Population-Based Estimated Reference Creatinine Values: A Novel Method of a Robust Electronic Acute Kidney Injury Alert System

Abstract

Background: Acute kidney injury (AKI) is common in hospitalized patients. Despite the progress that has been made in the last decade, early identification of AKI cases remains a challenge. In recent years, electronic AKI alert (e-AKI alert) systems have been tested and are usually based on changes in serum creatinine (Cr) values. However, these methods do not include one of the common scenarios, i.e. when there is no available preadmission Cr value available for a patient to compare and hence an e-AKI alert cannot be issued. Therefore, it is essential to have an alternative algorithm to produce e-AKI alerts in such scenarios. Method: We have developed e-AKI alert algorithms which compare serum Cr values at presentation with previous results, within KDIGO AKI guideline-specified classifications. However, where a comparator is not available, we have produced a ‘population-based reference Cr value' age and sex matched from 137,000 serum Cr values extracted from blood tests in general practice from our Telepath system. Results: Cr results were split by gender, and then within each group the Cr were stratified according to year of age. The median Cr for each individual year of age was identified and plotted versus age to give separate graphs for males and females that gave excellent fits (R²) to cubic regressions. Conclusion: Population-based estimated reference Cr measurements from community blood test results is a more robust method of baseline Cr value estimation in generating potential e-AKI alerts to help early recognition and treatment of AKI cases leading to improved outcome.