Purpose: To assess whether pre-existing chronic kidney disease (CKD) modified the relationship between the strategy for renal-replacement theraphy (RRT) initiation and clinical outcomes in the STARRT-AKI trial.

Methods: This was a secondary analysis of a multi-national randomized trial. We included patients who had documented pre-existing estimated glomerular filtration rate (eGFR) data prior to hospitalization, and we defined CKD as an eGFR ≤ 59 mL/min/1.73 m² . The primary outcome was all-cause mortality at 90 days. Secondary outcomes included RRT dependence and RRT-free days at 90 days. We used logistic and linear regression and interaction testing to explore the effect of RRT initiation strategy on outcomes by CKD status.

Results: We studied 1121 patients who had pre-hospital measures of kidney function. Of these, 432 patients (38.5%) had CKD. The median (IQR) baseline serum creatinine was 130 (114-160) and 76 (64-90) µmol/L for those with and without CKD, respectively. Patients with CKD were older and more likely to have cardiovascular comorbidities and diabetes mellitus. Patients with CKD had higher 90-day mortality (47% vs. 40%, p < 0.001) compared to those without CKD, though this was not significant after covariate adjustment (adjusted odds ratio [aOR], 1.05; 95% CI, 0.79-1.41). Patients with CKD were more likely to remain RRT dependent at 90 days (14% vs. 8%; aOR, 1.89; 95% CI, 1.05-3.43). CKD status did not modify the effect of RRT initiation strategy on 90-day mortality. Among patients with CKD, allocation to the accelerated strategy conferred more than threefold greater odds of RRT dependence at 90 days (aOR 3.18; 95% CI, 1.41-7.91) compared with the standard strategy, whereas RRT initiation strategy had no effect on this outcome among those without CKD (aOR 0.71; 95% CI, 0.34-1.47, p value for interaction, 0.009).

Conclusion: In this secondary analysis of the STARRT-AKI trial, an accelerated strategy of RRT initiation conferred a higher risk of 90-day RRT dependence among patients with pre-existing CKD; however, no effect was observed in the absence of CKD.