Abstract: PO0195
A Risk Score for Major Adverse Kidney Events One Year After AKI
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
- Bowe, Benjamin Charles, Veterans Health Administration, St. Louis, Missouri, United States
- Xian, Hong, Saint Louis University, Saint Louis, Missouri, United States
- Rigdon, Steve E., Saint Louis University, Saint Louis, Missouri, United States
- Xie, Yan, Veterans Health Administration, St. Louis, Missouri, United States
- Al-Aly, Ziyad, Veterans Health Administration, St. Louis, Missouri, United States
Background
Epidemiologic evidence suggests that those with AKI are at increased risk of post-AKI kidney disease, higher hospital resource utilization, and deathI. However, literature to support identification of those most at risk of these outcomes is limited. Here we pilot predicting risk of post-AKI MAKE.
Methods
In a cohort of 4.2 million United States Veterans, risks of MAKE within a year of discharge associated with an AKI were detailed using survival regression with inverse probability of treatment weighting. Risk factors for MAKE including demographics, clinical characteristics including diagnoses, medication use, and laboratory tests, as well as hospitalization parameters among those with an AKI were examined, and then a risk score was developed and evaluated following the Framingham Heart risk score algorithm.
Results
In the year after dischrarge form a hospitalization, compared to those without an AKI, those with an AKI were at an increased risk of a subsequent AKI (HR=1.47; 95% CI=1.45-1.49), incident eGFR less than 60 ml/min/1.73 m2 (1.23; 1.22-1.24), eGFR decline >30% (1.69; 1.67-1.71), receipt of kidney replacement therapy (2.41; 2.28-2.51), and MAKE (1.24; 1.23-1.25). Results were consistent in Fine and Gray competing risk models. Among those with an AKI, predictors of MAKE included age, albuminuria, bicarbonate, blood pressure before and during hospitalization, cardiovascular disease, cancer, chronic lung disease, dementia, diuretic use, baseline eGFR, hematocrit level, NSAID use, obesity, platelet count, pneumonia, serum creatinine trajectory during hospitalization, surgeries, and urinary tract infection. A risk score constructed using these predictors achieved an area under the curve (AUC) of 0.72, where corresponding probabilities of having a MAKE within a year of discharge ranged from 7.3% to 59.9% at the lowest and highest risk score values experienced in the cohort. Comparatively, use of KDIGO stage alone marginally predicted future risk of MAKE (0.52). Calibration plots suggested that models were well calibrated.
Conclusion
Use of EHR resulted in a moderate ability to identify those at increased risk of post-AKI MAKE. Further research is needed in identifying those who may benefit from post-AKI care.
Funding
- Veterans Affairs Support