Inability of providers to predict unplanned readmissions.

Area of Science:

• Geriatric Medicine
• Health Services Research
• Patient Safety

Background:

• Hospital readmissions incur significant patient distress and financial costs.
• Accurate identification of high-risk patients is crucial for readmission reduction strategies.
• This study evaluated the predictive accuracy of healthcare providers and a standardized tool for patient readmissions.

Purpose of the Study:

• To assess the ability of physicians, case managers, and nurses to predict 30-day readmissions in older patients.
• To compare provider predictions against the Probability of Repeat Admission (P(ra)) risk tool.
• To identify accurate methods for predicting patient readmission risk.

Main Methods:

• Older patients (≥65) discharged from a tertiary academic medical center were enrolled.
• Inpatient teams estimated readmission chance and reasons at discharge.
• The Probability of Repeat Admission (P(ra)) score was calculated for each patient.
• Readmissions were tracked via electronic medical records and patient/caregiver follow-up calls.

Main Results:

• 32.7% of 159 eligible patients were readmitted within 30 days.
• Physician predictions were closest to actual rates; case managers, nurses, and P(ra) overestimated readmissions.
• Predictive discrimination was poor across all provider groups and the P(ra) (AUC range: 0.50–0.59).
• No group accurately predicted the reasons for readmission.

Conclusions:

• Readmission rates were higher than anticipated, potentially due to thorough follow-up.
• Neither healthcare providers nor the P(ra) tool accurately predicted high-risk patients for readmission.
• Hospitals lack effective predictive tools to guide efforts in reducing readmissions.