The Health Consequences of Direct-to-Consumer Pharmaceutical Advertising

Direct to consumer (DTC) pharmaceutical advertising has grown rapidly since its 1998 deregulation. The welfare consequences of pharmaceutical advertising remain controversial. Advertisements could persuade individuals to consume products with little or no benefit, thus reducing welfare, or they could inform consumers of valuable treatments. Understanding this problem is further complicated by the persistent nature of treatments for chronic conditions - the short-run effects of advertising may have long-run consequences. We develop a simple dynamic model of pharmaceutical demand, which generalizes more commonly used static models. Using this framework, we derive short-run dynamic reduced form decision rules and a model of prescribing and health. These prescribing and health models are, in effect, estimated via probit and ordinary least squares with lagged dependent variables respectively. Variation in advertising expenditures from other therapeutic categories is used to identify the effect of advertisements on pharmaceutical demand. Parameter estimates and initial conditions are then used to simulate DTC advertising's long-run demand consequences. We implement this model using detailed medical record data regarding patients' cardiovascular health and cholesterol-reducing medication (statin) utilization. We find that advertising increases statin initiation among patients for whom guidelines indicate treatment and then increases treatment continuation. Advertising has no effect on patients when statin treatment is indicated, suggesting that advertisements are, in effect, informative. Furthermore, advertising effects are persistent and static models grossly underestimate advertising's long-run consequences. We then conduct policy simulations to measure the consequences of restricting or eliminating DTC advertising. Preliminary results suggest that eliminating statin advertisements would reduce long-run social welfare by more ten billion dollars.