Development and Assessment of a New Framework for Disease Surveillance, Prediction, and Risk Adjustment: The...

Area of Science:

• Health Services Research
• Medical Informatics
• Health Economics

Background:

• Current US disease risk-adjustment formulas use outdated diagnostic classification frameworks.
• The International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) is the current standard for diagnostic coding.

Purpose of the Study:

• To develop an ICD-10-CM-based classification framework for predicting healthcare payment, quality, and performance outcomes.
• To compare the predictive performance of the new framework against existing systems.

Main Methods:

• Physician teams mapped ICD-10-CM diagnoses into diagnostic items (DXIs): main effects, modifiers, and scaled variables.
• Stepwise and weighted least-squares estimation predicted 14 healthcare outcomes using DXIs.
• Model performance was validated and compared against the Agency for Healthcare Research and Quality Clinical Classifications Software Refined (CCSR) and Health and Human Services Hierarchical Condition Categories (HHS-HCC).
• The IBM MarketScan Commercial Claims and Encounters Database (2016-2018) was utilized.

Main Results:

• A total of 3223 DXIs were created from the ICD-10-CM codes.
• The DXI model achieved a validated R-squared of 0.589, outperforming CCSR (0.539) and HHS-HCC (0.428).
• The DXI system reduced underpayment for rare diagnoses by 83% compared to HHS-HCCs.

Conclusions:

• The novel DXI classification system demonstrated superior predictive accuracy for healthcare spending and utilization outcomes.
• This ICD-10-CM-based framework offers a significant improvement over existing diagnostic classification systems for risk adjustment.