oral cavity,
esophagus, urinary bladder, kidney, and pancreas. Although
cigarette smoking has been shown to be positively related to
the risk of adenomatous polyp development, a direct causal
relationship between cigarettes and colon cancer has not
been firmly established. While some studies were unable to
definitively establish an increase in risk of colon cancer
due to smoking, (Ferraroni et al., 1989; Nyr'en et al.,
1996) several other investigators have provided some
inferential evidence of a relationship.

A recent study by Slattery et a1. (1997), utilizing a
population-based case-control study conducted in Northern
California, Utah and Minnesota, observed a 50% increase in

9

colon cancer risk from smoking over a pack of cigarettes per
day among both men and women. The investigators also found
that those who smoked over 20 cigarettes per day and had a
larger body mass index were at greater risk than
participants who smoked the same amount but were smaller.
Chyou et al. (1996) observed 330 incident cases of colon
cancer among Hawaii Japanese men from a population of 7,945,
and noted that the risk of cancer increased with age,
alcohol intake, and pack-years of cigarette smoking.
Giovannucci et al. (1994), in a prospective study of
cigarette smoking U.S. women, found that although Relative
Risk is not significantly increased among women smokers in
general, it approaches 2.00 (CI=1.14-3.39) for those who
have smoked 10 cigarettes or more for 45 years. The authors
also note that, due to the lengthy induction period for
colon cancer (35 years), the number or cases related to
smoking may continue to increase due to the apparent
increase in smoking among younger women in the 1960's and
70's and again in the early 1990's. Finally, in what is
probably the largest population studied to date that has
been analyzed for the impact of cigarettes and colon cancer,
Heineman et al. (1995) followed 248,046 American veterans
for 26 years to clarify the relationship between tobacco and
colon cancer. Findings indicated, that when controlling for
social class and occupational activity, there was a Relative
Risk of 1.6 (CI=1.2-2.0) for colon cancer for those that
smoked an average of two packs per day or more, and a
Relative Risk of 1.4 (CI=1.2-1.8) for those who began

10

smoking by the age of 15, with the authors contending that
the issues critical to this study were the length of follow-
up and their choice of controls (veterans never using
tobacco).

Trends of consumption may also affect incidence of
colon cancer among Blacks. “Between 1960 and 1987 the age-
adjusted colon cancer mortality rate of Black men has
increased more per 100,000 than that of White men, from
196.3 to 269.2 in Black vs. 186.6 to 213.4 in White men”
(Wynder et al., 1994). In large part, this can be
attributed to increases in lung cancer, but the trend
differentials of smoking between Blacks and Whites may also
have impact upon colon cancer incidence. Wynder (1994)
notes that “The percentages of adult African-American men
who smoke cigarettes (54.6% in 1974 and 40.6% in 1985)
differ significantly from those of smokers among adult White
American men (42.3% in 1974 and 32.1% in 1985), reflecting a
lower percentage of ex-smokers among the African-American
male ever-smokers than among their White American
Counterparts".

Sorenesen et al. (1986) provide some insight into the
smoking habits of individuals based upon occupation. In a
sample of 3085 Minneapolis employees, smoking prevalence was
the highest among blue collar workers (45.7%) and lowest
among professionals (22.9%). What is even more revealing is
that blue collar workers were more likely to fail to quit in
work-sponsored cessation plans, and were more likely to have
smoked for a longer period of time based upon recent quit

11

rates. Data regarding this role of smoking in the etiology
of colon cancer, combined with data describing the role of
alcohol consumption, may point to the need for improved
methods for ascertaining accurate smoking and drinking
histories to explore the synergistic effects of the two
exposures.
OCCUPATION

In addition to examining cigarette smoking exposure to
investigate the increased risk of colon cancer, the
relationship between occupation and colon cancer has been
explored by various investigators, and although it is
generally not considered an occupational disease, excess
incidence of colon cancer has been reported in several
occupational classifications, with some emphasis placed upon
ethnic differences (Chow et al., 1994; Swanson et al., 1985;
Swanson et al., 1995; Lindsay et al., 1993; Courtney et al.,
1996; Garabrant et al., 1984; Lynge et al., 1988). However,
few investigators have explored the risk differential
between Blacks and Whites with respect to occupation and
colon cancer. In addition, there are few sources from which
these data may be obtained to perform this analysis.
According to Swanson and Burns (1995), the study of
occupation and its relationship to cancer incidence has
often been restricted to men, and more often than not, White
men. Again, this may have been due, in part, to the lack of
available data, but also may be attributed to the fact that
most large U.S. and international cancer registries
established linkages to Census occupational information from

12

the 1960's and 1970's, which included a smaller percentage
of women and Blacks in the U.S. and Europe. Examples of
these demographic limitations can be found in several large
population-based occupational cancer studies including The
Canadian Labour Force Ten Percent Sample Study (Lindsay et
al., 1993), the Cancer Environment Registry of Sweden (Chow
et al., 1994), and Data for the Danish National System
(Lynge et al., 1988), all of which utilized White males only
in their analyses.

Associations with colon cancer among occupational
subsets of the population have been difficult to establish
and often resulted in findings of relatively small increases
in risk. Sensitivity in measuring increased risk depends
upon several factors, all of which may be considered suspect
when discussing studies involving occupational exposures.
These factors include population size, the number of years
of accumulation of cases, the number of controls per case,
incidence of the tumor, the proportion of the population
exposed to the carcinogen, accuracy in attributing exposure
and disease status, and Type I and II errors (Siemiatycki et
al., 1981).

Limitations in establishing causality of a disease due
to occupational exposures are frequent and are a direct
result of the data resources available to researchers for
analysis. Typically, a variety of data sources have been
used to monitor trends of occupational illness and injury.
These include: employer reports, death certificates, birth
certificates, Worker's Compensation records, clinical

13

laboratory data, medical examiner reports, hospital
discharge records, and national health surveys (Baker et
al., 1989). Typically, these forms of data are collected
for reasons other than surveillance, and not likely to be
representative of the population at risk, and suffer from
temporal as well as data validity shortcomings. A study
evaluating the quality of death certificate data on
occupation (Turner et al., 1987), concluded that the
occupations and industries listed on death certificates were
accurate for only 54% of the individuals studied.
Additionally, other difficulties arise with regard to
attributing risk to an increased incidence of colon cancer
based upon occupation. Several issues that have been
difficult to control for include temporality, recall bias,
dose-response relationships, measurements of physical
activity, shortcomings of the instruments utilized to
measure dietary patterns, and finally, the myriad exposures
that occur outside of the workplace.

The study of the relationship between colon cancer and
occupation has generally focused upon the comparisons
between occupational groupings or categories, such as White
collar vs. blue collar, industrial vs. service, etc. There
has been little exploration into the differences between
races and gender within occupational categories. In
addition, those studies that have studied occupational risk
factors among women have been mortality studies, which are
limited by single death certificate entries regarding
occupation and lack information on other critical risk

14

factors such as smoking (Swanson et al., 1995). A
literature review by Burns et a1. (1991) revealed that
between 1984 and 1987, only 14 of 116 articles provided
information on non-Whites, and none of them evaluated
potential confounders.

