Abstract: PO0179
AKI Stage Is a Poor Predictor of Long-Term Cardiovascular Outcomes
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
- Griffin, Benjamin R., Iowa City VA Medical Center, Iowa City, Iowa, United States
- Wachsmuth, Jason, Iowa City VA Medical Center, Iowa City, Iowa, United States
- Yamada, Masaaki, Iowa City VA Medical Center, Iowa City, Iowa, United States
- Sambharia, Meenakshi, The University of Iowa Hospitals and Clinics, Iowa City, Iowa, United States
- Girotra, Saket R., Iowa City VA Medical Center, Iowa City, Iowa, United States
- Perencevich, Eli, Iowa City VA Medical Center, Iowa City, Iowa, United States
- Reisinger, Heather, Iowa City VA Medical Center, Iowa City, Iowa, United States
- Sarrazin, Mary Vaughan, Iowa City VA Medical Center, Iowa City, Iowa, United States
- Jalal, Diana I., Iowa City VA Medical Center, Iowa City, Iowa, United States
Background
Trials to improve long-term outcomes in post-AKI patients have enrolled patients primarily based on AKI stage; however, whether AKI stage is a reasonable predictor of cardiac morbidity and mortality is unknown. We developed predictive models for MACE readmissions in post-AKI patients and determined the utility of AKI stage within these predictive models.
Methods
VHA patient data for inpatient admissions was obtained between 2013 and 2018. AKI was defined as a creatinine increase of ≥0.3 mg/dL from baseline. The primary outcome was subsequent hospitalization for congestive heart failure (CHF), myocardial infarction (MI), or stroke (MACE), with follow-up of at least 2 years. Over 50 variables were considered for inclusion in the final model. Bootstrap modeling was used to determine the outcomes of 100 stepwise regressions using random sampling with replacement. Variables included in more than 60 were included in a final model using Cox regression censored for mortality. If not selected, AKI stage was forced into the model. Due to the association between cardiac disease and the primary outcome, and in order to evaluate risk factors for de novo cardiac disease, separate models were constructed for patients with and without pre-existing cardiac disease.
Results
A total of 241,781 Veterans with AKI were included. AKI stage did not meet selection criteria for either model. In patients without pre-existing cardiac disease, the final model included age, sodium, bilirubin, chronic lung disease (CLD), complicated diabetes mellitus (CxDM), atrial fibrillation (A-Fib), and proteinuria. AKI stage 3 (HR 1.12, CI 1.08-1.16) compared to AKI stage 1 was a weak predictor of subsequent MACE events. Similarly, in patients with prior cardiac disease, the final model included age, blood urea nitrogen, white blood count, CHF, MI, CLD, CxDm, A-Fib, cardiomyopathy, cardiac device, sleep apnea, complicated hypertension, valvular disease, major electrolyte abnormalities, and proteinuria. AKI stage 3 was again a weak predictor (HR 1.065, 1.03-1.10).
Conclusion
While AKI stage is commonly used to enroll patients into AKI survivorship clinics, it was not found to be a strong independent predictor of MACE events among post-AKI Veterans. Our findings may inform risk stratification for post-AKI follow up.