Screening for Cervical and Breast Cancer: Is Obesity an Unrecognized Barrier to Preventive Care?

Obese women have higher mortality rates for breast and cervical cancer than do thinner women (1, 2). Obesity also has important social, economic, and psychological consequences, including societal discrimination and poor self-perception (3-8). Whether these consequences influence the quality of medical care received by obese patients is unclear. Recent studies suggest that obese women receive preventive services, such as Papanicolaou (Pap) smears and clinical breast examinations, less often than normal-weight women (9, 10). Other studies suggest that physicians and other health care providers have negative attitudes toward and biases against obese patients, which may explain some of the disparities in care (11-14). Obese women also seem to have poorer self-esteem and body images than their thinner counterparts (5, 6). Poor self-concept may be influenced by ethnicity (6-8, 15). White women are more likely than black women of similar weight to perceive themselves as overweight (6), and this poor self-perception may influence attitudes toward screening with Pap smears and mammography. Because obesity is associated with higher mortality rates for cardiovascular disease and cancer of the cervix, breast, and colon, barriers to preventive screening and counseling in obese patients can have dire medical and economic consequences (2, 16, 17). In particular, barriers to Pap smears and mammography may contribute to the more than 50 000 deaths attributed to cervical and breast cancer each year (18). We used data from a nationally representative sample to examine screening with Pap smears and mammography among overweight and obese women. Methods Data Source The National Health Interview Survey is a continuing, in-person household survey of the civilian, noninstitutionalized U.S. population that is conducted by the Census Bureau for the National Center for Health Statistics (19). In 1994, the overall response rate was 94%. Approximately 116 000 persons (including children) from approximately 46 000 households responded to the core survey, which elicited information on sociodemographic factors, insurance coverage, basic health status, number of days spent hospitalized or home in bed, height, and weight. Respondents were also asked whether they had a usual place for health care, whether they had a usual provider, and what the provider's specialty was. In addition, a supplemental survey (Year 2000 Supplement) was administered to one randomly selected adult, 18 years of age or older, from 50% of the responding households (n =19 738). Respondents were queried about use of preventive health services, including Pap smears and mammography. Women were asked, About how long has it been since your last Pap smear test? and How long ago has it been since you had a mammogram? The response rate for the supplement was 88% (11 435 respondents). Our use of the National Health Interview Survey database was approved by the Committee on Clinical Investigations at Beth Israel Deaconess Medical Center, Boston, Massachusetts. Use of Papanicolaou Smears and Mammography Women 18 to 75 years of age who had not had a hysterectomy were considered eligible for analysis of screening with Pap smears. According to generally accepted guidelines (18), women who reported having a Pap smear in the previous 3 years were classified as having been screened. Women 50 to 75 years of age were considered eligible for analysis of breast cancer screening; those who reported having mammography in the previous 2 years were classified as having been screened (18). We limited our sample to women 50 to 75 years of age because mammography reduces mortality rates for breast cancer by 20% to 39% in this age group (20). Factors of Interest We defined our main variable of interest, body mass index (BMI), as body weight in kilograms divided by height in meters squared. We hypothesized that women with higher BMIs would be less likely to be screened. However, because we were not certain that a linear or dose-response effect existed, we used published definitions to classify women as underweight (BMI<18.5 kg/m2), normal weight (BMI, 18.5 to<25 kg/m2), overweight (BMI, 25 to<30 kg/m2), or obese (class I [BMI, 30 to<35 kg/m2], class II [BMI, 35 to<40 kg/m2], or class III [BMI 40 kg/m2]) (21). Women who were underweight made up approximately 3% of the sample and were included in all analyses; however, for reasons of simplicity, we did not report their results. We considered several factors thought to be highly correlated with preventive care as potential confounders in our analyses (22, 23). These included such sociodemographic factors as age (continuous variable), ethnicity/race (white, black, Hispanic, other), marital status (married, never married, divorced or widowed), education (less than high school, high school graduate, some college, graduate of a 4-year college or more), annual income (<$15 000, $15 000 to<$20 000, $20 000 to<$25 000, $25 000 to 50 000,>$50 000), insurance type (managed care, fee-for-service, Medicare, Medicaid, other, uninsured), and region of the United States (Northeast, Midwest, South, West). We adjusted for illness burden by using available surrogate markers, such as self-reported health status (excellent, very good, good, fair, poor), number of days hospitalized in the previous year (0, 1 to 7,>7), number of days spent in bed (0, 1 to 7, 8 to 30,>30), and number of visits to a physician (0, 1 to 2, 3 to 4,>4). In addition, we accounted for the specialty of the usual provider and for having a usual place to receive medical care (general internist or family practitioner, gynecologist, other specialist, other provider, no usual provider but usual place for care, no usual place for care). Statistical Analysis Using Wald chi-square statistics, we performed bivariable analyses to separately characterize factors associated with use of Pap smears, use of mammography, and BMI. Two-tailed P values less than or equal to 0.05 were considered statistically significant. To examine the relation between BMI and Pap smears or mammography, we built a series of multivariable logistic regression models for each separate outcome. Normal-weight women served as the reference. First, we developed an unadjusted model that included only categories of BMI as the independent variable. We then developed a fully adjusted model that controlled for factors previously shown or believed to affect screening: sociodemographic variables (except income), insurance type, illness burden, and provider specialty (22, 23). We performed several secondary analyses to examine the stability of our findings. First, because information on income was unavailable for more than 20% of respondents, we examined the confounding effect of income in a smaller group of women who had complete data. Second, to address the appropriateness of adjusting for variables that are potential intermediary factors and not true confounders of BMI, we developed a series of models by removing one factor at a time from the fully adjusted model and examined the confounding effect of that factor on BMI. For example, obesity may pose a barrier to care because obese women are denied health insurance as a result of discrimination; therefore, adjusting for insurance in this case may inappropriately mask differences in use of preventive care according to BMI. Third, to explore the potential effect of recall bias, we performed an analysis that adjusted for time since the respondents' last general physical examination (when Pap smears and mammography were most likely to have been performed). Finally, to test the hypothesis that the effect of BMI on use of preventive care may differ according to ethnicity/race, we introduced interaction terms between ethnicity/race and BMI in a subset of black and white women. All analyses used SAS-callable SUDAAN software, version 7.5 (Research Triangle Institute, Research Triangle Park, North Carolina), to obtain proper variance estimations that accounted for the complex sampling design (24). To reflect U.S. population estimates, results were weighted to adjust for nonresponse. We used Taylor series linearization to estimate standard errors (24, 25). For adjusted analyses, we converted the odds ratios into standardized risks or rates (weighted to the U.S. population) and subtracted the adjusted rates of normal-weight women from the rates of overweight and obese women to arrive at adjusted rate differences. Confidence intervals were computed by using the method of Flanders and Rhodes (26), adjusted for the complex design. Results Obesity and Papanicolaou Smears Of the 8394 women who were eligible for Pap smear analysis, 7857 had complete data on height, weight, and performance of Pap smears. More than 50% of these women had normal BMIs (Table 1). Overweight women (BMI, 25 to<30 kg/m2) and obese women (BMI 30 kg/m2) reported significantly lower rates of screening with Pap smears in the previous 3 years than did normal-weight women (78% and 78% compared with 84%, respectively; P <0.001). Heavier women were usually older, were less likely to be white or to have private health insurance, and had lower socioeconomic status. They reported a greater illness burden and were more likely to receive their usual health care from general internists and family practitioners than from gynecologists. Table 1. Characteristics of Women 18 to 75 Years of Age without Hysterectomy Who Were Eligible for Analysis of Papanicolaou Smears Table 2 shows the adjusted rates and rate differences for screening with Pap smears, according to BMI, for women who had complete data on all covariates (n =7405). Adjusted results were similar to unadjusted rates. Overweight and obese women reported similar rates of screening. However, these rates were significantly lower than those among normal-weight women, even after we controlled for sociodemographic factors, health insurance and access to care, illness burden, and provider specialty. Com