3065 — Improving the Detection of Medication Nonadherence
Voils CI, and Maciejewski ML, Durham VA Medical Center; Hoyle RH, Duke University; Reeve BB, University of North Carolina; Gallagher P, and Yancy Jr WS, Durham VA Medical Center;
Medication nonadherence is associated with increased emergency room visits, hospitalizations, and mortality, leading to excess healthcare costs estimated at $290 billion. The detection and management of nonadherence could be facilitated by self-report measures, yet the length and lack of reliability and validity of existing measures has prevented them from being routinely incorporated into clinical care. The purpose of this study is to develop a reliable and valid self-report measure of antihypertensive medication nonadherence that could be incorporated in clinical practice.
Extent of nonadherence was measured with four correlated indicators, whereas reasons for nonadherence were measured with 23 independent indicators in order to fully cover reasons for missing medications. Participants were 202 Veterans with a diagnosis of hypertension prescribed at least one antihypertensive medication. At baseline, participants completed the new measure and several comparison measures to inform construct validity. Three to 21 days later, participants completed the new measure again to provide evidence of test-retest reliability.
Participants were aged 64 on average, and 87% were male. Items assessing the extent of nonadherence produced reliable scores, alpha = 0.89, test-retest r = 0.64. As expected, items assessing reasons for nonadherence were marginally associated, average inter-item r = 0.28. As reasons for missing medications can change over even short periods of time, test-retest correlations of individual items ranged from 0.09 to 0.64; however, the total score was reasonably stable, r = 0.62. The two nonadherence constructs were related but distinct, r = 0.55. Supporting convergent validity, nonadherence was associated with medication self-efficacy (extent r = -0.46, reasons r = -0.55). Supporting discriminant validity, the nonadherence constructs were only moderately related to conscientiousness, social desirability, habit formation, and concurrently assessed blood pressure, all rs <0.31.
To improve the efficiency of detecting nonadherence, clinicians could use this measure to screen for nonadherence and then to query reasons for nonadherence when necessary. Future research is needed to examine longitudinal trajectories in these two nonadherence constructs and to provide further evidence of validity of the measure.
A two-step approach to detect problems with medication nonadherence could improve the efficiency and effectiveness of detecting and treating a common and costly health problem in the VA healthcare system.