Network Mapping Identifies Physician Influencers

Several years ago, scientists led by Harvard’s Nicholas Christakis showed how happiness and obesity can spread through social networks. The team showed for example, that when certain people gained weight, the chances increased that others in their social network would also gain weight. Similar analyses have mapped-out the impact of social networks on influenza outbreaks and cigarette smoking.

Now, Christakis has teamed-up with Larry Miller to create MedNetworks, a company that uses the same network mapping methods to identify physicians that most strongly influence their colleagues when it comes to prescribing drugs. If their work proves successful, they will surely find plenty of drug makers who will pay for their insights.

The Boston-based start-up relies on medical claims data for its mapping studies. It claims to be able to track growth in the popularity of a new drug within professional circles. It has identified physicians within these circles that appear to influence the prescribing behavior of others, in that after they begin prescribing a newly released drug, colleagues within three degrees of separation from them begin doing the same. “We’ve shown that we can predict adoption of pharmaceuticals among doctors,” Miller said in an interview.

To support its claims, MedNetworks cites a case study on the launch of Merck’s diabetes drug Januvia in the Raleigh-Durham area. In this example, prescribers that had Januvia adopters within one degree of separation in their social network were twice as likely to prescribe the drug as those who did not have Januvia adopters in their network.

As another example, MedNetworks reports that when a generic replacement drug for Pfizer’s blockbuster Lipitor became available, prescribers discontinued Lipitor “in clusters…and social network influence accounted for about 40% of the decline.” 

Aside from the lucrative opportunities to sell this data to pharmaceutical companies, MedNetwork executives believe their approach could also be used to identify opinion leaders within a community who would be most effective in spreading public-health messages. Suppose a particularly influential employee stopped smoking, for example. It’s possible that employee would be more effective in spreading the health-improving behavior through his or her company. Knowing these people might help public health officials gain maximal returns from their limited resources.

MedNetworks is currently attempting to identify such influencers among citizens in several California communities. The client is Healthways, a consultancy that helps employers control their health expenses. The effort is directed at finding people who are the most effective in spreading positive messages regarding cigarette smoking and obesity.