A predictive model for identifying low medication adherence among older adults with hypertension: A classification and regression tree model

Various individual characteristics may affect medication adherence; however, few studies have investigated the effect of interrelationships among these various individual characteristics on medication adherence. This cross-sectional study explored the interrelationships among risk factors for medication adherence and estab-lished a predictive model of low medication adherence among older adults with hypertension. Convenience sampling was used to recruit 300 older adults with hypertension. The following parameters were recorded: demographic and disease characteristics, health beliefs, self-efficacy, social support, and medication adher-ence of antihypertensive drugs. Classification and regression tree (CART) analysis was performed to develop a predictive model of low medication adherence. The CART model revealed that health belief, disease dura-tion, self-efficacy, and social support interacted to contribute to various pathways of low medication adher-ence. The predicted accuracy of the model was validated with a low misclassification rate of 26%. The proposed classification model can help identify risk cases with low medication adherence. Suitable health education programs based on these risk factors to manage and improve medication adherence for older adults with hypertension could be considered. (c) 2021 Elsevier Inc. All rights reserved.

Automated summary

This study explored the interrelationships among risk factors for medication adherence among older adults with hypertension. A predictive model was established, with interactions between health belief, disease duration, self-efficacy, and social support contributing to low medication adherence pathways. The model showed high accuracy and could identify cases with low medication adherence for targeted intervention.