Abstract: PO0199
An Automated, Open-Source Program to Standardize AKI Definition from Time-Stamped Creatinine Data
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
- Saran, Ishan, Yale University School of Medicine, New Haven, Connecticut, United States
- Aklilu, Abinet Mathias, Yale University School of Medicine, New Haven, Connecticut, United States
- Yamamoto, Yu, Yale University School of Medicine, New Haven, Connecticut, United States
- Biswas, Aditya, Yale University School of Medicine, New Haven, Connecticut, United States
- Wilson, Francis Perry, Yale University School of Medicine, New Haven, Connecticut, United States
Background
Though KDIGO guidelines specify a definition for AKI based on changes in serum creatinine, operationalizing this definition with real-world data requires multiple assumptions that leads to variation across studies. A standardized AKI flagging tool may increase inter-study validity.
Methods
We develoepd AKIFlagger, an open-source computational tool built in Python, R, and as a web application which implements a standardized AKI definition based on KDIGO guidelines while allowing for variational definitions of baseline creatinine. We applied the AKIFlagger to a dataset of patients hospitalized with COVID-19 while permuting various operational implementations of the guidelines.
Results
We demonstrate that subtle changes in definition can have a large impact on estimates of AKI prevalence and outcomes. Compared to a rolling window approach, using a baseline definition that leverages outpatient creatinine values and or imputes those values based on an eGFR back-calculation increases the size of captured patient populations by 20.7% and 57.1%, respectively. We characterize the predictive value of the different methods of identifying AKI by determining the sensitivity and specificity for stage progression and progression to death or dialysis. The approaches span sensitivities from 0.18 to 0.20 and specificities from 0.90 to 0.95 for stage progression, and sensitivities from 0.71 to 0.85 and specificities from 0.62 to 0.76 for progression to death.
Conclusion
Subtle differences in the definition of AKI can lead to drastic differences in which patient populations are captured by the definition. A standard mechansim to implement the KDIGO criteria is necessary for the field to accurately advance both clinical and basic science research. This standardized tool can be used by researchers to ensure definitions are uniform across studies.
Screenshot of the web interface for AKIFlagger
Funding
- NIDDK Support