Prescriber Profiling: Time to Call It Quits

Pharmaceutical marketing to physicians recently surpassed $23 billion per year, and the pharmaceutical industry currently employs 1 sales representative for every 5 office-based physicians (1, 2). In this setting, policymakers and patients are understandably concerned about commercial influences on physicians' prescribing decisions.

Real-time, physician-specific data on prescribing habits are a powerful tool that pharmaceutical sales representatives rely on when visiting a physician's office. A representative can quickly access a breakdown of pharmaceuticals prescribed by any physician on a handheld computer, enabling that representative to deliver a tailored marketing pitch to physicians selected for their current prescribing habits. Within weeks, the sales representative can monitor each physician's response to the pitch—as well as to inducements, such as meals, gifts, and drug samples—and can make repeated visits to achieve sales goals.

Given that prescriber profiling is so deeply embedded in pharmaceutical detailing, we have surprisingly little published research about its effect on sales. One recent market research report concluded that profiling improves profit margins by as much as 3 percentage points and the initial uptake of innovative drugs by 30%. Few details of the study are publicly available, and its quality is difficult to evaluate (3). However, other indirect indications of effectiveness exist—for example, the emphasis that pharmaceutical marketing consultants place on the value of prescribing profiles for enhancing sales force effectiveness (4–6).

How do pharmaceutical companies obtain such detailed prescribing data? Currently, retail pharmacies sell de-identified patient prescription records with limited physician identifiers to data intermediaries known as health information organizations (HIOs). By purchasing a comprehensive database of U.S. physicians from the American Medical Association (AMA), HIOs can link individual physicians to prescription records by using identifiers common to the pharmacy databases and …