A Randomized, Controlled Trial of Tai Chi for the Prevention of Falls: The Central Sydney Tai Chi Trial

Alexander Voukelatos, MA (Psychol); Robert G. Cumming, PhD; Stephen R. Lord, DSc; Chris Rissel, PhD


J Am Geriatr Soc. 2007;55(8):1185-1191. 

In This Article


This study was a randomized, controlled trial. The intervention was a 16-week community-based tai chi program. Participants randomized to the control group were placed on a 24-week waiting list. Fall outcomes were analyzed after 16 and 24 weeks. Balance was assessed at baseline and 16 weeks.

Recruitment of Study Participants

Recruitment was staggered across 10 periods between June 2001 and March 2003, with each period targeting a different geographical region in central and southeastern Sydney. Recruitment during each period involved advertisements placed in local community newspapers.

Participants were eligible for the trial if they were aged 60 and older, were living in the community, and had not practiced tai chi in the previous 12 months. Participants were excluded if they had a degenerative neurological condition such as Parkinson's disease, dementia, a severely debilitating stroke, severe arthritis, or marked vision impairment or if they were unable to walk across a room unaided.

Baseline assessments were conducted 2 weeks before the beginning of the tai chi program. Each participant was interviewed using a structured questionnaire to assess sociodemographic variables, falls history, adequate physical activity (defined as ≥30 minutes of moderate physical activity on ≥5 days a week),[7] health-related quality of life,[20] independent activities of daily living,[21] and falls efficacy.[22] Balance was assessed as described below.

A randomization list, whose details remained unknown to the research assistants, was prepared for each venue using randomly permuted blocks of four or six. After the baseline assessments had been conducted, the research assistants randomized participants into the intervention tai chi program or the waiting list control group. The Central Sydney Area Health Service Human Ethics Committee gave approval for this study, and informed consent was obtained from all subjects before their participation.

Tai Chi Intervention

The tai chi intervention consisted of a 16-week program of weekly 1-hour tai chi classes. All classes were community based and operated as normal, without any modification for this research project. Classes were restricted in size to between eight and 15 participants. tai chi instructors were recruited from among those running classes in the local community and were allocated to classes according to their availability. To be eligible to instruct a study tai chi program, instructors had to have at least 5 years experience teaching tai chi or have completed an accredited tai chi trainers' course, as well as having previously taught tai chi or other gentle exercise programs to older people. No restriction was made on the tai chi style taught by the instructors.

Twenty-two tai chi instructors, who conducted a total of 38 tai chi programs for the study, conducted classes at 24 community venues. The majority of classes involved Sun-style tai chi (83%), two classes involved Yang-style tai chi (3%), and the remainder involved a mixture of several styles (14%). Participants were not given any particular instructions about doing tai chi outside class time. Those who missed classes did not have extra catch-up classes, but instructors typically reviewed previous lessons at the start of each new class.

Participants were asked to pay AU$44 for the 16-week tai chi program. It was hoped that making this financial commitment would increase the likelihood of attendance at classes.

Control participants were instructed not to do any tai chi elsewhere during the 24 week study period. At the end of the study period, control participants were offered a 16-week tai chi program.

Falls Surveillance

Falls were defined as "unintentionally coming to rest on ground, floor, or other lower level."[23] Falls were monitored for 24 weeks. Participants were given a falls calendars and were instructed to record on the calendar each day for 24 weeks whether they had a fall. At the end of each month, participants were required to mail the prepaid-postage calendars to the study center. Participants who had not returned calendars within 2 weeks of the end of each month were contacted to establish their falls status.

Balance Measures

Four trained research assistants, blind to intervention status, administered six balance tests at baseline and 16 weeks later, at the conclusion of the tai chi program. These tests of balance have been used in a wide range of falls research studies and have good validity and test–retest reliability.[24,25,26]

The balance tests have been described in detail elsewhere.[24,25,26,27,28,29] Sway was measured using a swaymeter that measured displacement of the body at the level of the waist.[25,26] Two conditions were tested: participants standing on the floor and then standing on a foam rubber mat (40 cm × 40 cm × 15 cm thick). Leaning balance was measured using maximal balance range and coordinated stability tests—balance tests that require subjects to adjust their balance in a controlled manner when near the limits of their base of support.[25,27] Lateral stability was assessed by measuring maximal lateral sway with feet placed in a near-tandem position with eyes open.[25,28] Choice stepping reaction time was determined by assessing subjects' ability to step as quickly as possible onto one of four rectangular panels that were illuminated in a random order.[25,29] The average time for 20 steps was used in the analysis.

Statistical Power and Data Analysis

Previous studies of tai chi programs in the community have found relative reductions in falls and fallers of approximately 40%.[8,17] The total period for which participants were under observation for falls in the current study was 24 weeks (~6 months). Although most studies describe the incidence of falls over a 12-month period, one study found that 24% of older people will fall in a 6-month period.[30] The current study was designed to detect a 40% relative reduction (from 24% to 15%) in the proportion of people who have at least one fall during follow-up. A sample size of 284 per group (α=.05, power=80%) was necessary. With an estimated dropout rate of 25%, a final sample size of 379 per group was required.

The primary study outcome was falls during 16 weeks of follow-up. Secondary outcomes were falls during 24 weeks of follow-up and changes in balance during 16 weeks of follow-up. Data were analysed on an intention-to-treat basis. Analyses were done using STATA 8.2 statistical software (Stata Corp., College Station, TX).

Negative binomial regression models were used to calculate incidence rate ratios (IRRs) comparing falls rates in intervention and control groups during 16 and 24 weeks of follow-up.[31] The risk of falling was analysed using Cox regression models for one or more falls and the Andersen-Gill extension of the Cox model for two or more falls. The negative binomial and Cox models were repeated with adjustment for age, sex, venue, falls history, baseline sway on the foam mat, and adequacy of physical activity. The last two variables were the only baseline characteristics that differed appreciably between study groups. History of falls was included in multivariable models, because this is an established risk factor for future falls. Cumulative incidence ratios (relative risks) and chi-square tests were used to compare the proportions of fallers (participants who had ≥1 falls) and multiple fallers (participants who had ≥2 falls) in each group after 16 and 24 weeks.

Changes in balance over 16 weeks were assessed using forced-entry multiple linear regression analysis, with experimental group, age, sex, venue, falls history, and baseline adequacy of physical activity included as independent variables in the models.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.