Related Experiment Videos

Comparison of an Updated Risk Stratification Index to Hierarchical Condition Categories

George F Chamoun1, Linyan Li, Nassib G Chamoun

  • 1The Lown Institute, Boston, Massachusetts (G.F.C., N.G.C., V.S.); Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts (L.L.); and Department of Outcomes Research, Cleveland Clinic, Cleveland, Ohio (D.I.S.).

Anesthesiology
|October 10, 2017
PubMed

Insights

The rederived Risk Stratification Index (RSI) shows improved discrimination and prediction accuracy compared to Hierarchical Condition Categories (HCC) for mortality and length of stay, making it superior for quality metrics.

Area of Science:

  • Health Services Research
  • Health Economics
  • Medical Informatics

Background:

  • Two primary models, Risk Stratification Index (RSI) and Hierarchical Condition Categories (HCC), assess baseline risk using comorbidities and procedures.
  • HCC models are updated annually, while the RSI has not been rederived since 2010.
  • A direct comparison between the RSI and HCC models has been lacking.

Purpose of the Study:

  • To rederive the Risk Stratification Index using recent data.
  • To directly compare the performance of the rederived RSI against the contemporaneous Hierarchical Condition Categories model.
  • To evaluate the predictive accuracy and discrimination capabilities of both risk adjustment models.

Main Methods:

  • Reimplementation of the original Risk Stratification Index derivation procedures using Medicare data from 2007-2011.
  • Construction of Hierarchical Condition Categories using Center for Medicare and Medicaid Services-provided software on the entire dataset.
  • Comparison of model discrimination using C-Statistics and evaluation of predictive accuracy via calibration plots.

Main Results:

  • The rederived Risk Stratification Index demonstrated improved discrimination compared to its original derivation.
  • The RSI significantly outperformed the HCC model in predicting in-hospital, 30-day, and 1-year mortality, as well as hospital length-of-stay.
  • Both models showed linear predictive accuracy upon calibration, but the RSI exhibited less variance in its predictions.

Conclusions:

  • The enhanced discrimination and reduced variance in predictions establish the Risk Stratification Index as superior to the Hierarchical Condition Categories.
  • The rederived RSI offers a robust foundation for developing reliable care-quality metrics.
  • The RSI's improved performance supports its utility for objective provider comparisons in healthcare.
Abstract

Related Concept Videos