BACKGROUND: The appropriateness of psychotropic prescriptions in the elderly is a major quality-of-care challenge at hospital. Quality indicators have been developed to prevent inappropriate psychotropic prescriptions. We aimed to select and automatically calculate such indicators, from the Bordeaux University Hospital information system, and to analyze the appropriateness of psychotropic prescription practices, in an observational study. METHODS: Experts selected indicators of the appropriateness of psychotropic prescriptions in hospitalized elderly patients, according to guidelines from the French High Authority for Health. The indicators were reformulated to focus on psychotropic administrations. The automated calculation of indicators was analyzed by comparing their measure to data collected from a clinical audit. In elderly patients hospitalized between 2014 and 2015, we then analyzed the evolution of the appropriateness of psychotropic prescription practices during hospital stay, using methods of visualization, and described practices by considering patients' characteristics. RESULTS: Two indicators were automated to detect overuse and misuse of psychotropic drugs. Indicators identified frequent inappropriate drug administrations, but practices tended to become more appropriate after quality-of-care improvement actions. In the majority of patients (85%), there was no inappropriate administration of psychotropic drugs during hospital stay; for the remaining 15% with at least one inappropriate administration, physicians tended to limit overuse or misuse during hospital stay. Inappropriate administrations were more frequent in patients suffering from psychiatric disorders, dependence and associated complications or morbidities. CONCLUSIONS: The automated indicators are structuring tools for the development of a drug prescription monitoring system. Inappropriate psychotropic administrations were limited by physicians during hospital stay; some inappropriate prescriptions might be explained by clinical characteristics of patients.< Réduire