Borsos 2012.

Methods	Trial design: single‐centre RCT (parallel group) Location: Orthodontic Department of the Heim Pal Children's Hospital, Budapest Recruitment period: not stated Funding source: not stated Source of participants: patients attending clinic Study duration: not stated Time points at which follow‐up are reported: start of canine retraction to end of canine retraction, end of canine retraction to end of frontal retraction, end of frontal retraction to end of treatment, start to end of treatment
Participants	30 participants in total, mean age 14 years, range 12 years 6months to 17 years 5 months 15 in mid‐palatal implant group: 9 males, mean age 14.1 years ± 0.91, 6 females, mean age 14.1 years ± 1.67 15 in the transpalatal arch group: 4 males, mean age 13.8 years ± 0.89, 11 females, mean age 14.44 years ± 1.7 Inclusion criteria: 2 upper first premolar extraction therapy maximum posterior anchorage requirement in the upper arch post‐pubertal growth spurt and sufficient palatal bone morphology for the implant Exclusion criteria: poor oral hygiene, periodontitis, unwillingness to wear fixed appliances or to have the implant placed a medical history precluding gravidity, drug‐, nicotine‐ or alcohol addiction, diabetes, steroid‐ or chemotherapy, immunological or haematopoietic diseases, systematic osteopathy, general surgical contraindications or local radiotherapy
Interventions	Comparison 1: Mid‐palatal implants (Orthosystem, Strauman AG, Waldenburg, Switzerland) The surgical procedure followed the Strauman Institute protocol After 3 months a transpalatal bar was fixed to the implant and connected to the palatal surface of molar bands by laser welding The transpalatal bar was made of a 1.2 square stainless steel wire Comparison 2: Transpalatal arch (TPA) Goshgarian type TPA combined with a 0.017 x 0.025 inch heat treated stainless steel utility arch
Outcomes	Mesial movement (mm) of the upper first molar on superimposed lateral cephalometric radiographs measured during the canine retraction phase, the front retraction phase, the finishing phase, and overall treatment. Data used in this review were from the canine retraction phase Duration of canine retraction
Notes	Both groups were treated with Alexander brackets with a 0.018 x 0.025 inch slots
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote from paper: "...the allocation was carried out by using randomised blocks of six" Quote from correspondence: "We used block randomisation with blocks of six. So we had to wait, until a group of six patients were collected, then we carried out a randomisation. This was repeated five times. In practice, this means that you have to wait until six patients are collected ‐ meeting the inclusion criteria and are willing to participate in the study. As every patient, taken up to our Department, also the participants got a card number, when he first presented themselves. At the randomisation one will be drawn from the 20 possible blocks (closed envelopes), in all of which the two (A and B) groups are represented equally. The selected block would be assigned to the patients based on their card sequence number. (For example: we have six patients with card numbers 2, 4, 6, 7, 9, 12 and block 4 (AABBA) is selected – than patient 2, 4, 12 are candidates in group A and patient 6, 7, 9 in group B)"
Allocation concealment (selection bias)	Low risk	Quote from correspondence: "They were 20 pieces standard size postal envelopes stamped on the sealing. They were stored in the drawer of the head of the department in his closed office. Chief dental nurse was drawing and opening the envelopes (only one at once) in the presence of two witnesses (me and the head of dep.)"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote from paper: "...the cephalometric analysis was carried out anonymously, using an opaque marker in the approximate position of an implant in both groups"
Incomplete outcome data (attrition bias) mesial movement of upper first maxillary molar	Low risk	All randomised patients were accounted for
Selective reporting (reporting bias)	Unclear risk	Selective reporting of outcomes: insufficient information to permit judgement Selective reporting of data: no suggestion of incomplete reporting of data
Other bias	Low risk	Study appears to be free of other sources of bias