357 - Inter-Rater Reliability of Point of Care Lung US Scores in Preterm Infants: ​A Multicenter Prospective Study​

Publication Number: 357.5360

Delaney Wresch, Phoenix Children's Hospital, Phoenix, AZ, United States; Julianna Diddle, Phoenix Children's Hospital, Chandler, AZ, United States; Lauren Ridgway, Phoenix Children's, Phoenix, AZ, United States; Sara Mohamed, Baylor Scott White McLane Children's Medical Center, Temple, TX, United States; Manisha Singh, Baylor Scott and White, Temple, TX, United States; Vinayak Govande, Baylor Scott White McLane Children's Medical Center, Temple, TX, United States; Moraji Peesay, Georgetown University Hospital, Washington DC, DC, United States; Jennifer Shepherd, CHLA/USC, Los Angeles, CA, United States; Muppala Raju, Baylor Scott White McLane Children's Medical Center, Temple, TX, United States; Raza Bajwa, Baylor scott and white, Round Rock, TX, United States; Marc Ellsworth, Phoenix Children's Hospital, Phoeinx, AZ, United States; Kathie Laitano, Phoenix Childrens Hospital, Phoenix, AZ, United States; Hemananda Muniraman, Keck School of Medicine of the University of Southern California, Los Angeles, CA, United States

Background: The utility of point of care lung ultrasound (LUS) in NICUs is evolving with few studies reporting high accuracy of LUS scores in the diagnosis of respiratory distress syndrome (RDS) and prediction of bronchopulmonary dysplasia. Many published studies have relied on LUS experts to report and score the LUS. The training and skills to perform and score point of care LUS are not standardized with the inter-rater reliability likely to be variable.

Objective: To evaluate the inter-rater reliability of LUS performed by investigators and reviewers as part of a multicenter prospective study of point of care LUS in preterm infants.

Design/Methods: Multi-center prospective study (ClinicalTrials ID: NCT05925075) evaluating LUS in preterm infants ( < 32wk gestation). The LUS was performed by investigators (4 Neonatologists, 4 Neonatal-Perinatal fellows and 1 Pediatric Resident) after completing an online training and/or 10 independent scans. Each LUS image was scored using the Modified Brat classification (Figure 1). Images were then scored by one of two blind reviewers with expertise in LUS. Inter-rater reliability was analyzed with intraclass correlation coefficient (ICC) for all images and stratified based on center, study arm and individual investigators.

Results: A total of 1,950 LUS images from 325 LUS procedures performed at 3 centers were included in the analysis. LUS was performed and scored by 9 investigators, then reviewed and scored again by a blinded reviewer. Overall, there was excellent correlation with reviewers (ICC=0.90). LUS scores from RDS and the PDA arm of the study had very good correlation (ICC=0.81-0.89), while the steroid cohort had fair correlation (ICC=0.68). The two reviewers had excellent correlation with each other with an initial sample of 66 LUS scans (ICC=0.98). Amongst the investigators, 5 had excellent correlation (ICC=0.92-0.95), 3 had very good correlation (ICC=0.88-0.89), and 1 had good correlation (ICC=0.75) (Table 1).

Conclusion(s): Overall, there is excellent inter-rater reliability with multiple investigators with limited LUS experience after individualized training. The variability in scores within the steroid cohort could be related to the higher degree of lung pathology in patients receiving a course of low-dose steroids, resulting in higher LUS scores and more subjective interpretation. There continues to be a need for ongoing training and re-education amongst investigators and in patients with more severe lung pathology. Improving reliability of LUS scoring by clinicians will optimize the utility of point of care LUS as a real-time imaging modality.

Table 1. Intraclass Correlation Coefficient of Lung Ultrasounds


Figure 1. Modified Brat LUS Scoring
4 distinct LUS patterns identified from more aeration to less aeration (left to right). (0) A-line pattern. Normal lung tissue, score 0: presence of A-lines, pleural line sliding and < 3 B-lines; (1) Nonconfluent B-line pattern. Interstitial syndrome, score 1: absence of A-lines, > 3 B-lines, pleural line sliding; (2) Confluent B-line pattern. White lung, score 2: absence of A-lines, confluent B-lines, pleural line sliding; (3) Consolidation pattern. Subpleural consolidation > 5mm, score 3: absence of A-lines, confluent B-lines.