Debate in Print

Review of the Kilburn and Warshaw Chest Article—Airways Obstruction From Asbestos Exposure

Design

The reference group was studied in the early 1980s by a different group of investigators, for a different purpose.² This is similar to the risky but sometimes unavoidable use of historic control subjects in clinical trials. It should not be necessary to do this in large occupational morbidity surveys; with data on 8,720 asbestos-exposed subjects, it was necessary to identify a minimally exposed subgroup that was comparable in other important respects, such as age and smoking history. The

Data Collection

One reader read all the chest x-ray films “for consistency of interpretation.” This certainly eliminates between-observer variability, but it does not conform to standards for this kind of research. Furthermore, the reader not only knew the worker's exposure and health status, he rendered his reading in the worker's presence. A procedure more open to bias can scarcely be imagined.

Analysis

The reference group came from the Michigan population study described by Miller et al in 1986,² and the spirometry predictive equations used by Kilburn and Warshaw come from the same paper. But those equations were derived from younger subjects (about 20 years younger than Kilburn and Warshaw's subjects), and they were screened to exclude those with common respiratory symptoms or diagnoses, angina or diagnosis of coronary artery disease, certain physical examination findings, or body weight

Interpretation

There is no need to belabor the errors of interpretation. Proper analysis of the lung function data shows that the authors actually failed to demonstrate the purported differences on which they based their main conclusions.

Even without resort to statistics, there are anomalous features of the study. Here is a population with large numbers of old and long-service workers, in which there are no important differences in the lung function of those with and without asbestosis (what does this say