Prevalence and Prognostic Significance of Chloride Levels in Patients with Acute Medical Conditions: A Prospective Observational Study

Abstract

Chloride plays a considerable role in physiology. This study aimed to assess the association between serum chloride and prognosis in the population of adults with acute medical conditions. A prospective cohort study was conducted in an acute medical unit. Chloride levels at admission were the main exposure factor, categorized into hypochloremia, normochloremia, and hyperchloremia. The outcomes were in-hospital mortality and length of hospital stay (LOHS). A total of 798 patients were included. The mean age was 57.3 ± 18.3 years. The prevalence of dyschloremia was 40.9%. Restricted cubic splines revealed a linear association between hypochloremia and in-hospital mortality, as well as between hypochloremia and LOHS. After adjusting for age, sex, heart failure, diabetes, sodium, bicarbonates, creatinine, and diuretic use, hypochloremia was significantly associated with in-hospital mortality (OR = 2.23; 95% CI: 1.29, 3.86, p = 0.006), but not hyperchloremia (p = 0.57). Similarly, it was associated with a longer LOHS (β = 2.19; 95% CI: 0.01, 4.39, p = 0.05), but not hyperchloremia (p = 0.8). The interaction between chloride and sodium levels was not significant (p = 0.61). Subgroup analysis showed that the effect of hypochloremia on in-hospital mortality was consistent across subgroups. The prevalence of dyschloremia in this study was high at 40.9%. Hypochloremia increased the risk of in-hospital mortality and extended the LOHS. Differentiating the effects of chloride levels from those of sodium can enhance clinical risk stratification and enable a more targeted management approach for acutely ill patients. Recognizing this distinction is essential for optimizing prognostic assessment and tailoring treatment strategies accordingly.

Keywords: chloride; prevalence; prognosis; prospective observational study.

Figure 1

Restricted cubic spline curve showing the association between chloremia and mortality using multivariable logistic regression models. The black line represents the estimated odds ratio, and the shaded area indicates the 95% confidence interval. Each stratification was adjusted for age, sex, the presence of heart failure, diabetes mellitus, serum sodium level, serum bicarbonate level, serum creatinine level, and the use of diuretics. Legend: The adjusted odds ratios for mortality as a function of serum chloride levels (chloremia), estimated using multivariable logistic regression with restricted cubic splines. Serum chloride was modeled as a continuous variable, with a reference value set at 100 mEq/L (OR = 1) to represent baseline risk. The model is adjusted for age, sex, heart failure, diabetes mellitus, blood sodium, blood bicarbonate, blood creatinine, and diuretic use. The solid curve shows the point estimates of the odds ratios, and the shaded area represents the corresponding 95% confidence intervals, which widen at the extreme ends of the chloride distribution due to smaller sample sizes.

Figure 2

Restricted cubic spline curve showing the association between chloremia and length of hospital stay using multiple linear regression models. Each stratification was adjusted for age, sex, the presence of heart failure, diabetes mellitus, serum sodium level, serum bicarbonate level, serum creatinine level, and the use of diuretics. Legend: The restricted cubic spline curve showing the association between chloremia and length of hospital stay using multiple linear regression models. The data were adjusted for age, sex, the presence of heart failure, diabetes mellitus, serum sodium level, serum bicarbonate level, serum creatinine level, and the use of diuretics. The difference in hospital length of stay (LOHS) relative to normochloremic patients (~100 mmol/L). A value of zero represents no difference in LOHS. Positive values indicate a longer hospital stay for patients with hypochloremia, while negative values indicate a shorter stay for those with hyperchloremia. The shaded region represents the 95% confidence interval. The curve and the shaded areas around it represent the estimated values and their corresponding 95% confidence interval.

Figure 3

Non-linear interaction: Chloremia-Natremia: relationship between in-hospital mortality and chloremia at different natremia levels. Each stratification was adjusted for age, sex, the presence of heart failure, diabetes mellitus, serum sodium level, serum bicarbonate level, serum creatinine level, and the use of diuretics. Legend: Using fractional polynomial interactions, models of the relationship between chloremia and in-hospital mortality were plotted at different point values of natremia (105, 115, 125, 135, 145, 155, 165 mmol/L) and adjusted for age, sex, the presence of heart failure, diabetes mellitus, serum sodium level, serum bicarbonate level, serum creatinine level, and the use of diuretics (p = 0.61 for the interaction). While extreme values of natremia (e.g., 105 and 165 mmol/L) are rare and typically associated with critical illness, they were included for illustrative purposes to visualize interaction trends across the full range.

Figure 4

Subgroup analysis of the association between hypochloremia and hospital mortality. This forest plot presents the odds ratios (ORs) and corresponding 95% confidence intervals (CIs) for the association between hypochloremia and hospital mortality across various subgroups. Each point estimate (circle) indicates the OR for the respective subgroup, with horizontal lines representing the 95% CI. The vertical dashed line at OR = 1.0 represents no effect (null value). A point estimate to the right of this line suggests an increased hospital mortality, while a point estimate to the left suggests decreased odds. The P for interaction indicates whether the effect of hypochloremia and hospital mortality differs significantly across the subgroups shown. Subgroups include age (<65 vs. ≥65 years), sex, heart failure, diabetes, diuretic use, kidney injury, and bicarbonate (<20 vs. ≥20 meq/L).