The association between plasma osmolality and in-hospital mortality in the first 24 h after neonatal intensive care unit admission.

Background: Perturbation of osmolality is associated with increased mortality in adults and children in critically ill conditions. However, it is still unclear whether osmolality imbalance impacts the prognosis of critically ill infants. This study aimed to investigate the relationship between plasma osmolality and prognosis in critically ill infants within 24 h of admission.

Methods: This retrospective study enrolled 1,042 infants who had plasma osmolality data from 2010 to 2018. The initial plasma osmolality (within 24 h after admission) was extracted from the pediatric intensive care database (PIC V1.1). The locally weighted scatter-plot smoothing (LOWESS) and restricted cubic splines (RCS) methods were used to explore the approximate relationship between plasma osmolality and in-hospital mortality. Univariate and multivariate logistic regression analyses were used to further analyse this relationship. Kaplan-Meier analysis was applied to estimate the probability of hospital mortality within 90 days of admission. Subgroup analysis was employed to assess the impact of potential confounders (including postnatal days, gender, and gestational age).

Results: An approximately"U"-shaped relationship between plasma osmolality and mortality was detected. In the logistic regression model, plasma osmolality <270 mmol/L (low osmolality group) was significantly associated with in-hospital mortality (0.05; OR 2.52; 95% CI, 1.15-5.06). Plasma osmolality >300 mmol/L (high osmolality group) was also significantly associated with mortality (0.05; OR 3.52; 95% CI, 1.16-8.83). This association remained even after multivariable adjustments. The 90-day survival rate was lower in the abnormal plasma osmolality group (including high or low osmolality groups) than in the intermediate group (log-rank test, 0.05). The abnormal plasma osmolality group had a significantly higher incidence of all-cause mortality in the 0-7 postnatal days subgroup (high osmolality group, 0.05; OR 5.25; low osmolality group, 0.05; OR 3.01). Infants with abnormal osmolality had a significantly higher mortality rate in the female group (0.05). High osmolality was associated with a higher mortality rate in the preterm group (0.05).

Conclusions: Both hypoosmolality and hyperosmolality were shown to be independently associated with increased risk of in-hospital infant mortality in NICUs.