Li 2005.

Methods	Study design: RCT Number of study arms: 2 Length of follow‐up: 6 months
Participants	Setting: Legacy Health System, Portland, Oregon, USA Number of participants: 256 Number analysed: 188 Number lost to follow‐up: 68 Sample: people enrolled in HMO Age (years): mean 77.5 (SD 5), range 70 to 92 Sex: 70% female Inclusion criteria: age ≥ 70; physician clearance to participate; inactive (no moderate to strenuous activity in last 3 months); walks independently Exclusion criteria: chronic medical problems that would limit participation; cognitive impairment
Interventions	1. Group‐based Tai Chi: 1 hour, 3 a week for 26 weeks 2. Control group: low‐level stretching 1 hour, 3 a week for 26 weeks
Outcomes	1. Rate of falls 2. Number of people who experienced 1 or more falls (risk of falling) 3. Number of people who experienced 1 or more falls requiring medical attention
Duration of the study	52 weeks
Adherence	Adherence measured as class attendance 1. Group‐based Tai Chi group: median compliance; 61 sessions (range 30 ‐ 77). 92 (80%) attended 50+ sessions 2. Control group: median compliance; 61 sessions (range 35 ‐ 78). 87 (81%) attended 50+ sessions
Notes	Source of funding: National Institutes of Health, National Institute on Aging Economic information: not reported 6‐month fall data used as total over 12‐month period not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random numbers
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participants and personnel not blind to allocated group but impact of non‐blinding unclear
Blinding of outcome assessment (detection bias) Falls	Low risk	Falls reported by participants who were aware of their group allocation, using the same method on both groups. Fall diaries coded by blinded research assistant
Blinding of outcome assessment (detection bias) Fractures	Unclear risk	Not applicable
Blinding of outcome assessment (detection bias) Hospital admission, medical attention and adverse events	High risk	The only evidence for requiring medical attention was from self‐reports from participants
Blinding of outcome assessment (detection bias) Health related quality of life (self report)	Unclear risk	Not applicable
Incomplete outcome data (attrition bias) Falls and fallers	High risk	More than 20% of fall outcome data are missing (27%)
Selective reporting (reporting bias)	Unclear risk	Minimum set of expected outcomes was not reported (adverse events were not reported)
Method of ascertaining falls (recall bias)	Low risk	Prospective. Falls recorded on daily fall calendars, collected on a monthly basis