Leveraging Peer-to-Peer Networks in Pharmaceutical Marketing - Innovation and Marketing in the Pharmaceutical Industry - page 463

15.5.1 Mapping the Full Physician Network

Research has found different kinds of opinion leaders based on different techniques
used. Self-reported opinion leaders were found to be less affected by contagion
(Iyengar et al. 2011 ). Survey-based identification of opinion leaders does not lead to
a full mapping of the physician network, and hence reports lower social multipliers
than a method which maps the physician network more completely (as in Bhatia and
Wang 2011 using patient movements between physicians, or by using physician
referral network data, if available). Also, combining various sources of data such as
surveys, physical distances between physicians, physician referral patterns, and
other sources may lead to a more complete picture not just of the physician network
but also of the physician referral decision making process. Iyengar et al. ( 2011 ) find
that the tendency to adopt early is more pronounced among those who are central to
the network than the self-reported opinion leaders. Chritakis and Fowler ( 2011 )
argue that “To know whether a doctor is central, one must map the whole network,
not simply ascertain attributes of the doctors, such as their specialty or prescribing
behavior” (p. 214). This will lead not just to the identification of who is influential,
but also who is influenceable. If there is no contagion in individual level adoption,
then no increase in detailing to opinion leaders can lead to cascade effects.

15.5.2 Social Network Analysis

Various complex strategies for targeting also need investigation, and not just the
idea of targeting the most central or highest degree nodes (Valente et al. 2003 ).
Tracking the method of propagation of prescribing behavior or adoption may be
useful. For example, the possibility that it may spread via “complex contagion”
could be examined. Aral ( 2011 ) suggests studying how different types of physicians
are distributed in the network since it can affect cascades of social behavior and
contagion. Networks in which low-status physicians are clustered around high-sta-
tus physicians will possibly exhibit different adoption dynamics from isolated
peripheral clusters of low-status physicians distant from a densely connected core
of high-status physicians (as in prestigious clinics and hospitals such as Mayo
clinic, or John Hopkins hospital). If physician referral networks typically connect
competitors who typically service the same type of needs (such as PCPs or between
specialists of the same specialty), prescription referrals may not flow to peers as
easily as referral networks connecting physicians to specialists who do not directly
compete (Bhatia and Wang 2011 ).
All the current literature assumes that all the mechanism of influence for the
social ties is the same as that of professional ties. Depending on how the social ties
among physicians are formulated, it might lead to different levels of peer influence.
For example, the tie can be created socially, such as the case when two doctors get to
know each other at the same graduate school or in the same hospital for residency;
or the tie can be formulated professionally, such as the case when they work in the