Clinical guidelines and heterogeneous treatment effects: evidence from the roll-out of Percutaneous Coronary Intervention services in England

Abstract: Guidelines that govern how patients are cared for in health systems around the world are typically based on the results of clinical trials. Testing the effectiveness of treatments within randomised control trials provides a valid control group, and minimises the ethical concerns around using new treatments. However, whether the results from these trials are replicated in the real world, where clinical settings are more heterogeneous and patients potentially more complex, is rarely tested. In this paper we exploit a centralised decision in 2008 to roll-out a particular technology across England, Percutaneous Coronary Intervention (PCI) for AMI patients, to assess whether the impacts of PCI on the English population of AMI patients matches results from clinical trials. We instrument PCI use with variation in distance to PCI facilities over both time and space, generated as new facilities opened. Unlike estimates from clinical trials, we find no evidence that PCI reduces in in-hospital mortality, although there are reductions in patient length of stay and one-year readmissions for cardiovascular disease. We propose a number of explanations as to why we find no mortality effect, including improvements in non-PCI care, heterogeneous treatment effects, and doctor quality.