Purpose: To ameliorate conflicts of interest that arise when pharmaceutical representatives have unfettered access to physicians, the University of Pennsylvania implemented a policy regulating the physician-pharmaceutical representative relationship. Tenants of this policy include: 1) representative visits may occur by appointment only, 2) distribution of medication samples is prohibited, 3) promotional items (pens, etc.) cannot be distributed, and 4) meals and gifts to professional staff are prohibited. To inform institutions considering enacting such policies, we evaluate the effect of this intervention on physicians' prescription patterns.

Methods: Prescription data for internists at the University of Pennsylvania was collected for six months pre- and post-implementation of the new policy. Refill prescriptions were excluded. Medications in three drug classes were considered: angiotensin converting enzyme inhibitors (ACE-i), angiotensin receptor blockers (ARB), and proton pump inhibitors (PPI). Within each class, the percentage of prescriptions for each medication as well as overall use of branded and generic medications was compared using the two sample test of proportions.

Results: In the 6 months prior to policy implementation, there were 2128 prescriptions for ACE-i's; 6 months post-policy, there were 1911. Pre-policy 68.4% of these prescriptions were for generic ACE-i's compared with 73.5% post-policy (p=0.0004). Pre-policy, there were 1516 prescriptions written for ARBs; post policy, there were 1292. The use of Avapro/Avalide fell from 14.4% of ARB prescriptions pre-policy to 10.7% post-policy (p=0.003), while the use of Diovan increased from 28.7% to 33.1% (p=0.01). Post-policy, there was a non-significant decrease in the percentage of ARBs prescribed relative to ACE-I's (41.6% v. 40.3%, p=0.3). Pre-policy, there were 3050 prescriptions written for PPIs; post policy, there were 2864. Post-policy, there was decreased use of Aciphex (6.3% v. 4.3%, p=0.0005), Protonix (12.0% v. 8.8%, p=0.0001) and Nexium (28.6% v. 23.7%, p<0.0001), while there was increased use of generic omeprazole (13.0% v. 21.1%, p<0.0001).

Conclusions: The presence of a policy regulating the conduct of pharmaceutical representatives is associated with increased use of generic medications. The policy was not associated with increased ACE-i use at the expense of ARB use. Our results suggest that interaction with pharmaceutical representatives affects physicians' choice of class-equivalent drug but does not affect choice of drug class. Our results may usefully inform the decisions of academic medical centers considering enacting policies governing pharmaceutical representatives.