Aaron V. Sapp, MD

BACKGROUND: In the middle of the last decade several studies suggested an increased risk of thromboembolic events in women using the newer generation of oral contraceptives (OCs). These pills contain the newer progestins desogestrel, gestodene, and norgestimate. The authors of this meta-analysis attempted to address possible confounders and biases in these past studies and quantify the strength of the association.

POPULATION STUDIED: The meta-analysis included 9 case-control studies involving 3417 cases of thromboembolic disease with 9600 controls, as well as 3 cohort studies of more than 1 million women. Reported ages ranged from 15 to 49 years. All but one study (with 471 cases and 1772 controls) reported exclusion of women with a previous history of thromboembolism. Settings included hospitals, primary care offices and specialty clinics, primarily in Europe. Only studies comparing third-generation OCs to second-generation OCs were included in the final analyses.

STUDY DESIGN AND VALIDITY: The selected cohort or case-control articles were required to define cases as women with venous thrombosis or thromboembolism, and they had to contain sufficient information to determine relative risk and confidence intervals. The data were all obtained before November 1995. The study addressed possible heterogeneity among studies by stratified analysis of the results.

OUTCOMES MEASURED: The main outcome was the onset of venous embolism, comparing third-generation OC users to second-generation OC users. Because no randomized trial data were available, the investigators used odds ratios as an approximation of relative risk.

RESULTS: The study arrived at an overall adjusted odds ratio of 1.7 (95% confidence interval [CI], 1.4-2.0), a 70% higher association with thromboembolism for third-generation OCs compared with second-generation OCs. The highest odds ratio, 3.1 (95% CI, 2.0-4.6), was among first-time users. For short-term users (less than 1 year) the odds ratio was 2.5 (95% CI, 1.6-4.1), compared with 2.0 (95% CI, 1.4-2.7) for use longer than 1 year. Studies funded by the pharmaceutical industry had an odds ratio of 1.3 (95% CI, 1.0-1.7), while independent studies had an odds ratio of 2.3 (95% CI, 1.7-3.2). Age and confirmation of thromboembolism by vascular studies did not affect the odds ratios. These numbers translate to 1.5 additional incidents of thromboembolic disease per 10,000 woman years, approximately 4 additional deaths per 1,000,000 woman years. In other words, 25,000 women have to be treated over 10 years with third-generation OCs instead of second-generation OCs to expect one additional death.

BACKGROUND: In the middle of the last decade several studies suggested an increased risk of thromboembolic events in women using the newer generation of oral contraceptives (OCs). These pills contain the newer progestins desogestrel, gestodene, and norgestimate. The authors of this meta-analysis attempted to address possible confounders and biases in these past studies and quantify the strength of the association.

POPULATION STUDIED: The meta-analysis included 9 case-control studies involving 3417 cases of thromboembolic disease with 9600 controls, as well as 3 cohort studies of more than 1 million women. Reported ages ranged from 15 to 49 years. All but one study (with 471 cases and 1772 controls) reported exclusion of women with a previous history of thromboembolism. Settings included hospitals, primary care offices and specialty clinics, primarily in Europe. Only studies comparing third-generation OCs to second-generation OCs were included in the final analyses.

STUDY DESIGN AND VALIDITY: The selected cohort or case-control articles were required to define cases as women with venous thrombosis or thromboembolism, and they had to contain sufficient information to determine relative risk and confidence intervals. The data were all obtained before November 1995. The study addressed possible heterogeneity among studies by stratified analysis of the results.

OUTCOMES MEASURED: The main outcome was the onset of venous embolism, comparing third-generation OC users to second-generation OC users. Because no randomized trial data were available, the investigators used odds ratios as an approximation of relative risk.

RESULTS: The study arrived at an overall adjusted odds ratio of 1.7 (95% confidence interval [CI], 1.4-2.0), a 70% higher association with thromboembolism for third-generation OCs compared with second-generation OCs. The highest odds ratio, 3.1 (95% CI, 2.0-4.6), was among first-time users. For short-term users (less than 1 year) the odds ratio was 2.5 (95% CI, 1.6-4.1), compared with 2.0 (95% CI, 1.4-2.7) for use longer than 1 year. Studies funded by the pharmaceutical industry had an odds ratio of 1.3 (95% CI, 1.0-1.7), while independent studies had an odds ratio of 2.3 (95% CI, 1.7-3.2). Age and confirmation of thromboembolism by vascular studies did not affect the odds ratios. These numbers translate to 1.5 additional incidents of thromboembolic disease per 10,000 woman years, approximately 4 additional deaths per 1,000,000 woman years. In other words, 25,000 women have to be treated over 10 years with third-generation OCs instead of second-generation OCs to expect one additional death.

BACKGROUND: In the middle of the last decade several studies suggested an increased risk of thromboembolic events in women using the newer generation of oral contraceptives (OCs). These pills contain the newer progestins desogestrel, gestodene, and norgestimate. The authors of this meta-analysis attempted to address possible confounders and biases in these past studies and quantify the strength of the association.

POPULATION STUDIED: The meta-analysis included 9 case-control studies involving 3417 cases of thromboembolic disease with 9600 controls, as well as 3 cohort studies of more than 1 million women. Reported ages ranged from 15 to 49 years. All but one study (with 471 cases and 1772 controls) reported exclusion of women with a previous history of thromboembolism. Settings included hospitals, primary care offices and specialty clinics, primarily in Europe. Only studies comparing third-generation OCs to second-generation OCs were included in the final analyses.

STUDY DESIGN AND VALIDITY: The selected cohort or case-control articles were required to define cases as women with venous thrombosis or thromboembolism, and they had to contain sufficient information to determine relative risk and confidence intervals. The data were all obtained before November 1995. The study addressed possible heterogeneity among studies by stratified analysis of the results.

OUTCOMES MEASURED: The main outcome was the onset of venous embolism, comparing third-generation OC users to second-generation OC users. Because no randomized trial data were available, the investigators used odds ratios as an approximation of relative risk.

RESULTS: The study arrived at an overall adjusted odds ratio of 1.7 (95% confidence interval [CI], 1.4-2.0), a 70% higher association with thromboembolism for third-generation OCs compared with second-generation OCs. The highest odds ratio, 3.1 (95% CI, 2.0-4.6), was among first-time users. For short-term users (less than 1 year) the odds ratio was 2.5 (95% CI, 1.6-4.1), compared with 2.0 (95% CI, 1.4-2.7) for use longer than 1 year. Studies funded by the pharmaceutical industry had an odds ratio of 1.3 (95% CI, 1.0-1.7), while independent studies had an odds ratio of 2.3 (95% CI, 1.7-3.2). Age and confirmation of thromboembolism by vascular studies did not affect the odds ratios. These numbers translate to 1.5 additional incidents of thromboembolic disease per 10,000 woman years, approximately 4 additional deaths per 1,000,000 woman years. In other words, 25,000 women have to be treated over 10 years with third-generation OCs instead of second-generation OCs to expect one additional death.