Uping The RxAnte: An Adherence Predictive Model

Those of you that have heard me speak know that I look at this topic of predicting adherence both from an area of fascination along with the eye of a skeptic.  While I love the concept of predicting someone’s adherence and therefore determining how to best support them from an intervention approach, I also believe that the general predictors are pretty straightforward:

1. Number of medications
2. Plan design (i.e., cost)
3. Gender
4. Health literacy and engagement (see PAM score research)

And, this is a hot topic (see post on FICO adherence score).  You can see my prior posts on some different studies, on the Merck Estimator, and some notes from the NEHI event on this topic.  It generated a good dialogue on Kevin MD’s blog when I talked about paying MD for adherence.

I had a chance to talk with Josh Benner the CEO of RxAnte the other day.  It sounds very interesting, and they have an impressive team assembled.  In general, they’re focused on:

• Predictive modeling
• Decision rules
• Monitoring and managing claims to track adherence
• Evaluating effectiveness of interventions
• And creating a learning system

There are definitely some correlations to the work we do at Silverlink Communications around adherence.  We’re helping clients determine a communication strategy that might include call center agents, direct mail, automated calls, e-mail, SMS, mobile, or web solutions.  We’re looking at segmentation and prioritization.  We’re looking at past behavior and messaging.  The goal is how to best spend resources to drive health outcomes from primary adherence to sustaining adherence.  This is a challenge, and we all need to build upon the work that each other is doing to improve in this area.  We have a huge problem globally with adherence.