Both hypophosphatemia and hyperphosphatemia are associated with increased mortality in septic patients

OBJECTIVE: This study was intended to explore the relationship between level of serum phosphate and prognosis in septic patients. METHODS: Data were obtained from the public database, which were divided into 2 subgroups according to whether they were complicated with chronic kidney disease. Initial values of serum phosphate were extracted from patients on admission to hospital. Propensity score-matched analysis was performed. The relationship between hypophosphatemia, hyperphosphatemia and the severity of the disease in septic patients was explored separately. The lowess smoothing technique and the Kaplan-Meier method were utilized for a preliminary analysis of serum phosphate levels in relation to in-hospital mortality and 28-day survival. The initial values of serum phosphate were graded as level 1 (<1.5 mg/dL), level 2 (1.5-2.7 mg/dL), level 3 (2.7-4.5 mg/dL), level 4 (4.5-5.5 mg/dL), level 5 (5.5-6.5 mg/dL), level 6 (6.5-7.5 mg/dL) and level 7 (> 7.5 mg/dL). Multivariate logistic regression and cox regression was used to analyse the relationship between serum phosphate levels and mortality. RESULTS: There were 4059 cases (17.4%) combined with chronic kidney disease, including 419 cases (10.3%) with hypophosphatemia and 1091 cases (26.8%) with hyperphosphatemia. There were 19224 cases (82.6%) not combined with chronic kidney disease, including 3769 cases (19.6%) hypophosphatemia and 2158 cases (11.2%) hyperphosphatemia. After propensity score-matched, in-hospital mortality, 28-day mortality, risk of septic shock was significantly higher in the 2 subgroups of hypophosphatemia patients than in normophosphatemia patients. In-hospital mortality, 28-day mortality, risk of septic shock, occurrence of renal replacement therapy, occurrence of acute renal failure, and maximum clinical score were all significantly higher in the 2 subgroups of patients with hyperphosphatemia than in patients with normophosphatemia. Multivariate logistic regression was consistent with cox regression results. In septic patients without chronic kidney disease, hypophosphatemia was an independent risk factor for death. When serum phosphate was lower, the risk of death was higher. In all septic patients, hyperphosphatemia was an independent risk factor for death. When serum phosphate was higher, the risk of death was greater. CONCLUSIONS: Both hypophosphatemia and hyperphosphatemia are associated with increased mortality in septic patients and are independent risk factors for death.