Intra and inter-observer reliability and agreement on defining tachypnea in infants: A methodological comparison

Background: Measurement of respiratory rates (RR) in children often showed disagreement. We aimed to determine the reliability of RR measurement within and between observer(s) (intra- and inter-observers agreement), and to compare the reliability of different methods of RR measurement (visual versus using stethoscopes). Methods: Two observers performed RR measurements in infants whom reported to have Acute Respiratory Infections (ARI), by visual method and with stethoscope during home visit. Intra- and inter-observers and between method agreements were determined using Pearson correlation test and Bland-Altman plot. Kappa test was used to measure agreements in defining tachypnea. Results: During the study, there were 196 visits from 130 participants documented. The Pearson’s rho correlations were above 0.80 for all measurements. RR measurement with a stethoscope demonstrated a higher mean of 1.39 breaths per minute in observer 1 (limits of agreement -12.55 to 9.78) and 1.39 breaths per minute in observer 2 (limit of agreement -11.97 to 9.18), compared to the visual method. For tachypnea, the Kappa-agreements for intra-observer using visual count were 0.68 and 0.72, while for auscultation were 0.59 and 0.79, for observer 1 and 2, respectively. Inter-observer kappa agreements were 0.78 (visual count) and 0.79 (auscultation), while the between-method kappa agreement were 0.64 and 0.73. Conclusion: Although good agreements were obtained, the limits of agreements for all parameters were wide. When restricted to determine tachypnea, all agreements were weakened. This finding emphasises the need to review the current method in determining tachypnea to minimise under and over diagnosis of pneumonia in children. Keywords: respiratory rate; acute respiratory infection; children; visual; auscultation