Monday, October 24, 2011
Grand Ballroom AB (Hyatt Regency Chicago)
Poster Board # 32
(BEC) Behavioral Economics

Tuan Dinh, PhD and Peter Alperin, MD, Archimedes Inc., San Francisco, CA

Purpose: It is estimated that 30-50% of patients do not take medication as prescribed by their physicians, costing the American healthcare system billions of dollars annually in avoidable medical expenditure.  Medication non-compliance is a complex phenomenon, determined by the interplay of multiple factors, including patient-, condition-, and therapy-related health system factors, as well as social factors. We are developing a mathematical model for medication compliance that accounts for the underlying psychological processes of patient behavior. This will facilitate a deeper understanding of the effects of interventions designed to improve compliance as well as the resultant health and economic effects of these interventions.

Methods: The process of obtaining and taking medication is broken down into basic thought processes and actions. The pathway model of medication adherence is a synthesis of several psychological theories of medication compliance, including the Self-Regulatory Model and the Health-Belief Model. Standard questionnaires and scales such as Beliefs about Medicines Questionnaire, Illness Perception Questionnaire (IPQ), and Barriers in Diabetes Questionnaire are used to quantify key cognitive and psychological variables (e.g. perceptions about medical benefits and disease severity) and mental states (e.g. self-efficacy).  Correlations between psychological variables and mental states on medication adherence were derived from a meta-analysis of the literature. The output of this integrated model is medication adherence as a function of time. Each patient’s likelihood of adhering to medication recommendations changes over time, depending in part on his/her changing perception about disease severity, benefits of medication, and experience of disease symptoms and medication side-effects.

Results: We apply the current approach to model adherence to diabetes medication.   At the population level, the model reproduces the dependence of medication adherence on socioeconomic and clinical risk factors. At the individual level, the model captures the transient effects of life events and behavioral interventions on adherence. We use the model to examine the effects of  educational interventions designed to improve patient knowledge about disease severity on medication adherence.

Conclusion: We demonstrate that it is possible to construct a detailed, “mechanistic” mathematical representation of medication adherence. Such a model can be integrated with disease models to forecast health and economic effects of interventions aimed to improve medication adherence.