“Tai Chi Training for Attention Deficit Hyperactivity Disorder: A Feasibility Trial in College Students”

Abstract

Objective:

Many young adults are affected by attention deficit hyperactivity disorder (ADHD) and often desire non-pharmacological treatment options. Mind-body techniques might serve as complementary therapies to first-line stimulant medications, but studies are limited. Tai Chi is an increasingly popular practice that integrates movement with cognitive skills relevant to ADHD. We performed a feasibility trial of Tai Chi training in undergraduates to inform the design of a fully powered randomized controlled trial (RCT).

Method:

Undergraduates with ADHD were recruited, screened, enrolled, and assessed at baseline. They were assigned to three parallel seven-week intervention arms, Tai Chi, Active Control (cardio-aerobic fitness), and Inactive Control (no contact), with follow-up assessments. Feasibility of a larger clinical trial was evaluated, especially with respect to enrollment and retention. Additionally, potential clinical outcome measures were examined for practicality and reliability.

Results:

21 participants were assessed at baseline and 19 at follow-up (90% retention). The primary clinical outcome measure, self-reported inattention symptoms (Conners’ CAARS-S:L DSM-IV Inattentive Symptoms subscale), exhibited good test-retest reliability in controls (r=0.87, n=10) and correlated with reduced mindfulness (FFMQ acting with awareness subscale) at baseline (r=−0.74, n=20). Class attendance and self-reported daily practice time were variable. Randomization to group classes was hindered by the college students’ restricted schedules.

Conclusion:

The high retention rate and good data quality suggest that an RCT of Tai Chi for ADHD is feasible. Further measures are identified to improve enrollment rates, adherence, and randomization procedures. Future work might extend to other young adult populations and high school students.

Introduction

Attention deficit hyperactivity disorder (ADHD) begins in childhood and persists into adulthood in 30-70% of cases [1], [2]. Consistent with this, in a 2018 survey of US college freshmen 7.4% reported having ADHD, an increase from 5.0% in 2010 [3], [4]. Meta-analyses of RCTs show that stimulant treatments are effective in adult patients [5]–[7]. However, in many cases medications are not desired, tolerated, or sufficiently effective, and they can be subject to diversion and misuse [8]–[10]. Therefore, non-pharmacological options are urgently needed for this increasingly diagnosed condition [11]–[13]. Studies support cognitive behavioral therapy for adult ADHD [14], [15]. Exercise, neurofeedback, and meditation have also been studied [16]–[22]. The notion of a healthy mind in a healthy body, mens sana in corpore sano, is ancient and intuitively attractive, and there is a growing literature describing the potential benefits of combined motor and attention training for ADHD [23]–[28].

Tai Chi is an increasingly popular mind-body exercise that encourages mindful attention to the body in motion. It integrates low impact flowing movements with a diverse set of cognitive skills relevant to ADHD, including mental focus, somatic awareness, task shifting, goal setting, and imagery/visualization [29]–[32]. It is recognized as a safe low-cost complementary therapy, and it is practiced by millions of Americans for a variety of purposes [33], [34]. The majority of scientific studies to date have focused on physical outcomes such as gait, posture, and cardiovascular health, largely in middle-aged and elderly adults [35]. Preliminary studies also suggest that Tai Chi may enhance cognitive function and psychological well-being in the elderly [36]–[44]. A limited body of preliminary studies has begun to describe its effects in younger adults, for whom Tai Chi training may lead to improvements in physical and mental health, mindfulness, mood, perceived stress, and sleep quality [45]–[50]. There is also some evidence of improvements in salivary cortisol and blood immunological markers [51], [52]. Taken together, these results demonstrate that the physical and mental health effects of Tai Chi training are amenable to scientific study, and the observed improvements in cognitive function indicate that Tai Chi training may have the potential to serve as therapy for ADHD.

