To the Editor

Abstract

We would like to thank Dr. Dunning for his thoughtful evaluation of our article (May 2004). 1 Townsend MC Hankinson JL Lindesmith LA et al. Is my lung function really that good? Flow-type spirometer problems that elevate test results. Chest. 2004; 125: 1902-1909 Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar We agree with him on the need for the formal training of spirometry technicians and interpreters, and on the need for those individuals to recognize spirometry errors like those we described. All of our authors have been involved in teaching spirometry courses that have been approved by the National Institute for Occupational Safety and Health, and, in writing this article, it was our goal to introduce a different class of errors (equipment errors that develop during subject testing) into the discussion of spirometry quality assurance and into spirometry training courses. To be effective, training should cover all types of errors that spirometry users may encounter and address how to correct the problems. If errors with currently available spirometers are not discussed in the medical literature, the appearance of the errors, their probable causes, and their solutions will remain anecdotal, and will not come to the attention of many spirometry users. We wrote the article because the zero errors and contaminated sensor errors differ from the types of errors that are usually discussed in training courses, and these have the following four important ramifications:* 1To date, the focus of the pulmonary medicine community has been on errors in testing technique and leaks in volume spirometers, most of which lower the measured values. When these errors are corrected, there is no residual effect on the reported measurements, since inaccurate values are replaced by larger, more accurate values. Thus, the Occupational Safety and Health Administration/American Thoracic Society algorithm, which reports the largest FVC and FEV1 values, functions well. However, since the errors that we described elevate the test result values, they will be reported as the subject's measured values, displacing any lower but accurate results if they are not recognized and deleted. 2It is widely assumed that checking or setting the calibration of a spirometer guarantees the accuracy of subsequent results that have been recorded on that spirometer. However, we reported errors that develop during the subject test, after the calibration has been verified or set. The existence of these errors means that the shapes of spirograms and the temporal patterns of test results should be monitored continually, and that passing a calibration check, while essential, does not guarantee the accuracy of later subject tests. The clinical settings that produced the screening data shown in Figure 5 in our article assumed that the measurements were accurate because the spirometer was calibrated successfully, and no errors were flagged during the tests (ie, the temporal patterns of the results were not scrutinized). 3There is a common tendency for technicians to rely on computer diagnostic messages during testing. If no message is given, it is assumed that the test is acceptable. Many spirometers now being marketed have either no visual display or a very small display, which is a much different design than that of the spirometers available in 1978 when the Occupational Safety and Health Administration Cotton Dust Standard was promulgated. None of the errors that we presented in our article were flagged by the spirometer on which they were recorded, except perhaps as showing excessive variability. 4Finally, lack of reproducible results is generally assumed to be caused by the failure of the patient to perform the maneuvers in a consistent fashion. However, the zero errors and contaminated sensor errors that we presented in our article are likely to cause excessive variability, and this fact needs to be presented as an alternative explanation for poor reproducibility in our training courses.