PEAK FLOW VARIABILITY IN ASTHMA DIAGNOSIS - WHICH IS THE BEST THRESHOLD TO USE IN PRACTICE?

Introduction Although peak flow variability (PEFv) is one of the diagnostic tests in the NICE guidance for asthma (NG80), instructions on how to calculate PEFv are brief; PEFv can be expressed as amplitude% mean (A%M) or amplitude% highest (A%H, a simpler calculation). A positive test can be recorded if the average daily amplitude ≥20% (A%M20, advised in NG80), ≥15% (A%M15) or ≥10% (A%M10). We investigated the effect of changing the way a positive PEFv test is defined on the classification of adults and children with symptoms in keeping with asthma. Methods Adults and children with symptoms in keeping with asthma were referred by primary care for a thorough evaluation. Participants were asked to record PEF at least twice per day for two weeks with an electronic PEF meter. Participants completed a detailed panel of lung function tests and were treated with ICS for 8 weeks. All tests and response to treatment were evaluated by a panel of consultant respiratory physicians and participants were diagnosed with asthma or not asthma (or insufficient evidence). We calculated sensitivity and specificity of different PEFv thresholds in diagnosing asthma Results Of the 117 symptomatic participants recruited into the study, 91 (50 adults) completed PEF measurements for ≥4 days (median 10 days [IQR 8–12 days]). A significant correlation (R=0.927) was found between PEFv for patients who collected ≥10 days PEF and their first 5 days of PEF data. Of 47 adults with sufficient data, 31 were diagnosed with asthma. Of 31 children with sufficient data, 23 had asthma. Sensitivity and specificity of different thresholds of PEFv are shown in table 1. Only 3 adults with asthma had PEFv A%M20, indicating this threshold is too high. Expressing results as A%M or A%H made little difference in sensitivity or specificity, as did using the first 5 days of data rather than all data collected. Conclusion PEFv could be calculated as A%H10 for 5 days as part of the NICE CG80 diagnostic algorithm, as this is simpler to collect and to calculate, is more sensitive, and has little loss in specificity compared to A%M20 for 2 weeks.