1. The aim of this study was to determine the validity of various numerical indices of peak flow variability as tests for asthma.
2. Normal (n = 24) and asthmatic (n = 123) subjects aged 10-70 years were recruited from the community. Asthma was diagnosed by full clinical assessment and was active during the previous year. Subjects recorded their peak flow immediately after rising in the morning, at 18.00 hours and just before retiring at night for 12 days.
3. The most discriminating index was the highest peak flow variability which occurred within any 1 day during the 12-day recording period, calculated as the difference between the maximum and minimum peak flows expressed as a percentage of the minimum peak flow on that day (the proposed index). This index was <20% (90th centile) in 96% of the normal subjects and ≧ 20% in 89% of the asthmatic subjects.
4. Standard indices of mean peak flow, the forced expiratory volume in 1.0 s and its responsiveness to salbutamol, had much lower sensitivities than peak flow variability.
5. It is concluded that numerical indices of peak flow variability are highly valid tests for asthma. Using an upper limit of normal of 20%, the proposed index of peak flow variability discriminates better than other indices between asthmatic and normal subjects.