Associations of serum phosphate levels and intravenous phosphate exposure with mortality in patients with sepsis: a multi-database retrospective study

Abnormal phosphate concentrations are common in sepsis, but the prognostic relevance of serum phosphate and the association between recorded intravenous phosphate exposure and mortality remain uncertain. This retrospective cohort study used MIMIC-III, MIMIC-IV, and a single-center database from the First Affiliated Hospital of Chongqing Medical University. Adults with ICD-coded sepsis, a first ICU admission, and an ICU stay longer than 24 h were included. Admission phosphate was evaluated with restricted cubic spline regression for 28- and 90-day mortality. Recorded intravenous phosphate exposure was assessed using multivariable logistic regression in the 48- and 72-h landmark populations. Propensity-score matching and inverse-probability-of-treatment weighting were performed at both landmarks. A target-trial emulation was also performed. Cohorts 1 and 2 included 15,426 and 967 patients, respectively. Admission phosphate was nonlinearly associated with 28- and 90-day mortality (overall P < 0.001 for both outcomes). Recorded intravenous phosphate exposure by 48 h was associated with lower adjusted odds of 28-day mortality (OR, 0.799; 95% CI 0.704–0.907; P < 0.001). The 72-h estimate was similar (OR, 0.814; 95% CI 0.720–0.920; P < 0.001). The 48- and 72-h PSM and IPTW estimates were similar (OR, 0.801–0.847; all P < = 0.007). In cohort 2, the association was not statistically significant (OR, 0.766; 95% CI 0.441–1.328; P = 0.342). Admission serum phosphate provided nonlinear prognostic information for 28- and 90-day mortality in critically ill patients with sepsis. Recorded intravenous phosphate exposure at fixed landmarks was associated with lower 28-day mortality in MIMIC, but this association was not significant in cohort 2. These findings suggest that serum phosphate monitoring and individualized phosphate management should receive greater attention in patients with sepsis. Prospective studies are needed to further validate causality and determine the optimal intervention strategy.