Interrater agreement for sonographic stomach position classification in fetal diaphragmatic hernia across the North American Fetal Therapy Network

• Published in: Prenatal diagnosis 2022-03, Vol.42 (3), p.348-356
• Main Authors: Abbasi, Nimrah, Ryan, Greg, Ruano, Rodrigo, Cortes, Magda, Ye, Xiang Y., Shah, Prakesh S., Filly, Roy, Benachi, Alexandra, Johnson, Anthony
• Format: Article
• Language: English
• Subjects: Agreements; Chest; Classification; Diaphragm; Female; Fetal Therapies; Fetoscopy; Fetuses; Hernia; Hernias; Hernias, Diaphragmatic, Congenital - diagnostic imaging; Hernias, Diaphragmatic, Congenital - surgery; Humans; North America; Occlusion; Physicians; Pregnancy; Retrospective Studies; Stomach; Stomach - diagnostic imaging; Ultrasonography, Prenatal
• ISSN: 0197-3851; 1097-0223
• DOI: 10.1002/pd.5949

Objective To evaluate inter‐rater agreement for sonographic classification of stomach position (as a surrogate for liver herniation) in fetal left congenital diaphragmatic hernia (LCDH) among: (i) fetal medicine specialists from the North American Fetal Therapy Network (NAFTNet) centers within and without the fetal endoscopic tracheal occlusion (FETO) consortium and in comparison to an expert external reviewer (ER1); and (iii) among two expert ERs (ER1 and ER2). Methods Forty‐eight physicians from 26 NAFTNet centers and 2 ERs were asked to assess 13 sonographic clips of isolated LCDH and classify stomach position as “intra‐abdominal,” “anterior left chest,” “mid to posterior left chest,” or “retro‐cardiac" based on the classification published by Basta et al.8 Interrater agreement was assessed by determining proportion of stomach position ratings concordant amongst NAFTNet participants and ER1. Agreement for stomach position between ERs was calculated using kappa statistics. Results Agreement for stomach position was 69% (39%–85%; n = 19) and 54% (23%–92%; n = 29) among FETO and non‐FETO NAFTNet participants, respectively, when compared to ER1. Most disagreement in stomach position was related to a discrepancy of one position. ERs were in agreement for stomach position in 5 of 13 cases (38.5%) and inter‐rater agreement was highest for “anterior” stomach position. Conclusion Interrater agreement for stomach position assessment in CDH was poor across NAFTNet and indeed amongst expert reviewers. Key points What’s already known about this topic? Intrathoracic liver herniation has been identified as a poor prenatal prognostic indicator in congenital diaphragmatic hernia (CDH), however, this may be challenging to identify by ultrasound due to the similar sonographic appearance of liver, bowel and lung tissue. Sonographic stomach position classification as a surrogate for liver herniation has demonstrated good correlation with liver herniation estimation by magnetic resonance imaging and neonatal morbidity and mortality. Studies assessing reproducibility of this prognostic marker are lacking. What does this study add? When agreement for sonographic stomach position assessment was retrospectively determined on sonographic images of isolated left CDH across the North American Fetal Therapy Network, only moderate agreement was demonstrated among participants. Although sonographic stomach position assessment is promising in its simplicity and accessibility, this stud