Evidence based policy decisions through a Bayesian approach: the case of a statin appraisal in The Netherlands

It has often been suggested that Bayesian statistics is more congenial to the informational needs of policy makers than the standard frequentist methods. In order to illustrate this claim, we use both a Bayesian and a frequentist approach for revisiting a recommendation by the Dutch National Health Insurance Board that for all patients requiring lipid reduction, the cheapest alternative (Simvastatin) should be prescribed. We investigate whether Simvastatin and Atorvastatin, the most commonly used alternative, can be considered equivalent in terms of lipid control for patients with heterozygous familial hypercholesterolemia. Priors were elicited from GPs, cardiologists and internists. A systematic review for studies comparing Simvastatin and Atorvastatin was performed. The data from these studies were combined with the priors in a Bayesian meta-analysis. For comparability a frequentist meta-analysis was also performed. The two approaches lead to similar point estimates and 95% intervals. However, the Bayesian outcomes are easier to understand and interpret, and our Bayesian analysis leads to additional outcomes that would have more direct pertinence for policy makers, and which could help them to assess what the data have to say about the questions that are most relevant to the problems they face.