Association Between Serum Phosphorus and 28-Day Mortality in Patients with Bloodstream Infection: Potential...

Our objective was to investigate the association between serum phosphorus levels and 28-day mortality in patients with bloodstream infection (BSI), and to explore whether this association persists after adjusting for renal function and clinical severity.

This retrospective cohort study included 214 BSI patients. Patients were divided into hyperphosphatemia (≥2.2 mmol/L, n = 15) and control (<2.2 mmol/L, n = 199) groups. To address the small sample size and potential separation, multivariate Firth's penalized likelihood regression was utilized to evaluate the association with 28-day mortality. Restricted cubic spline regression explored the continuous relationship. Fine-Gray competing risk models, 1000-resample bootstrapping, and E-value analyses were conducted to ensure the robustness of the observed associations.

The 28-day mortality rate was significantly higher in the hyperphosphatemia group (80.0% vs. 39.7%, p = 0.005). After adjusting for age, sex, and estimated glomerular filtration rate (eGFR), hyperphosphatemia remained significantly associated with higher observed 28-day mortality (OR = 4.46, 95% CI: 1.36-18.54, p = 0.012). This association remained robust even after further adjustment for septic shock (OR = 4.74, 95% CI: 1.30-21.64, p = 0.017). Analyzed continuously, each 0.5 mmol/L increase in serum phosphorus was associated with 34% higher odds of mortality (OR = 1.34, 95% CI: 1.07-1.74, p = 0.01). Spline analysis confirmed a nonlinear relationship with a threshold at 2.2 mmol/L. Kaplan-Meier analysis demonstrated a severity-driven survival separation in the hyperphosphatemia group (Log-rank p < 0.001). The association remained highly robust after adjusting for early discharge competing risks (sHR = 4.62, p < 0.001) and in bootstrap validation (median OR = 4.80).

Serum phosphorus ≥ 2.2 mmol/L is associated with higher observed mortality in BSI patients, an association that remained evident after adjusting for renal function and clinical severity, including septic shock. However, given the small hyperphosphatemia subgroup (n = 15), limited statistical stability, and the potential for residual confounding, these findings should be considered hypothesis-generating rather than definitive, requiring prospective validation in larger, adequately powered cohorts. Rather than a definitive triage tool, serum phosphorus may serve as a simple, adjunctive marker for early metabolic assessment in severe infections.