Hospital readmission risk analytics: Identifying high-risk patients and reducing 30-day readmission rates through predictive analytics

Hospital readmissions within 30 days of discharge represent one of the most significant sources of preventable healthcare cost and avoidable patient harm in the United States. The Centers for Medicare & Medicaid Services (CMS) penalizes hospitals with excess readmission rates through the Hospital Readmissions Reduction Program (HRRP), creating both financial and quality improvement imperative.
This project developed and validated a logistic regression predictive analytics framework for identifying patients at elevated risk of 30-day readmission across five CMS-priority diagnostic categories. Using CMS HRRP and HCUP NRD data supplemented with CDC Social Vulnerability Index (SVI), the analysis produced risk stratification models with cross-validated AUC of 0.722, identified the strongest clinical and social predictors of readmission, and developed actionable reporting tools for discharge planning and care coordination teams.
500
Patients Analyzed    27.8%
Overall Readmission Rate    75%
High-Risk Readmit Rate    0.722
5-Fold CV AUC Score    8.0%
High-Risk Segment
Key findings: Social determinants: dual Medicare/Medicaid eligibility (OR=1.42), absence of documented follow-up (OR=1.35), high community SVI (OR=1.23) are among the most powerful predictors of 30-day readmission, comparable to clinical severity indicators. Prior hospitalization within 12 months is the single strongest predictor overall (OR=2.31). Targeted intervention in the highest-risk 8% of patients, who account for a disproportionate share of readmissions, could generate substantial improvements in population-level readmission performance.