The Truth is in the Data - Differences in the Same Measure Based on Different Sources among HVHC Members Using ICU...

Area of Science:

• Healthcare Informatics
• Health Services Research
• Clinical Data Analysis

Background:

• Intensive Care Unit (ICU) length of stay (LOS) is a key indicator of illness severity and cost in sepsis patient care.
• Discrepancies exist between electronic health record (EHR) data and Centers for Medicare and Medicaid Services (CMS) payment data regarding ICU LOS.
• Accurate ICU LOS measurement is crucial for understanding sepsis patient severity and associated healthcare costs.

Purpose of the Study:

• To investigate discrepancies in ICU LOS data between EHR and CMS claims for sepsis patients.
• To test the hypothesis that excluding intermediate ICU stays would improve data concordance.
• To refine the definition of ICU LOS for more accurate severity and cost assessments.

Main Methods:

• Linked 26,733 unique patient EHR records with CMS demographic, diagnosis, and revenue code data.
• Identified discrepancies in ICU days between EHR and CMS claims.
• Revised the CMS ICU definition by removing intermediate ICU LOS to test its impact on data agreement.

Main Results:

• Initial agreement between EHR and CMS data was 73%.
• After removing intermediate ICU LOS, the agreement between EHR and CMS data improved significantly to 86%.
• The exclusion of intermediate ICU stays led to a closer match between the two data sources.

Conclusions:

• Including intermediate ICU stays in CMS data analysis dilutes the accuracy of ICU LOS for estimating sepsis patient severity and cost.
• Ensuring consistency in clinical definitions across different data sources (EHR vs. CMS) is essential for reliable analysis.
• Early engagement with clinicians, analysts, and coders is vital for navigating data complexities from diverse sources.