The Successful Launch and Diffusion of New Therapies - Innovation and Marketing in the Pharmaceutical Industry - page 193

this dual-market phenomenon with the early adopters being patients who have
severe health problems and whose latent demand has accumulated prior to the new
drug's introduction, whereas the later adopters are patients with milder conditions
whose adoption may have been triggered by the launch itself.
Marketing scholars have also used diffusion models other than the Bass model to
characterize market penetration of pharmaceutical drugs. For instance, Desiraju
et al. ( 2004 ) examined the effect of market characteristics on the maximum penetra-
tion potential and diffusion speed for a new category of prescription drugs in both
developing and developed countries, using a logistic specification as in Van den
Bulte ( 2000 ). Van den Bulte and Lilien ( 2001 ) used a discrete-time hazard model to
show that several studies analyzing the diffusion of the drug tetracycline confounded
social contagion with marketing effects. That is, they showed that when marketing
efforts were controlled for in diffusion models, contagion effects disappeared,
underscoring the importance of controlling for potential confounds when studying
the role of social contagion in new drug diffusion.
The breakthroughs discussed above have helped to provide a better understanding of
the determinants of new drug diffusion. The developed models can be helpful in gaug-
ing the commercial potential of a new treatment in two main ways. First, after a new
drug is launched, these models can assist in making predictions of the drug's future
commercial potential (for instance, see Ofek's ( 2008b ) application of the Bass model in
forecasting the future diffusion of drug-eluting stents). However, these forecasts are
most reliable only after the inflection point—the point at which the growth in the cumu-
lative number of adopters starts to decline—has passed. A second way in which one
can use these diffusion models is to guesstimate the commercial potential of a new drug
using the diffusion path of a similar drug. Such a similar drug should resemble the focal
drug in its product characteristics, and the diffusion process must occur in similar mar-
ket conditions (see Ofek's ( 2005 ) application of the Bass model for this purpose in the
case of e-books and the background note in Ofek ( 2008a ); while some of us have used
this method inside pharmaceutical firms, unfortunately, no pharmaceutical application
exists in the public domain, to our knowledge).

7.1.2 Sales Models

Overall sales differ from adoption in that they encompass repeat purchases. Whereas
in durable markets (e.g., microwave ovens or refrigerators), for instance, repurchase
frequency is quite low, in many pharmaceutical markets (e.g., drugs for chronic
conditions, such as high cholesterol or hypertension) the repurchase rate is very
high. Given the high repurchase frequency in some markets, marketing scientists
have also developed models to forecast sales rather than adoption. The development
of models for sales rather than for adoption can assist in understanding the overall
dynamics in the market, and such models can potentially provide insight into the
relative roles of repeat purchase versus initial adoption in the sales of a new product.
The development of market-level sales models to forecast the commercial potential
of a new drug is also driven by the availability of data. Often, data on past sales are