There are four published studies of the effects of Tai Chi in subjects with ADHD. In a non-controlled pilot trial of 13 adolescents with ADHD, teacher ratings of symptoms improved following a 5-week Tai Chi course [53]. In a study conducted in China, children with ADHD randomly assigned to 12 weeks of Tai Chi training (n=16) showed improvements in hyperactivity, aggressiveness, vestibular function, proprioception, and learning abilities compared to controls (n=14) (cited in [54]). In a small pilot study, four children with ADHD were taught Tai Chi and tended to show improvements in ADHD hyperactive-impulsive symptoms [55]. In a biomarker development study, 34 children underwent an 8-week Tai Chi-based mindful movement intervention and exhibited improvements in ADHD symptoms that correlated with motor measures [56]. No studies have been reported in undergraduates with ADHD, who are subject to additional stress, in danger of drug misuse, and perhaps more open to complementary treatment options [57].

We have initiated a sequence of studies with the ultimate goal of rigorously evaluating Tai Chi training as therapy for ADHD in a fully powered randomized controlled trial (RCT). First, in an observational pilot study we compared typical university students in a semester-long Tai Chi course (n=28) to passive controls (n=44) and observed a reduction in symptoms of inattention [25]. Second, we surveyed undergraduates with ADHD (n=47) and without (n=49) to estimate the size of the pool of potential participants for an RCT. We were encouraged to find that over 30% of respondents with ADHD rated their likelihood of participating in an RCT at 80% or higher [58]. Third, in the present study we extend our work with a feasibility trial intended to help guide the design of an RCT [59].

The primary purpose of this study was to assess the feasibility of recruiting and retaining undergraduates with ADHD in a 7-week RCT of Tai Chi training versus Active Control and Inactive Control groups. We expected to recruit 45 participants and retain 80% through post-intervention assessment (Aim 1). Additionally, we sought to determine the data quality and statistical distribution of our primary outcome variable, the Conners Adult ADHD Rating Scales self-report long version (CAARS-S:L) DSM-IV Inattentive Symptoms subscale [60], as well as several secondary measures (Aim 2).

Methods

Study Design

Enrollment and retention benchmarks were the main focus of this work, along with pilot assessment of outcome measures. In this parallel group feasibility trial, participants were recruited, screened, assessed at baseline, and assigned to one of three arms: (1) Tai Chi, (2) Active Control, and (3) Inactive Control. Follow-up assessments took place after the 7-week intervention period. All procedures were approved by the University of Wisconsin-Madison Health Sciences Minimal Risk Institutional Review Board (protocol 2015-0807), and the trial was registered with ClinicalTrials.gov (NCT02688959). This paper was drafted in accordance with the Consolidated Standards of Reporting Trials (CONSORT) extension to randomized pilot and feasibility trials and the extension to randomized trials of nonpharmacologic treatments [61], [62].

Participants, recruitment, and enrollment.

Participants were UW-Madison undergraduates aged 18-23 diagnosed with ADHD. For generalizability, participants were not excluded for comorbid disorders or medication use. Participants were recruited by emails sent to students in the UW-Madison McBurney Disability Resources Center database with a diagnosis of ADHD. Recruitment also utilized a student newspaper website advertisement, a student radio station public service announcement, and flyers. All advertising pointed to a website, which provided further details including the consent form for review and an email address to contact the study team. Potential participants were offered $20 per test session and assured that they would not be asked to change their medications. A telephone screen confirmed undergraduate status, age, willingness to participate, and ability to provide documentation of ADHD diagnosis. Subsequent online screening obtained additional demographic information, ADHD presentation, and medication status. In addition to verbal and online consent obtained prior to the screenings, written consent was obtained when the participant arrived for the baseline assessment.

Allocation, assignment, assessments, and interventions.

Following recruitment and screening, participants were allocated to the Tai Chi, Active Control, and Inactive Control arms. During screening, participants indicated their availability for timeslots. To achieve adequate enrollment to evaluate feasibility parameters, randomization was compromised as follows. In case a participant was available for only one of the two active arms, instead of excluding them, they were allocated to that arm. If a participant was not available for either active arm, they were allocated to the Inactive Control arm. Participants were not informed of intervention class schedules prior to allocation, so at baseline assessment they were blind to their allocation. The PI concealed allocation from study staff involved in the assessments. Participants were informed of their allocation after baseline assessments were completed. Participants were asked not to reveal their assignment to staff at the follow-up assessment.

