Increased risk of thrombosis due to oral contraceptives: a further report.

In a previously reported case-control study of the relationship between oral contraceptives and thromboembolism, there were 461 cases and 1302 controls, individually matched on age, race, marital status, hospital, and date of admission. Initially, the control patients had not been matched with the cases for the presence or absence of six factors thought to predispose to or precipitate thromboembolic disease. The present paper reports the effects of taking into consideration these factors in the controls. Two methods of analysis (matched set, and logistic regression) gave closely similar results. Where the case series consisted of idiopathic cases, the revised estimate of the relative risk was reduced from 7.2 to 4.7 by these procedures; for predisposed cases, it was increased from 1.2 to 2.2. The explanation suggested in the previous report for the failure to find an increased risk for cases with predisposition receives support from these findings. Variation in the relative risk was examined for four separate diagnostic categories: venous thrombosis alone, pulmonary embolism alone, venous thrombosis and pulmonary embolism together, and myocardial infarction. The relative risk estimates were greater than unity for each thrombosis category for both predisposed and non-predisposed cases. The relative risk was not found to vary significantly according to age or smoking status.