Discontinuation Risk Models

Abstract

Medication discontinuation or non-adherence is a significant problem in healthcare around the world. When people stop taking their medicine, the result is often worsening health and potentially unnecessary hospitalization. Medication non-adherence doesn’t just impact the individual who is not taking his or her medicine; it impacts national economy, with experts estimating the cost of unnecessary medical treatment resulting from non-adherence at nearly $289bn per year [1]. Given the scale of the problem, it is imperative that payers, providers, and even patients look for solutions to curb the rates of discontinuations. In addition, manufacturers (life sciences companies) can lead with their own solutions and services to support patients on their products. In this paper we discuss the techniques to help predict discontinuations for patients on a particular life science company’s products, and how to develop risk profiles for individual patients. Finally, this paper will describe the process of constructing the decision rules for quantifying and classifying patient discontinuation risk in ways that lead to actions that care managers and nurse support services can act on. Decision tree models described in this paper can be both predictive and effective by being easy to implement and maintain.