Edited by: Catherine Kerr, Brown University, USA
Reviewed by: Guido P. H. Band, Leiden University, Netherlands; Teresa Dianne Hawkes, Air Force Research Laboratory, USA
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Tai Chi (TC) exercise improves balance and reduces falls in older, health-impaired adults. TC’s impact on dual task (DT) gait parameters predictive of falls, especially in healthy active older adults, however, is unknown.
To compare differences in usual and DT gait between long-term TC-expert practitioners and age-/gender-matched TC-naïve adults, and to determine the effects of short-term TC training on gait in healthy, non-sedentary older adults.
A cross-sectional study compared gait in healthy TC-naïve and TC-expert (24.5 ± 12 years experience) older adults. TC-naïve adults then completed a 6-month, two-arm, wait-list randomized clinical trial of TC training. Gait speed and stride time variability (Coefficient of Variation %) were assessed during 90 s trials of undisturbed and cognitive DT (serial subtractions) conditions.
During DT, gait speed decreased (
In healthy active older adults, positive effects of short- and long-term TC were observed only under cognitively challenging DT conditions and only for stride time variability. DT stride time variability offers a potentially sensitive metric for monitoring TC’s impact on fall risk with healthy older adults.
The ability to walk while simultaneously performing a secondary cognitive task – commonly referred to as a dual task (DT) – is essential to many activities of daily living such as successful ambulation while navigating complex environs and conversing with others. Increasing evidence from clinical practice, epidemiological studies, and clinical trials show that postural control, gait health, and cognition are interrelated in older adults (Montero-Odasso et al.,
Growing appreciation of the interdependence of cognitive and postural control processes has led to search for multimodal interventions combining motor and cognitive training for improving gait and preventing falls (Mirelman et al.,
The current study evaluates the impact of both long- and short-term TC training on gait speed and stride time variability during both undisturbed (single task) walking and walking with a cognitive DT challenge. Long-term training effects were assessed through observational comparisons of TC naïve healthy older adults and an age-matched sample of expert TC practitioners. Short-term effects of TC training were assessed by random assignment of the TC naïve healthy adults to either 6 months of TC plus usual care or usual care alone. Based on research to date, we predicted that (1) TC experts would exhibit greater walking speed and reduced variability, compared to controls, and that group differences would be greater under DT challenges; (2) TC-naïve older adults randomly assigned to 6 months of TC would subsequently exhibit greater walking speed and reduced stride time variability; and (3) improvements in walking speed and reduced stride time variability observed over 6 months would be greater in those randomized to TC compared to a usual care control, with between-group differences being greater under DT challenges.
We employed a hybrid study design that included a two-arm randomized clinical trial (RCT) along with an additional observational comparison group. The Institutional Review Board at Beth Israel Deaconess Medical Center approved this study. The RCT component of this study was registered at clinicaltrials.gov (NCT01340365). Gait outcomes reported herein are a subset of a larger battery of assessed outcomes including balance, cardiovascular, and cognitive outcomes. These latter outcomes are reported independently (Wayne et al.,
Sixty healthy older adults, aged 50–79, were randomized 1:1 to receive 6 months of TC training in addition to usual health care, or to usual health care alone (control group). Study participants randomized to usual care were offered a 3-month course of TC as a courtesy following the trial. Randomization was stratified by age (50–59, 60–69, 70–79 years) and utilized a permuted-blocks randomization scheme with randomly varying block sizes. Randomization was performed by the study statistician. All outcomes were assessed at baseline and 6 months, i.e., after completing 6 months of training. The primary staff overseeing assessment and analyses of the gait-related outcomes was blinded to treatment assignment. Recruitment spanned from March 2011 to March 2013. All follow-up procedures were completed by September 2013. Analysis was performed in 2014. Further details related to the design of the RCT component of this study are reported elsewhere (Wayne et al.,
Inclusion criteria were (1) age 50–79 years; (2) living within the Greater Boston area; and (3) willing to adhere to a 6-month TC training protocol. Exclusion criteria were (1) chronic medical condition including cardiovascular disease, stroke, active cancer; neurological conditions; or significant neuromuscular or musculoskeletal conditions requiring chronic use of pain medication; (2) acute medical condition requiring hospitalization within the past 6 months; (3) self-reported inability to walk continuously for 15 min unassisted; (4) regular TC practice within past 5 years; and (5) regular participation in physical exercise on average 4 or more times per week. Interested individuals underwent both an initial phone screen and an in-person screen at the BIDMC Clinical Research Center. Eligible individuals provided written informed consent and underwent baseline testing prior to randomization.
