Content uploaded by Lawrence J Cheskin
Author content
All content in this area was uploaded by Lawrence J Cheskin
Content may be subject to copyright.
Body Weight and Health Care Among
Women in the General Population
Kevin R. Fontaine, PhD; Myles S. Faith, PhD; David B. Allison, PhD; Lawrence J. Cheskin, MD
Objective: To examine the relation between body mass
index ([BMI] calculated as weight in kilograms divided
by the square of height in meters) and the use of medi-
cal care services among a nationally representative sample
of women.
Design and Setting: Multistage cluster-area probabil-
ity sampling survey. Data are from the Cancer Control
and Health Insurance supplements of the 1992 National
Health Interview Survey conducted by the National Cen-
ter for Health Statistics. Respondents were 6981 women
aged 18 years or older residing in the United States who
self-reported sociodemographic information and the use
of health care services.
Main Outcome Measures: Interval (#3 years vs .3
years) since most recent mammography, clinical breast
examination, gynecologic examination, and Papanico-
laou smear and the number of physician visits in the year
before the survey.
Results: When age, race, income, education, smoking,
and health insurance status were adjusted for, the BMI
was directly related to delaying clinical breast examina-
tions, gynecologic examinations, and Papanicolaou
smears. Obese women (BMI of 35) were more likely than
nonobese women (BMI of 25) to delay clinical breast ex-
aminations (odds ratio, 1.26; 95% confidence interval,
1.00-1.58), gynecologic examinations (odds ratio, 1.39;
95% confidence interval, 1.15-1.69), and Papanicolaou
smears (odds ratio, 1.29; 95% confidence interval, 1.04-
1.58). The BMI was not significantly related to delays in
mammography. It was also related to increased physi-
cian visits (P = .001).
Conclusion: Among women, an increased BMI is asso-
ciated with decreased preventive health care services,
which may exacerbate or even account for some of the
increased health risks of obesity.
Arch Fam Med. 1998;7:381-384
O
BESITY IS a major public
health problem in the
United States, with ap-
proximately 35% of
American women classi-
fied as overweight.
1
These women are at an
increased risk for diseases of many organ
systems, including some forms of cancer.
2
Indeed, obese women (body mass index
[BMI], calculated as weight in kilograms di-
vided by the square of height in meters, of
$35) have been shown to have signifi-
cantly higher rates of cervical, endome-
trial, ovarian, gallbladder, and breast can-
cers than nonobese women.
3
Although the
association between weight and certain
forms of cancer among women is strong,
other factors (eg, family history, age, smok-
ing status, and body fat distribution) ap-
pear to play an important role in determin-
ing cancer risk.
4
Nonetheless, the increased
cancer risk associated with obesity under-
scores the importance of preventive health
care examinations for obese women.
The examinations that were investi-
gated in this study (eg, mammography,
breast clinical examination, gynecologic ex-
amination, and Papanicolaou [Pap] smear
test) are critical strategies for the early de-
tection and treatment of some forms of can-
cer among women and form an important
element of the national health promotion
and disease prevention objectives contained
in Healthy People 2000.
5
It is estimated that
regular Pap smear tests followed by appro-
priate treatment could prevent nearly all
deaths from cervical cancer, and screening
mammography followed by timely treat-
ment could reduce the breast cancer mor-
tality by 30% for women aged 50 to 69
years.
6-8
Some evidence exists, however, that
obese women delay or avoid medical and
preventive care services.
9,10
For editorial comment
see page 385
BRIEF REPORT
From the Division of
Gastroenterology, Department
of Medicine, The Johns Hopkins
University School of Medicine,
and The Johns Hopkins
Bayview Medical Center,
Baltimore, Md (Drs Fontaine
and Cheskin); and the Obesity
Research Center,
St Luke’s-Roosevelt Hospital,
Columbia University College
of Physicians and Surgeons,
New York, NY (Drs Faith
and Allison).
