Results of observational studies performed in the 1960s and 1970s led to the conclusion that use of oral contraceptive agents and hormone replacement therapy should be avoided in women with a history of venous thromboembolism because of an increased risk of recurrent thromboembolism. Estrogen is the purported thrombogenic factor. Douketis and associates analyzed the literature to determine whether such a risk is valid and whether the current practice of not prescribing oral contraceptives or hormone replacement therapy in women with previous venous thromboembolism is justified.

A MEDLINE search was performed and each study was rated according to methodologic quality. The rating scale for methodologic quality was as follows: level 1—four out of four essential methodologic criteria were present and the potential for bias and inaccuracy was low; level 2—three of four essential methodologic criteria were present and the potential for bias and inaccuracy was moderate; level 3—two or fewer essential methodologic criteria were present and the potential for bias and inaccuracy was high. Because of recent concern about the thrombogenic potential of third-generation oral contraceptives, the results of studies with these agents were evaluated separately. A total of 35 studies were included in the analysis.

The authors found that the methodologic quality of the studies investigating the risk of venous thromboembolism in oral contraceptive users and in women receiving hormone replacement therapy was highly variable. No study was rated as demonstrating level 1 evidence; 14 were rated as level 2, and 21 as level 3.

Some of the methodologic errors included diagnostic suspicion bias from previous knowledge of oral contraceptive use, failure to use accurate diagnostic tests to evaluate venous thromboembolism and biased interpretation of findings. The use of subjective and inaccurate diagnostic methods combined with previous knowledge of oral contraceptive use would probably lead to greater diagnostic misclassification in oral contraceptive users than in nonusers and an exaggeration of the risk of thromboembolism with oral contraceptive use.

With respect to the oral contraceptive studies, the pooled risk ratios were 3.0 in case-control studies, 4.8 in retrospective cohort studies, 2.4 in prospective studies and 1.1 in randomized controlled trials. The association between oral contraceptive use and thromboembolism was weaker in studies that used objective diagnostic data than in studies that used subjective data. Mortality studies found no significant association or reported deaths from pulmonary embolism.

The authors examined nine studies of hormone replacement therapy and found that the majority were of such poor methodologic quality that definitive conclusions were precluded. They note that the biologic plausibility of a strong association between hormone replacement therapy and thromboembolism is weak. The risk with hormone therapy would be expected to be lower than that with oral contraceptives because of the physiologic doses used in hormone replacement therapy. It also has not been determined that hormone therapy causes changes in coagulation parameters.

The authors conclude that an association between oral contraceptive use and venous thromboembolism is probably valid, but the magnitude of the risk is likely exaggerated. Results of the pooled analysis demonstrate that avoidance of oral contraceptive use in women with a history of venous thromboembolism is prudent. Women with known thrombophilic status should avoid both oral contraceptive agents and hormone replacement therapy, whether or not they have previously had venous thromboembolism. Because of the paucity of data on an association between hormone therapy and venous thromboembolism, the authors believe more studies are needed before definitive conclusions can be made.

editor's note: The recommendation that third-generation oral contraceptive agents be avoided was based on the findings of three epidemiologic case-control studies that linked the use of agents containing desogestrel or gestodene with an increased risk of nonfatal venous thromboembolism, as compared with the risk associated with oral contraceptives that contain levonorgestrel. The increase was 10 to 15 cases of venous thromboembolism per 100,000 in users of oral contraceptive agents that contain levonorgestrel and 20 to 30 cases per 100,000 in women taking oral contraceptives that contain gestodene or desogestrel. A fourth study, which was not part of the original investigation that led to the recommendation that third-generation agents be avoided, evaluated 12 oral contraceptive formulations and demonstrated no risk differences for venous thromboembolism among the various agents, including those containing gestodene and desogestrel. Epidemiologic studies have demonstrated that the risk of thromboembolism is related to the estrogen component, because the progestational agent has little effect on clotting parameters and is not thrombophilic. For this reason, the biologic plausibility of the results of the three case-control studies is weak.—b.a.