Inclusion of less than five patients

Lack of clinical data on anchorage loss

Measurement of anchorage loss not by superimposition of lateral cephalograms or superimposition of study casts

Previous orthodontic treatment

Treatment in control group not specified

Inclusion of diseased patients, e.g., patients with systemic diseases, periodontal disease, and syndromes

Other treatment than en masse retraction and mini implants

Other sources of skeletal anchorage than orthodontic mini implants or micro implants

At least two review authors examined the titles and abstracts of the identified studies and reports independently. Reports which were clearly not relevant were excluded, whereas full-text documents were retrieved for all potentially relevant studies and eligibility was assessed according for the criteria defined in advance. Disagreements were resolved by open discussion occasionally arbitrated by an independent assessor (D.D.). A data extraction template was generated including the items’ study design, population, type of implants, number of implants, location of the implants, time points of observation, treatment duration, control intervention, measurement method, and primary and secondary outcomes as well as risk of bias (Additional file 1). Data extraction was performed independently by at least two review authors.

For qualitative and quantitative data analysis, the horizontal and vertical anchorage loss values associated with direct and indirect anchorage against a control measure were defined as primary outcomes. For qualitative data analysis, transversal anchorage loss, treatment duration, and implant failures with direct and indirect anchorage were defined as secondary outcomes.

A quality assessment of all selected full-text articles was performed according to the Cochrane Collaboration’s tool for assessing risk of bias (low, high, unclear) including the following domains: random sequence generation, allocation concealment, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias.

Quality assessment was performed in two different phases. In the first phase, quality assessment was conducted independently by at least two authors (A.P., C.B., K.B.) based on the published full-text articles. In the second phase, disagreements were resolved by discussion. A risk of bias table was completed for each included study.