Pharmaceutical sales are driven by a multiplicity of elements and factors. First, of course, is the drug’s efficacy and safety profile. Then, there’s sales force promotion, medical education, and managed care. On top of that, there are various marketing tactics such as DTC, conferences, peer-to-peer presentations (speaker programs – branded and unbranded), and more. The list goes on and on, creating a complex web that can push sales forward—or hold them back.
There’s an over-arching thread that ties these pieces together: relationships and networks between physicians. The networks of providers around key decision-makers have a significant impact on each of the above elements. This includes the referral network around the physician: both within the specialty, and among the multi-disciplinary team supporting the physician.
As Ivan Misner at Entrepreneur puts it, a referral network is “a system that works well because it ferrets out all those unpredictable, hidden, complex connections that exist between people in everyday life and in business.” When it comes to building a referral network in the pharmaceutical world, those connections run deep, and secondary data can help decode that. In this post, we will share how we, at 159 Solutions leverage longitudinal data capture to help build a more robust referral network that delivers measurable results.
In order to build a strong referral network, you need strong data. Unfortunately, traditional research methods have weak points that can lead you astray. Traditionally, primary market research determines influence by directly asking physicians who they are influenced by. This method often counts on a very small sample size. Additionally, results will be biased toward academically-minded physicians, as well as colored by individual perception. The result is an incomplete view of the true picture.
“The pharmaceuticals industry has seen an explosion in the amount of available data beyond that collected from traditional, tightly controlled clinical trial environments.”
- Harvard Business Review"
With the advent of patient-level claims data, we find an additional and complementary way to dig in a little deeper. Patient-level datasets usually provide a longitudinal view into treatment. This information can be analyzed to get a better sense of which specialties are influencing the patient at various points in their treatment journey. They also can be mined for insights into which physicians are often “referred” for second opinions. In addition, geographic markers help to zero in on influencers and referrals at a local level.
Claims data has some challenges, as well. Because nationwide coverage is incomplete, you end up with a biased capture of patients. The data will skew toward commercial care, with less information on those using public sources such as Medicare or Medicaid. With an inconsistent capture of patients, there is certainly an incorrect assessment of influence.
How do you then augment these patient-level datasets to avoid the common issues of the past? There are growing free data sources, such as DocGraph, that have a much higher capture rate for the Medicare population. By representing more patients, we can better understand and assess referrals between MDs and other entities. Then, when we layer this data on top of third-party patient-level data, new connections begin to crystallize.
Once we build up a rich and dynamic database, we have a wealth of information to guide us in our decision-making. We can then begin building you a robust and insightful referral network. Here are several key points that we consider as we move forward.
Understanding the treatment flow is the critical first step. We need to know how to best tease out relationships, and how to garner influence among various stakeholders in the market.
o Influence can be measured using a scoring system, based on several factors. These include patient volume, breadth of connections, depth of connections, types of connections, the weight of connected specialties, and more.
o Our experience shows that if the capture rates of patient databases are less than 50%, you may end up a significant number of false negatives. We can potentially avoid this kind of distortion by “stitching” together databases, and thoughtfully adjusting for bias.
o Is influence shown through claims in the clinical setting, pharmacy setting, or hospital setting? This is important because some vendors provide different databases for each, all possessing their own varying capture rates. It is crucial to ensure that MD connections are equally represented across these varied sources. If not, you may end up with physician clusters that are not actually independent from one another.
As you can see, there is a lot to consider when building up an effective referral network. From research tactics, to data sources, to connection quality, each piece has an impact that can bias your results. Further, referral networks are only one piece of a complex puzzle to determine influencers. We will discuss further variables to define and understand influence in a future post.
Back