Bicarbonate Use and Short-Term Mortality in Diabetic Ketoacidosis: Outcomes by Degree of Acidemia

The role of intravenous bicarbonate therapy in diabetic ketoacidosis (DKA) remains controversial. Current guidelines generally discourage bicarbonate use except in cases of severe acidemia, yet real-world practice varies widely. Evidence supporting specific arterial pH thresholds for bicarbonate administration is limited. This study evaluated short-term mortality associated with bicarbonate therapy versus no bicarbonate therapy across lower pH ranges in adults with DKA.

A retrospective cohort study using a multi-institutional U.S. healthcare database comprising 118,706,197 patients across 68 healthcare organizations (2005-2025). Adult patients hospitalized with DKA were stratified by venous or arterial pH at presentation (7.00-7.09, 6.90–6.99, 6.80–6.89, <6.80). Patients receiving intravenous bicarbonate were compared with those who did not. Propensity score matching was performed within each pH stratum to adjust for demographics, prior illness severity, and relevant comorbidities. The primary outcome was day 1-7 all-cause mortality.

There were 4436 patients with DKA and pH between 6.9 and 6.99 that received bicarbonate and 2770 where no bicarbonate was administered. Among patients with DKA and pH between 6.8 and 6.89 there were 2267 patients who received bicarbonate and 860 that did not. After propensity matching, in 2632 patients with pH 6.90–6.99, bicarbonate therapy was associated with significantly higher 7-day mortality compared with no bicarbonate therapy (4.1% vs 2.0%; RR 2.11; p<0.001). Findings were similar in the 7.00-7.09 group. In contrast, among 830 patients with more severe acidemia (pH 6.80–6.89), bicarbonate therapy was not associated with a mortality harm or benefit (4.1% vs 3.6%; RR 1.11; p=0.671). The number of patients with pH < 6.8 was too small to complete a valid analysis.

In this large real-world analysis of adults with DKA, bicarbonate therapy was associated with increased short-term mortality in patients with pH 6.90–6.99 and conferred no mortality harm or benefit in patients with pH 6.80–6.89. Despite potential residual confounding, these findings suggest that routine bicarbonate administration in DKA may not be harmful or necessarily beneficial when the pH is < 6.9 and provides no clear benefit in milder acidemia. Restriction of bicarbonate use to cases of severe acidemia may be warranted, pending prospective validation.