Abstract: PO0211
Diagnosing and Staging AKI in the Absence of a Baseline Serum Creatinine Value
Session Information
- AKI: Epidemiology, Risk Factors, and Prevention
November 04, 2021 | Location: On-Demand, Virtual Only
Abstract Time: 10:00 AM - 12:00 PM
Category: Acute Kidney Injury
- 101 AKI: Epidemiology, Risk Factors, and Prevention
Authors
- Buchkremer, Florian, Kantonsspital Aarau, Division of Nephrology, Aarau, Switzerland
- Segerer,, Stephan, Kantonsspital Aarau, Division of Nephrology, Aarau, Switzerland
Background
AKI is commonly diagnosed and classified from changes to serum creatinine according to the 2012 KDIGO criteria. When baseline creatinine is missing, the guideline recommends to back calculate it from an assumed MDRD-GFR of 75ml/min/1.73m2.
We describe an alternative method.
Methods
From NHANES 2015-2018 data we calculated distribution of serum creatinine values for the adult US population as a whole, and for gender, age and weight subgroups.
We then assessed mean values in an external validation cohort (NHANES 2011-14) for performance to predict baseline creatinine in comparison to back calculated MDRD values.
Results
Relative differences between back calculated MDRD and measured creatinine values in the validation cohort show a median bias of +8% and an interquartile precision range of 0% to +26% (Fig 1). Accuracy is rather low, with P15 and P30 values at 42% and 71%.
In contrast, our gender/age-based estimation eliminates bias to 0 % and improves precision, interquartile range of -6% to +13% (Fig 1). P15 increases to 58%, P30 to 86%.
The relative differences show a clear age dependency for MDRD, that is not present in our gender/age-based estimation (Fig 2).
Adding weight categories did not significantly improve our predictions.
Conclusion
We describe a simple method to estimate missing baseline creatinine values for assessing acute kidney injury.
Compared with the current standard approach our method shows no bias, more precision and improved accuracy in predicting baseline creatinine on a population level.