Serum Phosphate Trajectory Patterns and Clinical Outcomes in ARDS: a Multicenter Retrospective Cohort Study

Background Acute respiratory distress syndrome (ARDS) is characterized by marked biological and metabolic heterogeneity. Serum phosphate is essential for cellular energy metabolism; however, its longitudinal patterns and clinical relevance in ARDS remain unclear. We aimed to identify distinct serum phosphate trajectory patterns in ARDS and evaluate their prognostic value beyond static measurements. Methods In this multicenter retrospective cohort study, we analyzed adult ARDS patients from the MIMIC-IV database (derivation cohort, n = 8,058) and the eICU Collaborative Research Database (external validation cohort, n = 12,049). Latent class mixed modeling was used to define seven-day serum phosphate trajectory patterns after respiratory support initiation. The primary outcome was 30-day mortality. Associations were assessed using multivariable logistic regression for mortality and linear regression for ventilator-free days. Propensity score matching was performed as a complementary analysis to assess robustness rather than to establish causality. External validation used hospital mortality. Results Three distinct serum phosphate trajectory patterns were identified: Low, Normal, and Persistent High. Compared with the Normal trajectory, the Persistent High trajectory was associated with higher 30-day mortality (adjusted OR 4.74, 95% CI 3.52–6.39) and fewer 28-day ventilator-free days (adjusted β -7.38 days, 95% CI -8.72 to -6.04). These associations persisted after propensity score matching (OR 3.18, 95% CI 2.15–4.76) and in the external validation cohort (adjusted OR 1.65, 95% CI 1.39–1.95). The Low trajectory was not associated with increased mortality. Conclusions Longitudinal serum phosphate trajectory patterns identify a Persistent High group in ARDS that is independently associated with increased mortality and delayed respiratory recovery. Serum phosphate trajectories may provide a clinically accessible approach for risk stratification in ARDS.