Airways Obstruction From Asbestos Exposure: Possible Confounding Factors
Reader variability in classifying asbestosis from chest radiographs, particularly between ILO profusion categories of which the most critical is between negative and positive (ILO profusion 0/1 vs 1/0), could enhance or reduce differences in pulmonary functions. Using one consistent reader minimizes this variability. Interpreting an abnormal radiograph as normal, thereby placing a subject with asbestosis into the asbestos-exposed group, would decrease the exposed to asbestosis difference. Calling a normal radiograph abnormal would also decrease the difference between exposed and asbestosis. Such random misclassifications would cancel. Small differences in the classification of asbestosis were not critical because they were mostly of ILO profusion 1 and the large physiologic effects were observed that were consistent with our inferences and past interpretations. Whether variations in visibility may be confounding as mean differences in the intensity of peribronchiolar fibrosis must await morphometric studies of airways from lungs of subjects whose pulmonary functions were studied during life and accurate morphologic or biochemical quantitation of asbestosis.
Measurements of TLC from chest radiographs may underestimate volume due to failure to take in or to hold a complete inspired breath, assuming that radiographic technique is standard and planimetry is done systematically read only ventolin inhaler. Underestimation evidently did not occur to a substantial extent because TLC was increased in men who smoked whether they were asbestos exposed or had asbestosis.
There is no reason to believe that the relationships between pulmonary function measurements and asbestos exposure and radiographic signs of asbestosis are biased by workers’ health concerns. All these workers volunteered for study but wives made most of the appointments so that wellness or the workers’ lack of health concern vs high concern may have been offset by their wives’ concern.
For diseases with long latent periods, it would be ideal to use each subject as his own control, but this is seldom practical because it presupposes establishing surveillance before risk is known. Furthermore, studies of the natural history of diseases known to be preventable are not ethical.
Cross-sectional studies of large populations to visualize the natural history by comparing age соhorts may error by assuming that the average behavior in each age category mirrors the population trends that have been shown for unexposed subjects who never smoked. They further assume stable or diminishing exposures to asbestos decade to decade. This appears to be true for asbestos exposure. To assure realistic observations of the effects of asbestos exposure and asbestosis, groups must be large, matched for age, and analyzed to adjust for smoking behavior and other factors. The only other consistent factor, ambient air pollution, apparently averaged as no regional differences in function were found within the study.