Application of the missing-indicator method in matched case-control studies with incomplete data

A common practice in matched case-control studies with incomplete data is t o perform two analyses in parallel: a matched analysis of the complete pair s and an unmatched analysis of all subjects carried out after breaking the matching in the complete pairs. The missing-indicator method, which has the advantage of making use of the data in the incomplete pairs while still pr eserving the matching in the complete pairs, is recommended as an alternati ve method of analysis, it is shown here that its estimate of the odds ratio is a compromise between the odds ratios estimated by a matched analysis of the complete pairs and an unmatched analysis of the incomplete pairs. The method is illustrated using data from a matched case-control study of the r isk of childhood leukemia from exposure to residential electric and magneti c fields.