Selection bias in case-control studies on periodontitis: a systematic review

In case-control studies the frequency of the exposure of interest is compared between a group of diseased subjects and a group of controls to determine whether an association exists between disease and exposure. Case-control studies are useful, but can be subject to several sources of bias if poorly conducted. Selection bias, which results in a lack of comparability between the groups being studied, is one of the most harmful types of bias. The aim of this study was to assess the information reported on case-control studies of periodontitis to identify sources of selection bias. We conducted an electronic search in PubMed, EMBASE, and Web of Science and evaluated the occurrence of sources of selection bias in case-control studies published in English during the year 2004. In relatively few studies did the authors provided information on recruitment periods for cases and controls (31.1% and 20%, respectively), sampling methods (26.7% and 31.1%, respectively), or participation rates (8.9% and 6.7%, respectively). The source of control subjects was appropriate in 15.6% of the studies, and the strategy used to select the controls was adequate in only 8.9% of the studies. It may be concluded that case-control studies on periodontitis are frequently inadequately conducted and reported.