A total of 681 primary references were initially identified. After screening the titles and abstracts, forty-five references were left for full-text evaluation. Hand searching of the reference lists of selected studies did not identify additional papers. After full-text evaluation, eight papers were included in the final analyses. Two of the papers were included in the meta-analysis for treatment effectiveness and three of the papers were included in the meta-analysis for treatment efficiency (inter-rater agreement = 99%, kappa = 0.93). The flow diagram of literature search process is presented in Fig. 1.
In the eight included studies, the earliest study was published in 2005 , whereas the most recent study was published in 2018 . In terms of the geographic locations, four studies were conducted in America [10, 12,13,14], three in Europe [11, 15, 16] and one in Asia . In terms of study design, two of the studies were randomized controlled trials [16, 17], one study was prospectively clinical comparative study , and five studies were retrospectively clinical comparative studies [10,11,12,13,14]. All the investigations were performed in the universities. The treatments were conducted by clinicians specialized in orthodontics in the included studies except one study not reporting who conducted the treatment .
A total of 353 participants were treated with clear aligners, while another 353 participants were treated with fixed braces. The number of patients in each study ranged from 11 to 76. The gender ratio of all included studies was balanced between two groups except one study not reporting the gender of patients . The mean age of patients ranged from 15.5 to 35.2 years. Seven of the studies included nonextraction patients into the research [10,11,12,13,14,15,16] and one study included extraction patients . The included patients in three studies were Class I malocclusion [14, 15, 17], while the remaining five studies did not mention the classification of malocclusion of patients [10,11,12,13, 16]. Table 1 provides detailed study characteristics of the included studies.
Two studies evaluated the treatment effectiveness of two orthodontic appliances by using methods from the American Board of Orthodontics Phase III examination [10, 17]. The initial severity of malocclusion analyzed by the discrepancy index was controlled in two groups. The objective grading system (OGS) which consisted of the measurements of alignment, marginal ridges, buccolingual inclination, occlusal contacts, occlusal relations, overjet, interproximal contacts, and root angulation was used to systematically grade treatment effectiveness. The total number of points lost was the OGS score. One included study found that the clear aligners group significantly lost more OGS points than the braces group did on average . Another study found no significant difference between two groups . The result from meta-analysis illustrated that there was no statistically significant difference between two groups in OGS score. (WMD = 8.38, 95% CI [− 0.17, 16.93]; P = 0.05). High heterogeneity was evident among the included studies (P = 0.004, I2 = 88%). The forest plot of OGS score is presented in Fig. 2. Moreover, these two included studies found that clear aligners scores were consistently lower than braces scores in buccolingual inclination and occlusal contacts. Meanwhile no significant difference was found in scores for alignment, marginal ridges, inter-proximal contacts, and root angulation between two groups. One included study found that the scores for occlusal relationships and overjet were lower in clear aligners group than braces group , while another study did not . A case which lost 30 or fewer points received a passing grade for the ABO Phase III examination. Both included studies found that the passing rate of the ABO Phase III examination was lower in clear aligners group than the one in braces group.
One study evaluated the postretention dental changes of treated patients using OGS score. Two groups had no significant difference in total OGS score change between posttreatment and postretention time, but patients treated with clear aligners relapsed more than those treated with braces in alignment .
Two studies evaluated the treatment effectiveness of the two kinds of orthodontic appliances using the Peer Assessment Rating index (PAR) [11, 12]. The PAR score was used to assess eight components: maxillary anterior segment alignment, mandibular anterior segment alignment, anteroposterior discrepancy, transverse discrepancy, vertical discrepancy, overjet, overbite, and midline. The results from two studies showed that there was no significant difference in either total PAR score reduction or the changes of all eight components between two groups. Richmond et al.  determined that a reduction of 22 PAR points brought about great improvement for a case. Gu’s study defined the cases with a reduction of 22 PAR points and the cases with pretreatment PAR scores less than 22 points getting scores equal to 0 at the end of the treatment as great improvement and concluded that clear aligners group had a significantly lower rate of receiving great improvement than braces group . Lanteri’s study expanded the range of great improvement. They defined great improvement as PAR score reduction > 70% or a reduction in PAR score > 22 or PAR score = 0 in the end and found no significant difference between two groups .
Two included studies reported the treatment effectiveness on dental arches dimension [14, 15]. Grunheid et al.  found that clear aligners tended to increase mandibular intercanine width during alignment in contrast to braces. Pavoni et al.  found that braces produced significantly more transverse dento-alveolar width of maxillary intercanine and interpremolar, and more perimeter of maxillary arch width than clear aligners did, while two groups had similar effects on increasing intemolar width and maxillary arch depth.
Two studies focused on the effect of clear aligners on the proclination of mandibular anterior teeth [14, 16]. Grunheid et al.  found that treatment with braces significantly decreased the proclination of mandibular canines in contrast to treatment with clear aligners which tended to increase the intercanine width instead of decreasing inclination. Hennessy et al.  found that braces produced more mandibular incisor proclination during alignment than aligners did, but no statistically significant difference was found between two groups.
Four of the included studies found that clear aligners group had a shorter treatment duration than braces group did [10,11,12, 14] and three studies found no significant difference between two groups in nonextraction patients [13,14,15]. One study found that braces were more efficient than clear aligners in extraction patients . The data extracted from each included study about treatment duration (month) is presented in Table 2. Three studies were included in the meta-analysis [12,13,14]. The results illustrated that patients treated with clear aligners had a statistically significant shorter treatment duration than the patients treated with braces did. (WMD = − 6.31, 95% CI [− 8.37, − 4.24]; P < 0.001). Low heterogeneity existed among the included studies (P = 0.86, I2 = 0%). The result from meta-analysis is presented in Fig. 3.
All of the six included cohort studies were estimated to be of high quality. According to the Newcastle-Ottawa Scale, five studies were given 8 stars [10,11,12,13,14] and one study was given six stars . The other two included randomized controlled studies were estimated to be of moderate quality. According to the recommendations by Cochrane, one study had low risk of biases in six criteria and high risk of bias in one criterion . Another one study had low risk of biases in six criteria, high risk of bias in one criterion, and unclear risk of bias in one criterion . The results of quality assessment are presented in Tables 3 and 4.