Participants within both groups were encouraged to follow usual health care as prescribed by their primary care physicians. Participants in the TC group received 6 months of TC training in addition to usual care. All TC interventions were administered pragmatically at one of five pre-screened TC schools within the Greater Boston area that met specific guidelines described elsewhere (Wayne et al.,
Twenty-seven healthy older adults (age 50–79 years) currently engaged in an active TC training regimen, each with over 5 years of TC practice, were recruited for a single observational visit. No limitation was set on TC style. Eligibility and screening procedures for TC experts were identical to those for healthy adults enrolled in the RCT.
Participants reported to the Beth Israel Deaconess Medical Center’s Clinical Research Center (Boston, MA, USA) where they underwent in-person screening procedures. Participants with a mini mental state examination (MMSE) score ≥24 and no abnormal findings on an ECG were eligible to participate in the study. All outcome measurements were assessed at the Syncope and Falls in the Elderly (SAFE) laboratory at Beth Israel Deaconess Medical Center. Outcomes related to gait reported here were part of a larger battery of tests that lasted an average of 3.5 h. Changes in DT gait speed and gait variability from baseline to 6 months were
Gait was assessed along a 75-m long unobstructed hallway. Subjects were instructed to walk at their normal preferred walking pace and make wide turns at the ends of the hallway. Two 90 s walking trials were completed: undisturbed with no superimposed cognitive task (i.e., single task walking) and walking while completing a cognitive DT. The cognitive DT consisted of a serial subtraction exercise counting backwards by threes beginning at 500. Ultrathin, force-sensitive resistor footswitches were placed on each subject’s heel and toe to capture the temporal parameters of gait. Data were collected wirelessly at 1500 Hz with DTS Data Acquisition Software (Noraxon, Scottsdale, AZ, USA). The data for each trial was exported and analyzed in Matlab (Mathworks, Natick, MA, USA) to determine initial and final foot contact times for each stride. Stride times were calculated for each gait cycle as the time between initial heel strike of one foot and the subsequent heel strike of that same foot. Average gait speed was measured using the total distance walked during the trial. Stride time variability was calculated from the stride time time-series as the coefficient of variation (CV, 100 multiplied by the SD of the stride times divided by the mean of each subject’s stride times). To calculate these measures, the first five strides were removed to minimize potential gait initiation effects. A median filter was applied to the stride time time-series to remove large outliers, typically a result of turns at the ends of the hallway. This ensured that the steady state variability was analyzed for each time series. Average stride time and stride time variability was determined from the resultant time series.
Dual task costs were calculated using both absolute and proportional measures. Absolute DT costs (Abs. DT) were calculated for each participant as the difference in walking speed or stride time variability between undisturbed single task walking (ST) and DT walking. Proportional DT costs (% DT) were calculated for each participant’s walking speed or stride variability as: 100 × ((DT − ST)/ST). Both the number and the accuracy of serial subtractions during DTs were recorded.
