Volume calibration alone may be misleading

The use of spirometry is becoming more and more widespread in non-laborator y situations such as general practice or occupational medicine. In these no n-laboratory situations, volume calibration with a 3000 mi syringe is often the only feasible method to ensure that the spirometer produces valid and reproducible data. Sophisticated equipment to calibrate forced manoeuvres w ith standard waveforms are not present. In this study, we assessed whether volumetric calibration is a guarantee fo r valid and comparable spirometric results. Two portable spirometers were tested. On 8 consecutive test days, both spir ometers were calibrated with a 3000 mi syringe in accordance with the Ameri can Thoracic Society (ATS) guidelines. The comparability of the spirometric results (forced expiratory volume in 1 S, FEV1) was tested in two ways. Fi rstly, the spirometers were compared to each other using the results from 4 3 volunteers on the same 8 test days. The spirometers were presented in a r andomized order and volunteers were asked to perform a series of reproducib le manoeuvres in both spirometers. Paired observations were analysed, using Bland and Altman plots. Secondly, the spirometers were compared to a 'gold standard', a computer-driven syringe (CDS). Calibration with the 3000 mi syringe showed that both spirometers complied with the ATS criteria for volume calibration for diagnostic spirometry. How ever, paired FEV1 data obtained in subjects showed a systematic, volume-dep endent difference between the two spirometers (mean difference: 289 ml, P < 0.001, systematic difference: 8.6%, P < 0.0001). This systematic differenc e was confirmed by the comparisons with the CDS. Volume calibration may be misleading. The results from volume calibration m ay meet the ATS criteria, but this is no guarantee that data from forced ma noeuvres are accurate. If CDS equipment to simulate standard wave forms is not available, it is recommended that biological calibration is performed r egularly and, if possible, that paired data from two (or more) different sp irometers are compared. (C) 1999 HARCOURT PUBLISHERS LTD.