Am J Epidemiol 2003; 158:947-950.
Copyright © 2003 by Johns
Hopkins Bloomberg School of Public Health
ORIGINAL CONTRIBUTIONS |
Potential Bias due to Excluding Oral Contraceptive Users When Estimating Menstrual Cycle Characteristics
1 Department of Biostatistics and Research Epidemiology, Henry Ford Health System, Detroit, MI.
2 Epidemiology Branch, National Institute of Environmental Health Sciences, National Institutes of Health, US Department of Health and Human Services, Research Triangle Park, NC.
Women take oral contraceptives for contraception but also for menstrual dysfunction treatment. This raises the question of whether or not women with menstrual dysfunction are underrepresented in analyses of menstrual function because oral contraceptive users are excluded. To explore this, the authors examined the history of oral contraceptive use among 1,322 Black women and White women, aged 35–49 years, who had been randomly selected from a large health plan’s membership in Washington, DC, between 1996 and 1999. The women reported whether they took oral contraceptives during their teens, twenties, and thirties, and if so, the reason they took them (prevent pregnancy, medical problem, or both). They also reported their usual menstrual cycle length when not using oral contraceptives during these decades. The prevalence of oral contraceptive use strictly for medical problems was low for both Black women and White women (4–9% of women), and the distributions of usual cycle length were similar for women who did and did not take oral contraceptives. Thus, there was little evidence of substantial bias of estimates of cycle characteristics caused by excluding oral contraceptive users from analyses of menstrual function. However, our data indicate that, with only a few additional questions, information on usual menstrual cycle characteristics can be collected and used to evaluate bias in any given study.
bias (epidemiology); contraceptives, oral; menstrual cycle