McClanahan 1997.

Methods	Trial design: parallel (3 arms) Location: Indianapolis, USA (type of setting not reported) Number of centres: 1 Recruitment period: not reported Funding source: not reported but some authors associated with Procter and Gamble (the manufacturer of the stannous fluoride toothpaste)
Participants	Inclusion criteria: healthy adults; minimum 5 gingival bleeding sites; minimum 16 natural teeth (including 4 molars) Exclusion criteria: "rampant" caries; advanced periodontal disease; chronic dental neglect; serious medical condition Baseline plaque: (Quigley‐Hein Plaque Index) Gp A: mean 1.88 (SE 0.04); Gp B: mean 1.9 (SE 0.04) Baseline gingivitis: (Löe‐Silness Gingival Index) Gp A: mean 0.7 (SE 0.02); Gp B: mean 0.71 (SE 0.02); (gingival bleeding on probing or spontaneously ‐ number of sites) Gp A: mean 15.46 (SE 0.92); Gp B: mean 16.4 (SE 1.03) Baseline caries: not reported Age at baseline (years): Gp A: mean 35.5 (range 19‐71); Gp B: mean 36.5 (range 19‐70) Gender: Gp A: male 52 (34%), female 103 (66%); Gp B: male 60 (34%), female 114 (66%) Any other details of important prognostic factors: baseline staining (Meckel Stain Score): Gp A: mean 1.16 (SE 0.18); Gp B: mean 1.14 (SE 0.19) Number randomised: 378 (Gp A: 187; Gp B: 191) Number evaluated: 329 (Gp A: 155; Gp B: 174)
Interventions	Comparison: stannous fluoride* versus triclosan/copolymer/sodium fluoride versus sodium fluoride *We excluded this arm from our data extraction, risk of bias assessment and analyses Gp A (n = 187): twice daily brushing for 1 minute with toothpaste containing 0.3% triclosan, 2% copolymer, 0.243% sodium fluoride; all participants received thorough baseline oral prophylaxis Gp B (n = 191): as above but without triclosan and copolymer Duration of treatment: 6 months
Outcomes	Plaque (Quigley‐Hein Plaque Index), gingivitis (Löe‐Silness Gingival Index and Gingival bleeding on probing or spontaneously), adverse effects (Meckel Stain Scores and oral soft tissue status); assessed at 3 and 6 months' follow‐up
Notes	Sample size calculation: not reported Adverse effects: none observed; staining was reported as continuous data but no adverse events were reported as such
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "...randomly assigned...Subjects were separated by gender and by intervals of initial gingivitis scores. Within strata, subjects were assigned to treatment groups by random permutations of five" Comment: sufficient description of the method of sequence generation
Allocation concealment (selection bias)	Unclear risk	Quote: "...randomly assigned...Subjects were separated by gender and by intervals of initial gingivitis scores. Within strata, subjects were assigned to treatment groups by random permutations of five" Comment: unclear whether remote/central randomisation and no variation of block size
Blinding of participants (performance bias) All outcomes	Low risk	Quote: "Each subject received four 4.6 ounce uniquely labelled plain white tubes containing one of the following dentifrices...The study was conducted in a double‐blind fashion so neither the examiners nor subjects knew the identity of the dentifrices throughout the course of the trial" Comment: use of an identical control toothpaste meant that participants did not know which group they were assigned to
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Each subject received four 4.6 ounce uniquely labelled plain white tubes containing one of the following dentifrices...The study was conducted in a double‐blind fashion so neither the examiners nor subjects knew the identity of the dentifrices throughout the course of the trial" Comment: the examiner did not know which group the participants they were assessing had been assigned to
Incomplete outcome data (attrition bias) All outcomes	High risk	13% of randomised participants were not included in the final analysis (Gp A: 17%; Gp B: 9%). Reasons for attrition were not given but the rate was much higher in the triclosan/copolymer group than the control group. Also, if the missing participants had higher mean plaque/gingivitis scores in one group than the other, as the attrition rate increased, so would over/understatement of the mean difference
Selective reporting (reporting bias)	Low risk	Appropriate outcome measures were considered and reported in full, as described in the methods section
Other bias	Unclear risk	No mention of calibration of outcome assessor so it is unclear whether or not there was a risk of differential diagnostic activity