Evaluating Reminders for Medication Adherence and Side Effects in M-Health Environment

Introduction

Medication nonadherence leads to significant cost burden on healthcare systems (Cutler, Fernandez-Llimos, Frommer, Benrimoj, & Garcia-Cardenas, 2018). Not taking medications as prescribed, is a common, complex, and costly problem that contributes to undesirable treatment outcomes. An estimated healthcare expenses of $100-$300 billion per year in the US is due to medication nonadherence (Cutler et al., 2018; NEHI, 2009). According to the World Health Organization (Sabate, 2007), medication nonadherence needs to be addressed globally to improve the quality and reduce the cost of healthcare. Reducing the medication nonadherence for chronic diseases (Rajan, Seidmann, & Dorsey, 2013; Santo et al., 2017) alone could result in a 1:10 cost-to-savings ratio (Sabate, 2007). Medication nonadherence can be intentional, where the patient actively chooses to deviate from the treatment regimen, or unintentional, where the patient is forgetful in adhering to the regimen.

In healthcare literature, several interventions to improve medication adherence have been proposed (Choi, Lim, Kim, & Park, 2008; Kim, Combs, Downs, & Tillman Iii, 2018; Krebs et al., 2015; Maulucci & Somerville, 2011; McDonald, Garg, & Haynes, 2002; Oswald, 2018; Schreier et al., 2013) including those based on reminders. Although very promising, we find from a detailed analysis of the published literature that effectiveness of reminders varies widely and side effects of reminders and healthcare costs (Lokshina & Lanting, 2018) have not been addressed. Further in almost all of the evaluations of interventions, Average Medication Adherence (AMA), as the ratio of doses consumed to doses prescribed (Osterberg & Blaschke, 2005), is the predominantly used measure. Although a useful metric, different patients could achieve the same AMA with widely different consumption patterns and can have different health outcomes. Therefore, there is a need for objective measures to detect adherence improvements (Nieuwlaat et al., 2014). Recently, a single metric, termed Effective Medication Adherence (EMA), has been proposed to capture both the time gaps between doses and Average Medication Adherence (Varshney & Singh, 2013). EMA can indicate what fraction of doses is taken within the medically effective time interval. This is a better predictor of health outcomes than AMA alone. Further, the values of EMA over time can represent the pattern of medication adherence of the patient as steady, improving or declining at different rates (Varshney & Singh, 2013). The side effects, in terms of Undesirable Dose Events (UDE), where the patient is taking medications with less than medically safe time gap, needs to be evaluated also for different types of reminders.