Siemiatycki et al. (1981) indicate that few
occupational exposures are responsible for significant
proportions of any given type of cancer simply because it is
unusual for a large proportion of a population to be
occupationally exposed to any single substance. In
addition, these authors indicate that, apart from age and
sex, important cofactors such as socioeconomic status,
tobacco, and ethnic group are often not included in
occupational studies. The OCISS data utilized for this
study, however, is unique in that, for each subject, it does
include age, sex, ethnic group, and tobacco use history, as
well as detailed lifetime work histories.

Despite these limitations in studying the relationship
between colon cancer and occupation, several established and
probable carcinogens that are often encountered in the
workplace have been associated with elevated colon cancer
risk. Asbestos (Fredriksson et al., 1989; Selikoff et al.,
1979) has been shown to elevate cancer risk by two-fold in
some studies, while it has not been shown to elevate colon
cancer risk in others (Burns et al., 1994; Garabrant et al.,
1992). Additionally, synthetic fibers, grains, and wood
dust (Siemiatycki et al., 1986) have been implicated in
increased colon cancer risk.

15

Occupations that have demonstrated an increased
incidence of colon cancer include coke oven workers (Chan et
al., 1993), professional and technical workers,
administrative, executive and managerial workers, clerical
workers, sales personnel (Chow et al., 1994), chemical
processing and petroleum production workers, model makers
and patternmakers in the automobile manufacturing industry
(SMR=487) (Swanson et al., 1985) and construction workers
(Demers et al., 1994). Dubrow and Wegman (1983) undertook
an analysis of 12 major occupational disease surveillance
studies, utilizing only studies for which the association
was significant, the association was consistent across the
studies, and then the aggregate observed-to-expected ratio
was obtained by combining the results of the individual
studies. Findings indicated that across studies by
occupation, about 33% of these studies had significant
associations at the .05 level, indicating the difficulty of
finding a relationship in all but the largest studies.
Women were observed to have elevated risk when employed in
fabricated metals manufacturing, (OR=2.4, CI .7-8.0) and
rubber plastics manufacturing industries (OR=3.3, CI .6-
17.1), as were those working in utilities (OR=4.1, CI .8-
20.6).

Mortality studies of relationship between occupation
and colon cancer have not produced definitive results. A
study of 88,289 male and female chemical plant workers
revealed a slight increase in mortality from cancer of the
large intestine (Teta et al., 1990). A large Canadian study

16

(Lindsay et al., 1993) found increased mortality ratios for
colon cancer for individuals employed in clerical
occupations, coke oven workers exposed to polycyclic
aromatic hydrocarbons, and for locomotive engineers. The
Lindsay study suffered from several shortcomings. The
authors had no information on smoking or drinking, and
industry groupings were based upon incomplete data.

The Lindsay study did observe an association among
workers employed in occupations characterized by low levels
of physical activity and colon cancer (clerical workers).

In several studies, occupational exposures were not the
focus of the analysis, but rather the occupational physical
activity levels (Vena et al., 1985; Sandler et al., 1995;
Longnecker et al., 1995; Garabrant et al., 1984). Several
investigators have looked into the hypothesis of increased
incidence of colon cancer resulting from job-related stress,
with equivocal findings (Courtney, et al., 1993; Courtney et
al., 1996).

While no study can completely control for the large
number of difficulties in assessing occupational exposure
and increased risk of any cancer, the Occupational Cancer
Incidence Surveillance Study (OCISS) provides a unique
opportunity to combine population-based cancer registry data
with complete occupational and smoking histories, allowing
for the assessment of the differences in risk of colon

cancer experienced by the Black and White populations.

17

MATERIALS AND METHODS

The data file used in this study was constructed from
the 1984-1991 Occupational Cancer Incidence Surveillance
Study (OCISS), which is a population-based, case-referent
investigation of occupational risk factors for 11 cancer
sites diagnosed in Detroit Metropolitan area residents.

This study includes newly diagnosed cases, who were
identified through the Metropolitan Detroit Cancer
Surveillance System (MDCSS), a population-based cancer
reporting system for Wayne, Oakland, and Macomb counties
(Burns & Swanson, 1991). The MDCSS is a founding
participant in the National Cancer Institute's SEER Program
(Swanson & Brennan, 1981), and selects cases through the
MDCSS rapid reporting system, enabling investigators to
enroll participants into studies within 2 to 6 weeks after
diagnosis (Burns & Swanson, 1991).

Incident cancers occurring among Metropolitan Detroit
residents enrolled in the study include White and Black
cancer cases ages 40-84. This case-control analysis of
occupational risks is restricted to newly diagnosed patients
with cancer of the colon (ICD-9 CM 153), who were diagnosed
between November 1, 1984 and June 30, 1987. Overall,
approximately 20,000 interviews (each lasting approximately
fifteen minutes) were completed as part of the OCISS, with
cases or their spouse or other first degree relative being
interviewed by telephone within eight months of their cancer
diagnosis. The interviews captured smoking histories, adult
medical history, demographic information, residential

18

history, and a complete lifetime occupational history.
Racial background was self-reported for both cases and
controls.

Potential population controls were contacted using
lists of randomly generated telephone numbers. Interviewers
then obtained a working residential number and recorded the
age, race, and gender of all eligible adults in the
household, with one member of the household selected at
random to be interviewed (Swanson & Burns, 1995). Controls
were frequency-matched by age (within 5 years), gender and
area of residence (Macomb county, Oakland county, city of
Detroit, and Wayne county other than Detroit) (Burns &
Swanson, 1991). Individuals who reported ever having a
diagnosis of cancer were excluded.

The occupational history portion of the interview
instrument includes the occupational and industrial titles
of all jobs ever held, a description of the duties performed
on a daily basis, the duration of the position, and whether
the positions were full- or part-time. Occupational coding
was performed by a trained coder utilizing key words in the
descriptions of the positions held rather than by the job
title, thus enabling a more detailed level of coding.
Occupation and industry data obtained by the telephone
interview are coded according to the three-digit codes of
the 1980 Census Bureau classification. Completed interviews
underwent a final review by the supervisor of the
interviewing unit (Swanson Final Performance Report,
National Institute of Occupational Safety and Health, 1995).

19

This analysis focuses upon usual occupation and
industry, utilizing grouped codes, which are derived from
the 1980 Census Bureau codes. These codes were created by
combining appropriate 3-digit codes for occupations or
industries with probable similarities in work exposures, and
were created by the investigators including consultation
with an industrial hygienist, to increase the number of
cases included in each industrial and occupational groups
(Burns & Swanson, 1991). Occupational and industrial
histories were summed into two usual occupations and two
usual industries for each respondent. These are defined by
summing the total number of months for which a person
indicated employment in industries or occupations over their
entire work histories. The greatest number of months of
exposure was then chosen as the usual occupation or industry
for that respondent. All industrial and occupational codes,
which were categorized as either exposed or unexposed based
upon their potential exposure to known carcinogenic agents,
were utilized in this analysis.

