Validation d’une version française de l’outil AT-HARM10 pour la détection des hospitalisations liées au médicament

Abstract

Admissions of the elderly related to medication errors are frequent in hospital, more than half would be avoidable, but there is currently no validated method in French to identify them. The objective of this work was to validate the French version of the AT-HARM10tool in order to use it for patients admitted in our healthcare facilities. The tool has 10questions. A positive response to any of the first 3questions identify admissions that are unlikely to be drug-related. A positive response to one of the following 7questions identify possible medication-related admissions. For semantic and linguistic validation, we performed cross-validation with forward-backward translation. To clinically validate the method, we conducted a retrospective study including patients over 65admitted to short-stay units (UHCD) and to orthopedic surgery units in two French hospitals. Two hundred and sixty-six (266) patients were included; 166patients admitted to UHCD (mean age 86.0±5.7years; sex ratio 0.66; mean number of drugs prescribed 7.7±3.8) and 100patients admitted to orthopedic units (mean age 85.2±6.1years; sex ratio 0.43; mean number of prescribed drugs 6.4±3.6). We identified 55% of admissions probably related to medication in UHCD and 76% in orthopedic units (p<0.05). The most represented item was P5in both groups (Might [side] effects of the medications the patient was taking [prescribed or not prescribed] prior to hospitalization have caused the admission [including over-treatment]? The validated AT-HARM10tool is now integrated into our clinical pharmacy practices and medication reviews are offered as a priority to patients admitted for iatrogenic reasons.