Single-source vs. off-patent drugs: What is the role of physician preferences?

Rising drug expenditures have been a source of concern in most developed countries for several decades. Recently, the patent expirations of several blockbuster chronic drugs have presented new opportunities for cost containment, leading insurers and policy-makers to advocate moving patients from newer, single-source medications to similar off-patent drugs. An obstacle to this strategy is the fact that pharmaceutical companies heavily promote their single-source drugs to physicians, potentially leading them to favor these drugs over older therapeutic alternatives.

If advertising leads physicians to favor new drugs in their prescribing, an open question is whether it actually shifts their beliefs regarding the benefits of such drugs, or whether friendly relationships with pharmaceutical representatives create subtle incentives for supplier-induced demand. We shed light on this question by asking whether physicians are more or less likely to use single-source drugs for themselves versus when they prescribe to their own patients. If physicians are less likely to use the new single-source drugs themselves, this would support the story that they face external (and unobservable) incentives to prescribe new drugs. On the other hand, if they are more likely to use these drugs, this suggests that they believe these new drugs are superior.

We focus on statins (lipid lowering drugs), one of the most widely prescribed drug classes. It is widely believed that for the average patient, newer single-source statins offer little added benefit relative to the much less expensive, off-patent statins. As a result, many private insurers in the U.S. as well as public insurers outside the U.S. require special permission for coverage of the remaining single-source statins.

We use data from the Danish National Prescription Registry covering all statin treatment initiations (436,194) over the period 2004-2010 for the entire country of Denmark. The prescriptions data contains very detailed information about each prescription, including type of drug and quantity, along with socio-economic information of both patients and prescribers. Controlling for observable characteristics we find that physicians themselves are 40% more likely to initiate treatment with a single source drug compared to the overall population, with the finding being even more pronounced for primary care physicians. Further, primary care physicians who themselves initiate treatment with a single source drug are also more likely to start future patients on a single source drug. This indicates that physicians learn privately about attributes of the drugs that they convey to their prescribing patterns, and that a large share of physicians in Denmark find newer single-source statins to be preferable to the older alternatives.