ARCH FAM MED/ VOL 7, JULY/ AUG 1998
381
©1998 American Medical Association. All rights reserved.
at Johns Hopkins University, on October 27, 2008 www.archfammed.comDownloaded from
The purpose of this study was to examine the rela-
tion between the BMI and both the frequency of physi-
cian visits and the use of preventive health care services
among a nationally representative sample of women, ad-
justing for age, sociodemographic characteristics, smok-
ing, and health insurance status.
RESULTS
Selected characteristics of respondents are shown in Table
1. The mean age of the women was 46.2 years (median,
42 years), with a range of 18 to 97 years (3 women were
$99 years, but their actual age was not specified). The
mean number of physician visits in the 12 months be-
fore completing the survey was 5.35 (median, 2). Nearly
70% of the women reported having some form of health
insurance.
Of the 3105 women who answered the question
regarding the interval since their last mammography,
2657 indicated that it had been less than 3 years, and
448 reported that it had been more than 3 years. Of
the 4926 women who answered the question about
clinical breast examination, 4162 reported having had
1 less than and 764 more than 3 years before the sur-
vey. For gynecologic examination, of the 5924 women
who answered the question, 4416 reported having had
1 less than and 1508 more than 3 years before the sur-
vey. Finally, of the 6314 women who answered the
question regarding the interval since last Pap smear,
5126 reported having had 1 less than and 1188 more
than 3 years before the survey.
In the linear regression analysis adjusted for age, race,
education, family income, smoking status, and health in-
surance status, the BMI was associated with increased phy-
sician visits in the 12 months before completing the sur-
vey (b = .063, P = .002). That is, for every 1-unit increase
in the BMI, there was, on average, a 0.063-unit increase
in visits to a physician.
The fully adjusted odds ratios for having delayed
mammography, clinical breast examination, gyneco-
logic examination, and Pap smear testing (each for .3
years) as a function of the BMI are shown in
Table 2.
The BMI was directly related with a reduced probability
of having had a clinical breast examination (P = .04), gy-
necologic examination (P = .001), and Pap smear (P = .02)
in the previous 3 years. The BMI was not significantly
related to delays in mammography (P = .20). No evi-
dence of a nonlinear effect for the BMI was seen (ie, the
SUBJECTS AND METHODS
The data sources used were the 1992 National Health In-
terview Survey (NHIS), Cancer Control and Health Insur-
ance supplements, conducted by the National Center for
Health Statistics of the Centers for Disease Control and Pre-
vention, Atlanta, Ga.
11
The NHIS is a continuing annual per-
sonal interview survey of approximately 49 000 house-
holds in the civilian noninstitutionalized US population aged
18 years or older selected through a multistage cluster-
area probability sampling design. Data are collected on per-
sonal, sociodemographic, and health-related information
of family members and unrelated persons living in these
households.
12
The Cancer Control Supplement to the 1992 NHIS was
administered to a randomly selected subsample of 12 035
adult respondents. The response rate was 87%. The survey
ascertained information on health-related knowledge and at-
titudes, as well as the use of health care services. Because
our interest was in characteristics associated with delayed
preventive health care among women, analyses were re-
stricted to female respondents (N = 6981). To include health
insurance status as a covariate in the analyses, subjects’ re-
cords were linked with their corresponding responses in the
Health Insurance Supplement to the 1992 NHIS.
Self-reported weight and height data were used to com-
pute the BMI for each respondent. Self-reported weight is
highly correlated with measured weight.
13-16
The BMI is largely
independent of height (r = − 0.03), strongly related to weight
(r = 0.86), and a reasonable measure of body fatness.
16,17
The covariates examined in these analyses were those
that might influence the use of health care: age, race (non-
white or white), family income, education (years com-
pleted), smoking status (nonsmoker-former smoker or cur-
rent smoker), and health insurance status (not covered or
unknown or covered by private insurance, Medicare, or both).
The outcome measures were the number of physician
visits in the 12 months before completing the survey and the
interval since the most recent use of the following preven-
tive health care procedures: mammography, clinical breast
examination, gynecologic examination, and Pap smear. The
interval since the most recent procedure was defined by the
NHIS as follows: within the past year, 1 to 3 years ago, more
than 3 years ago, unknown specific interval (#3 years vs .3
years), not ascertained or don’t know, or unknown or re-
fused. These responses were recoded and treated as dichoto-
mous variables: 0 indicates 3 years or less; and 1, greater than
3 years. Responses coded as unknown or not ascertained were
treated as missing variables and excluded from the analysis.
