Walton 2002.

Methods	Trial design: 3‐year follow‐up, randomised parallel group trial Factorial design randomised trial where participants were randomly allocated to 1 of 4 treatment groups using balanced randomisation (retained by a ball or by a bar and clips, with or without a metal framework). The treatments were assigned randomly in blocks of 4 participants (via random permutations, generated in an Excel spreadsheet), using distinct randomisation within each of the 4 strata based jointly on gender and on ridge resorption
Participants	Inclusion criteria: edentulous participants with at least 1 year of experience wearing conventional complete dentures, medically and psychologically suited for implant surgery, able to complete the trial forms and to communicate orally in English, willing to commit to 2 years of participation in the trial after receiving new dentures (Walton 2002 and MacEntee 2005) but Walton 2003 states "be available for the duration of the trial" Exclusion criteria: received treatment previously with implants, need for additional pre‐prosthetic surgery, insufficient bone height for at least an 8.5 mm mandibular implant, or a history of systemic or neurologic disease or head and neck radiation, participants were not enrolled in the trial after implant surgery because the prosthodontist found that the implants diverged more than 15 degrees from each other, or that the implants were located less than 20 mm or more than 35 mm apart, because of evidence that such an orientation and location of implants could disturb the stability and maintenance of the implant overdenture Age at baseline (years): bar‐clip group: 61, ball‐spring: 63 (from MacEntee 2005); but Walton 2003 says 41 to 89 years (mean 65 years); and Walton 2002 says "ranged in age from 41.4 to 88.9 years, with a mean age of 64.4 years" Gender (M/F): bar‐clip group: 12/22, ball‐spring: 13/21 (from MacEntee 2005); Walton 2002 says "female participants (n = 41) outnumbered male participants (n = 23) by a ratio of almost 2:1" Number of mandibular overdentures randomised: 100 (mandibular 2‐implant overdentures) Number of mandibular overdentures evaluated at 1 year: 67 Number of mandibular overdentures evaluated at 2.98 years: 87 Note: number evaluated (participants): Walton 2002 says "of the 100 participants enrolled, 67 had their overdenture for a minimum of 1 year prior to the present analysis". 3 of these 67 were lost to follow‐up (1 died, 1 delayed treatment because of financial problems, and 1 was lost to follow‐up after implant placement surgery), leaving 64 continuing participants who had a mandibular overdenture for at least 1 year Note: MacEntee 2005 says 68/136; 87 participants had their overdentures for 2 to 4 years (mean 2.98)
Interventions	Mandibular 2‐implant overdentures ‐ splinted and unsplinted prosthodontic design Test group 1 (n = 34 at 1 year): Branemark® 2 mm single round gold bar joint patrices and single clip matrices Test group 2 (n = 34 at 1 year): Branemark® 2.25 mm ball patrices and titanium alloy spring matrices
Outcomes	Data were collected prior to implant surgery, and at 1 month, 1 year, 2 years, and 3 years after prosthesis placement, except for patient satisfaction which was not evaluated at year 3 Prosthodontic success by specific categorization Appendix 1 and 6‐field protocol Appendix 2 Prosthodontic maintenance by general categorization Patient satisfaction/quality of life assessed by a validated measure (VAS 100 mm scale after de Grandmont 1994) Cost (treatment time and/or material costs) measured in US dollars
Notes	Location: Canada Funding source: supported by Health Canada, National Health and Research Development Program (project number 6610‐2061‐403) and Nobel Biocare Canada Authors contacted by email (30 August 2015): reply received 14 September 2015. Unpublished information supplied for risk of bias table MacEntee 2005, Walton 2002, Walton 2003: all reports of same trial (Walton 2002) Details of ethical approval: not mentioned
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The treatments were assigned randomly in blocks of 4 participants (via random permutations, generated in an Excel spreadsheet)"
Allocation concealment (selection bias)	Low risk	Clarification from trial authors: "Statistician (Dr Ned Glick) completed the stratification and randomisation protocols, and provided sealed envelopes for participant allocation. The outer envelopes were each labelled with a stratum name (e.g. female, severe ridge resorption) and contained additional sealed envelopes specifying either number of implants or type of attachment system. Once a trial clinician had determined, according to pre‐defined criteria, the stratum to which the participant belonged, the trial dental assistant was instructed to confirm the stratum and then draw and open, in the presence of the clinician, an allocation envelope from that stratum. The assignment was then recorded on trial forms for that participant"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The trial did not report blinding, but we judge the risk of bias as low since it is an operative procedure and the outcomes are not likely to be influenced by lack of blinding of participants and personnel
Blinding of outcome assessment (detection bias) Objective outcomes	Low risk	The trial did not report blinding, however blinding is not possible due to the nature of the intervention, but we judge the risk of bias as low regarding the following outcomes due to being objective and are not likely to be influenced by lack of blinding: prosthodontic maintenance time or costs
Blinding of outcome assessment (detection bias) Patient‐reported outcomes	High risk	The trial did not report blinding, however blinding is not possible due to the nature of the intervention, but we judge the risk of bias as high regarding patient satisfaction due to being subjective and is likely to be influenced by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	High risk	Discrepancies exist between the 3 trial reports regarding the number of participants available for follow‐up at 3 years and the reasons for dropouts MacEntee 2005: "The results reported here are for the first 68 participants for whom data were available for 3 years" "2 died, 5 withdrew, 6 were otherwise lost to follow‐up, and 19 participants had less than 3 years in the trial since receiving new dentures" Walton 2003: "87 participants were available for follow‐up after 3 years" "Of the 13 participants who dropped out of the trial, 2 did so before overdenture placement (1 died and 1 was lost to follow‐up after stage‐one implant surgery). 1 participant in the bar‐clip group died after overdenture delivery. The remaining 10 (3 in the ball attachment group and 7 in the bar‐clip group) left the trial at the end of their 2 year commitment"
Selective reporting (reporting bias)	Low risk	The trial protocol is not available, but we judge the risk of bias as low due to the reporting of key outcomes that are expected to be reported for such a trial
Other bias	Low risk	None noted

F = female; IODs = implant‐retained overdentures; M = male; n = number; VAS = visual analogue scale.