Abstract

PIP: New studies linking use of oral contraceptives (OCs) containing a third generation progestogen with venous thromboembolism (VTE) have stirred controversy. Epidemiologists face the task of separating a direct effect of the OC from the confounding effect of a pre-existing factor (e.g., smoking). Only 1-2% of VTE cases are fatal. The percentage of VTE cases that are diagnosed and treated in the hospital is low and falling. A major challenge in the epidemiologic study of non-fatal conditions is that clinical and social factors will also affect whether or when or where the diagnosis is made. The newest studies of OCs and VTE were based on cases diagnosed and treated in the hospital. They found the risks for non-fatal VTE among OC users to be about 2-4 times higher than those for non-users and 1.5-2 times higher for those using levonorgestrel-containing OCs. In the past, laboratory and epidemiologic studies have pointed to the estrogen dose as the main risk factor for VTE and thus the lower estrogen dose OCs are less likely to contribute to VTE than the older higher estrogen dose OCs. Yet the new studies found an increased risk of VTE with lower doses of estrogen. There has been consistency in the findings, leading some to conclude that causation explains the findings. A re-analysis of the data from these studies found that the highest relative risks occurred with the recently introduced desogestrel-containing OC that has only 20 mcg ethinyl estradiol. Yet it would have been expected to be safer than the other OCs. The authors concluded that clinicians may have assumed that each new OC should be safer, and thus they have been selectively prescribing them to higher risk women. Since the association between the third generation OCs and VTE risk is weak and there is no biologic explanation for the association, one should not consider the association causal. Physicians should consider all the evidence before changing OC prescription practices. Clinical actions based on weak associations undermine their credibility and that of epidemiologists.