339 Hospital physicians’ and older patients’ agreement with individualised STOPP/START recommendations be explained by the hospital setting in the present study. In addition, potentially inappropriate medication (PIM) use was not associated with patients’ willingness to deprescribe one or more of their medications (74.3% without PIMs versus 79.9% with PIMs) in prior studies [24]. Female gender was associated with more PIM use (based on Beers criteria), especially benzodiazepines, Z-drugs and ≥3 concurrent psychoactive drugs, but not with willingness to deprescribe. We found no gender difference in PIM or PPO prevalence, but we did find an association between female gender and higher agreement with recommendations (both STOPP and START). This is an interesting new finding that needs to be confirmed in future research. Although patients’ reluctance to medication adjustments was an important reason for disagreement, factors within the attending physician and environmental constraints were also prevalent. Postponed recommendations to the GP (21% in total) were frequently associated with attending physicians feeling ill-equipped to take responsibility for suggested medication changes beyond their area of expertise, as we found for cardiovascular medication. These factors correspond relatively well with those found by Dalton et al., who investigated factors affecting prescriber implementation of computer-generated medication recommendations within the SENATOR trial [25,26]. Although the SENATOR-derived study significantly differs in methodology and outcome from our study, four important barriers for implementation were elucidated, of which some were partly overcome in our trial i.e. 1) computerised output leading to recommendations with low clinical relevance, thereby limiting their uptake; 2) the hospital environment with associated time constraints within the busy clinical environment and desire to devolve responsibility of managing older patients’ pharmacotherapy to GPs; 3) prescriber factors, particularly prescriber inertia and lack of awareness of the highly prevalent ADRs, reluctance to prescribe outside their therapeutic specialty; 4) patient factors, particularly the overriding focus on the patient’s acute status, where reviewing the prescribing recommendations was not a high priority for many attending physicians [25]. All pharmacotherapy optimisation recommendations that were discussed with the patient and the physician in our study, were already evaluated for appropriateness for the individual patient by the PT. This resulted in rejection of 603 out of 1059 (56.9%) STOPP/START signals generated by the CDSS during pharmacotherapy analysis in Dutch patients, based on information present in the patients’ medical records (results of this evaluation process are published elsewhere) [16,27]. Therefore, the category ‘computerised output’ was not applicable to our study, as all recommendations discussed were considered relevant to the patient by the PT. Additionally, our output was discussed face-to-face with both patient and attending physician, in contrast to providing a printed report with recommendations to the attending physician and nothing more. These factors would likely contribute to higher implementation rates than those found in the SENATOR trial (15%) and could explain the overall agreement of 60%we found in our study [26]. 4
RkJQdWJsaXNoZXIy MTk4NDMw