Early persistent-pattern acute kidney injury and short- and long-term mortality in critically ill patients with acute myocardial infarction: a 72-hour landmark cohort study using MIMIC-IV v3.1

Background Acute kidney injury (AKI) complicating acute myocardial infarction (AMI) is associated with mortality, yet most studies treat AKI as a static event. We examined whether early AKI trajectories were associated with short- and long-term mortality among critically ill patients with AMI whose kidney status was observable from 48 to < 72 h. Methods We conducted a retrospective 72-hour landmark cohort study using MIMIC-IV v3.1. Adults admitted to an ICU during a hospitalization coded for AMI were eligible; those who died or were discharged alive by 72 h were excluded. The primary comparison further required at least one derived KDIGO-stage record from 48 to < 72 h and excluded late-onset AKI and unobservable trajectories. Recorded trajectories were classified as no early AKI, transient AKI, or persistent-pattern AKI. Outcomes were 28-day and 1-year all-cause mortality. Cox models adjusted for demographics, AMI subtype, comorbidity, non-renal SOFA, vital signs, vasoactive drug use, and invasive ventilation. Results Among 7454 eligible patients, 6274 met the landmark criteria. The primary cohort comprised 3464 patients: 422 with no early AKI, 892 with transient AKI, and 2150 with persistent-pattern AKI. Another 2810 landmark survivors had late-onset AKI ( n = 195) or unobservable trajectories ( n = 2615). Compared with no early AKI, transient AKI was not associated with 28-day mortality (hazard ratio [HR], 0.91; 95% confidence interval [CI], 0.66–1.25) or 1-year mortality (HR, 0.97; 95% CI, 0.78–1.20). Persistent-pattern AKI was associated with higher 28-day mortality (HR, 1.54; 95% CI, 1.17–2.04) and 1-year mortality (HR, 1.33; 95% CI, 1.09–1.61). In the strict ICU-through-72-hour analysis, the corresponding estimates attenuated to 1.34 (0.99–1.83) and 1.18 (0.95–1.47). The 1-year association varied over time and was concentrated early after the landmark. Conclusions Persistent-pattern AKI was associated with higher mortality in the selected subgroup with observable trajectories from 48 to < 72 h. Attenuation under stricter ICU-observation requirements and residual AMI-specific confounding limit generalizability and causal interpretation. Dynamic AKI reassessment may provide prognostic information, but these data do not show that changing an AKI trajectory improves outcomes.