Effect of context on respiratory rate measurement in identifying non-severe pneumonia in African children

OBJECTIVE Cough or difficult breathing and an increased respiratory rate for their age are the commonest indications for outpatient antibiotic treatment in African children. We aimed to determine whether respiratory rate was likely to be transiently raised by a number of contextual factors in a busy clinic leading to inaccurate diagnosis. METHODS Respiratory rates were recorded in children aged 2-59 months presenting with cough or difficulty breathing to one of the two busy outpatient clinics and then repeated at 10-min intervals over 1 h in a quiet setting. RESULTS One hundred and sixty-seven children were enrolled with a mean age of 7.1 (SD ± 2.9) months in infants and 27.6 (SD ± 12.8) months in children aged 12-59 months. The mean respiratory rate declined from 42.3 and 33.6 breaths per minute (bpm) in the clinic to 39.1 and 32.6 bpm after 10 min in a quiet room and to 39.2 and 30.7 bpm (P < 0.001) after 60 min in younger and older children, respectively. This resulted in 11/13 (85%) infants and 2/15 (13%) older children being misclassified with non-severe pneumonia. In a random effects linear regression model, the variability in respiratory rate within children (42%) was almost as much as the variability between children (58%). Changing the respiratory rates cut-offs to higher thresholds resulted in a small reduction in the proportion of non-severe pneumonia mis-classifications in infants. CONCLUSION Noise and other contextual factors may cause a transient increase in respiratory rate and consequently misclassification of non-severe pneumonia. However, this effect is less pronounced in older children than infants. Respiratory rate is a difficult sign to measure as the variation is large between and within children. More studies of the accuracy and utility of respiratory rate as a proxy for non-severe pneumonia diagnosis in a busy clinic are needed.