Early Benefit Assessment of Pharmaceuticals in Germany

BACKGROUND: Since 2011, when the German Pharmaceutical Market Restructuring Act (AMNOG) came into effect, newly licensed pharmaceuticals must demonstrate an added benefit over a comparator treatment to be reimbursed at a value greater than the reference price. Evidence submitted by manufacturers is assessed by the Institute for Quality and Efficiency in Health Care (IQWiG) and subsequently appraised by the German Federal Joint Committee (FJC) as part of so-called early benefit assessments (EBA). This study aims to explain the decisions made, clarify the roles of the parties (manufacturers, IQWiG, FJC) involved, and guide manufacturers in developing future submissions by analyzing 42 EBAs concluded since January 2011. METHODS: We developed a variable list representing the essential components of the EBA: the rating decisions of manufacturers, IQWiG, and the FJC regarding each pharmaceutical's added benefit; the characteristics of the pharmaceutical; the characteristics of the EBA process; the types of evidence submitted; the methods used to generate evidence; and the pharmaceutical's maximum possible budget impact. We used Cohen's kappa to analyze agreement between the rating decisions of the different parties. The chi-square test and bivariate regression were used to identify associations between components of the EBA process and the rating decisions of the FJC. RESULTS: We observed a low level of agreement between manufacturers and the FJC (kappa = 0.21; 95% CI 0.107-0.31) and a substantial level of agreement between IQWiG and the FJC (kappa = 0.64; 95% CI 0.451-0.827) in their rating decisions. The characteristics of the EBA process--for example, duration of the process (P = 0.357), participation in the official hearing (P = 0.227), and the pharmaceutical's budget impact (P = 0.725)--did not have a significant effect on the rating decisions of the FJC. There was, however, an association between the type of evidence submitted and the FJC's rating decision when the manufacturer's dossier reported outcomes related to morbidity (P = 0.009) or adverse events (P < 0.001) but not mortality (P = 0.718) or quality of life (P = 0.783). CONCLUSIONS: While the FJC tends to disagree with the rating of benefit by manufacturers, it softens IQWiG's decisions, potentially to make the final outcome more acceptable. Concerns voiced that the FJC might be exceeding its statutory authority by taking cost or procedural considerations into account appear to be unfounded. Choosing appropriate evidence to submit for each endpoint remains a challenge, as submission of health outcomes evidently influences decisions.