Comparison of LACE and HOSPITAL Readmission Risk Scores for CMS Target and Nontarget Conditions

Area of Science:

• Health Services Research
• Clinical Informatics
• Predictive Analytics in Healthcare

Background:

• Accurate prediction of hospital readmissions is crucial for optimizing patient care and resource allocation.
• Existing risk prediction models like the LACE index and HOSPITAL score require evaluation across diverse patient populations and diagnostic groups.
• Understanding model performance variations based on specific diagnoses is essential for refining readmission risk stratification.

Purpose of the Study:

• To evaluate the utility and performance of the LACE index and HOSPITAL score in predicting 30-day readmission risks.
• To assess the accuracy of these models considering specific diagnostic categories, including CMS target diagnoses and cancer cohorts.
• To compare the predictive capabilities of the LACE index and HOSPITAL score in large academic medical center patient cohorts.

Main Methods:

• Retrospective analysis of 291,886 patient encounters from two academic medical centers (2011-2015).
• Data sourced from the Vizient Clinical Data Base and a regional health information exchange.
• Model performance assessed using Bayesian information criterion and area under the receiver operating characteristic curve (AUC) in overall, CMS diagnosis-based, and cancer diagnosis-based cohorts.

Main Results:

• Both LACE index and HOSPITAL score demonstrated good overall performance in predicting readmission risk.
• The LACE index exhibited slightly superior overall predictive accuracy (AUC 0.73 vs. 0.69, P ≤ 0.001).
• HOSPITAL score consistently outperformed the LACE index for four CMS target diagnoses, lung cancer, and colon cancer cohorts.
• Model performance varied significantly based on salient, diagnosis-specific risk factors.

Conclusions:

• Both LACE index and HOSPITAL score are valuable tools for predicting readmission risks in academic medical center populations.
• The choice of prediction model may depend on the specific patient cohort and diagnostic profile.
• Further research is warranted to refine these models for diagnosis-specific risk stratification to improve patient outcomes and healthcare efficiency.