Development of a multivariable prediction model for identification of patients at risk for medication transfer errors at ICU discharge

Introduction Discharge from the intensive care unit (ICU) is a high-risk process, leading to numerous potentially harmful medication transfer errors (PH-MTE). PH-MTE could be prevented by medication reconciliation by ICU pharmacists, but resources are scarce, which renders the need for predicting which patients are at risk for PH-MTE. The aim of this study was to develop a prognostic multivariable model in patients discharged from the ICU to predict who is at increased risk for PH-MTE after ICU discharge, using predictors of PH-MTE that are readily available at the time of ICU discharge.

Material and methods Data for this study were derived from the Transfer ICU Medication reconciliation study, which included ICU patients and scored MTE at discharge of the ICU. The potential harm of every MTE was estimated with a validated score, where after MTE with potential for harm were indicated as PH-MTE. Predictors for PH-MTE at ICU discharge were identified using LASSO regression. The c statisticprovided a measure of the overall discriminative ability of the prediction model and the prediction model was internally validated by bootstrap resampling. Based on sensitivity and specificity, the cut-off point of the prediction model was determined.

Results The cohort contained 258 patients and six variables were identified as predictors for PH-MTE: length of ICU admission, number of home medications and patient taking one of the following medication groups at home: vitamin/mineral supplements, cardiovascular medication, psycholeptic/analeptic medication and medication for obstructive airway disease. The c of the final prediction model was 0.73 (95%CI 0.67–0.79) and decreased to 0.62 according to bootstrap resampling. At a cut-off score of two the prediction model yielded a sensitivity of 70% and a specificity of 61%.

Conclusions A multivariable prediction model was developed to identify patients at risk for PH-MTE after ICU discharge. The model contains predictors that are available on the day of ICU discharge. Once external validation and evaluation of this model in daily practice has been performed, its incorporation into clinical practice could potentially allow institutions to identify patients at risk for PH-MTE after ICU discharge, on the day of ICU discharge, thus allowing for efficient, patient-specific allocation of clinical pharmacy services.