Multiple linear regression analysis was used to exam-
ine the relation between the BMI and the number of physi-
cian visits in the 12 months before completing the survey,
adjusting for the previously mentioned covariates. Logistic
regression analyses were used to examine whether the BMI
was related to delaying each of the 4 preventive care proce-
dures after controlling for the influence of the covariates. Af-
ter testing for linear effects (on the logit level), tests for non-
linear effects were conducted by adding the BMI squared to
the model. Finally, because of the heterogeneous age range
of this sample (
Table 1), interactions between the BMI and
age for each outcome variable were also tested.
Data were also tested by treating the BMI categorically
(quintiles). These results did not differ substantially from
analyses that treated the BMI as a continuous variable. There-
fore, given the advantages of using continuous variables in
epidemiological studies,
18
we report analyses that treated the
BMI as continuous. Separate odds ratios were computed for
BMIs of 35 and 40 relative to a BMI of 25 (the sample mean)
to estimate the relative odds of delaying care by obese and
severely obese women, respectively. Variability in sam-
pling associated with the estimated odds ratios was
assessed by 2-sided 95% confidence intervals, with P,.05
(2-tailed) considered statistically significant.
ARCH FAM MED/ VOL 7, JULY/ AUG 1998
382
©1998 American Medical Association. All rights reserved.
at Johns Hopkins University, on October 27, 2008 www.archfammed.comDownloaded from
BMI-squared term was not significant when it was added
to the models), and no significant interactions occurred
between the BMI and age for any of the outcome mea-
sures. Furthermore, there was no independent associa-
tion or trend between race and health care use.
COMMENT
These data support a direct relationship between the BMI
and an increased number of physician visits after age, race,
income, education, smoking status, and health insur-
ance status were taken into account. Thus, as the BMI
increased, the frequency of physician visits also in-
creased. Compared with women of average relative body
weight (BMI of 25), however, obese and severely obese
women were significantly more likely to delay clinical
breast examinations, gynecologic examinations, and Pap
smear testing, suggesting that body weight may play a
role in delaying these forms of preventive health care.
Thus, although obese women visit physicians more fre-
quently and, presumably, are prompted to undergo pre-
ventive services, they also appear to be the least likely to
use these services. Given their increased health risks, the
absolute decrease in preventive services is clearly inap-
propriate for this population.
Obesity may make women less likely to seek pre-
ventive health care, especially for procedures (eg, pelvic
examination) that involve disrobing and manual ma-
nipulation of their bodies.
9,19
This may explain why obe-
sity did not associate with delays in mammography, as
this procedure is less personally invasive and involves less
direct physical contact with a health care practitioner.
Delays in preventive health care might also occur be-
cause obesity may dissuade physicians from suggesting
or performing certain procedures, particularly pelvic ex-
aminations.
9,19
Physicians may be dissuaded by the tech-
nical difficulties associated with performing gyneco-
logic examinations of obese women.
19
Apart from the
difficulties physicians encounter when performing gy-
necologic procedures on obese women, they are also less
likely to perform such procedures on reluctant pa-
tients.
9
Thus, the association between body weight and
reluctance to undergo preventive health care examina-
tions may be mediated by attitudes toward one’s appear-
ance, discomfort with the procedures themselves, or the
nature of the interaction with the clinician.
9,10
Finally,
because obese women are more likely than nonobese
women to make physician visits, preventive services may
be of a lower priority for them.
Given the correlational nature of this study, it can-
not be presumed that the “causal” source of the ob-
served association lies with the women. The delay may
be initiated by the women, health care providers, or some
unmeasured covariate of obesity. For example, perhaps
associated disorders such as arthritis prohibit preven-
tive health care screening. Further investigations using
alternative designs will be fundamental in determining
possible causal pathways.
This study has some limitations. First, weights
and heights were not measured but were self-reported.
