The Contribution of Lab Results Data Linked to Claims Data in Identifying Chronic Kidney Disease Staging
Author(s)
Morgan Decker, MS, Lillian Hang, MBA, MPH, Stacey DaCosta Byfield, MPH, PhD;
Optum, Eden Prairie, MN, USA
Optum, Eden Prairie, MN, USA
OBJECTIVES: Analysis of ICD10 diagnosis codes via claims data offers a cost-efficient way to identify patterns in patient care and opportunities for targeted disease management. However, claims data can lack details in capturing accurate disease status. To highlight the benefits of including lab results with claims data to appropriately identify disease severity, this analysis compared chronic kidney disease (CKD) stage based on ICD-10 codes in claims data to CKD stage based on lab results data.
METHODS: This was a retrospective analysis of linked de-identified estimated glomerular filtration rate (eGFR) lab results and de-identified administrative claims for commercially insured and Medicare Advantage enrollees in the OptumLabs Data Warehouse. Subjects had ≥2 valid eGFR results (calculated by lab providers), ≥90 days apart between 01/01/2022-12/31/2023; and the first qualifying eGFR defined index date. Continuous enrollment for 360 days before and 30 days after index was required. Individuals were categorized into CKD stages based on: (1) eGFR values according to Kidney Disease Outcomes Quality Initiative: early-stage, stage 3, stage 4, and stage 5 (lab-based stage) and (2) ICD-10 codes on claims (claims-based stage). These stages were compared and Cohen’s Kappa statistic used to assess agreement (<0.4 indicates poor agreement).
RESULTS: Overall, 2,277,122 individuals with valid eGFR results were identified. Patient age averaged 65.1 (SD 14.4) years, 56% female and 40% commercial. Distribution by lab-based stage: 81% early, 17% stage 3, 2% stage 4, and <1% stage 5. Among those with lab-based stage ≥3, 69% had a corresponding or higher code-based stage and 31% had no or less severe code-based CKD staging (under-coding). The under-coding rate was 40% for commercial enrollees and 30% for Medicare Advantage enrollees. Lab-based and code-based stages had poor agreement (K=0.104).
CONCLUSIONS: This analysis shows that including eGFR lab results with claims data may enhance the ability to appropriately categorize patients by CKD stage.
METHODS: This was a retrospective analysis of linked de-identified estimated glomerular filtration rate (eGFR) lab results and de-identified administrative claims for commercially insured and Medicare Advantage enrollees in the OptumLabs Data Warehouse. Subjects had ≥2 valid eGFR results (calculated by lab providers), ≥90 days apart between 01/01/2022-12/31/2023; and the first qualifying eGFR defined index date. Continuous enrollment for 360 days before and 30 days after index was required. Individuals were categorized into CKD stages based on: (1) eGFR values according to Kidney Disease Outcomes Quality Initiative: early-stage, stage 3, stage 4, and stage 5 (lab-based stage) and (2) ICD-10 codes on claims (claims-based stage). These stages were compared and Cohen’s Kappa statistic used to assess agreement (<0.4 indicates poor agreement).
RESULTS: Overall, 2,277,122 individuals with valid eGFR results were identified. Patient age averaged 65.1 (SD 14.4) years, 56% female and 40% commercial. Distribution by lab-based stage: 81% early, 17% stage 3, 2% stage 4, and <1% stage 5. Among those with lab-based stage ≥3, 69% had a corresponding or higher code-based stage and 31% had no or less severe code-based CKD staging (under-coding). The under-coding rate was 40% for commercial enrollees and 30% for Medicare Advantage enrollees. Lab-based and code-based stages had poor agreement (K=0.104).
CONCLUSIONS: This analysis shows that including eGFR lab results with claims data may enhance the ability to appropriately categorize patients by CKD stage.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
RWD161
Topic
Real World Data & Information Systems
Disease
SDC: Urinary/Kidney Disorders