AbstractBackgroundRespiratory rate is often measured over a period shorter than 1 min and then multiplied to produce a rate per minute. There are few reports of the performance of such estimates compared with rates measured over a full minute.AimCompare performance of respiratory rates calculated from 15 and 30 s of observations with measurements over 1 min.DesignA prospective single center observational studyMethodsThe respiratory rates calculated from observations for 15 and 30 s were compared with simultaneous respiratory rates measured for a full minute on acutely ill medical patients during their admission to a resource poor hospital in sub-Saharan Africa using a novel respiratory rate tap counting software app.ResultsThere were 770 respiratory rates recorded on 321 patients while they were in the hospital. The bias (limits of agreement) between the rate derived from 15 s of observations and the full minute was −1.22 breaths per minute (bpm) (−7.16 to 4.72 bpm), and between the rate derived from 30 s and the full minute was −0.46 bpm (–3.89 to 2.97 bpm). Rates observed over 1 min that scored 3 National Early Warning Score points were not identified by half the rates derived from 15 s and a quarter of the rates derived from 30 s.ConclusionPractice-based evidence shows that abnormal respiratory rates are more reliably detected with measurements made over a full minute, and respiratory rate measurement ‘short-cuts’ often fail to identify sick patients.