Use of Combined Oral Contraceptives and Risk of Venous Thromboembolism: A Nested Case-control Analysis in German Claims Data

Background: Combined oral contraceptives (COC) are known to increase the risk of venous thromboembolism (VTE) but data on VTE risk of several 3rd and 4th generation progestogens is lacking.

Objectives: To compare the risk of VTE for 9 COCs including progestogens with as yet unclear risk of VTE.

Methods: Based on health claims data from the German Pharmacoepidemiological Research database (GePaRD) (~20% of the German population) we conducted a case-control study nested in a cohort of new COC users. The cohort included all girls/young women between 10-19 years (age at which COCs are reimbursable in Germany) with ≥1 COC dispensing between 2005 and 2017 after 1 year without such a dispensing. VTE was defined as a hospital diagnosis of pulmonary embolism or deep vein thrombosis (DVT), or an outpatient diagnosis of DVT with a dispensing of an anticoagulant. To each case, 10 controls were matched by statutory health insurance, age and time in cohort. Risk factors of VTE, comorbidity, educational attainment, potential other (off-label) indications, lifestyle factors and comedication were assessed at baseline, and comedication also on index date. Confounder-adjusted odds ratios (OR) of VTE were calculated by conditional logistic regression using levonorgestrel with low ethinylestradiol ( < 50μg) as reference. To assess potential effect modification, we also conducted analyses stratified by age, educational attainment, obesity, chronic comorbidity, acute infection, and immobilization at index date. Robustness of results and impact of potential misclassification of exposure were examined by sensitivity analyses. Additionally, we did a bias analysis on the confounder smoking which is largely underreported in the data.
This study was financed by the German Federal Institute for Drugs and Medical Devices (BfArM).

Results: Compared to levonorgestrel with low ethinylestradiol, the risk of VTE was 2-fold increased for COCs containing dienogest (OR 2.23, 95% confidence interval 1.77—2.80), cyproterone (2.15, 1.43—3.25), chlormadinone (2.06, 1.58—2.68), desogestrel (1.93, 1.44—2.61) and drospirenone (1.89, 1.41—2.55) and 5-fold increased for gestodene (5.05, 1.23—20.74). For norgestimate and nomegestrol the point estimates suggest a 2-fold and 40% increased risk (1.90, 0.62—5.81 and 1.41, 0.52—3.81). Sensitivity analyses indicated robustness of results and bias analyses showed that potential impact of misclassification of smoking is expected to be small.

Conclusions: Our study confirms that levonorgestrel with low ethinylestradiol ( < 50μg) is the COC with the lowest risk of VTE and suggests that for chlormadinone, the risk of VTE is 2 times higher and thus in the same range as for desogestrel (3rd-generation progestogen) and drospirenone (4th-generation progestogen).