Global cognitive function at baseline was assessed with the MMSE. Executive cognitive function was assessed using the trail making test (TMT B and TMT B-A) (Bowie and Harvey,
Cross-sectional measures of gait outcomes in TC experts and TC naïves were compared in a linear model controlling for age, gender, BMI, and physical activity and assuming equal variance across groups. For measures where the assumption of equal variance may have been violated, the larger of the two groups, the TC naïves, had greater estimated variance, leading to potential overestimates of pooled variance and conservative estimates of effect sizes and
For cross-sectional comparisons, we estimated that a sample size of 27 TC Expert and 60 TC naïve subjects would provide power to detect an effect size of 0.63. For the randomized trial, we estimated that the sample of 60 participants randomized 1:1, the study would have 80% power to detect a main effect of treatment if the true effect size was at least 0.74 based on a two-tailed test at
Demographic characteristics of the TC experts (
Randomized groups |
Observational group |
||
---|---|---|---|
Usual care ( |
Tai Chi ( |
Tai Chi experts ( |
|
AVG ± SD | 64.45 ± 7.42 | 63.94 ± 8.02 | 62.78 ± 7.57 |
Male | 11 (37.9%) | 9 (29%) | 13 (48.1%) |
Female | 18 (62.1%) | 22 (71%) | 14 (51.9%) |
White | 26 (89.7%) | 29 (93.5%) | 22 (81.5%) |
African-American | 3 (10.3%) | 0 (0%) | 1 (3.7%) |
Asian | 0 (0%) | 2 (6.5%) | 4 (14.8%) |
Non-Hispanic/Non-Latino | 29 (100%) | 30 (96.8%) | 26 (96.3%) |
Hispanic/Latino | 0 (0%) | 1 (3.2%) | 1 (3.7%) |
AVG ± SD | 16.19 ± 3.03 | 17.13 ± 3.41 | 18.44 ± 3.34 |
AVG ± SD | 29.21 ± 0.82 | 29.03 ± 1.17 | 29.07 ± 1.11 |
AVG ± SD | 59.93 ± 20.84 | 59.69 ± 22.03 | 53.07 ± 22.4 |
AVG ± SD | 29.54 ± 18.58 | 30.26 ± 20.01 | 28.09 ± 19.65 |
AVG ± SD | 26.54 ± 5.83 | 26.38 ± 5.19 | 23.54 ± 2.35 |
AVG ± SD | 4.0 ± 2.0 | 4.0 ± 2.0 | 6.0 ± 2.0 |
Tai Chi naïve adults randomized to TC plus usual care or usual care alone were comparable at baseline (see Table
A CONSORT flowchart detailing study recruitment, randomization, and retention for the randomized trial component of the study is shown in Figure
Adherence to the TC protocol was variable. Two participants in the TC group formally withdrew participation due to time commitments and an unrelated injury. Of the remaining 29 participants in the TC group, 21 (72%) were per-protocol – defined as attending 70% of classes and completing 70% of required home practice over the entire course of the trial (mean and median TC exposure hours were 59.9 and 62.4 h, respectively).
For those randomized to TC, 13 subjects were trained at a TC school teaching Yang style and the remaining 18 subjects were trained at a school teaching Wu style.
Compared to undisturbed walking, gait speed decreased during DT walking in both TC expert (
Outcome measure | Tai Chi expert ( |
Tai Chi naïve ( |
Between groups |
||
---|---|---|---|---|---|
Mean (95% CI) | Mean (95% CI) | Effect size | Mean difference (95% CI) | ||
Gait speed ST (m/s) | 1.12 (1.1, 1.2) | 1.12 (1.1, 1.2) | 0.008 | −0.001 (−0.08, 0.07) | 0.97 |
Gait speed DT (m/s) | 1.01 (1.0, 1.1) |
0.97 (0.9, 1.1)* | 0.28 | 0.041 (−0.03, 0.1) | 0.26 |
Stride time variability ST (CV %) | 1.75 (1.5, 2.0) |
1.90 (1.8, 2.0)** | 0.29 | −0.15 (−0.4, 0.1) | 0.25 |
