Kidney Week

Abstract: FR-PO1097

Evaluating a Computable Phenotype for CKD Detection in Adult Patients Treated in Primary Care

Session Information

• CKD: Epidemiology, Risk Factors, and Prevention - 2
  October 25, 2024 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: CKD (Non-Dialysis)

• 2301 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention

Authors

• Nguyen, Joseph D., University of Virginia, Charlottesville, Virginia, United States

Joseph D. Nguyen

• Kunaprayoon, Lalida, University of Virginia, Charlottesville, Virginia, United States

Lalida Kunaprayoon, MD

• Silverberg, Benjamin, University of Virginia, Charlottesville, Virginia, United States

Benjamin Silverberg, MD

• Mallawaarachchi, Indika V., University of Virginia, Charlottesville, Virginia, United States

Indika V. Mallawaarachchi, MS

• Ma, Jennie Z., University of Virginia, Charlottesville, Virginia, United States

Jennie Z. Ma

• Lyman, Jason A., University of Virginia, Charlottesville, Virginia, United States

Jason A. Lyman, MD, MS

• Scialla, Julia J., University of Virginia, Charlottesville, Virginia, United States

Julia J. Scialla, MD, FASN

Background

Chronic kidney disease (CKD) affects about 37 million Americans, particularly those with diabetes and hypertension, and is often detected via laboratory studies prior to symptom onset. This project aims to evaluate the accuracy of a computable phenotype to detect adults with CKD stage 3 or higher. The ultimate goal is to improve CKD recognition and evaluation by launching a best practice advisory (BPA) in patients likely to have CKD.

Methods

We developed an algorithm to identify probable cases of CKD in adult patients seen at a primary care clinic at one academic medical center between October 2023 and February 2024. The algorithm identifies patients with estimated glomerular filtration rate (eGFR)<60 ml/min/1.73m2 during the clinic visit, as well as on a prior date (90 to 730 days prior to the index test). If a patient identified by the algorithm is not yet enrolled into the EPIC CKD registry, the BPA alert is activated for probable CKD. If a patient was alerted multiple times, we analyzed the first alert. For patients flagged for probable CKD, we conducted chart review to collect: (1) laboratory parameters for CKD and (2) data relevant to the clinical impression of CKD that may also consider other elements like markers of kidney damage (e.g. albuminuria) or structural kidney disease (e.g. imaging abnormality).

Results

Of 277 probable CKD cases, 155 individuals were alerted. 133 of these individuals met stringent laboratory criteria for CKD stage 3 or higher (e.g. no intervening normal that disagree). 115 met laboratory criteria and were also deemed to have CKD by the clinical impression of the reviewer (74% of total alerted individuals). The remaining 26% either did not fully meet laboratory criteria (n=22) or were deemed to not clearly have CKD clinically (n=18). Most of these cases were due to either borderline eGFR near 60 ml/min/1.73m2, disagreement between different eGFR equations, recovered eGFR, or oscillating eGFR.

Conclusion

Our computable CKD definition demonstrated a relatively high rate of confirmed CKD, but clinician confirmation is advisable to ensure diagnostic accuracy prior to recommending further evaluation or treatment. Future directions include refining the BPA to include other measures of kidney damage in patients with equivocal eGFR.