Rowland 2007.

Study characteristics
Methods	2‐arm parallel RCT
Participants	Number recruited: 397 Sex (M:F): 156:241 Mean age: group 1: 14.8 years (SD 1 year 8 months), group 2: 15 years (SD 1 year 5 months) Inclusion criteria: all participants had preadjusted edgewise fixed appliances, with pretreatment records, treatment plan and study models available, and were willing to wear maxillary and mandibular retainers Exclusion criteria: single arch or sectional fixed appliance treatment; hypodontia requiring tooth replacement on the retainer; poor periodontal status; early debonding; transfer patients; learning difficulties; cleft lip or palate Setting: NHS specialist practice in Bristol, UK, treatment free of charge by 1 operator
Interventions	Two types of removable retainers Group 1: upper and lower Hawley retainers, worn 3 months full‐time then 12 h/day for 3 months Group 2: upper and lower clear plastic retainers ‐ worn full‐time for 1 week then 12 h/day. Full coverage design made from 1.5 mm 'Erkodur' blank Data collected after 6 months of retention
Outcomes	Stability: measured using: LII, intercanine width, intermolar width. Authors reported that tooth rotations, overjet and overbite analysis were measured, but no data were provided in the paper. Patient satisfaction assessed by questionnaire. Results of the most relevant questions provided in the paper Survival of retainers ‐ number of broken retainers recorded for each type of retainer Cost‐effectiveness was also reported, but was not one of the outcomes for our review. Time points: measured at 6‐month period of retention
Notes	Assessment of stability data provided as medians and IQRs. This could not be analysed using Review Manager 5 statistical software (Review Manager 2020), so we requested the raw data, which the study authors provided. Helen Worthington (author of this review) then converted these into data that could be put into Review Manager 5. Patient satisfaction data were dichotomised for full analysis as follows: responses related to whether retainers were worn away from home dichotomised by grouping "always" and "nearly always" together amount of discomfort data were dichotomised by grouping "never" and "on occasion" together comparison to fixed appliances was dichotomised by grouping together "much better" and "better", and grouping together "worse" and "much worse"
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used block randomisation (taken from "www.randomisation.com")
Allocation concealment (selection bias)	Low risk	Concealment done by ensuring randomisation was performed after consent was obtained, using the computer program www.randomisation.com, so the clinician and participant could not know what intervention would be generated
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessor blinded to the type of retainer used
Incomplete outcome data (attrition bias) All outcomes	Low risk	Number of dropouts clearly reported and described
Selective reporting (reporting bias)	High risk	Tooth rotations, overjet and overbite data measured but not reported. The authors also only reported the "most relevant" questions on the patient satisfaction questionnaire.
Other bias	Low risk	Pretreatment equivalence shown for both groups and no other apparent bias present