Fully independent validation and updating of a clinical pharmacy prioritizing risk score in an infectious disease hospital ward.

Abstract

The aim of this study was to validate and update the risk score originally developed at Hospital de Clínicas de Porto Alegre, verifying its performance in an infectious disease population. This is an observational study with consecutive selection of admission in a ward of participants with infectious diseases. Predictors were age, number of medications, intravenous drugs, potentially dangerous drugs, renal dysfunction, liver dysfunction, use of nasoenteral tube, nasogastric tube, gastrostomy feeding, jejunostomy feeding, oral enteral tube, total parenteral nutrition, cardiac or pulmonary dysfunction and immunosuppression. Outcome was defined as preventable prescription incidents by a clinical pharmacist. A GEE model was fit to make predictions each week. A total of 219 patients participated in the study, 79.25% of whom had prescription incidents in the first week of admission. Predictors of the updated model were number of drugs prescribed, number of intravenous drugs, use of tubes, truncated age at 36 years and week of hospitalization. The performance of the original model was poor. The updated model's discrimination and calibration were moderate (overall AUC 0.74). A calculator to apply the model is available at https://pedrobrasil.shinyapps.io/INDWELL/. The updated risk score enabled the user to make predictions at admission and throughout the weeks, allowing for a prioritized weekly update for clinical pharmacy intervention. The updated model has a moderate and satisfactory performance for infectious disease patients.