This analysis reports the association between the
occurrence of colon cancer among Black and White males and
females and (1) the usual occupations and industries in
which they were ever employed, and (2) the cigarette, pipe,
and cigar smoking histories of both the cases and controls.
The Mantel-Haenszel chi-square test was utilized to evaluate
risk differences between cases and controls and between
Whites and Blacks based upon work history. These analyses
included adjustment for the potential confounding effects of

20

age, gender, and pack-years of smoking. Work history data
represent usual industries or occupations that occurred
among at least 10 respondents and with odds ratios that were
elevated to a level of at least 1.5 among any of the
demographic subsets examined (i.e. males/females,
Blacks/Whites).

A multiple logistic regression model was used to
determine the maximum likelihood estimates of the odds ratio
(OR) for independent variables. The outcome (dependent)
variable in the model was case-control status. The main
explanatory variables of interest were binary indicators of
occupations and industries with adjustment for race and
pack-years of smoking history. Blacks, females, and persons
with zero pack-years of smoking were used as the referent
groups. Other factors may influence a diagnosis with colon
cancer and were included in the model as additional
explanatory variables. These factors included several
patient demographic variables (age and gender). Indicator
variables were created for approximately 30 of the usual
occupations or industries that exhibited increased odds
ratios based upon the Mantel-Haenszel chi-square tests. The
initial logistic models included all main effects and first-
order interaction terms, and a combination of backward and
forward elimination selection methods were utilized to
determine which explanatory variables remained significant.
The significance level of the maximum likelihood estimate
statistic for variables to be retained in the model was set
at 0.05, and ninety-five percent confidence intervals were

21

 

calculated for each odds ratio. Potential confounders,
pack-years of smoking and age, were controlled for in the
models. The SAS package v 6.12 (SAS Institute, Cary, North
Carolina, USA) was used to construct the datasets and
perform the chi-square tests and the logistic regression
analysis.

RESULTS

This study population includes 6,619 Black and White
subjects and is comprised of 2,879 colon cancer cases and
3,740 population controls. Interviews conducted with the
cases and population controls were carried out either with
the patient or a proxy. Among cases, 2129 (73.95%) of the
interviews were carried out with the patient, while 368
(12.78%) were proxy interviews of a living subject and 383
(13.27%) were proxy interviews of a deceased subject.
Population control interviews were carried out with the
actual study subject for 3,459 (92.49%) individuals, while
only 7.51% (281) were proxy interviews. The overall
response rate for the colon cancer cases was 95%, while the
response rate for eligible controls was 97.3%.

Demographic characteristics of the study population are
presented by race, gender, age, smoking status and history,
marital status, and educational history in Table 1.
Educational history categories include Elementary (grades K-
6), Jr. High (grades 7 & 8), High School (grades 9-12),
College or Technical (12+ years), and Graduate (16+ years).
As a result of the frequency-matched design of this study,
there are a greater number of controls than cases in this

22

 

Table l.

W

 

 

Variable Cases (N82879, 43.5%) Controls (N83740, 56.5%)
No. (%) lb. (‘)
Race 8 Gender
White Males 1103 38.3 1423 38.1
Black Males 278 9.7 371 9.9
White Females 1203 41.8 1529 40.9
Black Females 295 10.2 417 11.1
Total 2879 100% 3740 100
Age at Diagnosis
40-49 101 3.5 n/a n/a
50-59 451 15.7 n/a n/a
60-69 981 34.1 n/a n/a
70-79 966 33.6 n/a n/a
80-84 380 13.2 n/a n/a
Total 2879 100% n/a n/a
Smoking Status
Ever 1469 51.0 2103 56.2
Never 1410 49.0 1637 43.8
Tbtal 2879 100% 3740 100%
Pack Years of Smoking
Nonsmoker 1410 49.0 1637 43.8
1-20 427 14.8 685 18.3
21-40 408 14.2 591 15.8
41-60 310 11.0 432 11.6
61-80 118 4.0 160 4.2

23

Table 1 (cont.)

 

 

variable Cases (NI2879, 43.5%) Controls (W83740, 56.5%)
no. (%) no. (9)
81-100 92 3.0 99 2.6
101-120+ 86 3.0 98 2.6
Missing 28 1.0 38 1.1
Total 2879 100% 3740 100%
Marital Status
Never 141 5.0 170 4.5
Married 1628 56.5 2048 54.8
Divorced or
Separated 249 8.5 418 11.2
Widowed 848 29.5 1093 29.2
N/A Missing 13 .5 11 .3
Total 2879 100% 3740 100%
Educational History
Elementary 220 7.6 223 6.0
Jr. High 360 12.5 363 9.7
High School 1447 50.3 1976 52.8
College or
Technical 610 21.2 878 23.5
Graduate School 161 5.6 239 6.4
Missing 81 2.8 61 1.6
Total 2879 100% 3740 100%

 

24

sample. Distributions of gender, cigarette/pipe/cigar
smoking status, educational history, and marital status
among both Black and White cases and controls are similar.
Whites constitute 80.1% of the cases and 79% of controls.
Table 2 presents a comparison of the two independent
variables of interest in this study, age and smoking
history, as well as the employment status, among the White
and Black populations. As a group, Blacks tended to have
higher rates of cancer at earlier ages, (age groupings 40-
49, 50-59, and 60-69) until approximately 70 years of age
and older, at which point the White population has a higher
percentage of respondents with cancer of the colon. These
distributions are consistent with national cancer registry
data (Ries et al., 1998). Pack years of cigarette/pipe/
cigar smoking appear to be similar between the case and
control groups, as well as both the Black and White
populations. Among both Blacks and Whites, a greater
percentage of controls than cases were employed at the time
of their interview. A slightly greater proportion of cases
than controls within both races tended to be homemakers.
Unemployment, retirement, and disability rates appear
similar between cases and controls, as well as between
Blacks and Whites.

Odds ratios indicating the colon cancer risk associated
with usual occupational and industrial exposures are

presented in Tables 3 and 4. Mantel-Haenszel Chi—square

25

 

