How Physicians Respond to Information Shocks: The Role of Physician Characteristics in Boxed Warning Response

Boxed warnings are issued by the FDA when there are significant risks for serious adverse drug events. The considerable safety risks associated with boxed warning drugs make it vital to study prescriber response to these warnings. We aim to determine whether prescriber characteristics moderate post-boxed warning changes in prescribing trends. Boxed warnings may alter prescriber’s risk perceptions by providing updated risk information on a given drug. As perceived drug risks increase, there are often declines in boxed warning drug prescribing. However, acquiring new drug risk information is costly for prescribers: time spent learning is unpaid and comes with the opportunity cost of reduced leisure time and/or and compensated work activities (e.g. seeing patients).  Thus, the amount of time that prescribers spend learning will vary as a function of their cost and benefit of acquiring new information. We hypothesize that prescriber characteristics that suggest ease of acquiring new risk information (e.g. larger practice size, academic affiliation, urban setting), or higher benefit of information gathering (e.g. higher percentage of patients using the boxed warning drug or with a relevant diagnosis) will be correlated with faster and larger changes in post-warning utilization. We test our hypothesis for several boxed warning drugs in multiple drug classes. We perform a retrospective analysis of 2003-2012 National Ambulatory Medical Care Survey (NAMCS) data.  We use logistic regression analysis to calculate the change in odds of prescribing the boxed warning drug following the boxed warning. We examine the effect of prescriber characteristics on the speed and magnitude of changes in drug prescribing probability post-boxed warning.