(037) The public health impact of suspending cyproterone acetate (2mg)/ethinylestradiol (35µg) in France: A predictive modelling technique to estimate adverse drug reactions prevented by regulatory action

Background: Pharmacovigilance (PV) actions such as marketing authorisation (MA) suspension can significantly impact public health (PH) by preventing adverse drug reactions (ADRs). However, methodologies to quantify the impact of PV actions in terms of morbidity and mortality are yet to be developed. Products containing cyproterone acetate/ethinylestradiol (CA/E), used to treat acne in females of childbearing age, were suspended in France in 2013 due to increased risk of thromboembolism (TE). The products were returned to the market after eight months following European Medicines Agency review. This temporary suspension provides an opportunity for measuring the PH impact of a PV action.

Objectives: To estimate the PH impact of a temporary suspension of CA/E in France using a novel predictive modelling method which estimates the number of ADRs prevented.

Methods: Prescriptions of CA/E were counted by 10-year age group and sex for the 24-month period around MA suspension date in IQVIA’s IMS-France electronic health record database, and annual exposure was estimated. The reason for MA suspension (TE) was obtained from the National Agency for the Safety of Medicines and Health Products (ANSM) records. The frequency of TE used in the model was as per the Summary of Product Characteristics (SmPC). A systematic review was conducted to identify studies that may facilitate an estimation of the real-world post-approval frequency of TE, as this may be different. PH impact was estimated by modelling product usage figures and ADR risk.

Results: There were 4,325,863 active female patients of childbearing potential (aged 10-49 years) in the IMS-France database at the time of MA suspension. Between 12 months pre- and 12 months post-suspension, average monthly exposure to CA/E decreased from 8.49 to 1.80 prescriptions per 100,000 females (aged 10-49). TE was categorised as ‘rare’ in the SmPC (≥ 0.01%, < 0.1%). Based on this frequency and the average exposure to CA/E prior to MA suspension, it is estimated that 0.03-0.27 cases of thromboembolism may have been prevented during the 8-month suspension amongst this IMS-France population of females. Only case-control studies were identified by systematic review, so the real-world frequency of TE could not be estimated.

Conclusions: This predictive modelling technique has potential applicability in other settings where a measurable change in drug utilisation may be obtained, such as withdrawal of MA or restriction of prescribing. Further refinement of the method, with particular focus on how to estimate real-world ADR frequency (e.g. using electronic health records), is ongoing.