Ao¢0nudonn Aaoodfionh AaOOnunhn flwocuuanun Aoocuunmau arcanuccnn H.908

xvowcesu .amdcuea .anacmna 1ammcmomn .ooucmam .«omocmo noxqeusom
.am.oaa Aumcmo Aamcme Aamcmoa .ancmm .aecno ooanuuaa
.omvcsoo .amcvnmn Aaeecemu .aoocomow .asncomofl Awevcoaofl consume
.avcom .avcam .om.mu 1a~.mm .aucov .ascsn oososdsucs
xaumcoon .aeuccmn .aom.vfla .aomcnona .pomcaom .oancume cascadsu
soapusuun no case an nan-um anon
.oooucnona Aaooncaos .oooucnsm .aooucamum .aooucwnan .ooeucuonu Hanna
.aa.ma .mu.¢a .oH.e .aacme .aflcew xaavvn oceans:
.aa.sa .aaco .oacm .wn.soa .ancom Aoncss +o~HIHoH
.aacma .aHCHH .aacm .amcasa .ancmm .ovcco ooHuHm
Aoucun .an.o~ .a~.- .amcoem .amvoefl .amvoofi condo
.ao«.~nfl .aoflcom .am.~m .amscoao .omaenmn Awascmmw amuse
Aymavaflu .aoavmma Apmflcmm Asmacmms Aaoacome .mva.-m oenan
aoucwsu .aumcssfl .wsavmm .aoscovm .asacoom Aaencmmm onus
.amvcooo .amccmmm .eemcsom .amecsmmm xanccemma .amevmofia noxosucoz “w
mnquIm usmqo\emwm\euveusmwu no anus» xusm
xaooacuonu .aooncoes .aooucnsm .aooacamum .aooacuman Aaoonvoonn «nuns
Anacuna Iascoo .aaacao .aaflcmmm Amocosm .aeflcnmm venom
.amucomn .ammcmaw .sflnvmsa .awm.smo~ .aomcflmm .«mncoms msuos
.asnvmom .amncmon Aamncoom .aemcomsa .ammvamoa .anncmms mouoo
xvoacamm .amd.¢¢a .wmacmoa Assacflom .amacmcm .amflcwmn mmuom
.aocsm 1amoao .aecom .wocamu Auscmnu Aancos melee

sedbueunn us out

 

flflOOHOh IQ“ dOHOQOO COCO “ECONOE 30“ Havana .010
.«unuusc nan-«n .omnmuzc nouns:

 

.‘a e. r c .4. e. . of .«s. .«e «a s‘. e . . . < . e e: e .i . 04< e . e;- ’ e u o . . e.

.N CHAIR

mo.vme «season one .mndxoaa mo eueehlxuem .3OH>ueucH us

on. now ocuusflua .

 

 

oo.v qu.o mm.o n m mo.H .ms.o HH.H am no mococh
-.o .om.o Hm.H s s om.H .mn.o mm.o oH AH noxuoz Hnuuom
Hm.v .uo.o sw.H HH m mH.H .oe.o om.o «a on noucHam
om.“ .ms.o v¢.H pm ea No.0 .mn.o mm.o on we um>Hun
H~.H .un.o «o.o oH on mn.~ .mo.H om.Ho om me uounuono ocHsuar
~¢.H .uo.o «8.0 we mo on.~ .no.H om.Ho am no souHcau
Hm.» .mm.o mm.H m m om.¢ .nH.o Ho.o w v ocHoHoz
He.mmH .v~.o an.o H o m¢.c .sm.o oo.~ AH «H umucHum
nu.o .mo.o Ho.o H m no.~ .vo.o on.H mH mH ouauuo Hume
mm.“ .sH.o sm.o v m m~.m .Ho.H on.~o oH HH Hues HuHoom
mzoueemnooo

om.o .Hu.o mH.H n m mm.o .mm.o mm.H s m monum>om
~m.¢ .sH.o om.o m m an.H .n¢.o ss.o mH on «Hum:
a~.m .v~.o Hm.H m m Hm.~ .nm.o mv.H es en mcHuuoHunam Hand:
o~.m .v¢.o mm.H s m se.~ .so.o m~.H AH mH nuance
mm.m .me.o os.H s o o~.H .sn.o so.o oH en sumuHHHz
no.8 .mo.o hm." MH 8 me.H .so.o mo.o >4 mm nououm unusuuodmn
ms.H .mo.o mm.o m m ms.m .am.o mo.~ m m uoOH>umm HuHoom
mm.on .mo.o mv.v m o on.m .mm.o so.~ m v .6“: .oqu
on.o .-.o mH.H a m ms.m .cm.o ms.H s m unsHo
Hm.n .no.o s~.o H n nn.m .Nm.o mo.~ «H m coHuHHom
ms.Hnm .oo.o oo.m H o ~¢.n .mo.o mm.H NH HH mcHuHuuo>o¢
us.¢s .so.o oH.m c H o«.~ .mm.o mH.H pH oH Hosanna
mm.vm .s~.H oo.HHo m H o~.n .om.o as.H mm oH uoHum uuo
mo.sH .oH.o um.H m m vs.~ .mm.o mm.H ow Hm uoHuHHHua
Hu.n .«s.o 84.8 H n em.n .nv.H mn.~o we on ocHucHum
ou.vH .nn.o mH.u n n «m.H .mm.o Hn.H on on uuchHnuux

Ho «mm so some Houucoo Ho amm so some Houucoo

.HonHuav «scene Hanna's. unsung seasonan

 

.n CHACH

27

mo.vmo «Hoodoo one

.oconan no nusohrxusm .3OH>uousH us

one nan counsnoc «

 

 

so.v .sm.o Ho.~ 4H m He.H 486.0 ~m.o o4 em oucach
mo.H .eo.o o~.o H m nm.m .om.o no.H 6H mH uoxuoz Haumom
Hm.m .mo.o mm.o H m m¢.H .sm.o ~m.o an «e umuchm
-.~H .nm.o so.H m N Ho.H .Hm.o ms.o om ooH um>Huo
mm.n .ss.o mo.H mH MH so.H .ms.o mH.H Hm mm uouaumdo mcHnomz
m~.~ .mm.o ~¢.H an mm m¢.H .ns.o mo.H Ho as nouHcan
m~.nH .No.o mm.o o H mm.¢ .mm.o oo.H oH m mcHuHoz
oH.n .eo.o en.o H e um.» .mm.H sH.no pH 8 noucHum
mo.m .ms.o «o.~ HH 5 H~.~ .vn.o sm.o m nH muauum Home
m~.n .nm.o Nn.H m oH v~.¢ .mo.o no.H HH m xuo: HmHuom
moneamaooo

oo.mH .so.o mH.H H H o~.¢ .sm.o mm.H o m amouo>am
om.H .on.o co.o 0H NH AH.~ .He.o «8.0 oH nH «Hum:
He.e .eo.o so.H m s nm.~ .mm.o m¢.H em mm ocHuouHunam Huumz
om.m .mo.o oo.n m e mm.H .me.o sm.o mH mH umnnsm
c~.m~ .mH.o SH.~ N H mm.H .mo.o ms.o Hw on suuuHHHz
ms.H .os.o mH.H me we om.~ .mm.o -.H pH pH umuoum unassuudon
sn.~ .nn.o «8.6 m «H sm.Hm.ss.o mm.m m H umoH>umm HuHuom
sH.~n .om.o m~.m v H me.~H.mm.o -.m m n .uux .ouH:
-.mH .mm.o mm.u e N ~¢.n .m~.o mm.o m o unsHo
H¢.v .ne.o om.H m m -.m .sc.o sm.H m m conHHom
Hs.HH .mn.o vo.~ m m os.n .mo.o Hm.H HH a mcHuHuum>o¢
ms.n .ss.o mo.H mH NH mm.u .mm.o Ho.H m s Hosanna
Hm.m~ .m~.o mm.~ m H mv.n .mo.H ma.Ho on mH aoHam uao
s~.oH .mo.H mn.¢o s a Hv.~ .ns.o mn.H em Ha uoHuHHHua
oH.m .om.o mH.~ «H a os.n .-.H «H.~o on om mcHueHum
mm.H .HH.o on.o e «H mo.~ .so.H me.Ho as He auuHcHsoaz

