Comparative Diagnostic Performance of Stewart Versus Henderson-Hasselbalch Methods for Acid-Base Disorders in Critically Ill Patients: A Prospective Observational Study

Background: Acid-base disorders are prevalent in critically ill patients, but traditional Henderson-Hasselbalch analysis may miss complex disorders. The Stewart physicochemical approach offers potentially superior diagnostic accuracy. Objective: To compare the diagnostic accuracy of Henderson-Hasselbalch versus Stewart methods for detecting acid-base disorders in hospitalized patients and assess their prognostic implications. Methods: This 11-month prospective observational study included 70 consecutive patients with arterial blood gas analysis from intensive care and pulmonology departments at Laghouat Mixed Hospital, Algeria (June 2024 - April 2025). Acid-base disorders were classified as simple, mixed, or complex using both methods by two independent reviewers. The primary outcome was diagnostic concordance between methods; secondary outcomes included mortality prediction and clinical correlations. Inter-rater reliability was assessed using Cohen's kappa. Results: Among 258 hospitalized patients, 70 (27.1%, 95% CI: 21.9%-32.8%) had acid-base disorders requiring blood gas analysis. The Stewart method identified significantly more disorders than Henderson-Hasselbalch (70 vs 23 disorders, p<0.001). Complex disorders predominated (47.1%), followed by mixed disorders (42.9%). Inter-rater agreement was excellent for Stewart method (κ=0.92) and moderate for Henderson-Hasselbalch (κ=0.58). Metabolic acidosis was associated with highest mortality (78.9% vs 45.1% for other disorders, p=0.012). A prognostic score incorporating disorder type, service, and base excess achieved good discrimination (AUC=0.812). Conclusions: The Stewart method identifies significantly more acid-base disorders than traditional Henderson-Hasselbalch analysis, particularly complex disorders. However, the clinical significance of this increased detection requires validation through interventional studies.