Participants in the active arms were assigned to classes that met twice weekly for seven weeks during the Fall semester of 2016 at UW-Madison sports facilities 1.1-1.4 km from central campus. The Tai Chi class met Tuesdays and Thursdays 4:00-5:00 pm and was open only to participants in this study. The class was taught by an instructor with over twenty years’ experience and emphasized experiential learning with two weeks of introductory sessions on gait, posture, and Tai Chi principles followed by instruction in the 24-form Yang style sequence [63], [64]. The instructor was aware that the participants were diagnosed with ADHD. The Active Control class met Mondays and Wednesdays 6:15-7:15 pm. This “Box Blast” class was offered through UW-Madison Recreational Sports and was available to all University community members. The class, taught by an undergraduate, emphasized non-contact kickboxing for cardio-aerobic fitness [65]–[67]. The instructor was not informed of the participants’ ADHD status. Students in both classes were given access to online instructional videos and encouraged to practice at home. Participants in the Inactive Control arm did not receive any training and were only contacted for screenings and assessments.

Evaluation of Feasibility (Aim 1)

Recruitment, enrollment, and retention.

Assuming 5% prevalence [68] among the 29,504 undergraduates at UW-Madison [69], we estimated a population of approximately 1475 students with ADHD. A recruitment rate of 3% in one semester would yield 45 participants. We expected to retain at least 80% of participants through post-intervention assessment.

Expectations, adherence, acceptability, and safety.

In the online screen all participants were asked how they expected their ADHD symptoms would change in response to each of the three assignment options. We targeted 80% class attendance and monitored attendance by instructor’s records (Tai Chi only). Every day during the intervention period, class participants were texted and asked how many minutes they practiced, including time in class and at home. At the follow-up assessment, participants in the intervention arms responded to written questions about their satisfaction with their classes and any adverse reactions. Finally, all participants were asked to share any thoughts about the study in an open-ended written response.

Clinical Outcome Measures (Aim 2)

Self-report and objective measures were obtained. Participants were assessed double blind at baseline before assignment and single blind at follow-up (Table 1).

Table 1.

60 minute Assessment Session

Welcome and Consent (3 min)

Demographics (1 min)

Begin Cardiac Recording (2 min)

Neurocognitive (9 min)

Postural Stability (7 min)

Conners (5 min)

GRE Verbal Reasoning Excerpt (12 min)

Pittsburgh Sleep Quality Index (3 min)

Five Facet Mindfulness Questionnaire (5 min)

SF-36 (3 min)

Debrief (2 min)

Times are typical. There were no time limits placed on self report and GRE excerpt.

Primary clinical outcome measure.

Symptoms of inattention tend to predominate in young adults diagnosed with ADHD [70], [71]. Therefore, the CAARS-S:L DSM-IV Inattentive Symptoms subscale was chosen a priori as the primary outcome measure to help guide the data collection and analysis components of this feasibility trial [60]. This subscale consists of 9 of the 66 CAARS-S:L 4-point Likert-style items and has a range of 0-27 with a higher score indicating more severe symptom levels. Raw scores are converted to T-scores by sex with a mean of 50 and standard deviation of 10 for a non-clinical population aged 18-29 [60]. To assess internal validity, the CAARS-S:L inconsistency index was calculated by summing the absolute differences between 8 pairs from the 66 items expected to yield similar responses. A score of 8 or higher may indicate a noncompliant or unmotivated respondent [60]. However, given that we had two datasets from each participant, we included cases with a score of 8 if the participant was “consistently inconsistent” in these response pairs, i.e. if the root mean square difference in the 8 pair differences between baseline and follow-up was < 1.5.

Secondary clinical outcome measures.