Ho amm mo mono Houucoo Ho amm so mono Houucoo

.vvenun. annexes .nanua. anger seasonan

 

28

statistics were calculated between Black and White males and
females for all usual occupations and usual industries
collected in the study, however, the results of only 16
industry and 12 occupational groups are presented here.
These industries or occupations were chosen based upon each
grouping having a minimum of 10 observations or an odds
ratio of at least 1.50 among either Blacks or Whites. Table
3 presents an analysis of usual occupations for both cases
and controls among Blacks and Whites separately, and adjusts
for age, pack years of smoking, and gender, in an effort to
examine the potential influence of race on case-control
status. Among Whites, the odds ratio for those working in
the printing industry was significantly elevated, as was the
odds ratio for social workers. Also in Table 3, a number of
usual occupations and industries were also elevated for
Whites, but not significant, including printing occupations,
and usual industries of religion, miscellaneous
manufacturers, and social services. For Blacks listed in
Table 3, several usual occupations and industries exhibited
elevated odds ratios including machinists, apparel industry
workers, advertisers, miscellaneous manufacturers, and
printers, however, they were not significant. Blacks in car
sales and department store industries showed significantly
elevated levels of risk.

A similar comparison for industry and occupation groups
was done comparing males and females, controlling for race,
age at interview and pack-years of smoking in Table 4. The
highest odds ratios for usual industry for males are found

29

 

among individuals employed in printing, miscellaneous
manufacturing, and social services, as well as those with a
usual occupation of printers. Printing is the only industry
and occupation significantly elevated for males. Elevations
in the odds ratios are also seen among females in Table 4 in
the usual industries of printing, utilities, car sales,
advertising, clubs, miscellaneous manufacturing, and rubber
manufacturing. Usual occupations for females showed excess
risk in real estate and finance. The only usual occupation
or industry that was significantly elevated for women was
utilities. However, it is important to recognize, that in
this sample that contains 3,444 females, over 77% (2,676)
indicated having a usual occupation of housewife and at the
time of interview, 1,776 females had a current work status
of homemaker (Table 2), thus, creating difficulties with
small cell sizes for most of the usual groupings for the
female population.

Table 5 displays the results of the multivariate
logistic regression analysis. Univariate analyses were run
on all variables thought to exhibit an increased risk of
colon cancer and are not presented. Variables were selected
for the model based upon their level of significance
comparing backward and forward elimination techniques.

Usual occupations or industries were selected for univariate
analyses based upon those presented in Tables 3 and 4.

Usual occupations or industries that remained in the model
met the minimum 0.05 level of significance and are presented
in Table 5. No interactions or higher order terms were

30

Table 5.

 

 

- 8 8
95%
Variable Parameter Standard Odds Ratio Confidence
Estimate Error Intervals

Intercept -.9557 .1323

Age 2 40 1.00
Age 2 50 .5822 .1315 *1.790 1.383, 2.316
Age 2 60 .6514 .1249 *1.918 1.502, 2.450
Age 2 70 .8430 .1256 *2.323 1.816, 2.972
Age 2 80 1.0888 .1403 *2.971 2.256, 3.911
Black Race 1.00
White Race .0175 .0628 1.018 0.900, 1.151
Female 1.00
Male .0324 .0539 1.033 0.929, 1.148
0 pack years 1.00

1-20 pack years -.2390 .0736 *0.787 0.682, 0.910
21-40 pack years '-.1520 .0765 90.859 0.739, 0.998
41-60 pack years '-.1681 .0855 90.845 0.715, 0.999
61-80 pack years -.1415 .1299 90.868 0.673, 1.120
81-100 peck years .0610 .1529 1.063 0.788, 1.434
101-120 pack years .0047 .1553 0.995 0.734, 1.350
Machine Industry .3015 .1608 1.352 0.987, 1.853
Printing Industry .7819 .2445 *2.186 1.354, 3.529
Utility Industry .5796 .2789 91.785 1.033, 3.084
Car Sales Industry .7851 .2946 *2.193 1.231, 3.906
Miscellaneous

Manufacturing 1.3807 .5871 93.978 1.259,12.571
Metal Fabricators .3970 .2244 1.487 0.958, 2.309

 

*p<.05, 4<.10

31

necessary given the fit yielded from this more parsimonious
model. The overall model fit the data with a P < 0.0001.
After controlling for other observable variables through the
logistic regression analysis, including patients
demographics, smoking patterns, and work history, the
results show that the odds of increased cancer risk based
upon race, gender, and smoking, are equivocal and are not
significantly different from 1.00. Both race and gender
were forced into the model despite nonsignificant P-values
from log-likelihood ratio tests due to the fact that they
were main variables of interest. Additionally, no increased
risk is seen for any duration of smoking history, relative
to those who indicated no history of cigarette, pipe, or
cigar smoking.

However, some individual coefficients are worthy of
note, with positive and significant odds ratios for
increasing age beyond age 50, and several occupations and
industries. As expected, age group independent variables
consistently yielded positive parameter estimates,
increasing for each group relative to the study referent
group (age 40-49), so that by age 80, individuals are at
nearly three times the risk of colon cancer as are those 40-
49 years of age. Other factors besides age may have also
influenced cancer incidence among this population.
Occupational or industrial exposures in this multiple
regression indicated that the odds of having a usual
industry of miscellaneous manufacturing carried nearly four
times the risk as those not employed in these types of

32

industries. Examples of occupations that fell into this
category included Census Bureau classifications 391
(miscellaneous manufacturing industries) and 392 (not
specified manufacturing industries). Other industries that
exhibited positive and significant parameter estimates
included those employed in the printing industry, the car
sales industry, and the utilities industry. Other usual
occupations or industries that remained in the logistic
model were not associated with case status.
DISCUSSION

This case-control study was designed to test the
hypothesis of a potential association of Black race with an
increased risk of colon cancer, controlling for known or
suspected risk factors, including pack-years of tobacco use,
occupational exposures, age, and gender. The literature
provides sparse information about the combination of these
relationships. Past studies on this subject have been
limited by the use of death certificate data to establish
occupation, small sample sizes, samples containing only
White males, and limited smoking histories. This paper
presents the results of this study, which highlight some of
the difficulties in examining colon cancer risk, race,
occupational risk factors, and the impact of tobacco use.
The study utilizes OCISS data, which contains outcome
information provided from hospital abstracts of incident
cancer cases, along with a lifetime history of occupational
information, complete lifetime tobacco use histories
obtained by interview. Furthermore, the sample is a large,

33

population based, racially diverse, representative group of
subjects from which logical inferences could be made with
respect to the goals of this paper.

