Commentary

Incorporation of MIs is widely accepted and increasingly implemented in the orthodontic practice. When the success of MIs is compromised, they will need to be removed or replaced as they have lost the ability to work as stationary anchors against reaction forces.1 Meta-analyses are able to combine the outcomes from several studies into a single quantitative estimate or summary effect size to identify patterns, although homogeneity is hard to find because of clinical and methodological differences among the studies.2 Hong et al. justified conducting a meta-analysis on the combined OR of the success of orthodontic MIs with respect to patient and MIs factors in order to reduce the obstacles for decision-making that follow from conflicting results between the studies, even as they address similar prognostic factors. The primary outcome - success of MIs - was defined as ‘the absence of clinically detectable mobility when the orthodontic force applied by a clinician was sustained regardless of the predetermined period.’ Through the meta-analysis, the authors were able to identify patterns by combining the outcomes of different non-randomised studies. The authors favourably restricted the population under consideration to patients receiving MIs in the buccal posterior regions only, since these are less variably inconstant, which adds homogeneity to their study. This meta-analysis followed the Cochrane Handbook for Systematic Reviews of Interventions3 and the PRISMA statement.4

Its study selection and data extraction were valid as two independent authors performed them and a consensus procedure for addressing disagreements was implemented to solve discordance in article selection. Although the review provided by the electronic literature search of three databases (PubMed, Scopus and Web of Science) with appropriate keywords and MESH terms is extensive, the inclusion of a European database (eg Embase) would have further strengthened the search by identifying other published records that may have not been retrieved. A potential limitation to this study is that the authors added limits in their search strategy to eliminate non-English publications, studies containing microbiology or patient satisfaction, animal and in vitro studies, radiographic evaluations, case reports, meta-analysis, narrative and systematic reviews. Excluding non-English records was justified by the authors pursuant to Morrison et al.'s5 findings that these studies have a tendency to exaggerate treatment effects and to be of lower quality. However, this is still a controversial issue.6,7 On the one hand, the Cochrane Collaboration recommends searching without language restrictions to avoid introducing language bias3 in the results, in spite of the fact that RCTs are more often published in English journals.8 On the other hand, the likelihood of publication bias may have been reduced with the trim-and-fill method after finding bias through the Begg's and Eggar's tests, as was the case with the subgroup of retrospective studies assessing age factors. These statistical tests were correctly performed and reported in Table 4 of the review.

The number of studies according to each reason of exclusion was reported. Nonetheless, the citations were listed neither in the manuscript nor in the appendix.

The scientific quality of the included studies was assessed properly in formulating clinical recommendations because study design, study quality and sample size were considered in the subgroup analyses. This followed the 9-star NOS assessment, the quality assessment recommended by the Cochrane Collaboration to score non-randomised studies.9 This method has previously raised concerns as it may produce highly arbitrary results.10 However, in this case, only two prospective studies in this analysis had a high score (7-9 points), whereas the remaining prospective studies and all the retrospective studies had a medium score (4-6 points).

The methods used to combine the results of studies in a subgroup meta-analysis were appropriate since the p-values of the Q-test were lesser than 0.10, and I2 values were less than 30%. Therefore, low heterogeneity was found.

The source of support for this study was clearly acknowledged in the systematic review (the Korean Ministry of Education). However, the conflict of interest of each included study was not explicitly stated in the review.

Only five main systematic reviews were found in the dental literature related to orthodontic MIs, and no meta-analyses were found prior to 2012.11,12,13,14,15 The following can be concluded from agreement and disagreement between Hong et al. and other similar studies. In 2012 a meta-analysis16 found similar results to Hong et al. regarding success in the jaw of insertion and sex, whereas results differed for age, MIs length and diameter. In 2014, the same prognostic factors as Hong et al. were analysed by a meta-analysis17 that found similar results regarding success in the jaw of insertion, sex and age, whereas results differed for MIs length and diameter.

The summarised review presented herein provides valuable recommendations regarding the clinical implications of prognostic factors of MIs published in The Korean Journal of Orthodontics(IF=1.173). This journal may be considered for being regularly searched for articles for inclusion by Evidence-based Dentistry.18 The meta-analysis of primarily medium-quality primary studies concluded that MIs inserted in the maxilla, patients who are over 19 years of age, and long and large MIs were predictors of higher success of MIs, whereas no difference was found between male and female patients.