Association for Public Policy Analysis & Management

When a new pharmaceutical product is approved for marketing, the full range of benefits and risks are not known. Nevertheless, new drugs are often widely adopted, despite typically high prices. Evidence on the safety and effectiveness of a treatment accrues as the product is used in clinical care and the information generated from experience has been found to affect physicians’ treatment decisions. A medical reversal occurs when evidence that a treatment is ineffective or unsafe reverses recommendations for the treatment.

Most research into medical reversals focuses on the affected drug or others in its class, but it is unknown how responses to these information shocks are related to demand for other new drugs. In this paper, we ask how physician preferences for new drugs change following a medical reversal of a drug already in use. The extent of spillovers across drug types informs inventors and regulators about potential consequences of pursuing products that may prove ineffective or unsafe, and the importance of monitoring marketed products.

We linked physicians’ responses to an information shock discouraging prescribing of the oral antidiabetic Avandia (in the TZD class) with their future prescribing of a new class of oral anticoagulants. We examined whether physicians who responded more strongly to the information shock were less likely to adopt the subsequently approved drugs.

We first modeled the probability of TZD prescribing (among all diabetes prescriptions) following the FDA safety alert for Avandia. We used multi-level mixed effects logistic regression models that included random effects for individual physicians’ levels of prescribing relative to their peers. We next assessed the relationship between the physician-specific effects and use of new oral anticoagulants. We modeled the probability of prescribing new oral anticoagulants (Pradaxa and Xarelto) versus warfarin (the conventional treatment) in each month of 2011 (the first full year the new products were available).

We found no difference in the use of new oral anticoagulants based on how physicians responded to a safety-related information shock for oral antidiabetics in the TZD class. Our findings were robust to different model specifications, sample constructions, and variable definitions. These results suggest that the effects of a medical reversal for pharmaceutical products do not spill over across drugs in different therapeutic areas.

Over the last decade, there have been several legislative actions affecting the authorities and resources available to the FDA for the approval and post-market monitoring of prescription drugs. Recently, the 21^st Century Cures Act expanded the types of data the FDA may consider when approving new treatments. These provisions were aimed at improving access to innovative drugs, but have raised concerns as some worry they are compelling the FDA to laxer standards. In this context, post-market monitoring and surveillance play a key role in protecting patient safety. Physician responses to regulatory actions in turn provide feedback about factors influencing adoption and use of new drugs. This study suggests that responses are narrow and do not spill over to decisions about the use of subsequently approved drugs to treat different diseases.