Assessment of Appropriateness of Antimicrobial Therapy in Resource-Constrained Settings: Development and Piloting of a Novel Tool—AmRAT

Abstract

Inappropriate antimicrobial prescribing is considered to be the leading cause of high burden of antimicrobial resistance (AMR) in resource-constrained lower- and middle-income countries. Under its global action plan, the World Health Organization has envisaged tackling the AMR threat through promotion of rational antibiotic use among prescribers. Given the lack of consensus definitions and other associated challenges, we sought to devise and validate an Antimicrobial Rationality Assessment Tool—AmRAT—for standardizing the assessment of appropriateness of antimicrobial prescribing. A consensus algorithm was developed by a multidisciplinary team consisting of intensivists, internal medicine practitioners, clinical pharmacologists, and infectious disease experts. The tool was piloted by 10 raters belonging to three groups of antimicrobial stewardship (AMS) personnel: Master of Pharmacology (M.Sc.) (n = 3, group A), Doctor of Medicine (MD) residents (n = 3, group B), and DM residents in clinical pharmacology (n = 4, group C) using retrospective patient data from 30 audit and feedback forms collected as part of an existing AMS program. Percentage agreement and the kappa (κ) coefficients were used to measure inter-rater agreements amongst themselves and with expert opinion. Sensitivity and specificity estimates were analyzed comparing their assessments against the gold standard. For the overall assessment of rationality, the mean percent agreement with experts was 76.7% for group A, 68.9% for group B, and 77.5% for group C. The kappa values indicated moderate agreement for all raters in group A (κ 0.47–0.57), and fair to moderate in group B (κ 0.22–0.46) as well as group C (κ 0.37–0.60). Sensitivity and specificity for the same were 80% and 68.6%, respectively. Though evaluated by raters with diverse educational background and variable AMS experience in this pilot study, our tool demonstrated high percent agreement and good sensitivity and specificity, assuring confidence in its utility for assessing appropriateness of antimicrobial prescriptions in resource-constrained healthcare environments.