Comparison of Pooled Risk Estimates for Adverse Effects from Different Observational Study Designs: Methodological Overview
Although hormone replacement therapy (HRT) provides health benefits, concerns about harmful effects have been raised, including an increased risk of venous thromboembolism. The purpose of this review is to assess the risk for venous thromboembolism associated with HRT. Data searches involved the MEDLINE database from January 1982 to December 1997, and bibliographies of retrieved articles. Seven case-control studies and 1 prospective cohort study considering venous thromboembolism as the outcome of interest were selected and rated according to methodological criteria. Results of case-controls studies were pooled to determine an overall estimate. The pooled risk ratio (95% confidence interval) for venous thromboembolism among current users of HRT compared with non-users (never-users and past-users combined) was 2.1 (1.4 to 3.0) in the case-control studies (n = 7). Based on one prospective cohort study, the risk of primary pulmonary embolism associated with HRT was 2.1 (1.2 to 3.8). No association was found with past use of HRT. The risk of venous thromboembolism was higher within the first year of treatment. There was no strong evidence for a dose-response relationship with estrogen. No striking difference in risk of venous thromboembolism was observed between unopposed and opposed regimens. The small number of women using transdermal estrogen therapy precluded any firm conclusions about the impact of the route of estrogen administration. Current use of oral estrogen as HRT is associated with a 2- to 3-fold increased risk of venous thromboembolism. However, the absolute risk remains low, and these findings need to be weighed against the potential benefits of treatment. Whether transdermal estrogen therapy is safe with respect to venous thromboembolism has yet to be adequately investigated.