Interrater Reliability of Point-of-Care Cardiopulmonary Ultrasound in Patients With Septic Shock: An Analysis of Agreement Between Treating Clinician and Expert Reviewers

Background: Cardiopulmonary ultrasound (CPUS) is commonly used to assess cardiac function and preload status in patients with septic shock. However, the reliability of CPUS findings at the point of care is unknown. Objective: To assess interrater reliability (IRR) of CPUS in patients with suspected septic shock between treating emergency physicians (EPs) vs emergency ultrasound (EUS) experts. Methods: Single-center, prospective, observational cohort enrolling patients (n = 51) with hypotension and suspected infection. Treating EPs performed and interpreted CPUS for cardiac function parameters (left ventricular [LV] function and right ventricular [RV] function and size) and preload volume parameters (inferior vena cava [IVC] diameter and pulmonary B-lines). The primary outcome was IRR (assessed by Kappa values [κ] and intraclass correlation coefficient [ICC]) between EP and EUS-expert consensus. Secondary analyses examined the effects on IRR of operator experience, respiratory rate, and known difficult views on a Cardiology-performed echocardiogram. Results: IRR was fair for LV function, κ = 0.37, 95% confidence interval (CI) 0.1–0.64; poor for RV function, κ = −0.05, 95% CI −0.6–0.5; moderate for RV size, κ = 0.47, 95% CI 0.07–0.88; and substantial for B-lines, κ = 0.73, 95% CI 0.51–0.95 and IVC size, ICC = 0.87, 95% CI 0.2–0.99. Involvement of ultrasound-trained faculty was associated with improved IRR of RV size (p = 0.002), but not other CPUS domains. Conclusions: Our study demonstrated high IRR for preload volume parameters (IVC size and presence of B-lines), but not for cardiac parameters (LV function and RV function and size) in patients presenting with concern for septic shock. Future research must focus on determining sonographer and patient-specific factors affecting CPUS interpretation in real-time.