Recovery From Acute Kidney Injury Requiring Dialysis in Patients With Cirrhosis: A Population-Based Study.

Abstract Rationale & Objective The decision to initiate kidney replacement therapy (KRT) for acute kidney injury (AKI) in cirrhosis remains controversial as it is unclear which patients will benefit. We sought to characterize factors associated with recovery from AKI requiring KRT in patients with cirrhosis to inform shared clinical decision-making. Study Design Population-based retrospective cohort study. Setting & Participants Adult patients from Ontario, Canada identified, using administrative data, to have cirrhosis at the time of admission to hospital with AKI (based on serum creatinine) requiring KRT (01/01/2009–12/31/2016) and followed until 12/31/2017. Exposures Demographics and comorbidities prior to admission. Outcomes Kidney recovery defined as the absence of KRT for at least 30 days. Analytical Approach The cumulative incidences of kidney recovery, death, and liver transplantation were calculated at 1, 3, 6, and 12 months and independent predictors of kidney recovery were evaluated using Fine and Gray competing risk regression models that generated subdistribution hazards ratios (sHR). Results Overall, 722 patients were included (median age 61 years [IQR 54-68]; MELD-Na 26 [IQR 22-34]; 66% male; 52% had viral hepatitis, 25% non-alcoholic fatty liver disease, 18% alcohol-associated liver disease). The cumulative incidences of kidney recovery at 1, 3, 6, and 12 months were 3%, 22%, 25%, and 26%, respectively. Higher MELD-Na score (sHR 0.72 per 5 units, 95%CI 0.65-0.80), acute-on-chronic liver failure (sHR 0.61, 95%CI 0.43-0.86), and sepsis (sHR 0.57, 95%CI 0.41-0.81) were associated with a lower hazard of kidney recovery while those on a liver transplant waitlist (sHR 3.10, 95% CI 1.96-4.88) and who were admitted to a teaching hospital (sHR 1.48, 95%CI 1.05-2.08) were more likely to experience kidney recovery. Limitations Observational design, AKI etiology not identified. Conclusions Kidney recovery from KRT occurred in only one-quarter of patients and was very unlikely after 3-months. These findings provide information regarding prognosis that may guide decisions regarding KRT initiation and continuation.