[Validation of a method for detecting hospitalizations for falls related to adverse drug reactions in geriatric departments]. / Validation d'une méthode de détection des hospitalisations pour chute liées à un effet indésirable médicamenteux en service de gériatrie aiguë.

Fall is one of the five main causes of drug-related hospital admissions (DRA) in France. A standardized chart review method, to identify DRA adapted to elderly patients, has recently been developed by Thevelin et al. Our first aim was to assess the reliability of this method for detecting DRA for falls in elderly subjects. Our second aim was to assess the feasibility of this method and to evaluate its reliability for assessing causality, the contribution of DRA to hospitalization, and the avoidability of DRAs in elderly patients hospitalized for a medication-related fall. A retrospective observational study was conducted on 16 patient cases admitted to the hospital for falls in May 2022, in the geriatric department of a French university hospital. Six healthcare professionals (pharmacists, pharmacologists, and geriatricians) assessed a method for detecting DRA individually and then in multidisciplinary pairs of raters. Inter-rater agreement (individually and in pairs) was assessed for DRA detection, causality, avoidability, and contribution of the DRA to hospitalization. A κ > 0,4 was considered a satisfactory threshold for agreement. The mean age was 86 years. When the assessment was done individually, detection of DRA-related hospitalizations (κ = 0,46; p < ,001), and DRA contribution to hospitalization (κ = 0,50; p < ,001) were moderately concordant. The causality assessment (κ = 0,09; p = 0,24) did not agree, and the avoidability assessment (κ = 0,63; p < ,001) agreed substantially. When the evaluation was done in pairs, detection of DRA-related hospitalizations (κ = 0,47; p < ,001) was moderately concordant between pairs. Avoidability assessment (κ = 0,79; p < ,001) concurred substantially. The assessment of causality (κ = 0,29; p = 0,01) and DRA contribution to hospitalization (κ = 0,38; p < .001) agreed fairly well. This study validated, individually and in pairs, the reliability of the method to identify DRA in the context of falls. This method will be of great use in research and epidemiological studies.