Key points are not available for this paper at this time.
This paper presents the use of a matching method for analysis of prospective epidemiologic studies which generate a relatively small number of cases of a given disease. In such studies, matching in the manner presented may be of a given disease. In such studies, matching in the manner presented may be superior to other methods of analysis in removing bias due to confounding. A multivariate minimum-distance matching algorithm is used for computer selection of controls for the cases and the method is illustrated by application to a prospective cardiovascular disease study. The main advantages are that the results are easy to understand and interpret, and that there is no possibiligy of bias due to confounding variables when close matches are obtained. In contrast, the results of multivariate analysis may be difficult to understand and interpret, and the removal of bias due to confounding is dependent on the assumptions of the mode. It is recommended that the matching method should be used in addition to and at times stead of other methods of analysis whenever a study involves a relatively small number of cases and a large source of controls of controls on whom relevant study variables have already been documented.
Smith et al. (Wed,) studied this question.