We have observed that the results of pulmonary function tests obtained at one site, in general, may not be considered 'acceptable' at another site--in part because of known or suspected variability in equipment and techniques. We sought to document the presence or absence of such variability in our metropolitan area.
Methods & materials: We compared the test results from 5 trained healthy subjects (3 men and 2 women) studied in 13 Denver-area pulmonary function laboratories in a randomized order and at approximately the same time of day.
Results: We performed analysis of variance on commonly reported parameters and found no significant difference for FVC (p = 0.11), FEV1 (p = 0.075), FEF25-75% (p = 0.41), and FRC by helium dilution (p = 0.22). However, marked differences between certain sites could be clinically important. In addition, we found a statistically significant difference for DLCO (p less than 0.001) and TLC (p = 0.024). Six different brands of pulmonary function equipment were used by the 13 hospitals, and differences in the number of trials performed, sequence of testing (eg, FRC determinations were sometimes done first, sometimes last), and calculation of the DLCO breath-hold time.
Conclusion: We conclude that although the FVC, FEV1, FEF25-75%, and FRC measured by helium dilution were not statistically different in healthy trained subjects in the 13 hospitals studied, clinically important differences may exist. The DLCO and TLC were statistically different. To minimize variability and improve comparability, hospitals in a given area should give consideration to adopting standardized techniques, using comparable equipment, and adopting common reference equations.