The cohort design is more desirable than the case-control design because it provides a direct estimate of incidence rates, and therefore of relative risk, while the case-control design yields only an approximation of the relative risk provided certain assumptions are met, namely, the cases must be representative of all cases of lung cancer and the controls must be representative of the population that gave rise to the cases. These assumptions are not met in many case-control studies. Unfortunately, there are defects of one sort or another in all the studies of these types included in this review. However, the defects in the ten negative studies are less important than those in the three positive studies. The failure to adjust comparisons for differences in smoking habits is one of the sources for confounding when searching for small differences in risk.
Only two of the cohort studies made an attempt to establish comparability of smoking habits; both showed no substantial difference between the persons with pleural plaques and the comparison group, and there was no increased risk of lung cancer among the persons with plaques in either study. These two studies were also the only ones taking latency (interval between onset of asbestos exposure and diagnosis of lung cancer) into account. None of the four case-control studies controlled for smoking habits, despite the fact that this information could have been easily obtained.
A significant problem in evaluating evidence in the published literature arises from publication bias. This results from the tendency of authors, and to a lesser degree editors, to shrink from reporting negative results. Therefore, the literature commonly begins with a positive report or two on a particular issue, followed by a series of negative reports when other investigators attempt to repeat the initial results. This phenomenon is exemplified by the literature reviewed here with the exception that one negative autopsy study, that by Meurman in 1966, antedated the positive cohort studies of Fletcher and Edge. Meurmans data were limited to small parts of a very long article on asbestos bodies and pleural plaques in Finnish autopsies. The likelihood of publication bias in the literature reviewed here lends undue prominence to the three positive studies included.
While it is difficult to prove a negative proposition, pragmatism requires that we make decisions based on available evidence. The weight of the evidence considered in this review indicates that lung cancer risk is not elevated among persons with asbestos-related pleural plaques in the absence of asbestosis. To confirm this conclusion, a large unbiased study of an asbestos-exposed cohort with pleural plaques, defined by strict criteria, and an unexposed cohort with good control of confounders is desirable.