Stride time variability DT (CV %) | 2.11 (1.8, 2.4) | 2.55 (2.3, 2.8) | 0.57 | −0.44 (−0.8, −0.5) | 0.027 |
Abs. DT cost speed | −0.11 (−0.2, −0.1) | −0.15 (−0.2, −0.1) | 0.34 | 0.042 (−0.02, 0.1) | 0.18 |
% DT cost speed | −9.50 (−13.9, −5.1) | −13.24 (−16.1, −10.4) | 0.34 | 3.74 (−1.7, 9.1) | 0.17 |
Abs. DT cost variability | 0.36 (0.03, 0.7) | 0.65 (0.4, 0.9) | 0.35 | −0.29 (−0.7, 0.1) | 0.16 |
% DT cost variability | 23.49 (3.3, 43.7) | 40.82 (27.7, 54.0) | 0.35 | −17.33 (−42.3, 7.6) | 0.17 |
Linear models adjusting for variation due to age, gender, BMI, and activity revealed that average walking speeds were very similar in the TC naïve vs. TC expert groups during both undisturbed single task and DT walking. Group differences in these outcomes, as well as derived differences in absolute or percent DT costs, were not statistically significant (see Table
Random-slopes model with shared baseline adjusting for variation due to age, gender, BMI, and activity also revealed trends toward reduced stride time variability following 6 months of TC. A small improvement in DT stride time variability (effect size = 0.2) was estimated with TC training, but no significant differences between groups were observed (see Table
Outcome measure | Tai Chi ( |
Usual care ( |
Between groups |
||||||
---|---|---|---|---|---|---|---|---|---|
Baseline |
6 months |
Within-group |
Baseline |
6 months |
Within-group |
Effect size | Mean difference (95% CI) | ||
Mean (95% CI) | Mean (95% CI) | Mean (95% CI) | Mean (95% CI) | ||||||
Gait speed ST (m/s) | 1.12 (1.1, 1.2) | 1.16 (1.1, 1.2) | 0.074 | 1.12 (1.1, 1.2) | 1.17 (1.1, 1.2) | 0.076 | −0.011 (−0.07, 0.05) | 0.71 | |
Gait speed DT (m/s) | 0.97 (0.9, 1.0) | 1.03 (1.0, 1.1) | 0.97 (0.9, 1.0) | 1.03 (1.0, 1.1) | 0.016 | 0.002 (−0.06, 0.06) | 0.94 | ||
Stride time variability ST (CV %) | 1.96 (1.8, 2.1) | 1.74 (1.5, 2.0) | 0.11 | 1.96 (1.8, 2.1) | 1.79 (1.5, 2.0) | 0.22 | 0.079 | −0.044 (−0.3, 0.3) | 0.77 |
Stride time variability DT (CV %) | 2.58 (2.3, 2.9) | 2.29 (2.0, 2.6) | 2.58 (2.3, 2.9) | 2.46 (2.2, 2.8) | 0.39 | 0.19 | −0.17 (−0.5, 0.1) | 0.27 | |
Abs. DT cost speed | −0.15 (−0.2, −0.1) | −0.14 (−0.2, −0.1) | 0.63 | −0.15 (−0.2, −0.1) | −0.14 (−0.2, −0.1) | 0.65 | 0.002 | 0.0 (−0.06, 0.06) | 0.99 |
% DT cost speed | −12.92 (−16.7, −9.2) | −11.32 (−15.0, −7.7) | 0.43 | −12.92 (−16.7, −9.2) | −11.33 (−16.0, −6.7) | 0.52 | 0.001 | 0.014 (−5.4, 5.4) | >0.99 |
Abs. DT cost variability | 0.62 (0.3, 0.9) | 0.61 (0.3, 0.9) | 0.95 | 0.62 (0.3, 0.9) | 0.62 (0.3, 0.9) | 0.98 | 0.007 | −0.007 (−0.4, 0.4) | 0.97 |
% DT cost variability | 39.48 (20.9, 58.1) | 38.00 (20.7, 55.4) | 0.90 | 39.48 (20.9, 58.1) | 40.13 (22.1, 58.2) | 0.96 | 0.037 | −2.13 (−23.8, 19.6) | 0.85 |
Both the TC and usual care group increased DT walking speed (
To our knowledge, this study presents the first evidence that TC has the potential to positively impact DT stride time variability in healthy older adults. Cross-sectional comparisons revealed a significant degree of lower stride variability during dual tasking in TC experts, compared with TC naïve healthy older adults. Moreover, TC-naïve adults that were exposed to 6 months of TC training exhibited within-group significant improvements in DT stride time variability. By contrast, we observed limited impact of TC on stride variability during quiet walking or of TC on gait speed during single task walking.
