412 CHAPTER 5 value in a hospital setting (~40%), considerably high compared to other riskprediction models for drug-related harm or regular drug safety monitoring [69– 75]. Similarly, the adjudication guide used within OPERAM for DRA assessment had a PPV of 42% [60]. Therefore, these tools may be valuable screening tools to select older patients at risk for potential drug-related harm. However, increasing the availability of structured, electronic patient data (e.g. medical conditions) is necessary before clinical practice implementation. The increased use of digital health combined with the rapidly evolving science of machine learning allows exploring new methods for developing more sophisticated predictive models. For instance, a recent systematic review concluded that prediction models using EHR data perform better than administrative data [75]. Another study demonstrated that artificial intelligence could identify patients at high risk of heart failure or death after myocardial infarction [76]. Thus, the availability of large, real-time patient data combined with computer sciences is a promising strategy for future differentiation between older people at high or lower risk for drug-related harm, thereby increasing the PPV of existing screening tools based on medication use. Until then, an integrated approach based on known highrisk factors (e.g. polypharmacy, multimorbidity, older age, high-risk medication), expert opinions (including frailty assessment) and patient preferences remain the next best option to select patients who are at the highest risk for drug-related harm (Figure 3) [6]. 4.2. When, where and how to performmedication optimisation interventions We focused on medication optimisation in a hospital setting for older people with multimorbidity and polypharmacy. Our choice to study an in-hospital intervention is understandable given the association between inappropriate prescribing and the increased risk of drug-related hospital admissions. In addition, the hospital is a high-risk environment prone to medication errors, including those resulting from care transitions. However, preventing medication-related harm is better than cure. In addition, the average hospital stay of patients ≥70 years is only five to seven days in the Netherlands, the shortest among European countries, impeding long-term monitoring and follow-up [77]. Therefore, whether the acute hospital setting is the most appropriate setting to conduct medication reviews could be questioned. Nevertheless, the hospital environment facilitates specific geriatric and non-geriatric expertise (e.g. cardiovascular, pain management) with easy access to diagnostics, laboratory values and therapeutic drug monitoring. For example, a comprehensive geriatric assessment (CGA) has proven benefits in a clinical setting, while the benefits of CGA in primary care remain uncertain [78,79].
RkJQdWJsaXNoZXIy MTk4NDMw