The association of serum chloride level and mortality in patients with sepsis or septic shock in the emergency department. A retrospective registry study.
Hyperchloremia is known to be associated with patient's poor outcomes such as acute kidney injury and higher mortality. However, except for hyperchloremic acidosis, the causal relationship is unclear and controversial. And the subjects were ICU patients and postoperative patients. There are few studies on the relationship between serum chloride concentration and prognosis in the emergency department(ED) patients with sepsis or septic shock. We hypothesized that higher initial serum chloride concentrations in patients with sepsis and septic shock visiting ED will result in higher mortality. A design of this study is retrospective registry study. We analyzed the registry that enrolled 2264 patients with sepsis and septic shock who visited 21 university hospital’s EDs in the Republic of Korea from September 1, 2014 to December 31, 2017. Patients who visited ED through other hospitals were excluded from the final analysis considering the change in chloride concentration due to the effect of unknown initial fluid. 1341 patients with sepsis and septic shock were finally included. Patients were divided into three groups according to the quintiles of serum chloride concentration: Hyperchloremia, normochloremia and hyperchloremia. We also analyzed the relationship between 28-day mortality and chloride level using multivariable logistic regression analysis to adjust confounding effects. Secondary clinical outcomes were a length of ICU stay and a length of mechanical ventilator use. The mean age, SOFA, APACHE and 28-day mortality rate of total subject patients were 67.6, 8.3, 20.6 and 20.8%, respectively. The 28-day mortality rate were 17.6%, 24.8% and 26.4% (p=0.002) in the order of three groups: normochloremia(98.4 ± 2.8, n = 795), hyperchloremia(107.3 ± 3.7, n = 270) and Hypochloremia(89.4 ± 4.3, n = 276). After multivariable logistic regression analysis with age, gender, SOFA, pH and lactate, hypochloremia group showed significantly higher odds ratio(1.59; 95% confidence interval 1.11-2.27) for 28 days of death, but hyperchloremia was not significant. Other secondary outcomes were not significantly associated with chloride level. The 28-day mortality rate was significantly higher in the hyperchloremic or hypochloremic patients with sepsis or septic shock who visited ED. However, when the confounding factors were adjusted, only the hypochloremic group was significantly associated with higher 28-day mortality. This is noteworthy because it was rarely addressed in previous ED-based studies. Further studies are needed to determine whether the selection of different fluids according to serum chloride concentration improves the clinical outcome.
