Content Validation of a Medication Adherence Risk Score ( <scp>MARS</scp> ) for Mortality Risk Prediction: A Delphi Consensus

Abstract

The Medication Adherence Risk Score (MARS) is an objective adherence-based score, derived from transactional dispensary data. This study provides expert consensus to validate the use of the MARS as a predictor of mortality risk. A systematic review and assessment of South African health legislation and professional guidelines informed the development of consensus statements regarding the relevance, validity, predictive power, and implementation feasibility of the MARS. A two-round modified Delphi approach using a 5-point Likert scale was employed. Consensus was reached if at least 80% of respondents selected "agree" or "strongly agree." Nonconsensus statements were revised based on qualitative feedback and subsequently reevaluated in a second round. Thirteen clinical, actuarial, and academic experts were successfully recruited. Consensus was achieved for 20 of the 28 statements in Round 1. Five of 18 (27.8%; risk score components), 1 of 3 (33.3%, validity/predictive power), and 2 of 7 statements (26.5%, real-world implementation) did not initially reach consensus. Nonconsensus statements were revised to provide clarity, after which consensus was achieved for all statements by the end of Round 2. Expert consensus supports the inclusion, exclusion, and weighting of MARS risk factors. The panel agreed that the score offers predictive value for mortality outcomes, and that real-world application for enhancing risk stratification and reducing healthcare costs is feasible. Future statistical analysis is necessary to demonstrate its relationship with mortality risk. Developing practical implementation strategies will facilitate incorporation into clinical practice and public health interventions.