Horwitz 2005

Methods	Randomised, parallel group study. Patients and outcome assessors blind. 3 withdrawals at 1 year (2 in the AmF/SnF2; 1 in the CHX group), none at 3 months.
Participants	Adults. Age 34 to 79 years. One‐stage implants. 33 enrolled and results given for 33 at 3 months, 30 at 1 year.
Interventions	2 groups. Amine fluoride/stannous fluoride mouthwash (10 ml) twice per day for 60 seconds versus chlorhexidine mouthwash following toothbrushing for 3 months. Recall first and third month. Study duration: 12 months (data at 3 months).
Outcomes	Implant failure, bone width, bone to implant distance, radiographic bone height (3 and 12 months), staining index, patient compliance. Patient subjective evaluation questionnaire (visual analog scale): taste, change in taste, desire for future use, overall satisfaction at 3 months. 3‐month data used.
Notes	36 packages were randomised, but 33 patients enrolled. Mouthrinses used for 3 months after 1‐stage surgery. No indication on plaque and mucosal inflammation parameters. Radiographic data and bone data at implant level.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Computer randomisation was done before commencement of the study by assigning either CHX or AmF/SnF2 to unmarked packages numbered 1 to 36".
Allocation concealment (selection bias)	Low risk	"De‐coding of the mouthwashes was performed only after the completion of all data acquisition and termination of the study".
Blinding (performance bias and detection bias) All outcomes	Low risk	"De‐coding of the mouthwashes was performed only after the completion of all data acquisition and termination of the study".
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss at follow‐up at 3 months when date were taken.
Selective reporting (reporting bias)	Unclear risk	Important variables related to plaque or mucosal inflammation parameters are missing.
Other bias	Unclear risk	This study was partially supported by a grant from GABA International, Ltd, Munchestein, Switzerland. Bone data and data on radiographic bone level given at implant level. Written but not possible to have data at patient level.
Blinding of the patient?	Low risk	"Computer randomisation was done before commencement of the study by assigning either CHX or AmF/SnF2 to unmarked packages numbered 1 to 36".