A62-16 Mind the Gap: Abdominal Versus Chest Observation for Pediatric Respiratory Rate Assessment

Rationale

Respiratory rate (RR) is a fundamental pediatric vital sign, yet measurement suffers substantial inter-observer variability. Direct visual observation and manual counting, the primary method in resource-limited settings, lack standardization regarding the optimal anatomical landmark for assessment. Whether chest or abdominal motion provides more reliable estimates remains unclear. Recent advances in video-based technology enable quantitative analysis of respiratory motion, raising questions about systematic differences between video-derived RR and routine bedside clinical measurements. This study examined: (1) which anatomical site (chest versus abdomen) provides more reliable visual RR assessment in young children, and (2) whether video-derived RR systematically differs from clinical RR.

Methods

Videos of spontaneous breathing from 120 children presenting to an emergency department (ED) with respiratory distress were analyzed. Three trained annotators marked chest and abdominal peak excursions using a custom platform (Figure 1). Respiratory waveforms were generated from peaks, and RR calculated by fast Fourier transformation. Video-derived chest and abdominal RR were compared with ED nurse measurements, obtained by chest auscultation (reference standard). Mean absolute error (MAE) was used to assess agreement and identify the anatomical site most aligned with clinical practice.

Results

Video-derived abdominal RR demonstrated greater accuracy compared to chest-derived RR (MAE: 12.9 vs 14.2 breaths/min; median: 10.1 vs 11.8, p = 0.042), with significantly less negative bias relative to the reference (−6.2 vs − 8.9 breaths/min; p = 0.0025). Both methods systematically underestimated the reference RR in approximately two-thirds of measurements. In underestimation-only cases (n = 78), chest-derived RR showed greater underestimation magnitude than abdominal RR (mean 16.6 vs 14.6 breaths/min; p = 0.004). Age-stratified analysis revealed abdominal superiority was significant only in infants <1 year (MAE: 18.4 vs 22.0 breaths/min, p = 0.049), with no significant differences in older age groups. Measurement accuracy improved with age, decreasing from approximately 20 breaths/min in infants to 7-9 breaths/min in school-age children, with significant inverse correlations between age and error for both chest- and abdominal-derived RR (Spearman r = −0.39 and −0.35; both p < 0.001).

Conclusions

Abdominal RR demonstrated better agreement with bedside clinical RR than chest observation, particularly in infants under one year. This suggests the abdomen may be a preferable anatomical landmark for visual RR assessment in young children. Both video-based methods systematically underestimated clinical RR, likely reflecting inherent differences between real-time triage assessment and post-hoc controlled observation. This systematic bias should be addressed in the future development of video-based decision-support tools designed to complement, rather than substitute, clinician judgment.

This abstract is funded by: Ministry of Science and Technology Israel