Response to “Behavioral science and reasons for non-adherence to medication”

We thank Dr. Molloy for his commentary, which raises two important issues to consider when addressing medication nonadherence. First, he suggests that our measure should classify reasons for nonadherence as intentional or unintentional because this distinction might indicate whether behavior is influenced more by reflective or automatic processes. This distinction could dictate which intervention approaches are applied to improve medication adherence. Second, Dr. Molloy suggests that we classify reasons for nonadherence as addressing capability, opportunity, and motivation—three conditions for health behavior at the individual and population levels.

A common theme underlying these two points is that interventions must be matched to determinants of nonadherence. Indeed, our motivation for assessing reasons for nonadherence follows from our interest in informing interventions aimed at improving adherence; once we know why patients do not take their medications, then we can design, test, and select appropriate intervention strategies. This goal has a practical measurement implication: a scale measuring reasons for nonadherence must possess construct validity by providing comprehensive coverage of the construct (1, 2). Thus, we agree with Dr. Molloy that reasons for nonadherence should address capability, opportunity, and motivation (which subsumes reflective and automatic processes)(3). Although we did not organize the reasons component of our measure around this behavior framework a priori, we could easily reclassify this component in this manner or in any other manner that would facilitate intervention efforts. Our initial validation work was conducted for antihypertensive medications, so we encourage researchers interested in adapting this scale for other medications or health conditions to determine whether additional reasons for medication nonadherence are relevant. As well, some reasons relevant to antihypertensive medications may not be relevant to other conditions or medication classes.

The issues raised by Dr. Molloy and the field as a whole underscore a dire need for further research to: (1) determine whether some reasons for nonadherence are dominant in specific conditions, medication classes, or treatment settings; (2) examine whether and to what extent reasons for nonadherence change over time between and within patients; and (3) examine heterogeneity of treatment effects to determine which interventions work and for whom. Before embarking on research to address these issues, a reliable and valid measure of medication nonadherence that comprehensively assesses reasons for nonadherence is needed. We hope that our new measure will help advance research addressing the complex problem of medication nonadherence.