AHRQ Study Examines Risk Models for Predicting Hospital Readmissions
Wednesday, April 27, 2016 8:49 AM
A risk prediction model designed to forecast hospital readmissions based on patients’ electronic health records (EHR) for entire hospital stays was only modestly better than a model based on EHR data for just the day of admission, according to a new AHRQ-funded study. The study on predicting 30-day, all-cause hospital readmissions was based on data from nearly 33,000 hospital stays at six Dallas-Fort Worth hospitals from 2009 to 2010.
The authors stated their model was the first to use comprehensive EHR data from entire hospital stays. That model was shown to be only modestly better at predicting risk of readmission despite including many additional clinically relevant prognostic factors. The authors noted customized models that are disease-specific may be more effective in predicting readmission compared with the common multi-condition, readmission risk prediction model.
The study “Predicting All-Cause Readmissions Using Electronic Health Record Data From the Entire Hospitalization: Model Development and Comparison” appeared in the Journal of Hospital Medicine. Access the abstract for more details.