BackgroundThe use of sodium bicarbonate for metabolic acidosis has been a topic of debate, primarily due to the lack of clinical efficacy evidence. This study aims to identify which types of patients with various acid-base balance parameters can benefit from sodium bicarbonate therapy. MethodsPatients diagnosed with metabolic acidosis were screened from a large multi-center critical care database to form a retrospective cohort. Mortality curves, logistic regression analysis, simulation methods, and propensity scores were used to compare data between sodium bicarbonate (SOB group) and non-treated (Non-SOB group) patients. ResultsThere was an interaction between baseline chloride, anion gap levels and sodium bicarbonate therapy on patients' in-hospital death. As chloride levels increased, the in-hospital mortality curves of the SOB group and Non-SOB group gradually converged, with the difference narrowing from approximately 20 % to 10 %, and then gradually widened with the increase of the anion gap. Furthermore, when patients had high chloride levels (≥112 mmol/L), those in the SOB group exhibited a higher incidence of hypernatremia, hypokalemia, and hypocalcemia at 24 h, and a lower incidence of hyperchloremia. Patients in SOB group also had a lower simulated mortality. Among patients treated with sodium bicarbonate, those with low chloride had more difficulty in normalizing pH compared to those with high chloride. ConclusionsThis study identified an interaction between baseline chloride and sodium bicarbonate therapy on patient survival. Hyperchloremic metabolic acidosis may potentially benefit from sodium bicarbonate therapy. Further prospective randomized controlled studies are warranted.