Adjusted for age, education, income, marital status, health-related conditions (diabetes, joint symptoms, low back pain, headache/migraine, oral health), health status and limitation of activity (special equipment, limitations in functional activities), health behaviors (leisure time physical activity, daily activities), health care access and utilization (health care provider contacts, immunizations, and HIV test), cancer screening (skin exams), selected menstrual and reproductive information (live birth and birth control pill use), health insurance, and risk of getting cancer in the future.
BMI and Cervical Cancer Screening among White, African-American, and Hispanic Women in the United States
Version of Record online: 6 SEP 2012
2006 North American Association for the Study of Obesity (NAASO)
Volume 14, Issue 3, pages 526–527, March 2006
How to Cite
Lipnick, R. J. and Kao, T.-C. (2006), BMI and Cervical Cancer Screening among White, African-American, and Hispanic Women in the United States. Obesity, 14: 526–527. doi: 10.1038/oby.2006.68
- Issue online: 6 SEP 2012
- Version of Record online: 6 SEP 2012
In their study, Wee et al. (1) used data from the 2000 National Health Interview Survey to examine the relationship between BMI and cervical cancer screening. They found that severely obese white women (BMI ≥ 40 kg/m2) were less likely to undergo Pap smear screening (relative risk, 0.92; 95% confidence interval, 0.83 to 0.99) compared with white women of normal weight, and that BMI was not associated with screening in African Americans or Hispanics. The article did not clearly state whether women with a history of cervical cancer or women with a Pap smear test for diagnostic purposes were excluded from the study. It is also unclear what the rationale was for the selection of variables adjusted in the models from many variables available in the data.
Using the same data, we conducted similar analyses. Our analyses included 7839 white and 2071 African-American women, ages 18 to 75 years old (the same age range as that in the study by Wee et al.). The same measure used for cervical cancer screening, the utilization of a Pap smear during the previous 3 years, was used. However, we excluded women who had a history of cervical cancer and women who had had a Pap smear for a reason other than routine physical or pregnancy examination, in addition to excluding women with a history of hysterectomy. In the logistic analyses, demographic variables were always included in the model because of their possible complex effects on study results. Other variables were considered as potential confounders based on their relationship with Pap smear screening and with BMI in the women who had had a Pap smear in the past 3 years. Based on the full model with the potential variables, the variables, without changing the odds ratio for BMI substantially (<10%), were excluded using a backward elimination strategy to increase the precision of the model. The SAS-callable SUDAAN 9.01 (RTI, Inc., Research Triangle Park, NC) was used for this analysis. We also converted odds ratio to relative risk using the same approximation method used by Wee et al. Instead of finding an association in white women as shown in the study of Wee et al., we found that African-American women with moderate obesity were about twice as likely not to have had a Pap smear screening in the previous 3 years compared with African-American women with normal BMI (relative risk, 1.93; 95% confidence interval, 1.03 to 3.01; Table 1). No associations were found between BMI and screening for white women.
|BMI (kg/m2)||Women with Pap smear||Women without Pap smear||Relative risk||95% CI||Women with Pap smear||Women without Pap smear||Relative risk||95% CI||Women with Pap smear||Women without Pap smear||Relative risk||95% CI|
|<18.5||237||58||0.86||0.58 to 1.23||210||51||0.87||0.59 to 1.27||27||7||0.26||0.04 to 1.40|
|18.5 to 24.9||3709||857||1.00||Reference||3207||731||1.00||Reference||502||126||1.00||Reference|
|25 to 29.9||2007||408||0.98||0.83 to 1.15||1458||311||0.94||0.79 to 1.13||549||97||1.07||0.73 to 1.50|
|30 to 34.9||950||226||1.15||0.95 to 1.39||644||166||1.13||0.90 to 1.39||306||60||1.18||0.76 to 1.75|
|35 to 39.9||394||91||1.15||0.85 to 1.52||271||59||1.01||0.72 to 1.37||123||32||1.93||1.03 to 3.01|
|≥40||258||81||1.24||0.92 to 1.63||147||56||1.27||0.89 to 1.77||111||25||1.29||0.70 to 2.11|
Our analyses suggest that study results may vary depending on which study subjects are included and how covariables in the models are selected.