Systematic reviews in dentistry have problems including split-mouth and cross-over trials in meta-analysis

Article type
Authors
Carrasco-Labra A1, Brignardello-Petersen R2, Araya I3, Yanine N3
1Evidence Based Dentistry Unit, Faculty of Dentistry, Universidad de Chile, Chile, Department of Clinical Epidemiology & Biostatistics, McMaster University, Canada
2Evidence Based Dentistry Unit, Faculty of Dentistry, Universidad de Chile, Chile, Department of Health Policy, Management & Evaluation, University of Toronto, Canada
3Department of Oral & Maxillofacial Surgery, Evidence Based Dentistry Unit, Faculty of Dentistry, Universidad de Chile, Chile
Abstract
Background: Split-mouth and cross-over trials are frequently used in dentistry. There is some evidence about systematic reviews' authors having problems trying to incorporate these types of studies in meta-analysis.

Objectives: To determine to what extent authors are including the evidence from split-mouth and cross-over trials in systematic reviews and meta-analysis in dentistry.

Methods: All the systematic reviews published by the Oral Health Group in the Cochrane Database of Systematic Reviews up to the issue 3, March 2011 were included. Two researchers examined independently and in duplicated the references list of included and excluded studies. Each reference was searched in MEDLINE to determine whether the trial corresponded to a split-mouth or cross-over trial. When one of these studies was detected, the reasons for exclusion from the review were registered. When these studies were included in the systematic review, it was evaluated whether they were included in the meta-analysis. Also the methods section of each review was assessed and the approaches for including split-mouth and cross-over with the rest of the evidence were recorded.

Results: A total of 114 systematic reviews from the Oral Health Group were retrieved. From these, 51 (44.7%) reviews had split-mouth or cross-over trials in the reference list of included or excluded studies. Of the reviews that included these types of design 19 (57.6%) were not able to include them in the meta-analysis. The most frequent reasons for exclusion due to methodological flaws were inadequate data analysis or inappropriate design (n = 40) and the condition of non-randomized allocation of the treatment to time/side (n = 20).

Conclusions: Authors face many challenges when trying to incorporate the evidence from split-mouth or cross-over trials in systematic reviews. Even when these studies are included, most of the times they do not contribute to the estimation of the treatment effect.