Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Public concern has recently grown about the safety of the U.S. pharmaceutical supply due to well-publicized drug withdrawals or adverse reactions. One of the most notable examples of this are concerns raised over the risks associated with the general class of non- steroidal anti-inflammatory drugs, particularly the Cox-2 inhibitors Vioxx, Bextra and Celebrex. In response to a sense of crisis, the Institute of Medicine conducted a study of the oversight options available to the Food and Drug Administration (FDA). The IOM report argues that substantial improvements are required in the way that post-marketing safety issues are identified and handled. Currently, the FDA relies upon mandated physician advisory letters and changes to product labeling as its primary tools for conveying safety concerns to physicians and patients. The most restrictive labeling tool is mandating a "boxed warning" (also known as a black box warning). However, as the IOM report noted, the FDA's decision making occurs in a diverse information environment, which also includes direct to consumer advertising (DTCA), general media coverage, scientific journal publications, adverse event reports, and direct to physician marketing. Thus, while a great deal of regulatory effort goes into determining whether to issue a The most restrictive labeling tool is m boxed warning, no research has been conducted to determine whether, given the large number of alternative sources of information on safety, whether the boxed warning mechanism has any marginal effect on pharmaceutical use. This is particularly important since if the FDA is to devote substantial resources and time into revising its safety system (as the IOM report recommends) it would be inefficient to devote effort to improving the efficiency of a monitoring tool which has little or no marginal effect on the behaviors of interest.

This paper will examine the effects of public information - including media reports, DTCA, physician-directed marketing, clinical research, adverse event reports, FDA advisory actions and FDA black box warnings – on the use of the Cox-2 inhibitors Vioxx and Celebrex to estimate: 1) the flow of patients into physician practices, and then 2) the likelihood of drug prescribing. The primary data for this study are taken from a geographically diverse national research network of 140 primary care practices in 33 states, with several million patients. Brand specific advertising data is collected at the national and local level, data on direct to physician marketing is collected national, data on media coverage, professional journal publications and FDA actions is collected from public data sources. We estimate bivariate probit models of visit- seeking and NSAID prescription seeking at the patient level, and estimate the residual effect of FDA actions on patient flow and prescriptions.