Second, respondents were asked to recall the number
of physician visits in the previous 12 months and the
intervals since their most recent use of health care ser-
vices. The accuracy of recall cannot be verified and
provides another possible source of bias. Furthermore,
the reasons for the physician visits are not known.
Finally, because these data were obtained from an
observational study, the associations observed could
be due to residual confounding or confounding from
unmeasured variables.
The effect of obesity on cervical, ovarian, endo-
metrial, and breast cancer mortality among women is
well documented.
3,4
If detected early, however, many
of these cancers might be successfully treated.
6-8
Therefore, screening tests for this high-risk population
are of particular importance. As this study illustrates,
although obese women visit physicians more often
than nonobese women, they appear less likely to use
preventive health care services. Delays in the use of
these services may account for some of the increased
risk of mortality from these cancers.
Table 1. Selected Characteristics of Respondents
Characteristic Respondents, No. Value*
Age, y 6981 46.2 ± 18.6
White race, % 6981 79.9
Weight, kg 6981 69.8 ± 26.5
Height, cm 6964 163.1 ± 7.1
Body mass index, weight (kg)/
[height (m)]
2
6805 25.1 ± 5.4
Physician visits 6947 5.3 ± 14.4
Married, % 6974 50.2
Education level, %
,12 y
23.5
12 y 6960 37.9
.12 y 38.7
Family income, %
,$20 000
42.5
$20 000-$40 000 5745 29.3
.$40 000 28.2
Smoking status, %
Former or nonsmoker
2972
44.4
Current smoker 55.6
Health insurance, %
Covered
6981
67.8
Not covered or unknown 32.2
*
Values are expressed as mean ± SD unless otherwise indicated.
Table 2. Adjusted Odds Ratio for Delaying
Mammography, Clinical Breast Examinations,
Gynecologic Examinations, and Papanicolaou Smears
According to Body Mass Index (BMI)*
Form of Preventive
Health Care
Obese (BMI of 35)
(n = 203)
Severely Obese
(BMI of 40) (n = 135)
Mammography 0.81 (0.59-1.12) 0.73 (0.45-1.19)
Clinical breast examination 1.26 (1.00-1.58)† 1.42 (1.00-1.99)†
Gynecologic examination 1.39 (1.15-1.69)† 1.63 (1.23-2.19)†
Papanicolaou smear 1.29 (1.04-1.58)† 1.46 (1.07-1.98)†
*
Data are given as odds ratio (95% confidence interval). Odds ratios are
relative to the average BMI, calculated as weight (kg)/[height (m)]
2
,ofthe
sample (25.1). Odds ratios have been adjusted for age, race, education,
income, smoking status, and health insurance status.
†
Significantly different from 1.00 (
P,
.05).
ARCH FAM MED/ VOL 7, JULY/ AUG 1998
383
©1998 American Medical Association. All rights reserved.
at Johns Hopkins University, on October 27, 2008 www.archfammed.comDownloaded from
An interesting finding was the discrepant results for
mammography vs breast examinations because these tests
are usually conducted in parallel. No clear answer ex-
ists why obesity was unrelated to delays in mammogra-
phy. Practice or referral services may underlie this dis-
sociation. Mammographies have traditionally only been
done when ordered by a physician. Perhaps once pa-
tients reach this stage of physician referral, their obesity
status no longer acts as a barrier to screening. More-
over, the incidence of patient self-referral for mammog-
raphy may be increasing in certain states. Perhaps re-
moving the physician from the referral process plays a
role in this nonsignificant finding. Further studies will
be needed to address such issues.
Given the association between the BMI and the num-
ber of physician visits, physicians may have a unique op-
portunity to present this information, alleviate fears about
the procedures, and encourage the use of preventive care
services. The already remarkably high prevalence of obe-
sity in the United States seems likely to continue or even
increase in the years to come.
5
Greater emphasis on pre-
ventive services has the potential to reduce the risk of a simi-
lar magnitude of increase in associated health problems.
Accepted for publication August 15, 1997.
