Control-initiated case-control studies

Abstract

In a control-initiated study the investigator begins with existing data on a non-diseased group and then selects cases to obtain comparison data; thus the usual order of case and control selection is reversed. This paper discusses the parallels and differences between control-initiated studies and more traditional case-control designs. Special attention is given to problems of selection bias and matching. A sample-size formula for control-initiated studies is also provided.