269 OPERAM: cluster randomised controlled trial - SI Definitions of underuse, overuse and misuse in Table 5 Underuse, overuse and misuse were based on START and STOPP criteria version 2, and using an algorithm run on the trial database. The START criteria were used to detect drug underuse (i.e., potential prescribing omissions); each STOPP criterion was categorized as either measuring overuse or measuring misuse (i.e., potentially inappropriate medication). In total, 30 of 34 START criteria and 65 or 80 STOPP criteria were included and measured, as some criteria required data that were not available (mainly (i) laboratory measurements that were not available at two months in this pragmatic RCT, and (ii) the implicit STOPP criteria A1, A2, and A3 that require evaluation at patient-level by a trained clinician. We developed and validated an algorithm for the measurement of the following outcomes: drug underuse, drug overuse, drug misuse. The algorithm was developed from previous experience and reports from our team related to the automated detection of STOPP and START criteria [8,9]. Research Team statisticians and programmers (Prof. Dimitris Mavridis and Mr Agapios Panos, University of Ioannina, Greece) developed an R package that provided automated evaluation for each criterion (https://github. com/agapiospanos/StartStopp). In summary, detection was performed by using a validated algorithm (that was applied to the research database), based on the STOPP and START criteria. Drug-drug interactions were assessed using a validated consensus-based list of 66 drug-drug-interaction criteria that we have recently published [10]. Once again, research team statisticians and programmers developed an R package that evaluated patient data for drug-drug interactions based on these criteria, using ATC coded medication lists (https://github.com/agapiospanos/DDI). This algorithm identifies combinations of ATC codes and was pilot tested in several rounds to check for accuracy in the detection. 3
RkJQdWJsaXNoZXIy MTk4NDMw