3.1 Study selection
The study selection process is illustrated in Figure 1. A total of 583 articles were identified, and upon review of the titles and abstracts, 420 were excluded, leaving 19 articles. After reading the full texts, 8 studies were included in the present review for qualitative and quantitative synthesis. The 8 remaining studies included three RCTs[14, 32, 33] and five CCTs[15, 37-40]. Information about the excluded records is summarized in Additional file 2.
3.2 Study characteristics
The characteristics of the included studies are presented in Table 1, and the relevant data extracted from the included articles are shown in Table 2. Three RCTs[14, 32, 33] and five CCTs[15, 37-40] were included in the present review. Six studies investigated direct implant anchorage for canine retraction, while two studies used indirect implant anchorage. Four studies[14, 38-40] compared TADs with dental anchorage in a split-mouth study. Two studies[32, 33] compared TPAs with TADs in the maxilla in a parallel study; one[33] also compared lingual bars with TADs in the mandible. The other one[15] compared midpalatal implant-reinforced TPAs with conventional TPAs in the maxilla. One study[37] compared midpalatal implants with dental anchorage in the maxilla and TADs with dental anchorage in the mandible. Three studies[14, 38, 39] inserted implants in both the maxilla and the mandible when the patients’ ANB angle was between 2° and 4° but in only the maxilla when the ANB angle was greater than 5° as a part of camouflage treatment.
3.3 Risk of bias assessment
Three RCTs[14, 32, 33] were considered to have a high risk of bias because none of them reported using an appropriate strategy for blinding the participants or personnel. The study by Davis et al, 2018[14], used a computer-generated program to randomly allocate the sides only, and did not perform allocation concealment. The randomization method used in the study by Sharma et al, 2012[32], involved random numbers generated by a computer, but the allocation was performed by alternation, which leads to a high risk of bias for allocation concealment. The third study[33] did not report the use of any randomization method, which resulted in an unclear risk of bias. Blinding of the outcome assessment was also difficult in the studies because the TADs could be observed in lateral cephalograms. However, Sharma et al, 2012[32], removed the miniscrew implants and TPAs before obtaining the cephalometric radiographs that were taken after the completion of canine retraction, resulting in a low risk of bias. Davis et al, 2018[14], used guide wires to differentiate the right and left sides on the lateral cephalograms, resulting in a high risk of bias. Gökçe et al, 2012[33], did not report using a process for blinding the assessor, resulting in an unclear risk of bias. The quality assessment results of the RCTs are summarized in Figure 2.
Five CCTs[15, 37-40] were assessed using the ROBINS-I[36] tool. The study by Hedayati et al, 2007[37], gave inadequate information regarding the patient inclusion criteria, Thiruvenkatachari et al, 2006[38], and Thiruvenkatachari et al, 2008[39], only inserted implants in the maxilla when the ANB angle was greater than 5° as a part of camouflage treatment, which led to a moderate risk of bias in selection. In the measurement of outcomes, Thiruvenkatachari et al, 2006[38], and Thiruvenkatachari et al, 2008[39], used wires identifiers, and Chaudhary et al, 2014[40], used CBCT-generated 2D cephalometric with implants clearly seen on it, which led to a serious risk of bias in the measurement of outcomes. Borsos et al, 2012[41], used an opaque marker in the approximate position of the implant in both groups, which led to a low risk of bias. Finally, Hedayati et al, 2007[37], did not present information regarding outcome measurements. Therefore, the overall bias across studies was serious bias in three studies[38-40], moderate bias in one study[37] and low bias in another study[41]. The risk of bias information for the included CCTs is summarized in Table 3.
3.4 Primary outcome measures
3.4.1 Mesial molar movement (anchorage loss)
Seven studies[14, 15, 32, 33, 37, 38, 40] were qualified for meta-analysis, and the total and subgroup analysis results are given in Figure 3(a, b). In the maxilla, the results showed a total mean difference of 1.56 mm (95% CI: 1.14 to 1.98), with statistical significance (P<0.00001). Subgroup analysis showed a mean difference of 1.74 mm (95% CI: 1.32 to 2.17, P<0.00001) in the direct group and a mean difference of 0.93 mm (95% CI: -1.04 to 2.90, P=0.35) in the indirect group. In the mandible, the results showed a total mean difference of 1.62 mm (95% CI: 1.24 to 2.01), with statistical significance (P<0.00001). Subgroup analysis showed a mean difference of 1.45 (95% CI: 1.13 to 1.78, P<0.00001) in the direct group. Only one study[37] included a mandibular indirect group; the results showed a mean difference of 2.73 mm (95% CI: 1.98 to 3.48, P<0.00001). In both the maxilla and mandible, the direct and indirect groups showed substantial heterogeneity, with I2> 50%.
3.4.2 Distal canine movement
Four studies[14, 33, 39, 40] were qualified for meta-analysis of both maxillary and mandibular data, and the results are given in Figure 3(c, d). In the maxilla, the results showed a total mean difference of 0.43 mm (95% CI: 0.16 to 0.69), with statistical significance (P=0.001); I2=0. In the mandible, the results showed a total mean difference of 0.26 mm (95% CI: 0.02 to 0.49), with statistical significance (P=0.03); I2=0.
3.5 Secondary outcome measures
One study[14] included mesial tipping of the maxillary and mandibular molars with direct TADs. The results showed a mean tipping of 0.30° degrees in the TAD group and a mean tipping of 2.45° in the conventional anchorage group in the maxilla (P=0.000); in the mandible, the values were 0.1875° and 2.6875° (P=0.001), respectively.
One study[37] included vertical displacement of the maxillary and mandibular molars with indirect TADs. The results showed a mean intrusion of 0.33 mm in the study group and 0.95 mm in the control group in the maxilla; in the mandible, the results showed a mean intrusion of 0 mm in the study group and 1.02 mm in the control group. With consideration of the cephalometric error (-0.55 mm), slight maxillary molar extrusion and mandibular molar intrusion were observed.
One study[40] included distal tipping of the canines in the maxilla and mandible with direct TADs. The results presented tipping of 9.51° in the study group and 6.51° in the control group in the maxilla (P=0.106); in the mandible, the results showed tipping of 7.88° in the study group and 4.34° in the control group (P=0.057).