The Association Between Serum Sodium Variation During Pediatric Intensive Care Unit Admission and the Risk of Clinical Outcomes in Critically Ill Children: A Retrospective Cohort Study.

内容摘要

Dysnatremia at Pediatric Intensive Care Unit (PICU) has been associated with adverse outcomes; however, the impact of sodium variation during PICU stay remains underexplored. We aimed to assess the association between dysnatremia and sodium fluctuations during PICU admission and the probability of prolonged PICU stay (≥7days) and mortality. This retrospective cohort study included 966 critically ill children (2019-2022) with at least two serum sodium measurements during PICU admission. Patients were classified as normonatremic, hyponatremic, hypernatremic, or mixed dysnatremic based on all sodium values. Sodium fluctuation was defined as the difference between the highest and lowest sodium levels during admission and categorized into quartiles. Associations of dysnatremia categories and sodium fluctuation with prolonged PICU stay and PICU mortality were evaluated using multivariable logistic regression. During the PICU admission, 542 patients developed dysnatremia. The PICU-acquired dysnatremia was independently associated with higher odds of prolonged PICU stay after adjusting for major confounders. The PICU-acquired hypernatremia (OR:5.23;95%CI:1.72-15.92) and mixed dysnatremia (OR:2.89;95%CI:1.14-7.29) had significantly higher odds for PICU mortality. Even mild sodium fluctuations (4-8mmoL/L) during PICU admission were significantly associated with higher odds of prolonged PICU stay. The highest (≥15mmol/L) fluctuations in serum sodium levels during PICU admission were also significantly associated with higher odds of PICU mortality (OR:2.74;95%CI:1.11-6.79). Our findings underscore the importance of careful monitoring of serum sodium levels during PICU admission to improve clinical outcomes. These results should be interpreted in light of the retrospective design and the unavailability of standardized illness-severity scores (e.g., PRISM or PIM) in the dataset.