© 2008 American Thoracic Society
Study on Twins: A Good Tool for Infectious Diseases?To the Editor:We read with great interest the reanalysis of the Prophit Survey (1) of the 1950s by Dr. van der Eijk and colleagues (2). Comstock, in his 1978 reanalysis of that survey, inferred from the 2.5-fold higher tuberculosis concordance rate among monozygotic twins that inherited susceptibility, rather than environment, is a major risk factor for tuberculosis in humans (3). However, a drastically different conclusion was reached with careful subgroup analysis of the old data by van der Eijk and coworkers (2). The difference between monozygotic and dizygotic twins in concordance for tuberculosis was confined to 106 twin pairs involving a smear-positive index case. The odds ratio of concordance was proportional to the intensity of exposure (sputum smear positivity, physical proximity between members of a twin pair, contagiousness of disease, and living together). The fact that drastically different conclusions were reached by two groups of astute researchers using well-established multivariate analysis is perhaps of equal interest as the conclusions themselves. van der Eijk and coworkers attributed such difference to the failure of the previous analysis to take into account strong imbalance (or interaction in statistical terms) of variables within subgroups (2). Indeed, the tendency of monozygotic twins, necessarily of the same sex, to stay in closer proximity could have confounded the opposite conclusion in the previous analysis. Models are just approximations to the reality. No doubt, erratic results could come out when the usual orthogonality assumption broke down with interacting variables in the multiple logistic regression model. There is, however, a more fundamental methodological issue. Although twin studies have frequently been employed to distinguish the effects of heredity from those of environment, the interaction between nature and nurture is sometimes too complicated for such a tool. Unless twins are reared apart from each other, the identical gender and greater likeness of monozygotic twins generally place them in closer proximity than dizygotic twins. This, in turn, poses a major limitation in the study of an infectious disease like tuberculosis, especially in relation to the transmission of disease between members of a twin pair. In this regard, the original design of the Prophit Survey was probably suboptimal for the very question it sought to answer. For that, statistical tools cannot help much, and pitfalls might well be expected.
Department of Health
Grantham Hospital FOOTNOTES Conflict of Interest Statement: Neither author has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
