Confirming the findings that touched off the 1995 "pill scare," an analysis of a large British database shows that use of oral contraceptives containing a "third-generation" progestogen (desogestrel or gestodene) is associated with a higher risk of venous thromboembolism than use of pills containing levonorgestrel.1 As further evidence of this association, the researchers found that subsequent to the pill scare, when British women's use of third-generation formulations dropped sharply, the number of venous thromboembolisms occurring among oral contraceptive users was lower than would have been expected if use had not changed.
Since 1995, when a British advisory committee warned that third-generation pills are a poor choice for women at risk of venous thromboembolism, the question of whether these pills pose a greater risk than earlier versions has been the subject of intense controversy, and research on the issue has produced conflicting results. However, this analysis overcomes a number of methodological problems of previous work and has been called perhaps "the most important paper yet published on this vexed subject."2
The researchers used Britain's General Practice Research Database to identify women aged 15-39 who received pill prescriptions between January 1993 and December 1999. In all, the analyses include nearly 400,000 person-years of oral contraceptive use, roughly evenly divided between pills containing levonorgestrel and those containing third-generation pro-gestogens. The investigators conducted a cohort analysis to compare the risk of venous thromboembolism in the periods leading up to the pill scare (January 1993 to October 1995) and immediately thereafter (January 1996 to December 1999); they also conducted a case- control analysis to compare the risks associated with the different types of pills.
In the years preceding the pill scare, third-generation pills accounted for 63% of oral contraceptive use among British women. In the later period, however, only 18% of pill use involved these preparations. The decline occurred in all age-groups but was especially steep among younger women. For example, 82% of teenage pill users in 1993-1995 took third-generation pills, compared with 11% in 1996-1999; among women in their late 30s, by contrast, the proportions were 56% and 18%, respectively.
A total of 106 women in the study population developed venous thromboembolism--42 users of oral contraceptives containing levonorgestrel and 64 users of third-generation pills. Seventy-one of these women were in the earlier cohort and 35 in the later one. During both periods, the crude incidence of venous thromboembolism was higher among women using third-generation pills (37-41 cases per 100,000 person-years) than among those using pills with levonorgestrel (20-23 per 100,000).
After adjusting the data for women's age, the analysts calculated incidence ratios to compare the risks by pill type and by cohort. They found that for the two periods combined, the risk of venous thromboembolism was twice as great for women taking third-generation pills as for those using pills containing levonorgestrel (incidence ratio, 1.9). Furthermore, the incidence of venous thromboembolism associated with each type of pill was the same in both periods.
In additional analyses, the researchers calculated the number of venous thromboembolisms that would have been expected in 1996-1999 if the age-specific distribution of types of pill used had not changed. These results show that in the absence of the shift to levonorgestrel, 44 pill users would have developed venous thromboembolism--nine more than actually did.
For the case-control analysis, each woman who had venous thromboembolism was matched with up to six women who were the same age, went to the same physician and were using the pill at the time that the woman with venous thromboembolism received her diagnosis; data were available for 569 controls. The researchers performed conditional logistic regression analyses to assess the risks associated with each type of pill, adjusting for body mass index, smoking, duration of pill use and switching of pill types.
In these analyses, women taking third-generation pills had twice the odds of venous thromboembolism of those using pills containing levonorgestrel (odds ratio, 2.3). According to the researchers, the difference between the results in the cohort and case-control analyses is attributable to the additional adjustments for confounding factors in the latter.
The results were similar in the years preceding the pill scare (2.2) and immediately thereafter (2.8). They also were similar for pills containing gestodene (1.9) and those with desogestrel (2.0-2.8, depending on the estrogen dose). High body mass and smoking were associated with increased risks of venous thromboembolism, and the analysts found evidence that doctors may have taken these factors into account when deciding which type of pill to prescribe after 1995.
As the researchers note, a number of studies have yielded conflicting results regarding the association between venous thromboembolism and the use of third-generation oral contraceptives. In particular, a widely publicized analysis based on the same database used in this study found no difference in the incidence of the condition before and after the pill scare, leading to the conclusion that third-generation pills were associated with no higher risk than pills containing levonorgestrel.3 However, the researchers contend that several features of their methodology improve on the approach of the earlier study: For example, they restricted the analysis to third-generation pills and formulations containing levonorgestrel (rather than including all combined oral contraceptives), and they more adequately controlled for important confounding factors.
A commentary accompanying the study concurs about the strength of this study. "As well as answering the previous report," the commentator writes, "it provides vital evidence on several controversial matters."4--D. Hollander
1. Jick H et al., Risk of venous thromboembolism among users of third generation oral contraceptives compared with users of oral contraceptives with levonorgestrel before and after 1995: cohort and case-control analysis, BMJ, 2000, 321(7270):1190-1195.
2. Skegg DCG, Pitfalls of pharmacoepidemiology: oral contraceptive studies show a need for caution with databases, BMJ, 2000, 321(7270):1171-1172.
3. Farmer RDT et al., Effect of 1995 pill scare on rates of venous thromboembolism among women taking combined oral contraceptives: analysis of General Practice Research database, BMJ, 2000, 321(7259):477-479.
4. Skegg DCG, 2000, op. cit. (see reference 2).