Factors Associated with Early Kidney Recovery After KDIGO Stage 3 Acute Kidney Injury Requiring Renal Replacement Therapy: The RENRECOV Study

Acute kidney injury (AKI) is a common and serious condition in hospitalized patients, with KDIGO stage 3 AKI requiring renal replacement therapy (RRT) resulting in mortality rates of up to 50%. Among survivors, the degree and timing of renal recovery are key determinants of long-term outcomes, influencing risks of chronic kidney disease, cardiovascular events, and dialysis dependence. However, the recovery of renal function after AKI is highly variable, and there is currently a lack of validated predictors of early renal recovery, particularly in Latin American populations. The objective of this study was to identify clinical factors associated with early renal recovery in patients with KDIGO 3 AKI who required RRT.

We conducted a multicentre, retrospective cohort study across six tertiary hospitals in Bogotá, Colombia, from 2016 to 2024. Adults (aged 18 years and over) with KDIGO 3 AKI who initiated RRT during their hospitalisation were included in the study. ERR (primary outcome) was defined as successful discontinuation of RRT within seven days; absence of ERR corresponded to RRT need beyond seven days. The baseline creatinine was estimated to an eGFR of 75 mL/min/1.73 m² (MDRD) when unknown. Descriptive analyses and group comparisons used χ² and t tests. Univariable and multivariable logistic regression identified independent predictors of ERR, reporting adjusted odds ratios (aOR) with 95% confidence intervals (CI). The model's capacity for discrimination was evaluated using the receiver operating characteristic (ROC) curve and Nagelkerke's R².

Following a thorough review of the 3,677 records, it was determined that 599 patients met the inclusion criteria (mean age: 63 ± 15 years; 56.7% male). ERR occurred in 163 cases (27.2%), whereas 436 cases (72.8%) did not recover by day 7. Patients with ERR had higher pre-RRT urine output (741 ± 835 vs. 523 ± 633 mL), higher haemoglobin (11.9 ± 2.8 vs. 10.9 ± 2.8 g/dL), and lower prevalence of chronic kidney disease stages 3–4 (5.8% and 5.8% vs. 20.2% and 10.4%). Anuria was a less frequent indication for RRT in the ERR group (40.4% vs 54.5%). In multivariable analysis, factors independently associated with ERR included: pre-RRT urine output ≥ 500 mL (aOR ≈ 1.54; 95% CI 1.05–2.26; p = 0.028), higher haemoglobin prior to RRT (p = 0.002), surgical admission vs cardiovascular (aOR ≈ 2.06; 95% CI 1.01–4.22; p = 0.047), and centre-level differences. The model's performance was modest, with an area under the curve (AUC) of 0.64 and a R² of 0.073. The model's sensitivity was 0.60, while its specificity was 0.59.

In patients with KDIGO 3 AKI requiring RRT, higher pre-RRT urine output, higher hemoglobin, and surgical admission were independently associated with early renal recovery. Despite modest discriminative power, these easily measurable clinical variables may help guide individualized strategies to preserve residual kidney function and inform RRT discontinuation decisions. Prospective, multicenter studies using standardized definitions of renal recovery are needed to validate these findings and improve prognostic models in AKI care.