This study was supported in part by research grants
DK26687, DK51716-01, R29DK47256-01 A1, and
T32DK37352 from the National Institutes of Health, Na-
tional Institute of Diabetes, Digestive, and Kidney Dis-
eases, Bethesda, Md.
Moonseong Heo, PhD, provided statistical assistance.
Corresponding author: Kevin R. Fontaine, PhD, De-
partment of Medicine, The Johns Hopkins University School
of Medicine, 333 Cassell Dr, Suite 1640, Baltimore, MD
21224-6805 (e-mail: krf@welchlink.welch.jhu.edu).
REFERENCES
1. Kuczmarski RJ, Flegal KM, Campbell SM, Johnson CL. Increasing prevalence of
overweight among US adults: the National Health and Nutrition Examination Sur-
veys, 1960 to 1991.
JAMA.
1994;272:205-211.
2. Allison DB, Pi-Sunyer FX. Obesity treatment: examining the premises.
Endocri-
nol Pract.
1995;1:353-364.
3. Garfinkel L. Overweight and cancer.
Ann Intern Med.
1985;103:1034-1036.
4. Deslypere JP. Obesity and cancer.
Metabolism.
1995;44(suppl):S24-S27.
5. US Dept of Health and Human Services, Public Health Service.
Healthy People
2000: National Health Promotion and Disease Prevention Objectives—Full Re-
port, With Commentary.
Washington, DC: US Dept of Health and Human Ser-
vices, Public Health Service; 1991. DHHS publication PHS 91-50212.
6. US Preventive Services Task Force. Screening for breast cancer.
Am Fam Phy-
sician.
1989;39:89-106.
7. Kerlikowske K, Grady D, Rubin SM, Sandrock C, Ernster VL. Efficacy of screen-
ing mammography: a meta-analysis.
JAMA.
1995;273:149-154.
8. Devesa SS, Young JL, Brinton LA, Fraumeni JF. Recent trends in cervix uteri can-
cer.
Cancer.
1989;64:2184-2190.
9. Adams CH, Smith NJ, Wilbur DC, Grady KE. The relationship of obesity to the
frequency of pelvic examinations: do physician and patient attitudes make a dif-
ference?
Women Health.
1993;20:45-57.
10. Olson CL, Schumaker HD, Yawn BP. Overweight women delay medical care.
Arch
Fam Med.
1994;3:888-892.
11.
1992 National Health Interview Survey Cancer Control Public Use Record.
Hy-
attsville, Md: National Center for Health Statistics; 1994.
12. Massey JT, Moore TF, Parsons VL, Tadros W. Design and estimation for the Na-
tional Health Interview Survey: 1985-1994.
Vital Health Stat 2.
1989;110. DHHS
publication PHS 89-1384.
13. Jeffrey RW. Bias in reported body weight as a function of education, occupa-
tion, health and weight concern.
Addict Behav.
1996;21:217-222.
14. Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among
women.
N Engl J Med.
1995;333:677-685.
15. Heymsfield SB, Allison DB, Heshka S, Pierson RN. Assessment of human body
composition. In: Allison DB, ed.
Handbook of Assessment Methods for Eating
Behaviors and Weight-Related Problems: Measures, Theory, and Research.
San
Diego, Calif: Sage Publications Inc; 1995:515-560.
16. Palta M, Prineas RJ, Berman R, Hannan P. Comparison of self-reported and mea-
sured height and weight.
Am J Epidemiol.
1982;115:223-230.
17. Romieu I, Willett WC, Stampfer MJ, et al. Energy intake and other determinants
of relative weight.
Am J Clin Nutr.
1988;47:406-412.
18. Zhao LT, Kolonel LN. Efficiency loss from categorizing quantitative exposures
into qualitative exposures in case-control studies.
Am J Epidemiol.
1992;136:
464-474.
19. Lubitz RM, Litzelman DK, Dittus RS, Tierney WM. Is obesity a barrier to physi-
cian screening for cervical cancer?
Am J Med.
1995;98:491-496.
ARCH FAM MED/ VOL 7, JULY/ AUG 1998
384
©1998 American Medical Association. All rights reserved.
at Johns Hopkins University, on October 27, 2008 www.archfammed.comDownloaded from
