Background The diagnosis of occupational asthma requires objective confirmation. Analysis of serial measurements of Peak Expiratory Flow (PEF) is usually the most convenient first step in the confirmatory process. We previously described a statistical method of analysis comparing mean 2-h values on work and rest days which required the worker to wake at similar times on rest and work days. This was achieved in only 43% of records. We describe a new method of timepoint analysis without this restriction and overcoming a theoretical problem with the original analysis (the assumption that the variance of the waking reading was the same as the variance at other times of the day).
Methods Workers were asked to measure PEF approximately 2-h from waking to sleeping for 3–4 weeks. 236 PEF records from workers with independently diagnosed occupational asthma, and 320 from asthmatic controls were available. Readings were grouped by the time since waking, in an attempt to correct for changes in diurnal variation induced by changes in shift and waking time. Daily PEF measurements were meaned into matching 2-h time segments. The pooled SD for rest day measurements (excluding waking readings) was obtained from a one-way ANOVA. Timepoints with mean workday PEF statistically lower (at the Bonferroni adjusted 5% level) than the restdays were counted, after adjusting for the number of contributing measurements at each point.
Results A minimum of four analysable timepoint comparisons per day was needed. 78% of records were suitable for analysis. Records with one or more timepoints statistically worse on workdays gave a sensitivity of 71% against independently diagnosed occupational asthma and a specificity of 93% in non-occupational asthmatics.
Conclusion The removal the requirement to wake at similar times on work and rest days increased the utility of timepoint analysis for the diagnosis of occupational asthma from 43–78% without compromising sensitivity or specificity. Statistical validity was also improved.