Association of inter-clinician diagnostic disagreement with increased mortality among patients with septic shock: a retrospective cohort study

Background Septic shock mortality remains high despite dissemination of Surviving Sepsis Campaign guidelines. These guidelines recommend early treatment with antibiotics, fluids, and vasopressors. A prerequisite for early treatment is early identification of sepsis as the shock etiology. We hypothesized that early disagreement between clinical teams regarding shock etiology would be associated with treatment delays and increased mortality among septic shock patients. Therefore, we aimed to quantify the association between inter-clinician disagreement regarding shock etiology and mortality in septic shock patients, and to determine whether delays in septic shock treatment mediate this relationship. Methods Multicenter, retrospective cohort study of 2,233 septic shock patients meeting Sepsis-3 organ-dysfunction criteria across eight acute care hospitals (2018–2023). Shock etiologies documented by clinicians within the first 48 ICU hours were identified using regular expressions and compared using Cosine Similarity Scores. Disagreement was defined as a score < 1. An inverse-probability-of-treatment-weighted multivariable logistic regression tested the association between disagreement and mortality adjusting for demographics, comorbidities, SOFA score, blood culture positivity, and ICU specialty. Multivariable Fine-Gray models assessed whether disagreement affected administration of antibiotics, bolus fluids, or vasopressors. Mediation analysis assessed whether differences in time to antibiotics, bolus fluids, or vasopressors explained any relationship between disagreement and mortality. Results Two hundred and thirteen patients (9.5%) had inter-clinician disagreement on shock etiology. Disagreement was present for 13.3% of patients at the academic medical center and 7.2% at community sites ( p < 0.001). Effect modification of the association between disagreement and mortality by blood culture positivity was significant (OR 2.37, 95% CI 1.03–5.45, p = 0.042), with a stratified 1.99-fold increased adjusted odds of ICU mortality in positive blood culture patients (95% CI 0.98–4.01, p = 0.056; linear-probability delta-risk + 16.9% points, 95% CI + 1.1 to + 32.7, p = 0.036). Notably, disagreement was not associated with different rates of antibiotic, fluid, or vasopressor administration in the blood culture-positive subgroup. Disagreement-associated mortality was not mediated by time to antibiotics, fluids, or vasopressors in the subgroup. Conclusion Early inter-clinician disagreement on shock etiology is a prognostic marker of increased ICU mortality among Sepsis-3 septic shock patients with positive blood cultures. The association is not mediated by timing of antibiotics, fluid boluses, or vasopressors.