Abstract: PUB055
Kidney HEALTH: Harnessing e-Records for Longitudinal Kidney Disease Surveillance and Population Health Research Through an Electronic Health Record (EHR)-Integrated Registry
Session Information
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Paulus, Amber, Virginia Commonwealth University Department of Internal Medicine, Richmond, Virginia, United States
- Street, Sarah E., Virginia Commonwealth University Department of Internal Medicine, Richmond, Virginia, United States
- Christensen, Johanna L., Virginia Commonwealth University Department of Internal Medicine, Richmond, Virginia, United States
- Kidd, Jason M., Virginia Commonwealth University Department of Internal Medicine, Richmond, Virginia, United States
- Gupta, Gaurav, Virginia Commonwealth University Department of Internal Medicine, Richmond, Virginia, United States
Background
Chronic kidney disease (CKD) remains underdiagnosed across clinical settings. To improve early detection, monitoring, and intervention, VCU Health developed Kidney HEALTH, an EHR-integrated registry to classify and track patients with or at risk for kidney disease.
Methods
A multidisciplinary team developed the registry using Agile methods and Epic tools (Reporting Workbench, SmartForms, and flowsheets). Adults ≥18 with qualifying kidney-related diagnoses, labs, or procedures over the past five years were categorized into four cohorts: AKI, AKD, CKD, and NKD, per the KDIGO-aligned Levey framework. CKD required ≥3 months of sustained criteria. Registry logic and SmartForms were designed to enable clinical decision support launching in August 2025.
Results
Patients were initially staged by eGFR (N=53,625) to build the CKD registry, as urine protein tests varied. UACR, used in kidney failure risk equation (KFRE), was reported in 3.4% (1,809) vs. UPCR in 25.2% (13,539). Only 1.9% (1,036) had abnormal UACR, while 3.6% (1,954) had abnormal UPCR. These abnormal results define the longitudinal cohort. UPCR use increased by stage, highlighting systemic preference despite UACR’s role in risk stratification.
Conclusion
The Kidney HEALTH registry provides a scalable EHR-integrated framework for identifying and tracking patients with kidney disease. By staging patients initially through eGFR to overcome variability in proteinuria assessments, we revealed significant underutilization of UACR, the preferred input for risk stratification via KFRE, despite widespread UPCR use. These findings inform future CDS deployment and highlight critical gaps in CKD evaluation that must be addressed to advance data-driven kidney care.
Figure 1. Cohort Classification Framework for Kidney HEALTH Registry Based on Functional and Structural Criteria and Timeframe