Abstract
Objectives. The aim of this study was to assess preferential prescribing of OC according to different thrombotic risk factors.
Material and methods. The control group in an ongoing Danish case-control study on stroke and OCs collected in 1994 and 1995 underwent a control-only analysis concerning the occurrence of thrombotic risk factors among users of different types of OC. Specific attention was given to differences between OCs with second and third generation progestagens. The association between specific risk factors and the pill types was assessed crude and after multivariate analysis with confounder control for age and other risk factors, in order to identify risk factors, which after these corrections still had a significant confounding influence on the prescribing of OC. Results. Users of OCs with third generation progestagens had a significantly higher proportion of familial thrombotic disposition (23.1%) than users of OCs with second generation progestagens (7.1%) (p = 0.01). After correction for age and other risk factors this difference was still highly significant (p = 0.002). Among users of third generation pills the proportion of short time users (< 1 year) (22.4%) was significantly higher than the per cent among users of OCs with second generation progestagens (5.5%) (p < 0.001). This difference was still significant after correction for age and other risk factors (p < 0.001).
Smoking, years of schooling, migraine, and body mass index did not differ significantly between the two pill groups.
Conclusion. In Denmark, women with familial thrombotic disposition are four times more likely being prescribed OCs with third versus second generation progestagens compared with women without such a disposition. At the same time users of OCs with third generation progestagens include significantly more short time users than users of OCs with second generation progestagens.
For thrombotic diseases where familial disposition or duration of use of OCs play a role for the pill-associated risk, these differences may significantly influence the thrombotic risk measures in case-control studies and non-randomized cohort studies unless confounder control is conducted for this selection.