Screening and selection

Two reviewers participated in selection of studies (MT and MP). At the initial phase of selection, abstracts and titles of articles were screened by one reviewer (MT) to exclude articles that clearly were not related to wide-diameter dental implants. The previously described inclusion and exclusion criteria were applied when including articles for full-text screening. When there was a doubt as to the relevance of the article, due to insufficient information in the abstract, the full-text article was analyzed together with a second reviewer (MP).

Search results

The search yielded 553 potentially relevant articles in PubMed, 303 in Web of Science, 35 in Cochrane Central Register of Controlled Trials, 64 from Google Scholar, and 19 not identified in the above search strategies from hand searching of reference lists of selected articles (Fig. 1). After screening the abstracts of the articles, 38 articles were selected for full-text screening from PubMed, 26 from Web of Science, 6 from Cochrane Central Register of Controlled Trials, 16 from Google Scholar, and 17 from hand searching of reference list. After elimination of duplicate articles, a total of 57 articles were selected for full-text screening.

Predetermined inclusion and exclusion criteria were applied to the selected full text by two reviewers (MT and MP); article selection was completed independently and then in unison. Areas of ambiguity were resolved through discussion. As for two articles that needed clarification of patient selection criteria, the authors were contacted via email and given 2 weeks to respond. One author responded with the requested information. Six articles remained for further review.

A total of 51 studies were excluded after the full-text review. These included 19 studies that did not document different lengths of wide implants studied, 15 studies that used graft materials in or around the implant sites, 4 studies that immediately placed implants in extraction sites or upon removal of failed implants, 2 studies that immediately loaded the implants, 10 studies that were retrospective, and 1 study that did not receive a clarification from the author.

Data extraction

One researcher (MT) extracted the data, and a second researcher (DPD) independently checked the data extraction for accuracy and completeness. The disagreements were resolved by discussion. The data extraction form was pilot tested on a representative sample before applying it to the selected articles.

Statistical analysis

The forest plot was used to determine pooled wide-diameter implant survival rate of the selected studies. The funnel plot was used to determine the possibility of publication bias of the selected studies. Heterogeneity of the data was analyzed to determine if the data from the selected studies can be analyzed and if the random effects model can be used in the meta-analysis. In addition, a meta-regression (type III test of fixed effects) was used to evaluate the effects of location, length, and surface on the implant survival.