Beeler PE, Eschmann E, Schneemann M, Blaser J

Beeler PE, Eschmann E, Schneemann M, Blaser J.. solitary tertiary care facility were used to identify DDIs that were regarded as a high-priority for contextualized alerting. A panel of DDI specialists developed algorithms that include drug and patient characteristics that impact the relevance of such warnings. The algorithms were then implemented as computable artifacts, validated using a synthetic health records data, and tested over retrospective PX-866 (Sonolisib) data from a single urban hospital. Results Algorithms and computable knowledge artifacts were developed and validated for a total of 8 high priority DDIs. Screening on retrospective real-world data showed the potential for the algorithms to reduce alerts that interrupt clinician workflow by more than 50%. Two algorithms PX-866 (Sonolisib) (citalopram/QT interval prolonging providers, and fluconazole/opioid) showed potential to filter nearly all interruptive alerts for these mixtures. Summary The 8 DDI algorithms are a step toward addressing a critical need for DDI alerts that are more specific to patient context than current commercial alerting systems. Data generally available in EHRs can improve DDI alert specificity. qualitative study on prescribers connection with electronic medication alerts showed that when alerts failed to provide contextual information, prescribers bypassed the alert and then searched for the relevant data that they needed.26 Quality improvement projects have found that making DDI alerts more appropriate to clinical context can improve alert acceptance. Daniels et al15 PX-866 (Sonolisib) observed a reduction in the override rate from 93.9% to 46.8% after making nearly a third (30.2%) of DDI alerts more contextual and suppressing another 16.5% of alerts. Similarly, Muylle et al27 were able to reduce alerts that interrupt clinician workflow (interruptive alerting) for potassium increasing potential DDIs by 92.8%, PX-866 (Sonolisib) with no statistically significant effect on the pace of hyperkalemia, by restricting alerts to only cases where a recent potassium value was 5 mMol/L. Objectives The primary target audience for this work is informatics leaders at health systems who are responsible for and/or interested in implementation and evaluation of CDS systems. The overarching goal of our work is to develop and validate algorithms that use data available in electronic health records (EHRs) as contextual info to provide higher specificity to DDI alerts that are frequently overridden. In this study, we describe the creation of algorithms that were validated on both synthetic and real-world EHR data. The algorithms apply to adult patients exposed to 1 or more of the interacting drug combinations. We expect the DDI algorithms to have significant impact because the involved drug combinations account for a disproportionate share of alert overrides.15,21 Moreover, we provide both logic circulation diagrams and computable artifacts the DDI algorithms that should help others to adapt them to their setting. METHODS Recognition of high priority DDIs and algorithm development High priority DDIs were identified based on the most frequently overridden alerts at the former University of Arizona Medical Center (right now Banner University Medical Center Tucson) over a 3-month period (Table 1A and B). A necessary condition for the DDIs to be high priority for this project was that more than 1000 overrides were observed. Drug specialists on the research team (4 pharmacists trained in DDIs: AVR, DCM, JRH, and Tgfb3 PDH) also assessed if evidence for potential harm existed in the literature, if the alert lacked specificity as implemented in the health system, and if the alert would be amenable to contextualization using EHR data, such as laboratory ideals, diagnoses, duration of use, and other.