In a subsequent full-scale efficacy trial, it will be important to capture secondary measures to characterize the population sampled, control for confounds, and identify biomarkers that respond to the intervention. We therefore collected measures expected to be relevant to Tai Chi training and ADHD, and we examined their reliability and relation to the primary outcome measure of inattention. Demographic, diagnostic, and treatment data were acquired. Participants were assessed for ADHD symptoms (CAARS-S:L, all scales) [60], general health (RAND 36-Item Health Survey 1.0., SF-36) [72], sleep quality (Pittsburgh Sleep Quality Index, PSQI) [73], and mindfulness (Five Facet Mindfulness Questionnaire, FFMQ) [74]. Participants completed 12-item excerpts from the verbal reasoning section of a practice Graduate Record Exam (GRE) as well as the Flanker test of attention and inhibitory control and the Dimensional Change Card Sort test of executive function (NIH Toolbox) [75]. Participants also performed a postural stability test and were monitored for heart rate [76]–[81].

Analytical Methods

Enrollment and retention were compared to a priori goals. Descriptive and inferential statistics for outcome measures were calculated using R version 3.3.2 [82]. Central tendency and variability are presented as mean ± standard deviation. Missing data were imputed by mean substitution, and data points greater than three standard deviations from the mean were excluded. Normality was evaluated by visual inspection and the Shapiro-Wilk test, and non-normal measures were log transformed prior to correlation and inferential analyses. Bivariate correlations, including test-retest reliability in controls, were evaluated by Pearson r. Linear mixed effects modeling was used to calculate group x time inferential statistics [83]. Group equivalence at baseline was evaluated by one-way ANOVA.

Results

This feasibility trial is summarized in the CONSORT flowchart shown in Figure 1 [61]. Here we describe results pertaining to feasibility (Aim 1) as well as clinical outcome measures evaluated for a future full-scale RCT (Aim 2).

Figure 1.

Flowchart of participants in feasibility trial of Tai Chi for ADHD.

Feasibility (Aim 1)

Of 47 potential participants who inquired about the study in response to recruitment, 21 were enrolled and assessed at baseline, and 19 were retained through post-intervention assessment (Figure 1). Baseline demographics are presented in Table 2 along with details regarding diagnosis and medication. Ten participants reported a comorbid disorder. Of the 21 enrolled participants, 15 reported that they learned of the study through emails that had been sent by the campus disability resource center to 276 students (45% female) with a diagnosis of ADHD. The other participants had learned of the study by word of mouth (5) and a flyer (1).

Table 2.

Baseline Demographics, Diagnosis, and Medication

	n=21
Age (years)	20.7 ± 1.5
Female	67%
White	86%
Age of first diagnosis (years)	13.3 ± 5.1
Age of most recent diagnosis (years)	18.6 ± 3.3
Presentationa
Inattention	12
Hyperactivity/Impulsivity	1
Combined	7
Take Medication b	16
Amphetamine	10
Lisdexamfetamine	4
Methylphenidate	5
Adherence and Response (1 high - 5 low)
Take as prescribed	2.12 ± 0.72
Effective	1.75 ± 0.58
Tolerated	1.81 ± 0.75
Want to decrease	2.9 ± 1.1

^a1 unknown

^b2 used multiple types

The PI, blind to baseline assessment results, allocated participants as described above. Participants and research staff were blind to allocation at baseline assessment. Participants were assigned to Tai Chi (n=9), Active Control (n=5), or Inactive Control (n=7). Research staff were blind to assignment at post-intervention assessment.

Expectations of improvement prior to assignment due to each intervention (1=low to 4=high) were Tai Chi = 2.6 ± 1.0, Active Control = 2.5 ± 0.9, and Inactive Control = 1.2 ± 0.4 (n=21). Reported daily practice time for Tai Chi was 10 ± 10 minutes (range 0-33, n=9) and for Active Control 5 ± 8 minutes (range 0-18, n=4). Participants in the Tai Chi arm attended 3.7 ± 2.2 (range 1-8) of 14 classes. Active Control (“Box Blast”) attendance was not recorded. Class satisfaction responses are presented in Table 3. Two participants in the Tai Chi arm reported muscle soreness and one reported knee soreness. Two participants in the Active Control arm reported muscle soreness. Finally, participant open-ended responses concerning the study are presented in Supplementary Materials.

Table 3.

