Answers are grounded in this paper and cite their sources.
Retrospective cohort analysis shows a J-shaped relationship between serum bicarbonate and 28-day ICU mortality in adults with sepsis, indicating important thresholds.
Sepsis is a leading cause of death in critically ill adults. This retrospective multicenter cohort study included 16,091 patients with sepsis from 208 United States intensive care units (ICUs) (2014-2015) and investigated the association between early serum bicarbonate (HCO3-) levels and 28-day ICU mortality, with particular focus on potential nonlinear relationships. Using the eICU Collaborative Research Database, we stratified patients into HCO3- tertiles (T1: 2.00-19.80, T2: 20.00-23.90, T3: 24.00-51.00 mmol/L). Serum HCO3- exposure was defined as the first available HCO3- measurement obtained within 24 hours after ICU admission. The association between HCO3- and 28-day ICU mortality was assessed using multivariable logistic regression, generalized additive models, and two-piecewise regression. Multivariable models adjusted for demographics, illness severity, comorbidities, infection site, and key laboratory variables. Nonlinear analysis identified a J-shaped mortality relationship with an inflection at 26.4 mmol/L: below 26.4 mmol/L, each 1 mmol/L increase in HCO3- was associated with 4% lower odds of 28-day ICU mortality; above 26.4 mmol/L, each 1 mmol/L increase was associated with 5% higher odds. Adjusted models supported a nonlinear, threshold-dependent association between HCO3- and 28-day ICU mortality, with the lowest observed mortality in the range of 24.0 to 26.4 mmol/L. This study identified a J-shaped dose-response relationship between serum HCO3- and 28-day mortality. Below 26.4 mmol/L, higher HCO3- levels were associated with lower 28-day ICU mortality, whereas above this threshold, higher levels were associated with higher mortality. Adjusted models suggest a prognostic range of 24.0 to 26.4 mmol/L associated with the lowest observed mortality.
No takes yet. Share an insight, caveat, or question.
Du et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: