Harradine 1998.

Study characteristics
Methods	Setting: Bristol Dental Hospital, UK Design: parallel (2 arms) No. of centres: 1 Study duration: 5.5 years (best estimate)
Participants	Inclusion criteria: "All patients had previously undergone orthodontic treatment, but on entry into the study were no longer wearing any orthodontic appliances or retainers. Orthodontic treatment comprised active treatment in the upper arch only with either removable appliances or a single arch fixed appliance, with no treatment or premolar extractions only being carried out in the lower arch. All patients had crowded third molars." Exclusion criteria: not reported Patient sampling: N = 164 selected Group 1 (only numbers of those available at follow‐up reported, n = 44): sex and mean age of group not reported Group 2 (only numbers of those available at follow‐up reported, n= 33): sex and mean age of group not reported Overall age reported across both groups at start: 14.1 years ±  16.2 months Overall sex reported across both groups at start: 90 female, 74 male Dropouts: 87 (55%)
Interventions	Orthodontic intervention: extractions Extraction of third molars versus no active treatment (control)
Outcomes	Crowding, arch length
Notes	Funding source not cited
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate method of randomisation
Allocation concealment (selection bias)	Unclear risk	Unclear if allocation sequence concealed from those assigning patients to participant groups
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not possible to blind participants and personnel due to the different interventions used
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Adequate blinding of assessor Quote: "The third molar status was unknown to the digitizer"
Incomplete outcome data (attrition bias) All outcomes	High risk	High dropout rate with only 45% of the original participants completing the study
Selective reporting (reporting bias)	Low risk	All outcomes reported
Other bias	Unclear risk	Unclear due to recall rate