electronic health records, neurocritical care, readmissions, systems of healthcare



  1. Peacock, Sarah
  2. Siegel, Jason
  3. Harmer, Emily
  4. Alejos, David
  5. Freeman, W. David


ABSTRACT: Background and Purpose: Thirty-day hospital readmissions have been shown to be a measure of quality and result in higher mortality and increased costs. Readmissions are a target for hospitals and payers; thus, several centers have developed predictive readmission scores to identify high-risk patients. The purpose of this study was to evaluate the current hospital-wide readmission risk calculator and the ability of this tool to predict 30-day readmissions in the neurocritical care population. Methods: A retrospective chart review was performed that included 340 consecutive patients admitted to our neuroscience critical care unit. Data including readmission scores, reason for admission, length of stay, and whether they were readmitted were recorded. Results: After removing patients without readmission scores or who died at the end of the original admission, the records of N = 279 patients were analyzed. Patients were more likely to be readmitted if they were initially emergently hospitalized or if there was a history of malignancy. Readmitted patients had a longer original hospital length of stay. Furthermore, 65.8% of the patients who were given a "low risk" for readmission were readmitted within 30 days. Conclusions: This small set of data in a specific patient population found that the current risk prediction score was inaccurate in predicting readmission in the neuroscience intensive care unit population. Further evaluation is needed of a larger patient population to generalize these results for all neuroscience intensive care unit patients. To design an accurate readmission risk tool, centers should create unique readmission scores based on less heterogeneous patient populations.