Variability of respiratory rate measurements in children suspected with non-severe pneumonia in north-east Tanzania

Florida Muro, Neema Mosha, Helena Hildenwall, Frank Mtei, Nicole Harrison, David Schellenberg, Raimos Olomi, Hugh Reyburn, Jim Todd
Publication year: 


Measurement of respiratory rate is an important clinical sign in the diagnosis of pneumonia but suffers from inter-observer variation. Here we assess the use of video recordings as a quality assurance tool that could be useful both in research and in training of staff.


Respiratory rates (RR) were recorded in children aged 2 - 59 months presenting with cough or difficulty breathing at two busy outpatient clinics in Tanzania. Measurements were repeated at 10-minute intervals in a quiet environment with simultaneous video recordings that were independently reviewed by two paediatricians.


859 videos were sent to 2 paediatricians. 148 (17.2%) were considered unreadable by one or both. For the 711 (82.8%) videos that were readable by both paediatricians, there was perfect agreement for presence of raised RR with a kappa value (κ) of 0.85 (p<0.001); and in 476 (66.9%) cases both paediatricians agreed on the RR within 2 breaths per minute (+/- 2 bpm). A reported illness of 5 days or more was associated with unreadable video recordings (OR=3.44, CI 1.5-6.08; P<0.001). The multilevel model showed that differences between observers accounted for only 13% of the variability in RR.


Video recordings are reliable tools for quality assurance of RR measurements in children with suspected pneumonia. Videos with a clear view of respiratory movements may also be useful in training primary healthcare staff.