Reinsch 1992.

Methods	RCT (cluster‐randomised by senior centre. 2 x 2 factorial design) Study design: Cluster‐RCT Number of study arms: 2 Number of clusters: 16 Length of follow‐up: 12 months
Participants	Setting: Los Angeles County and Orange County, California, USA Number of participants: 230 Number analysed: 230 Number lost to follow‐up: 0 Sample: recruited from 16 senior centres Age (years): mean 74.2 (SD 6.0) Sex: 80% female Inclusion criteria: aged > 60 Exclusion criteria: none listed
Interventions	Randomised into 4 groups: 3 intervention groups (1 group‐based balance and strength training, 1 cognitive‐behavioural training, 1 exercise and cognitive training) and 1 control group (discussion group). Only the group‐based balance and strength training and control group were included in this review 1. Group‐based balance and strength training: no special equipment used; 1 hour, 3 a week for 52 weeks 2. Control group: health and interest discussion group, 1 hour, 1 a week for 52 weeks
Outcomes	1. Number of people who experienced 1 or more falls (risk of falling)
Duration of the study	52 weeks
Adherence	Dropout/noncompliance defined as missing ⅓ or more of the classes taught at their centre 1. Group‐based balance and strength training: 13/57 noncompliance (44/57 compliance) 2. Control group: 8/50 noncompliance (42/50 compliance)
Notes	Source of funding: NIH, AARP Andrus Foundation, Roosevelt Warm Springs Foundation Economic information: not reported MacRae paper includes a subset of results for only 2 arms of the study, in Los Angeles county only Number of clusters allocated to intervention: 4; number of clusters allocated to control: 4; number of clusters analysed (intervention): 4; number of clusters analysed (control): 4
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "randomly assigned to treatments"
Allocation concealment (selection bias)	High risk	Quote:" A biostatistician not involved in the study randomized general practices into the intervention or control group by using computer‐generated random numbers. After the randomization, the general practitioners enrolled patients for the study according to the inclusion and exclusion criteria". The method of concealment is not described and assume the recruiting general practitioners were unblinded and may have had knowledge of participant characteristics
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participants and personnel implementing the intervention not blinded to allocated group, but impact of non‐blinding unclear
Blinding of outcome assessment (detection bias) Falls	Unclear risk	Falls reported by participants who were aware of their group allocation. Blinding of research assistant not described
Blinding of outcome assessment (detection bias) Fractures	Unclear risk	Not applicable
Blinding of outcome assessment (detection bias) Hospital admission, medical attention and adverse events	Unclear risk	Method for recording medical attention and adverse events was unclear. Appears to be self‐report
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	Low risk	No missing fall data
Selective reporting (reporting bias)	High risk	Falls were measured, but number of falls was not reported
Method of ascertaining falls (recall bias)	Low risk	Prospective. Monthly diaries plus weekly phone calls or visits
Cluster‐randomised trials	Unclear risk	Individual participant recruitment was undertaken after group allocation. The method of concealment is not described and it is likely that recruitment was undertaken by a person who was unblinded and may have had knowledge of participant characteristics; baseline characteristics of clusters were not reported; missing outcomes for clusters or within clusters were not reported; did not account for the clustered design in the analysis; results comparable with individually‐randomised trials