OBJECTIVES: To assess trends in the use of evidence to support regulatory actions in Europe, and to identify a model which quantifies public health impact of a subset of these regulatory actions in terms of morbidity.

METHODS: Types of evidence used during regulatory decision-making were assessed for products withdrawn in Europe between 1999-2016 for safety reasons; patterns over time were evaluated. Quantification of impact of regulatory actions was performed for a subset of products (withdrawn/revoked in France/Germany/United Kingdom, 2012-2016). Annual exposures for each product were estimated using IQVIA Medical Research Data electronic health records. Reasons for regulatory action (adverse drug reactions [ADRs] of interest) were obtained from EMA records. Systematic review identified research quantifying ADRs, allowing estimation of absolute risk in exposed patients. Public health impact, expressed as annual number of ADRs prevented due to action, was estimated by modelling exposure and ADR risk.

RESULTS: In Europe between 1999 and 2016, 35 products were withdrawn; 91.4% of withdrawals (n=32) were supported by spontaneous reports and/or published case reports or case series, 60.0% (n=21) were supported by RCTs, 48.6% (n=17) were supported by non-randomized trials, 40.0% (n=14) were supported by observational studies. Meta-analyses contributed least frequently (n=11, 31.4%). The number of evidence sources cited in regulatory decisions increased over time, particularly for observational research. A sub-study involving 6 withdrawals/revocations for safety reasons in France, Germany and/or UK met inclusion criteria. Absolute risks were 0.1–41.25%. To estimate impact of withdrawal/revocation, average annual exposure was multiplied by absolute risk, providing crude numbers of ADRs prevented annually.

CONCLUSIONS: Spontaneous reports remain the most cited evidence source in regulatory decisions. However, a shift towards more epidemiological research supporting decision-making was observed. We successfully quantified public health impact of withdrawal/revocation in terms of serious morbidity. The method can be applied to products in other settings.