Forschung & Innovation
Publikationen
Development and validation of a risk prediction tool for drug-related problems in pre-operative elective surgical patients (mediPORT)
PMU Autor*innen
Stephanie Clemens, Wanda Lauth, Olaf Rose, Georg Zimmermann, Peter Gerner, Christina Duckelmann, Maria Flamm, Johanna Pachmayr
Alle Autor*innen
Stephanie Clemens, Clara Simon, Wanda Lauth, Olaf Rose, Georg Zimmermann, Peter Gerner, Christina Duckelmann, Maria Flamm, Johanna Pachmayr
Fachzeitschrift
PLOS ONE
Kurzfassung
Background Drug-related problems (DRP) in pre-operative care can harm patient outcomes. This study aimed to develop and validate a pre-operative risk prediction tool (mediPORT) to calculate the probability of DRP in admitted patients.Methods Elective surgery patients aged >= 18 years admitted to the pre-anaesthesia clinic and participating in a medication review by pharmacists were included in this case-control study. Routinely reported patient variables were included in a backward stepwise logistic regression to determine the most relevant predictors (minimum Akaike Information Criterion) of DRP. Performances using the area under the receiver operating characteristic curve (AUC) were assessed to test the model. Internal validation was performed using a 10-fold cross-validation procedure.Results The target population consisted of 11,176 participants, of whom 284 cases with >= 1 DRP and 980 controls without DRP were drawn. Most relevant predictors for DRP were age, number of drugs at admission, body mass index, sex and renal function. These factors were included in the final five variable model. A correlation between renal function and occurrence of DRP was found. Age and number of drugs frequently appeared in all models of the backwards elimination and represented an alternative two variable model. The AUC for predicting DRP were 0.823 (CI 95% 0.766-0.879) for the five-variable model and 0.872 (CI 95% 0.835-0.909) for the two-variable model. In the validation model, sensitivity was 77.6% and specificity was 76.5% for the five-variable model and 81.3%, 75% for the two-variable model, respectively.Conclusions Resulting equations can be used by hospital admission to identify patients at high risk, for whom a precise assessment of medication is critical.
Keywords
MODELS, Clinical pharmacist