Development and Validation of a Predictive Model for Short- and Medium-Term Hospital Readmission Following Heart...

Area of Science:

• Cardiovascular Surgery
• Health Services Research
• Predictive Analytics

Background:

• No existing models predict hospital readmission after heart valve surgery (HVS).
• Coronary artery bypass surgery has predictive models, highlighting a gap in HVS research.

Purpose of the Study:

• To identify and validate factors predicting short- and medium-term hospital readmission post-HVS.
• To develop a clinical tool for identifying high-risk HVS patients.

Main Methods:

• Utilized the Premier Inpatient Database (2007-2011) with a diverse sample of US hospitals.
• Employed a generalized estimating equation model accounting for hospital clustering.
• Examined patient, hospital, and clinical factors in 38,532 HVS patients.

Main Results:

• Identified 7.8% and 12.8% readmission rates at 1 and 3 months, respectively.
• Developed a 3-month readmission model with fair discrimination (C-statistic, 0.67) and good calibration.
• The simplified REVEaL model identified 5 key predictors: packed Red blood cells, End-stage renal disease, Valve surgery type, Emergency admission, and hospital Length of stay.

Conclusions:

• Successfully described and validated key predictors for HVS readmission.
• The developed models enable clinicians to identify high-risk HVS patients.
• Aims to improve post-discharge care and follow-up for at-risk individuals.