Univariate and multivariate analysis of risk factors for ovarian cancer: case-control study, Mexico City.

Female residents of the southern part of Mexico City diagnosed with epithelial ovarian cancer between 1989 and 1992 (n = 172) were interviewed regarding reproductive events. The same information was obtained from the controls (n = 441). There was an elevated risk for ovarian cancer associated with an increased number of abortions, relative risk = 3.66 (95% confidence interval = 1.02 - 13.45) for women with four fetal losses. Ovarian cancer risk was shown to decrease as parity increased, reaching a relative risk of 0.23 (95% confidence interval = 0.09 - 0.55) in women with four or more live births in the postmenopausal group. This negative association seems less evident in overall cases. Logistic regression methods were used and abortions and high estimated number of ovulatory cycles were found to be associated with an increased risk of ovarian cancer. The inclusion of nonreproductive variables such as educational status, in the logistic model, showed that cases had a lower educational level than that of controls.