Li 2012.

Study characteristics
Methods	RCT
Participants	Setting: facility, USA N = 195 Sample: recruited from four Oregon cities (Eugene, Corvallis, Salem, and Portland) by means of newspaper advertisements, referrals from neurologists or physical therapists, and information distributed to local Parkinson’s disease support groups (37% women) Age (years): Mean (SD) Tai Chi intervention group 68 (9), strength training intervention group 69 (8), control group 69 (9) Inclusion criteria: diagnosis of PD; Hoehn and Yahr stage 1 to 4; age of 40 to 85 years; at least one score of 2 or more for at least one limb for the tremor, rigidity, postural stability, or bradykinesia items in the motor section of the Unified Parkinson’s Disease Rating Scale; stable medication use; ability to stand unaided and walk with or without an assistive device; medical clearance for participation; willingness to be assigned to any of the three interventions Exclusion criteria: current participation in any other behavioral or pharmacologic study or instructor‐led exercise program; cognitive impairment (Mini–Mental State examination score <24); debilitating conditions or vision impairment that would impede full participation in the study; unavailability during the study period. Disease severity at baseline: HY stage 1 to 4; UPDRS motor score mean (SD) 15.2 (5.9)
Interventions	Exercise 1. Exercise: Tai Chi Group supervised by a Tai Chi instructor (60 min, 2x/week for 24 weeks) 2. Exercise: strength training (lower limb). Group supervised by an instructor (60 min, 2x/week for 24 weeks) 3. Control: Stretching. Group supervised by an instructor (60 min, 2x/week for 24 weeks)
Outcomes	1. Rate of falls 2. Number of fallers 3. Number and type of adverse events 4. Quality of life (PDQ‐8) (Li 2014) Economic analysis reported in Li 2015: 1. Cost of delivering the intervention 2. Cost of health service use 3. Incremental cost per QALY gained 4. Incremental cost per fall prevented Other outcomes reported but not included in this review
Duration of the study	9 months
Funding source	National Institiute of Neurological Disorders and Stroke
Notes	Fall data collected: during the 6‐month intervention period and at the 3‐month follow‐up period by monthly falls diaries Economic analysis reported in US dollar (price year 2011)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of generating the randomisation list not described. Quote: "...randomly assigned to one of the interventions, in a ratio of 1:1:1, without stratification, with the use of permuted‐block randomization."
Allocation concealment (selection bias)	Low risk	Allocation concealment was described by central allocation. Quote from protocol: "Concealment of allocation will be implemented. The randomization schedule, generated by the project data analyst, will be kept by a project staff who will deliver it, in a sealed envelope, to a research assistant who will then assign qualified individuals to intervention groups.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and intervention (exercise) delivery personnel not blinded to group allocation, but were unaware of which was the control group. Quote: “To reduce potential expectation bias, participants will be informed that the study will be comparing three different exercises and that they will be assigned to an exercise group at random.” And “Because of the behavioral trial, blinding instructors will not be possible. However, the instructors will not be provided with any information related to the objectives of the study..."
Blinding of outcome assessment (detection bias) Falls and fallers	Unclear risk	Personnel collecting fall information not blinded to group allocation, but were unaware of which was the control group. Quotes: “Because of the behavioral trial, blinding instructors will not be possible. However, the instructors will not be provided with any information related to the objectives of the study, nor will they participate in any outcome assessments.”
Blinding of outcome assessment (detection bias) Fractures	Unclear risk	Insufficient information to permit judgement. No fractures reported and unclear how data regarding fractures was collected.
Incomplete outcome data (attrition bias) Falls	Unclear risk	Data not available to assess.
Incomplete outcome data (attrition bias) Fallers	Unclear risk	Data not available to assess.
Selective reporting (reporting bias)	Low risk	The study protocol is available (NCT00611481) and all of the study’s pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way.
Method of ascertaining falls (recall bias) Falls and fallers	Low risk	The study used concurrent collection of data about falling with monthly, or more frequent, follow‐up by the researchers. Quote: ”Falls were monitored by means of daily “fall calendars” that were maintained by the study participants and collected monthly throughout the intervention or until a participant withdrew from the study.”