Overwhelming evidence suggests that there exists a physician's tendency in recommending unnecessary medical treatments to the patients. This paper discusses this issue by providing a theoretical model with one physician and many patients who are uncertain about an underlying state of disease. The physician decides a preliminary test for the patients who need it. We show that the optimal test always reveals the state of disease, but not always the healthy state. The physician's advice may induce people undertaking unnecessary treatment as well as refraining from doing it when is needed. A policy intervention imposing a minimum information standard is very effective in reducing overtreat- ment but does not influence the number of tested people. Moreover, we show unintended consequences of releasing relevant news with or without the policy intervention.