Acceptability of Interventions

Satisfaction (1 high - 5 low)	Tai Chi (n=9) mean ± s.d.	Active Control (n=4) mean ± s.d.
Assigmnent to class	2.6 ± 1.1	2.8 ± 1.5
Class content	2.2 ± 1.0	2.0 ± 0.8
Teacher	1.9 ± 0.8	2.2 ± 0.5
Helped to reduce ADHD symptoms	2.7 ± 1.0	2.8 ± 0.5
Felt motivated to participate in class	2.8 ± 1.3	2.0 ± 0.8
Class location was convenient	3.7 ± 1.3	2.8 ± 1.5
Class time was convenient	3.0 ± 1.1	3.2 ± 1.0
Enjoyed being with other students in class	2.2 ± 1.0	2.2 ± 0.5
Easy to practice outside of class	2.7 ± 1.4	4.0 ± 0.8

Responses at follow-up after 7-week intervention period. All p’s > 0.1 (two sample t-test).

Clinical Outcome Measures (Aim 2)

Self-report, cognitive, and physiological measures, were obtained for the 21 participants enrolled at baseline and the 19 who were retained through follow-up (Table 4). The primary outcome variable at baseline was 68 ± 11 (n=20, CAARS-S:L DSM-IV Inattentive Symptoms). Its test-retest reliability was evaluated for the Active Control and Inactive Control participants who were assessed at baseline and follow-up (r = 0.87, p = 0.0012 uncorrected, n = 10). The CAARS-S:L inconsistency index was 5.6 ± 1.9 (range 2-8) at baseline and 5.2 ± 1.8 (range 2-8) at follow-up. Four participants scored 8 at baseline and/or follow-up, but because their response patterns were similar at both times (r.m.s. difference in 8 pair differences of 1.1 to 1.3 vs. 0.92 ± 0.37, n = 19), their results were included in the subsequent analysis. One subject scored 8 at baseline and withdrew prior to follow-up, so we excluded their CAARS results. Finally, in this small feasibility trial our aim was not to test effectiveness, but we note that there was no significant group x time interaction in the primary outcome measure comparing Tai Chi vs. Active Control (β_group:time ± standard error = −0.4 ± 3.0, p = 0.9, df = 11) nor Tai Chi vs. Inactive Control (−1.7 ± 2.4, p = 0.5, df = 13).

Table 4.

