Oral contraceptives and venous thromboembolism: a systematic review and meta-analysis.

BACKGROUND: An association between oral contraceptive (OC) use and venous thromboembolism (VTE) has long been recognized. However, no summary estimates of the increase in VTE risk associated with OC use have been available since 1995, and no meta-analyses have evaluated the VTE risk of new preparations containing drospirenone. OBJECTIVE: The aim of the study was to carry out a meta-analysis to summarize existing evidence on the association between venous VTE and OC use, and to investigate how such an association may vary according to the type of OC, OC user characteristics, study characteristics and biases. METHODS: Relevant cohort or case-control studies were searched in MEDLINE and other electronic databases up to May 2010, with no language restriction. Data were combined using a generic inverse-variance approach. Meta-regression in addition to stratification was used to explore potential predictors of the summary estimate of risk. RESULTS: Sixteen cohort and 39 case-control studies were included in at least one comparison. Overall, the odds ratio (OR) of OC users versus non-users was 3.41 (95% CI 2.98, 3.92). This estimate was based upon nine cohort studies evaluating approximately 12 000 000 person-years, and 23 case-control studies including approximately 45 000 women. VTE risk for OC users was significantly lower in studies evaluating 'all VTE cases' than in those evaluating 'idiopathic VTE only' (OR 3.09 and 4.94, respectively). Among the carriers of genetic mutations G20210A and Factor V Leiden (FVL), OC users showed a significantly increased VTE risk compared with non-users (OR 1.63; 95% CI 1.01, 2.65, and OR 1.80; 95% CI 1.20, 2.71, respectively). When the newest OCs containing drospirenone were compared with non-drospirenone-containing OCs (except those containing levonorgestrel only), VTE risk did not significantly increase (OR 1.13; 95% CI 0.94, 1.35). CONCLUSIONS: This meta-analysis confirms that OC use significantly increases VTE risk. The strength of this association, however, varies according to the generation of OC, type of outcome and presence of a genetic mutation, with ORs ranging from 3 to 5