As DT stride variability is associated with falls in the elderly (Visser,
Results of a recent 5-year prospective cohort study of healthy community-dwelling adults, ages 70–90 years and without gait impairments, provide context for the relevance of our findings. In adjusted models that accounted for age, gender, and fall history, baseline DT stride time variability (average CV was 2.97 ± 1.47%) was the only performance-based measure that predicted falls (Rate Ratio 1:11) (Mirelman et al.,
Prior studies evaluating the impact of TC on gait performance in older adults are quite variable in design and findings. We are not aware of any studies that evaluated the impact of TC on stride time variability and only a small number of studies have evaluated outcomes utilizing a DT paradigm. In one recent study of older adults (mean age 87.7 years) living in assisted living facilities, Manor et al. (
Our observed differences in outcomes based on stride time variability but not gait speed are not surprising. These features of gait have been shown to be independent and reflect different neuromotor processes (Hausdorff,
Our finding that the impact of TC on gait was more apparent during DT vs. undisturbed walking conditions was also consistent with our hypotheses. DT-related changes in gait may result from interference caused by a competition between the attention demanded by gait and the attention demanded by a concomitant task, in our case, serial subtractions recited aloud (Verhaeghen et al.,
The potential of TC to reduce cognitive–motor competition is supported by other studies in which more challenging DT activities have better discriminated potential cognitive–motor benefits of TC. For example, in a randomized trial comparing short-term TC training to more traditional balance training, when exposed to simulated slips (experimentally shifted supporting force plates) older adults exposed to TC showed reduced tibialis anterior reaction time and reduced occurrence of co-contraction with antagonist muscles (Gatts and Woollacott,
Our study has a number of important limitations. First, samples for both the cross-sectional comparison and RCT were small, and could have resulted in type II errors. Because we considered this an exploratory study, we included statistical evaluations of outcomes without adjusting for multiple comparisons. When multiple comparisons were accounted for with Bonferonni adjustments, none of the outcomes were statistically significant. Findings in this study were intended to generate hypotheses to explore in future studies. Thus, the long- and short-term effects of TC on DT stride parameters in active healthy adults will need to be confirmed in larger, adequately powered studies. For our RCT, it is also possible that lack of more robust findings are due to the fact that 6 months of TC is insufficient training time to impact DT stride variability, and/or that more provocative DT challenges (e.g., more complex mental tasks and/or stride variability while negotiating turns or obstacles) are needed to observe any therapeutic impact of TC. Finally, our use of a non-active wait-list comparison group does not control for participant expectancy or psychosocial support afforded by participation in active TC programs. Future studies will require active comparison groups (e.g., alternative group exercise programs) that control for these factors.
With respect to our cross-sectional study, comparisons between TC experts and naïves may be confounded by differences between groups other than TC exposure. While linear models that included potential confounders (i.e., age, gender, BMI, physical activity) suggested an association with TC even after these factors were taken into account, other factors, including training in other martial arts could not be fully accounted for.
Another limitation of this study is lack of objective independent measures of proficiency in TC, which may have varied considerably especially in our expert TC group. In our cross-sectional comparison, TC experience among experts ranged from 10 to 50 years. A regression analysis indicated a slightly inverse relationship between years of training and DT stride variability, but this relationship was not statistically significant (
In healthy active older adults, trends toward positive TC effects on gait were most obvious only under cognitively challenging DT conditions, and only for stride time variability. DT stride variability offers a potentially sensitive metric for monitoring the impact of TC on fall risk with healthy aging. These findings also support the value of neurophysiological research evaluating how mind–body exercises like TC impact cognitive–motor interactions and confirm previous findings, which suggest that TC reduces the risk of falls in older adults (Wolf et al.,
PW, JH, LL, CP, and BM conceived and designed the study. ML acquired the study data with oversight from VN. ML and BG analyzed the study data. BG and EM performed statistical analysis. PW drafted and revised the manuscript. All authors reviewed the manuscript and approved the final version.
Peter M. Wayne is the founder and sole owner of the Tree of Life Tai Chi Center. Peter M. Wayne’s interests were reviewed and are managed by the Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict of interest policies. The Tree of Life Tai Chi Center did not provide payment or services for any aspect of this study. The other co-authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Danielle Berkowitz. This work was supported by grant R21 AT005501-01A1 from the National Center for Complementary and Alternative Medicine (NCCAM), National Institutes of Health (NIH). Dr. LL was supported by grant R03-AG25037 from the National Institute on Aging (NIA) and the Irving and Edyth S. Usen and Family Chair in Geriatric Medicine at Hebrew SeniorLife. Dr. VN was supported by R21 AT005501-01A1. Dr. CP was supported by a grant (NSC 102-2911-I-008-001) from Ministry of Science and Technology of Taiwan. Dr. BM was supported by grant 5 KL2 RR025757-04 from the Harvard Catalyst and grant 1K01 AG044543-01A1 from the National Institute on Aging. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NCCAM, NIA, or NIH.