Acute kidney injury (AKI) is associated with increased mortality in critically ill patients. This study aimed to evaluate the association between AKI and 28-day mortality, with a focus on measurable AKI-related complications (serum bicarbonate, blood urea nitrogen–BUN, potassium and fluid balance) as potential mediators.

We extracted data for critically ill patients admitted between 2008 and 2019 from the MIMIC-IV database whose ICU length of stay exceeded 48 h. Causal inference methods were employed to evaluate the association between AKI and 28-day mortality through marginal structural Cox models. We used the survival mediational g-formula to estimate the extent to which mortality in severe AKI (stage 2/3) could be reduced by shifting complication variables to match those of patients without severe AKI.

Among 28,549 patients, 70.2% experienced severe AKI for at least one day. Marginal structural Cox models demonstrated that severe AKI was independently associated with a higher hazard of death within 28 days. Mediation analysis revealed that serum bicarbonate concentration and fluid balance explained 33.2% and 7.4%, respectively, of the increased mortality observed in patients with severe AKI ( P   <  0.001). Furthermore, kidney replacement therapy (KRT) was associated with a 17.6% reduction in mortality attributable to AKI ( P   <  0.001). Serum potassium and BUN levels were not significant mediators of mortality related to severe AKI.

Serum bicarbonate levels and fluid balance mediate approximately 40% of the mortality risk attributable to AKI. KRT is beneficial for AKI-associated mortality. Despite controversial results regarding the benefit of bicarbonate administration in AKI, the focus should remain on how managing metabolic acidosis—possibly beyond simple bicarbonate replacement—in combination with improved fluid balance management can enhance outcomes in critically ill patients with AKI.