I investigate whether a decrease in a prescription drug's out-of-pocket cost increases risky behaviors due to ex-ante moral hazard. In a difference-in-difference setting, I leverage the staggered implementation of U.S. state-level policies lowering insulin out-of-pocket costs.  For individuals with diabetes, insulin dosage must increase as carbohydrate intake increases. Using household-level scanner data, I find that the out-of-pocket cost reduction results in a temporary 4.8% increase in carbohydrate purchases. The results show that the increase in carbohydrates is driven both by fibers (4.1%) and sugar (9.3%), with only the latter causing blood sugar spikes. Although I find a temporary increase in other dietary components— such as fats and overall caloric intake—, a diet score analysis reveals no changes in diet healthiness. Hence, ex-ante moral hazard appears to be present but has a limited role. In addition, I find evidence of improved disease monitoring and better medication adherence.