Clinical Outcome Measures

	Tai Chi		Active Control		Inactive Control		All Subjects at Baseline (n=21)			Controls (n=10)
	Baseline (n=9, 5 F)	Follow-up (n=9, 5 F)	Baseline (n=5, 4 F)	Follow-up (n=4, 3 F)	Baseline (n=7, 5 F)	Follow-up (n=6, 4 F)			vs. DSM-IV inatten. a	Follow-up vs. Baseline
	mean ± s.d.	mean ± s.d.	mean ± s.d.	mean ± s.d.	mean ± s.d.	mean ± s.d.	mean ± s.d.	range	r	r
ADHD self report CAARS-S:L fg
DSM IV Inattentive a	68.7 ± 7.6	68.1 ± 9.1	77 ± 14	76.2 ± 9.9	62.3 ± 9.8	65.3 ± 8.2	68 ± 11	49 to 90	1	0.87**
DSM IV Hyperactive/Imp.	58 ± 11	57 ± 11	54 ± 16	57 ± 16	63 ± 11	59 ± 12	59 ± 12	36 to 82	−0.0014	0.75*
Cognitive function, reaction time (s)
Flanker mean	0.41 ± 0.10	0.392 ± 0.087	0.50 ± 0.11	0.41 ± 0.14	0.53 ± 0.10	0.475 ± 0.097	0.47 ± 0.11	0.3 to 0.67	−0.13	0.55
Flanker s.d. cd	0.107 ± 0.049	0.101 ± 0.036	0.162 ± 0.091	0.115 ± 0.027	0.128 ± 0.052	0.114 ± 0.027	0.127 ± 0.063	0.058 to 0.31	0.17	−0.39
Card Sort mean d	0.54 ± 0.33	0.41 ± 0.18	0.51 ± 0.19	0.44 ± 0.12	0.61 ± 0.28	0.54 ± 0.2	0.55 ± 0.28	0.17 to 1.4	−0.47*	0.81**
Card Sort s.d. bde	0.175 ± 0.057	0.23 ± 0.19	0.25 ± 0.12	0.28 ± 0.17	0.27 ± 0.18	0.31 ± 0.26	0.23 ± 0.13	0.064 to 0.65	−0.61**	0.46
Postural stability i
mean velocity (mm/s) eyes open cd	15.1 ± 9.7	13.8 ± 3.2	15.5 ± 4.6	13.5 ± 1.7	20 ± 12	14.2 ± 2.2	16.5 ± 9	9.1 to 41	−0.18	0.086
closed de	15.4 ± 6.0	15.8 ± 5.7	17.9 ± 6.4	13.4 ± 1.6	19.9 ± 7.5	18.3 ± 2 .0	17.2 ± 6.4	9.8 to 33	−0.15	0.71
rms distance (mm) eyes open bcd	6.5 ± 4.0	6.5 ± 2.8	7.2 ± 2.2	5.3 ± 1.9	7.7 ± 2.6	7.3 ± 2.4	7.0 ± 3.2	3.1 to 17	−0.45	−0.035
closed e	5.7 ± 2.7	5.7 ± 2.3	7.3 ± 4.7	4.43 ± 0.94	9.9 ± 2.5	9.0 ± 3.7	7.2 ± 3.5	2.9 to 15	−0.32	0.64
mean frequency (Hz) eyes open d	0.48 ± 0.16	0.44 ± 0.13	0.414 ± 0.062	0.51 ± 0.17	0.392 ± 0.071	0.37 ± 0.16	0.44 ± 0.12	0.28 to 0.74	0.031	−0.075
closed	0.54 ± 0.10	0.537 ± 0.081	0.54 ± 0.15	0.61 ± 0.10	0.369 ± 0.07	0.44 ± 0.15	0.49 ± 0.13	0.29 to 0.72	0.33	0.6
centroidal frequency (Hz) eyes open	1.69 ± 0.16	1.65 ± 0.14	1.54 ± 0.17	1.64 ± 0.17	1.45 ± 0.11	1.55 ± 0.19	1.59 ± 0.18	1.3 to 2.1	0.12	0.76*
closed	1.65 ± 0.15	1.64 ± 0.10	1.58 ± 0.12	1.646 ± 0.053	1.460 ± 0.097	1.59 ± 0.12	1.58 ± 0.15	1.3 to 1.9	0.4	0.89**
Cardiac during postural stability task j
rate (bpm) eyes open	101 ± 14	103 ± 15	99 ± 18	93 ± 14	86.3 ± 10	98 ± 14	95 ± 15	67 to 120	0.37	0.42
closed	101 ± 16	104 ± 19	100 ± 18	91 ± 15	88.3 ± 8.7	98 ± 14	96 ± 15	74 to 130	0.31	0.54
variability, rmssd (ms) eyes open	13.9 ± 7.0	11.5 ± 5.4	12.9 ± 6.0	17.6 ± 6.7	17.1 ± 3.5	14.2 ± 7.0	14.9 ± 5.5	4 to 26	−0.23	0.43
closed	13.0 ± 6.7	11.2 ± 6.2	11.8 ± 4.7	14.8 ± 4.8	16.2 ± 5.5	14.0 ± 7.4	13.9 ± 5.7	4 to 24	−0.21	0.48
PSQI - daytime dysfunction deh	1.11 ± 0.60	1.78 ± 0.67	1.8 ± 1.3	1.50 ± 0.58	1.43 ± 0.98	1.2 ± 1.2	1.38 ± 0.92	0 to 3	0.57**	0.52
FFMQ - acting with awareness	19.8 ± 5.2	20.1 ± 4.8	16.2 ± 8.7	14.2 ± 6.5	21.1 ± 4.6	20.5 ± 5.2	19.4 ± 6.0	8 to 29	−0.74***	0.82**
GRE - verbal reasoning (% correct)	57 ± 20	59 ± 25	46 ± 13	52 ± 12	42 ± 15	49 ± 16	49 ± 18	17 to 92	0.4	0.55
SF-36 - Physical Functioning bcdek	92.5 ± 6.5	88 ± 16	98.0 ± 4.5	100 ± 0	97.9 ± 3.9	98.3 ± 2.6	95.8 ± 5.7	85 to 100	−0.26	0.93***

