Ovarian Cancer Risk Factors in a Defined Population Using Rare Event Logistic Regression

Background: this study evaluated the role of family history of cancer andgynecologic factors in relation to the etiology of ovarian cancer in a low socioeconomicpopulation in iran.methods: from 2007-2009 we conducted a screening program on women withinsurance coverage provided by the imam khomeini relief foundation. a total of 26788women participated in this study of whom 76 cases had ovarian cancer and 26712 wereconsidered as controls. we used rare event logistic (relogit) regression analysis witha prior correction method that used the zelig package in r to obtain odds ratioestimates and confidence intervals.results: ovarian cancer was more frequent among postmenopausal thanpremenopausal (odds ratio: 2.30; confidence interval: 1.17-4.49) women. we observedincreased risk for this disease in women with histories of hormone replacement therapycompared to those with no history (odds ratio: 2.36; confidence interval: 1.13-4.91). agreater increase in ovarian cancer was observed in women with family histories of breast(odds ratio: 2.88; confidence interval: 1.44-5.77), ovarian (odds ratio: 11.27; confidenceinterval: 5.63-22.54) and all cancer sites (odds ratio: 2.95; confidence interval: 1.71-5.08).however, the use of oral contraceptive pills was significantly associated with lower riskfor ovarian cancer (odds ratio: 0.47; confidence interval: 0.28-0.79). there was noassociation between ovarian cancer and age, marital status, occupation, education level,age at menarche, age at first pregnancy and number of pregnancies.conclusion: ovarian cancer was considered a rare event. thus we deemed itnecessary to explore the associated risk factors using relogit with a prior correctionmethod. the risk factors for ovarian cancer were menopause, history of hormonereplacement therapy and family history of cancer of the breast, ovaries and other sites.oral use of contraceptive pills showed a protective effect on risk for ovarian cancer.