Peak flow rate records in surveys: reproducibility of observers' reports

Records of peak expiratory flow rate (PEFR), commonly used in hospital in the management of asthma, have not been evaluated as a method of identifying cases of asthma in population surveys. Four observers were asked to report on whether asthma was present or absent in 61 graphs of PEFR recorded two hourly for four weeks during surveys of working population. Agreement within individual observers was measured using a subset of 29 graphs which had been copied and distributed at random among the set of 61; agreement was good, from 90% in one observer to 100% in two. Agreement between observers was measured on the basis of all 61 graphs. Agreement occurred between all four observers in 69% of graphs, between at least three out of four in 97%, and, when pairs of observers were examined, between 72% and 93% of graphs. Graphs assessed as showing asthma demonstrated more within day PEFR variability (expressed as the number of days in which the difference between maximum and minimum readings was at least 15%) than graphs assessed as not showing asthma. Some graphs with little within day variability were assessed as showing asthma, apparently because they demonstrated between day PEFR variability.