Bader 2013.

Methods	Trial design: parallel (2 arms) Location: clinical centres of university dental schools, USA Number of centres: 3 Recruitment period: April 2007 to September 2008
Participants	Inclusion criteria: aged 21 to 80 years; minimum of one coronal or root surface cavitated caries lesion; minimum of 12 teeth with exposed coronal or root surfaces; ability to read and understand study materials in English (the caries criteria were meant to ensure the inclusion of participants with an elevated risk of experiencing caries) Exclusion criteria: more than 10 teeth with caries lesions; type IV periodontitis (pocket depths or attachment loss greater than 6 mm); long‐term antibiotic use; requiring antibiotic treatment before dental treatment; history of head and neck radiotherapy; history of adverse reaction to either the active or placebo ingredients; serious illness that would interfere with participation; plans to leave the area within the following 3 years; no telephone; co‐inhabiting with someone already enrolled in the study Baseline caries: (D₂FS) Gp A: mean 18.8 (SD 12.8); Gp B: mean 18.5 (SD 12.5) Age at baseline (years): Gp A: mean 46.3 (SD 13.5); Gp B: mean 47.7 (SD 13.7) Gender: Gp A: 62.2% female; Gp B: 67% female Any other details of important prognostic factors: fluoride exposure (toothpaste and professionally applied topical fluoride) was similar in both groups; all 3 study areas had fluoridated water Number randomised: 691 (Gp A: 344; Gp B: 347) Number evaluated: 669 (Gp A: 331; Gp B: 338) ‐ numbers including data from all 3 centres (see comment in risk of bias table below under 'Other bias')
Interventions	Comparison: xylitol lozenges versus placebo lozenges Gp A (n = 344): one peppermint flavoured lozenge (1 g xylitol) dissolved in the mouth five times per day (total dose = 5 g xylitol per day) Gp B (n = 347): as above but containing sucralose (considered to be inert, i.e. neither causes nor prevents caries) instead of xylitol Duration of treatment: 33 months
Outcomes	Caries: cumulative decayed or filled surfaces (D₂FS) increment (root and coronal surfaces combined) from baseline through the three follow‐up examinations (expressed as annualised increment). Assessed at 12, 24 and 33 months Adverse effects
Notes	Sample size calculation: "80% power to detect a 20% reduction in the D₂FS increment assuming a two‐tailed test with a type I error rate of 5%. The target sample size of 750 allowed a 10% attrition rate per year" Adverse effects: "all adverse effects...similar for the two study groups." Data not reported in a usable format Funding source: grants from the National Institute of Dental and Craniofacial Research (NIDCR) Declarations/conflicts of interest: none of the authors reported any disclosures The information and quotes used in this table and the risk of bias table below are from both the papers listed under Bader 2013 in the reference section (one paper is dedicated to design/methods)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization was carried out using a web‐based randomization application process. Allocation assignments were stratified by site and age group (≥ 50, < 50 yrs.) in permuted blocks of varying sizes within each stratum" Comment: this is an appropriate method of random sequence generation
Allocation concealment (selection bias)	Low risk	Quote: "permuted blocks of varying sizes within each stratum" Comment: the use of varying sizes of random permuted blocks implies that a reasonable attempt was made to prevent those admitting participants from knowing upcoming assignments. We feel that this was probably done properly in this study
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The placebo lozenge was identical in size and color to the active lozenge...Both the active and placebo lozenges were peppermint flavored" and "Staff and participants were blinded to treatment assignment" Comment: participants and personnel would not know to which group a participant was assigned
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The placebo lozenge was identical in size and color to the active lozenge...Both the active and placebo lozenges were peppermint flavored" and "Staff and participants were blinded to treatment assignment" Comment: participants and personnel would not know to which group a participant was assigned
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 3% of randomised participants were not included in the final analysis (Gp A: 4%; Gp B: 3%) Comment: we do not believe that any of the above could pose a risk of bias significant enough to have led to a distortion of the true intervention effect
Selective reporting (reporting bias)	Low risk	All raw data is publicly available on the study website (www.xactstudy.org). The authors also provided us with the mean and SD for the 33 month increment
Other bias	Low risk	The study authors report that there was a problem with adherence data at one of the three centres arising due to protocol violation (despite the exclusion criterion, 10 participants who were co‐inhabiting with another participant had been randomly assigned). However, this was transparently reported and the authors reported results both including and excluding the problematic centre's data. The results including this centre were analysed per protocol (therefore five participants from each group were excluded from the analyses for the same reason) and were virtually identical to those excluding the centre. There does not appear to be any risk of bias "To the extent possible, the same examiner who conducted the baseline examination performed all follow‐up examinations" and "Primary and back‐up examiners and recorders from all three clinical centers participated in a four‐day training and calibration session with a reference‐standard examiner, as well as refresher sessions before the 12‐, 24‐ and 33‐month examinations." We consider that the risk of differential diagnostic activity was low