Abstract
An adverse drug reactions avoidability tool called the Liverpool ADR avoidability assessment tool (LAAT) was recently developed (for research purposes), and subsequently validated with mixed interrater reliability (IRR). We investigated the comparative IRR of this tool in an inpatient cohort to ascertain its practical application in this setting.The patient population was comprised of 44 ADR drug pairs drawn from an observational prospective cohort of patents with ADR attending a Weill Cornell Medicine-affiliated tertiary medical Centre in Doha Qatar (Hamad General Hospital). Using the LAAT, and modified Hallas tools, 4 independent raters (2 Clinical Pharmacologists, and 2 General Physicians) assessed and scored the 44 ADR-drug pairs. Agreement proportions between the rating pairs were evaluated as well individual/overall kappa statistics and intraclass correlation coefficients. We evaluated the weight of each of the 7 questions on the LAAT tool to ascertain its determinative role.Across 44 ADR-drug pairs, the overall median Fleiss kappa using the LAAT, and modified Hallas tools were 0.67 (interquartile range (IQR) 0.55, 0.76), 0.36 (IQR, 0.23-0.71) respectively. The overall percentage pairwise agreement with the LAAT and modified Hallas tools were 78.5%, and 62.2% respectively. Exact pairwise agreement occurred in 37 out of 44 (range 0.71-1), and 27 of 44 (0.53-0.77) ADR-drug pairs using the LAAT and modified Hallas tools respectively. Using the LAAT tool, the overall intraclass correlation coefficient was 0.68 (CI 0.55, 0.79), and 0.37 (CI 0.22, 0.53) with the modified Hallas tool.We report a higher proportion of "possible" and "definite" avoidability outcomes of adverse drug reactions compared with the modified Hallas, or that reported by developers of the LAAT tool. Although initially developed for research purposes, our report has suggested for the first time a potential applicability of this tool in clinical environment as well.