Negative Tests and the Efficiency of Medical Care: What Determines Heterogeneity in Imaging Behavior?

We develop a model of the efficiency of medical testing based on the frequency of negative CT scans for pulmonary embolism. The model is estimated using a 20% sample of Medicare claims from 2000-2009. We document enormous heterogeneity in testing conditional on patient population. Less experienced physicians and those practicing in high spending areas test more low-risk patients. Assessing the efficiency of current practices requires calibration assumptions regarding the costs of testing, the benefits of treatment and the likelihood of false positives. While we cannot tell whether any particular testing decision was mistaken in the context of our model, we find that collectively-given these additional calibration assumptions-there are systematic differences between doctor testing practices and the recommendations of our model of optimal testing. According to our model, 90-99% of doctors test even when costs exceed expected benefits; optimal testing thresholds would increase social welfare by 20-35%. Shifting doctor practice to weight risk factors differently could increase net welfare in our model by 275%.