Entwicklung einer Vorgehensweise zur automatisierten Erkennung eines Bedarfs an klinisch-pharmazeutischer Betreuung aus GKV-Routinedaten mittels Data-Mining

Abstract

Identifying patients receiving polypharmacy who are in need of pharmaceutical care - a predictive model This study was performed to identify aspects of patients receiving polypharmacy who are at risk of hospitalization. A retrospective database analysis of routine health insurance data from 2005-2010 was performed. Patients on polypharmacy were included. A predictive model was derived using SPSS 19.0 and logistic regression (n=44.108). The final model was validated using a second data set (n=45.739). Of 45.739 patients on polypharmacy 88.1 R0were using 5-8 drugs and 39.6R0were admitted to hospital within one year. Compared to using the number of drugs as a solely predictor (>13 drugs: n=489, PPV=59.9 20the model identified a larger group of patients with a higher probability of hospitalization (n=1.161, PPV=71.6¥20AUC=65.2µ2095R0CI 64.7-65.7 The strongest predictors for hospitalization were number of drugs per year, age, drug costs and the use of metamizol, opioids, loop-diuretics, phenprocoumon und clopidogrel. The derived predictive model improves identification of patients and helps addressing phramaceutical care to patients in need.