The Cotwin Method and Experimental Science

A strict evaluation of cause and effect has for long been regarded to require the tools of experimental design. The main approach is to constitute two or more subject groups by randomization or equating procedures, whereupon one or more groups will be manipulated by the experimenter, whilst one groups is left unmanipulated and thus serves as a control group. There are numerous experimental designs aimed at quantifying the effect of manipulation, one of which is the cotwin control method. The advantage of the method is the minimizing of the intrapair variance by the genetic identity between the members of a pair. Experiments, however, have limitations in problems where the presumed causal factor cannot be manipulated but exists in certain population groups and not in others. In comparing such groups the method of randomization in its true sense cannot be employed. Even matching or equating may cause bias because the variable or variables which induced the presumed causal factor in certain individuals may not be known or measurable and are thus not equatable. In such cases a modified cotwin method may be used, which employs twins where the one member has the presumed causal factor and his cotwin not. The results of a comparison between the members of such discordant pairs will not be biased by genetic differences that may be correlated to the presumed causal factor. The greater resemblance between the twins in a pair in early environmental influences will further enhance the reliability of the method. This application of the cotwin method is of great value in epidemiological research in connection with evaluation of the effect of environmental exposures such as air pollution, tobacco smoking, etc. It has been applied in this sense in Denmark, USA, and Sweden, and examples will be given.