Peer Influences on Diffusion of New Treatments

Variation in practice patterns among physicians, across geographic areas, and even within a local area, has been widely documented.  The reasons for the variation are less clear.  The diffusion of new treatments is also of interest to researchers, and to the companies promoting such treatments, and there is an intersection of the two topics in understanding why doctors practice as they do.  An article in Journal of the American Medical Association Network Open examines the spread of a new therapy for cancer among physicians.   (JAMA Article)   The study used data from 2005 to 2010 to examine use of the new drug among over 3250 oncologists in 250 communities for Medicare patients.  In particular they were interested in understanding how peer use of the product might affect an oncologist’s own use.  The drug in question was approved in 2004 for colorectal cancer but was soon approved for advanced stage lung, breast, brain, kidney, cervical and ovarian cancer.  The use of the drug showed significant variation across oncology practices.  The authors examined social and other connections among physicians to determine if there was some peer influence that affected the rate of adoption of the new therapy.  They looked at doctors who were relatively early adopters of the drug, in 2005 and 2006, and those who were not, and analyzed whether doctors who were connected to those who were early adopters were more likely to become extensive prescribers of the drug themselves.  They excluded doctors treating less than 30 Medicare patients with the relevant cancers and constructed physician networks by identifying physicians who shared treatment of patients on a regular basis.

Use in the 250 communities in the baseline period of 2005 to 2006 varied from 0% to 31% of all patients with the eligible cancers and so the researchers divided the communities into tertiles of use.  They next analyzed use in the 2007 to 2010 period by doctors who had not prescribed the drug in the baseline period.  For doctors whose peers had the lowest tertile of early adoption, the rate of use was 10%; for doctors with peers in the middle tertile, it was 9.5% and for those with peers in the highest use tertile, it was 13.6%.  So the study appears to support the notion that having peers in a community who adopt a new treatment faster has a spillover effect.  If a new treatment has a clear advantage in outcomes compared to existing treatments (and many cancer therapies don’t meet this criteria) it could be very beneficial to use physician networking and peer influence effects to increase appropriate uptake.  But if the therapy doesn’t have such clear advantages, these networking effects may just be promoting more marginal use.  In any event, understanding the extent to which these effects exist, how they work and how they may be effectively countered or enhanced is useful for payers and policymakers.