29EMF Interpreter Variability Of Lung Point-of-Care Ultrasound Rubric in a Population of Non-Critically Ill COVID Patients

Study Objectives: Lung point-of-care ultrasound (L-POCUS), a novel and radiation-free diagnostic tool, could aid in COVID-19 prognosis. Early studies have yielded scoring rubrics focused heavily on hospitalized populations including the critically ill. Operator characteristics of this novel technology in non-critically ill, ambulatory COVID patients has not been described and is an important consideration for dissemination. The purpose of our study was to determine to the inter-rater reliability of an L-POCUS scoring rubric in a population of non-oxygen dependent patients. Methods: This was a cross sectional study design of patients at three academic institutions in the Northeast, Midwest, and West. We included subjects with respiratory complaints who tested positive for COVID-19 and maintained oxygen saturation ≥92% for two hours after presentation to the emergency department as part of a larger project focused on describing L-POCUS prognostic characteristics in a non-critically ill COVID pneumonia population. L-POCUS was performed on seven lung windows on each side of the chest: two anterior, two lateral, and three posterior. All clips were obtained with a curvilinear probe or a linear probe using machine settings to enhance lung findings ("nerve" or "lung"). The scoring rubric ranged from 0 to 6 for each lung field with 0 being normal lung and 6 indicating severe lung pathology from COVID. We divided lung findings into pleural and parenchymal with the score per lung field representing the sum of the two parts. Pleural findings included normal (0 points), blurring, indenting, or thickening (1 point), and discontinuity (2 points). Parenchymal findings included normal (0 points), B lines (1-3 B lines equaled 1 point, >3 B lines equaled 2 points, coalescing or “waterfall” B lines equaled 3 points), and subpleural consolidation (4 points). As discontinuous pleura necessarily accompanies subpleural consolidations per definition, lung fields with subpleural consolidations automatically scored 6 points. Clips, collected and scored at bedside by an expert sonologist, were randomly selected for scoring by other operators of differing experiences: a resident, a faculty member without ultrasound fellowship training, an ultrasound fellow, and a second expert. Scores were then analyzed using the intraclass correlation coefficient (ICC) using the R package “ICC” to determine inter-rater reliability between the initial expert rater and all other raters. Results: A total of 50 clips lasting 6 seconds each were chosen for scoring, 49 with the culvilinear probe and 1 with the linear probe. The calculated Intraclass Correlation Coefficient (ICC) for expert raters was 0.71 (0.55, 0.83, p<0.0001) 0.83). Moderate agreement between all raters was found with an ICC of 0.72 (0.62, 0.81). The faculty member without ultrasound fellowship training and the fellow disagreed the most from the group and resulted in the highest variability. A Loess graph demonstrates less variability at low scores than high scores. Conclusion: The L-POCUS rubric for scoring lungs infected with COVID in an ambulatory population revealed moderate to good agreement among a diverse group of operators. Greater variation at higher scores reveals ambiguity in definitions of lung pathology in COVID. This warrants future studies refining criteria for lung findings and correlating to clinical implications. [Formula presented]