This paper emphasizes three findings. First, the risk
of colon cancer increased steadily with age, and this age-
related increase in risk occurred for each age-group,
controlling for race, gender, pack-years of smoking, and
occupational history. Age-related increases in incidence
and mortality, as well as colon cancer risk, are well
documented. The Surveillance, Epidemiology, and End Results
(SEER) Program, indicates a rate of 7.9 cases per 100,000
for those ages 40-44, which increases by over a factor of 4
(33.3 per 100,000) by age 50, to over 214 (214.1 per
100,000) for individuals between 70 to 74 years of age (Ries
et al., 1994). In this study, a higher percentage of Blacks
were diagnosed with colon cancer at earlier ages than Whites
at corresponding ages, until the age of 70, when the
percentage of Whites with colon cancer was higher than for
Blacks. This information is supported by prior findings.
Cooper et al. (1994), found in a database of 75,266 1987
U.S. Medicare patients, that risk levels between Blacks and
Whites were similar until 69 years of age, but beyond age
70, risk levels of Whites were greater than those of Blacks.

Second, although this study does present some
interesting leads for future investigation into the
association between race and certain occupational exposures,
one cannot conclude from the data presented that the risk of
colon cancer is consistently greater in any one usual

34

industrial or occupational grouping for Blacks than for
Whites. In actuality, a greater number of usual industries
and occupations had significant or elevated but not
significant, increases in risk for Whites than for Blacks.
It does appear that for certain usual occupations and
industries, there is a consistent increase in risk of colon
cancer, regardless of race, gender, or smoking histories.
There is some evidence for an association between colon
cancer and specific industries, such as printing,
miscellaneous manufacturing, and utilities. Alternatively,
observed risk levels among occupational subgroups among
Blacks and Whites may be explained by the inability of these
data to measure certain confounders, such as dietary or
physical activity patterns, in relation to the main
exposures of interest.

Third, consistent with earlier research (Potter et al.,
1993; Heineman et al., 1995; Giovannucci et al., 1994), the
results of this study do not support the association between
colon cancer due to smoking for either Blacks or Whites,
regardless of pack-year histories. This was consistent in
men and women and in Blacks and Whites.

Although interview data are commonly used in cancer
research, these data have several relevant limitations.
Reliance upon survey data limits the ability to accurately
assess specific exposures to known carcinogenic agents.
Although these data do contain complete lifetime
occupational and smoking histories, the levels of physical
activity or intensity of exposures to known risk factors

35

cannot be measured. Additionally, these data cannot account
for activity levels not related to the respondent's
occupations, further limiting the assessment of any
relationship between occupation, physical activity, and
colon cancer.

While a large number of industrial and occupational
categories are available for analysis from these data, this
may also be considered a limitation. A large number of
comparisons across occupational and industrial groups,
especially among the Black population, were limited by small
numbers within cells. This problem was more common among
female subjects. The ability to accurately assess an
increase in risk of colon cancer in females compared to
males, or between women across various occupations, is
severely limited in this study due to the fact that over
three-fourths (77%) of the colon cancer cases and controls
combined had a usual occupation of housewife. Another
difficulty in attempting to attribute an increased risk in
colon cancer to a specific occupation or industry is the
limited data available on potential confounding variables.
No data was collected on diet, alcohol consumption, or
family history, all of which have been positively associated
with increases in colon cancer risk (Potter et al., 1993).
Other variables that were unavailable include income,
physical activity, and body mass index.

In terms of public health significance, the range of
occupational and industrial exposures in this study may be
seen as a strength. The relatively large sample size allows

36

for comparisons of many occupational risks. This study
represents an investigation of occupational exposures
through interview, and provides a complete lifetime
occupational assessment, rather than single entry from death
certificates. Analyses that utilized OCISS data were
performed by Schade and Swanson (1988), and showed a 30-50%
error rate for death certificate employment data compared to
interview data. Also, one study indicates that the death
certificate occupation is the least likely to be the most
recent occupation, is somewhat more likely to be usual
occupation, and even more likely to be occupation in the
last 15 years (Turner et al., 1987). This points to the
validity of using data from interview respondents to
determine usual occupation and industry, and to using usual
occupation as the baseline for the analysis of occupational
exposures in this study, rather than most recent occupation.
Additionally, because this population-based sample is
obtained from hospital abstracts of incident cancer cases,
these data can be considered highly accurate regarding the
cancer diagnosis, and also fairly representative of the
working pepulation in this region of the country. The high
overall response rate among cases (95%) and controls
(97.3%), as well as a minimal reliance upon proxy interview
(15.6%) of all cases and controls, increases the reliability
of the occupational data. Because this study is not
restricted to White males, as many occupational assessments
have been in the past (Lindsay et al., 1993; Lynge et al.,
1988), a comparison between the White and Black population

37

is possible, and allows for the testing of differences
between these two racial groups, improving the
generalizability of the findings. Finally, complete
lifetime tobacco use histories are obtained for all OCISS
subjects. As a result, the occupational risks take into
account the cigarette smoking history of the subjects and
its potential confounding effects (Burns & Swanson, 1991).

The overall results of this case-control study do not
indicate that Blacks have a significantly higher risk of
colon cancer than Whites when controlling for occupational
exposures, gender, age, and tobacco histories. While there
were increases in occupational risk among Blacks
(machinists, car sales, department stores) and Whites
(printing, social work), the analysis across multiple
specific usual occupational and industrial groupings show
different workplace risks for Blacks and Whites. Since a
greater number of usual industries and occupations were
limited by small sample sizes for Blacks, a larger sample of
Blacks may have elucidated clearer results with regards to
risk of colon cancer associated with certain industries and
occupations.

Consistent with earlier research, the exposure to any
level of cigarette smoking did not appear to increase the
risk of colon cancer, however, it was very important to have
this variable available to control for its potential
confounding effects, especially because of the possibility
of the synergistic effects that smoking may have with
certain occupational exposures such as asbestos. This study

38

also would have benefitted from more precise measures of
other potential carcinogens and biologic agents, as well as
diet and physical activity.

The implications of these findings are three-fold.
First, although these data do not necessarily support an
increased risk of colon cancer among Blacks, the population-
based incidence and mortality data indicate that certain
segments of the Black population, especially those
approaching 50 years of age, need to be more aggressively
targeted for colorectal cancer screening. Based upon the
1997 Clinical Guidelines and Rationale by the American
Gastroenterological Association, ‘Screening for colorectal
cancer and adenomatous polyps should be offered to all men
and women without risk factors, beginning at age 50'
(American Gastroenterological Association, 1997). This is
especially relevant for the Black population, given the fact
that their incidence rate by age 50 is 50.2 per 100,000
compared to 28.4 per 100,000 for Whites, and also, their
rate of mortality due to colon cancer is higher in every
single age category than that of Whites (Ries et al., 1998).
Second, the results of this paper support the lack of a
relationship between tobacco use and colon cancer. This
finding was consistent across age, race, and gender, and
increased levels of exposure did not exhibit any dose-
response relationship.

Finally, these data continue to support the need for
continued study of the Black population, utilizing larger
sample sizes. Even with the high quality of data obtained

39

through a case-referent methodology, this study was limited
in its ability to detect increases in risk in many
occupations among the Black population due to small cell
sizes and points to the need for continued aggressive

surveillance of this population.

