Kidney Week

Abstract: FR-PO0019

Development of a Point-of-Care Model to Estimate Urine Creatinine Using Urine Dipstick Data

Session Information

• Artificial Intelligence and Digital Health at the Bedside
  November 07, 2025 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Artificial Intelligence, Digital Health, and Data Science

• 300 Artificial Intelligence, Digital Health, and Data Science

Authors

• Tummala, Snikitha, Johns Hopkins Medicine, Baltimore, Maryland, United States

Snikitha Tummala

• Xue, Jiashu, Johns Hopkins Medicine, Baltimore, Maryland, United States

Jiashu Xue, MSc

• Moledina, Dennis G., Yale School of Medicine, New Haven, Connecticut, United States

Dennis G. Moledina, MD, PhD, FASN

• Coca, Steven G., Icahn School of Medicine at Mount Sinai, New York, New York, United States

Steven G. Coca, DO, MS

• Menez, Steven, Johns Hopkins Medicine, Baltimore, Maryland, United States

Steven Menez, MD

• Bitzel, Jack, Johns Hopkins Medicine, Baltimore, Maryland, United States

Jack Bitzel

• Thiessen Philbrook, Heather, Johns Hopkins Medicine, Baltimore, Maryland, United States

Heather Thiessen Philbrook

• Parikh, Chirag R., Johns Hopkins Medicine, Baltimore, Maryland, United States

Chirag R. Parikh, MD, PhD, FASN

Background

Urine creatinine is essential to adjust various urine biomarkers for urine concentration changes but while biomarkers can be measured at the point-of-care, urine creatinine is often unavailable or delayed. Urine specific gravity is widely available as part of urine dipstick analysis. Here we created an equation to predict urine creatinine concentration from urine dipstick specific gravity.

Methods

Patients from a prospective cohort with outpatient urine tests were included. Linear regression models with a gamma distribution were developed to predict urine creatinine from urine specific gravity alone (univariable model) and with age, gender, diabetes, proteinuria and glucosuria (multivariable model).

Results

The study population (n=426) had a mean (SD) age of 53.46 (16.36) years; 40.8% were male, and 23.2% had diabetes. The median [IQR] urine creatinine was 88.2 [52.78, 128.90] mg/dl, and median [IQR] specific gravity was 1.015 [1.011, 1.020]. The percentage of predictions within 30% of observed values (P₃₀) was 53% for the univariable model and improved to 58% for the multivariable model. In a subset of 287 participants with urine creatinine between 50-200 gm/dl, the P₃₀ for the multivariable model reached 68%. The predicted urine creatinine values from the univariable and multivariable models are presented in table 1.

Conclusion

This study demonstrates that urine creatinine can be estimated rapidly and inexpensively from urine specific gravity and other common dipstick measurements. This approach offers a promising avenue for point-of-care urine creatinine assessment, especially in resource-limited settings.

Funding

• NIDDK Support

Digital Object Identifier (DOI)

• doi: 10.1681/ASN.2025sdqyygne