Data sourcesPubMed, Embase, Cochrane Library and LILACS database, review of references cited in included articles and a manual search of leading orthodontic journals. No language restrictions were imposed in the search. Study authors were contacted when necessary.Study selectionRandomised controlled trials (RCTs) and controlled clinical trials (CCTs) in healthy patients that directly compared the adverse effects following treatment using buccal and lingual appliances. Studies involving single arch or dual arch appliances were considered. Studies on patients with systemic diseases, animal studies and in vitro studies were excluded. The primary outcomes of interest to the authors were a list of adverse effects: pain, caries, eating and speech difficulties and oral hygiene.Data extraction and synthesisTwo authors reviewed the titles and abstracts of all studies identified through the search without blinding to names of authors or publication dates. Selected articles from searches were evaluated independently by two authors against established inclusion criteria, disagreements were resolved by consensus or by consulting a third author. Two authors independently assessed the risk of bias using the Cochrane Collaboration's tool (randomised trials) and the Newcastle-Ottawa Scale for non-randomised studies. The level of agreement between the authors was assessed using the Cohen kappa statistic. A meta-analysis was performed to provide pooled effect estimate (expressed as odds ratio) as well as 95% confidence interval. The outcomes of interest were pain, caries, eating difficulties, speech difficulties and deficient oral hygiene. Heterogeneity was quantified using I2 statistic and potential causes explored. Publication bias was assessed using a funnel plot.ResultsEight articles were included; three RCTs and five CCTs. One RCT was considered to be at high risk of bias, one moderate risk and one low risk. Of the non-randomised studies, four were low risk and one was high risk of bias. Six studies involving a total of 131 patients were included in a meta-analysis. The lingual appliance was associated with significant pain in the tongue (OR=28.32, 95% CI 8.6-93.28), difficulty in maintaining oral hygiene (OR=3.49, 95%CI 1.02-11.95) and greater speech difficulty (OR = 9.39, 95% CI 3.78-23.33) compared to buccal appliances. On the other hand, patients with lingual appliances had decreased pain in the lips and cheeks. There was no difference between the two appliances with regards to caries risk.ConclusionsLimited available evidence indicates that lingual orthodontic appliances are associated with increased pain in the tongue, speech difficulties and difficulty in maintaining oral hygiene.