40

LI ST OF REFERENCES

LIST 0? REFERENCES

American Cancer Society (1996). Cancer Facts and

Figures Amer1can_Cancer_Soc1et¥1_Inci

American Gastroenterological Association (1997).
Colorectal cancer screening: Clinical guidelines and

rationale. Gatroenterolos¥1_112. 594-642.

Baker, E. L. , Honchar, P. A. & Fine, L. J. (1989).
Surveillance in occupational illness and injury: Concepts

and content Amer1can_Journal_of_2ublic_Healthi_12. 9-11

Bostick, R.M., Potter, J.D., Kushi, L.H., Sellers,
T.A., Steinmetz, K.A., McKenzie, D.R., Gapstur, S.M., &
Folsom, A.R. (1994). Sugar, meat, and fat intake, and
nondietary risk factors for colon cancer incidence in Iowa

women. Cancer_Causes_and_Controli_5. 38-52.

Burns, P.B., & Swanson, G.M. (1991). The occupational
incidence surveillance study (OCISS): Risk of lung cancer by
usual occupation and industry in the Detroit Metropolitan

area. Amer1can_Journal_of_Industr1al_Med1c1nei_12. 655- 671

Chau, N., Bertrand, J.P., Mur, J.M., Figeredo, A.,
Patris, A., Moulin, J. J. & Pham, Q.T. (1993). Mortality in

retired coke oven plant workers. British_fl9urnal_gfi_
Industrial_Medicinei_Efl 127- 135

Chow, W.H., Malker, H.S.R., Hsing, A.W., McLaughlin,
.K., Weiner, J. A. Stone, B. J. Ericsson, J. L. E. & Blot,
J (1994) Occupational risks for colon cancer in Sweden.

Journal_of_0ccunaLional_Med1cinei_36(6). 647-651.

Chyou, P. H. Nomura, A. M. & Stemmermann, G. N. (1996).
A prospective study of colon and rectal cancer among Hawaii

Japanese men Annals_of_Enidem1olog¥1_6(4) 276- 82.

Coates, R.J., Greenberg, R.S., May-Ting-Liu, Correa,
P., Harlan, L.C., Reynolds, P., Fenoglio-Preiser, C.M.,
Haynes, M.A., Hankey, B.F., Hunter, C.P., & Edwards, B.K.
(1995). Anatomic site distribution of colon cancer by race
and other colon cancer risk factors. Diseases_gfi_the_gglgn
and_Bectum.

Committee on Diet and Health, Food, and Nutrition
Board, Commission on Life Sciences. (1989). Die;_and

National Academy Press: Washington, D.C.

Cooper, G.S., Yuan, Z., Landefeld, S., Johanson, J.P.,
& Rimm, A.A. (1995). A national population-based study of

41

incidence of colorectal cancer and age: Implications for
screening in older Americans. CancerL_15(3).

Courtney, J.G., Longnecker, M.P. Theorell, T. &
Gerhardsson de Verdier, M. (1993). Stressful life events

and the risk of colorectal cancer. Epidemiology‘_§_, 407-
414.

Courtney, J. G. Longnecker, M. P. & Peters, R. K.
(1996). Psychosocial aspects of work and the risk of colon

cancer. Epidem1olos¥1_1(2), 175-180.

Dales, L.G., Friedman, G.D., Ury, H.K., Grossman, S., &
Williams, S. R. (1978). A case- -control study of
relationships of diet and other traits to colorectal cancer

in American Blacks Amer1can_Journal_of_En1demiologxi
192(2), 132- 144.

Demers, R.Y., Burns, P.B., & Swanson, G.M. (1994).
Construction occupations, asbestos exposure, and cancer of

the colon and rectum. Q9nInal—9f—QQQHDBLiQnal_Medicin§L
15(9).

Dubrow, R., & Wegman, D.H. (1983). Setting priorities
for occupational cancer research and control: Synthesis of
the results of occupational disease surveillance studies.

Journal_Qf_the_Nat1onal_Cancer_Inst1tutei_11(6).

FerrarOni, M., Negri, E., La Vecchia, C., D'Avanzo, B.,
& Franceschi, S. (1989). Socioeconomic indicators, tobacco
and alcohol in the aetiology of digestive tract neoplasms.

Internat1onal_1ournal_of_En1demiolog¥1_18(3) 556- 561.

Fredriksson, M., Bengtsson, N.O., & Hardell, L. (1989).
Colon cancer, physical activity, and occupational exposures.
Sangeri_fll. 1838-42.

Garabrant, D. H. Peters, J. M. Mack, T. M. & Bernstein,
L. (1984). Job activity and colon cancer risk. American

Journal_of_Enidemiolog¥i_112(6).

Garabrant, D. H. Peters, R. K. & Homa, D. M. (1992).
Asbestos and colon cancer: Lack of association in a large

case- -control study Amer1can_Journal_of_Enidemiologxi_135.
843— 853.

Giovannucci, E., Colditz, G.A., Stampfer, M.J., Hunter,
D., Rosner, B.A., Willett, W.C., Speizer, & F.E. (1994). A
prospective study of cigarette smoking and risk of
colorectal adenoma and colorectal cancer in U. S. women.
Journal_of_the_Nat1onal_Cancer_Inst1tutei_86(3).

Giovannucci, E., Stampfer, M.J., Colditz, G., Rimm,
E.B., & Willett, W.C. (1992) Relation of diet to the risk

42

of colorectal adenoma in men. Journal_gfi_§he_flag1gnal
Cancer_1nst1tutei_15, 75.

Grahm, 8., Marshall, J., Haughey, B., Mittelman, A.,
Swanson, M., Zielney, M., Byers, T., Wilkinson, G., & West,
D. (1988). Dietary epidemiology of cancer of the colon in

Western New York. Amer1can_Journal_of_Epidemiclosxi_128.
490- 503.

Heineman, B.F., Zahm, S.H., McLaughlin, J.K., & Vaught,
J. (1995). Increased risk of colorectal cancer among
smokers: Results of a 26- -year follow- up of U. S. veterans and
a review. Internat1onal.19urnal_of_sanceri_52 728- 738

Lindsay, J. P. Stavraky, K. M. & Howe, G. R. (1993).
The Canadian labor force ten percent sample study: Cancer

mortality among men, 1965- 1979. Journal_gf_Qccupat1gnal
Med1cinei_35(4). ~

Longnecker, M.P., De Verdier, M.G., Frumkin, H., &
Carpenter, C. (1995). A case- -control study of physical
activity in relation to risk of cancer of the right colon

and rectum in men Internat1onal_Journal_of_Enidem1ologxi
24(1).

Lynge, E., & Thygesen, L. (1988). Use of surveillance
systems for occupational cancer: Data from the Danish

national System Internat1onal_Journal_of_Enidem1olos¥1
11(3), 493 499.

Manousos, 0., Day, N.E., Trichopoulos, D.,
Gerovassilis, P., & Tzonou, A., (1983). Polychronopoulou.
Diet and colorectal cancer: A case control study in Greece.
Internat1onal.19urnal_of_9ancer1_32. 1- 5.

