Association between serum phosphate levels and length of hospital stay in infants with neonatal sepsis: a retrospective cohort study.

BACKGROUND: Research on the association of serum phosphate levels with the severity and prognosis of neonatal diseases is limited. Neonatal sepsis is the primary cause of neonatal mortality. Therefore, in this study, we aimed to investigate the association between serum phosphate levels and neonatal sepsis outcomes. METHODS: This retrospective cohort study was conducted using the Pediatric Intensive Care (PIC) database (2010-2018). Neonatal sepsis was diagnosed based on ICD-10 codes. Serum phosphate levels within 72 h of sepsis diagnosis were selected. Outcomes were severe sepsis, in-hospital mortality, length of hospital stay (hospital Los.), and ICU stay (ICU Los.). Covariates included demographic and clinical characteristics as well as serum biomarkers. Multi-variable regression and subgroup analyses were performed to explore the association between serum phosphate levels and neonatal sepsis outcomes. RESULTS: A total of 120 infants with neonatal sepsis were included, and their characteristics were analyzed according to serum phosphate tertiles. The median age was 3.0 (1.0, 17.0) days. The proportion of male infants was 62.5% (75/120). The proportion of preterm infants was 44.2% (53/120). The incidence of late-onset sepsis was 60.8% (73/120). Multivariate linear regression analysis showed that the serum phosphate level at sepsis diagnosis was associated with hospital Los. and remained so after adjustment for all covariates. Each 1 mmol/L decrease in serum phosphate level prolonged the length of hospital stay by 7.53 days. The low serum phosphate group had an 11.75-day longer hospital stay than the high serum phosphate group. Curve fitting showed a negative linear correlation between serum phosphate levels and hospital Los. in infants with neonatal sepsis and no significant interactions were observed in the subgroup analysis; however, serum phosphate levels were not independently associated with ICU Los. In addition, no correlation was found between serum phosphate levels and mortality or the severity of sepsis in neonates. CONCLUSIONS: Hypophosphatemia is highly prevalent in neonates with sepsis. Lower serum phosphate levels are associated with prolonged hospital stay in neonates with sepsis. Strengthening the monitoring of serum phosphate levels in neonates with sepsis and maintaining the homeostasis of phosphate in the body can promote refined management of neonatal sepsis and help improve the prognosis of the patients.