126 Chapter 5 The wards included were the Covid-19 ward, pulmonary ward, and internal medicine ward, with 24, 33, and 32 beds, and 26, 58, and 56 nurses, respectively. The latter represent the total number of nurses working on these wards, not the full time equivalent, and do not include nursing students. Managers’ enthusiasm and their expressed need for a different approach to PU prevention were key factors in selecting these wards. Figure 1 Visualization of DRAAI’s web application at the start of the trial, presenting at (high) risk patients from a specific ward with real-time feedback on follow-up actions of the predictions with the status of the nursing care plan (at-risk of PU). The intervention – DRAAI DRAAI was developed based on existing literature [16, 17, 25], expert opinions, and retrospective analysis of PU patients within our hospital. Dozens of predictors were selected from routinely collected data in the electronic patient records, including open text, continuous, and categorical fields. Examples of these predictors include age, expected remaining length of stay, mobility level, prescribed medications, planned surgeries, and laboratory results such as albumin levels. DRAAI predictions were validated using nursing diagnoses of PU [26], medical diagnoses, and PU observations recorded during prevalence measurements. These measurements are conducted three times a year according to hospital protocol, adhering to the EPUAP statement on PU prevalence monitoring [27]. The predictive performance of the final model, measured by the area under the ROC curve, exceeded the predictive performance of the Waterlow risk assessments (0.80-0.82 vs. 0.68-0.70). A detailed description of the development, validation, and calibration of DRAAI is provided in a separate publication [28].
RkJQdWJsaXNoZXIy MTk4NDMw