Mayberry, R.M., Coates, R.J., Wesley, M.N., Click,
L.A., Chen, V.W., Austin, D.F., Redmond, C.K., Fenoglio-
Preiser, C.M., Hunter, C.P., Haynes, M.A., Muss, H.B.,
Wesley, M. N. ,Greenberg, R. S. & Edwards, B. K. (1995). A
determinants of black/white differences in colon cancer
survival Journal_of_the_National.§ancer11nst1tute

Miller, A.B., Howe, G.R., Jain, M., Craib, K.J.P., &
Harrison, L. (1983). Food items and food groups as risk
factors in a case- -control study of diet and colo- rectal
cancer Internat1onal.19urnal_of_Canceri_32. 155-161.

Miller, B.A., Kolonel, L.N., Bernstein, L., Young, Jr.
J.L., Swanson, G.M., West, D., Key, C. R. Liff, J. M.
Glover, C. S. Alexander, G. A. et al. (1996). Rag1aiLELhn19

'. ‘ 0‘ 0 ‘ o 9‘ r ‘0 , a ‘= '33- °' 14 .004

Cancer_1nst1tute. NIH Pub. No. 96-4104. Bethesda, MD.

Murata, M., Takayama, K., Choi, B.C., & Pak, A.W.
(1996). A ested case-control study on alcohol drinking,

43

tobacco smoking, and cancer. Canch_ne;egt1gn_and
WM), 557-65.

Nyr'en, 0., Bergstrom, R., Nystrom, L., Engholm, G.,
Ekbom, A., Adami, H.O., Knutsson, A., & Stjernberg, N.
(1996). Smoking and colorectal cancer: A 20—year follow-up
study of Swedish construction workers. Journal_gf_;he
WW. 1302- 7

Potter, J.D., Slattery, M.L., Bostick, R.M., & Gapstur,
S.M. (1993). Colon cancer: A review of the epidemiology.

E .3 . J . E . .

Ries, L. A. G., Kosary, C.L., Hankey, B.F., Miller, B.A

& Edwards, B K.(1998).SEEE_Cancer_Stet19_t1c_s_Renien.
WW Bethesda. MD.

' I

Ries, L.A.G., Miller, B.A., Kosary, C.L., Harras, A., &
Ewards, 5.x. (1994). W
W NIH Pub.
No. 94-2789. Bethesda, MD.

Sandler, R. S. Pritchard, M. L. & Bangdiwala, S. I.
(1995). Physical activity and the risk of colorectal

adenomas. Ep1dem1glggy1_§(6).

SAS Institute, Inc. (1990). SAS Procedures Guide,
Version 6, Third Edition, Cary, NC: SAS Institute Inc.

Selikoff, I.J., Hammond, B.C., & Seidman, H. (1979).
Mortality experience of insulation workers in the United

States and Canada. 1943-1976. Ann_NX_Acad1_Sc111_110, 91-
116 .

Siemiatycki, J., Day, N.E., Fabry, J., & Cooper, J.A.
(1981). Discovering carcinogens in the occupational
environment: A ovel epidemiologic approach. Journal_gf_the
Wire).

Siemiatycki, J., Richardson, L., & Gerin, M. (1986).
Associations between several sites of cancer and nine
organic dusts: Results from an hypothesis— generating case-
control study in Montreal 1979- 1983. Amer1gan_gguznal_gfi
Epidemielesx1_123 235-249

Slattery, M.L., Potter, J.D., Friedman, G.D., Ma, K.N
& Edwards, S. (1997). Tobacco use and colon cancer.
WW9), 259-64.

° I

Sorensen, G., & Pechacek, T. (1986). Occupational and
sex differences in smoking and smoking cessation. Journal

WW6).

44

Swanson, G.A., Gridley, G., Greenberg, R.S.,
Schoenberg, J.B., Swanson, G.M., Brown, L. M. , Hayes, R.,
Silverman, D., & Pottern, L. (1993). A comparison of blacks
and whites in three areas of the United States. Nutx111gn
and_Caneeri_ZQ. 153- 165.

Swanson, G.M. (1988). Cancer prevention in the
workplace and natural environment: A review of the etiology,
research design, and methods of risk reduction. Cancer1
52(8).

Swanson, G. M. (1995). Occupational cancer

surveillance: New approaches. Einel_2erformanee_8enert1

\. '09. 9‘ . r o ,-, o O 9. or. e f 2.. 9‘. 9

5-ROl-OHOZO67.

Swanson, G.M., Belle, S.H., & Burrows, R.W. (1985).
Colon cancer incidence among modelmakers and patternmakers
in the automobile manufacturing industry. Journal_gf

Qeeunetional_Med1cinei_21(8)

Swanson, G.M., Belle, S.H., & Satariano, W.A. (1985).
Marital Status and Cancer Incidence: Differences in the

Black and White Populations. Cancer_Researgh1_15, 5883-
5889.

Swanson, G.M., & Burns, P.B. (1995). Cancer incidence
among women in the workplace: A study of the association
between occupation and industry and 11 cancer sites.
Jeurnal_ef_Qeeupet1enal_end_Enr1renmental_Med1e1nei_31(3)

Taplin, S.H., Barlow, W., Urban, N., Mandelson, M.T.,
Timlin, D.J., Ichikawa, L., & Nefcy, P. (1995). Stage, age,
comorbidity, and direct costs of colon, prostate, and breast
cancer care. Jeurnal_ef_theINatienal_Cancer_Inst1tutei
&1(6).

Teta, M.J., Schnatter, A.R., Ott, M.G., & Pell, S.
(1990). Mortality surveillance in a large chemical company:
The union carbide corporation experience, 1974- 1993.
Amer1ean_Jeurnal_of_Industr1al_Med1e1ne._11 435- 447.

Turner, D. W. Schumaker, M. C. & West, D. W. (1987).
Comparison of occupational interview data to death

certificate data in Utah Amer1ean_leurnal_ef_1ndustr1el
Med1e1ne1_12. 145- 151

U. S. Department of Health, Education, and Welfare.
Public Health Service, Office of the Assistant Secretary for

Health. Office on Smoking and Health. Smgk1ng_and_flealth1_a
Renert_ef_the_Surgeen_Seneral.

45

Vena, J.B., Grahm, S., Zielezny, M., Swanson, M.K.,
Barnes, R. E. & Nolan, J. (1985). Lifetime occupational

exercise and colon cancer. American_lgurnal_gf
En1dem1olos¥+_122(3)

Willett, W.C., Stampfer, M.J., Colditz, G.A., Rosner,
B. A. & Spiezer, F. E. (1990). Relation of meat, fat, and
fibre intake to the risk of colon cancer in a prospective

study among women Neu_England_Journal_of_Medicine1_lzl.
1664- 1672.

Wynder, E.L., & Hoffman, D. (1994). Smoking and lung

cancer: Scientific challenges and opportunities. Cancer
Research1_fifi. 5284-5295.

46

MICHIGAN STATE UNIV. LIBRARIES
1111111111111NWWI“111WllllWlllHWl
31293017718374