^*p<0.05,

^**p<0.01,

^***p<0.001

^aPrimary outcome measure

^{b c}One outlier >3 s.d. from mean excluded at ^b baseline or ^c follow-up

^{d e}Non-normal distribution (Shapiro-Wilk test p<0.05) at ^d baseline or ^e follow-up. Both baseline and follow-up were log transformed prior to calculation of inferential statistics.

^fOne participan’s CAARS-S:L responses were excluded by the inconsistency index criterion.

^{g h}Item responses missing in the ^g CAARS-S:L (5, 0.2% of all responses) and ^h PSQI (6, 1.0% of all responses) were imputed by mean substitution.

^{i j}Due to technical difficulties, data were not acquired for ⁱ postural stability in 2 baseline and 1 follow-up sessions and for ^j cardiac in 3 baseline and 1 follow-up sessions.

^kDue to an editing error in creating the self-report instrument, only 22 of the 36 items in the SF-36 were obtained, which prevented scoring four of its subscales.

Six secondary measures exhibited good reproducibility, i.e. significant follow-up vs. baseline correlations (Table 4). Four secondary measures were found to significantly correlate at baseline with the primary measure (Table 4) including FFMQ Acting with Awareness (Figure 2). Significant group differences at baseline were observed for two postural stability measures (mean frequency eyes closed, p = 0.027, and centroidal frequency eyes open, p = 0.034, one-way ANOVA), but for none of the other clinical outcome measures (p > 0.05).

Figure 2. Inattention and Mindfulness.

At baseline, those participants who reported higher symptom levels (primary outcome measure, CAARS-S:L DSM IV inattention scale) reported lower levels of the 'Acting with awareness' subscale of the Five facet mindfulness questionnaire (FFMQ). r = −0.74, n=20, p<0.001 uncorrected, p<0.05 corrected for multiple comparisons.

Discussion

The overarching goal of this trial was to assess the feasibility of an RCT to determine the effectiveness of Tai Chi training as therapy for ADHD. The primary objectives were to enroll 45 participants and retain 80% through follow-up within a single semester. In addition, we expected participants in active arms to attend 80% of classes. While recruitment and adherence goals were not met, retention for follow-up assessments exceeded expectations. This feasibility trial provided valuable insight for future work concerning recruitment, allocation, and adherence, as well as acceptability of the experimental and control interventions. We also evaluated primary and secondary clinical outcome measures.

Feasibility Lessons Learned

We enrolled about half of our target sample size. Of the enrolled participants, 71% learned of the study through emails sent to clients of the campus disability resource center. This suggests direct email to the campus body might be an effective recruiting tool. Word of mouth yielded 24% of participants, and this might be enhanced through use of social media and a more extended recruiting period. Additionally, the multiple step screening process might have been a hindrance for this busy and potentially distractible population. Starting with 47 email inquiries from potential participants, only 28 individuals advanced to the next step of a phone screen. It may be more effective to first advertise a link to an online screen.

In this trial, only one Tai Chi class and one Active Control class were available. This limited scheduling options for the participants and made it impractical to allocate them entirely randomly to interventions. Instead, the allocations were based in part on scheduling availability. Sufficient resources and lead time would permit the scheduling of multiple Tai Chi and Active Control classes.

Participants in the Tai Chi arm generally attended fewer than half of the classes. This might be addressed in future work through improvements in scheduling and location as participants reported poor satisfaction with the class time and place. It might also be helpful to increase participant compensation and tie it more closely to attendance. In our earlier survey, undergraduates with ADHD reported that they would prefer a mixed class [58]. Therefore, opening the Tai Chi class to students without ADHD might improve attendance by study participants. We used text messaging to track daily practice in this trial, and this might be modified to provide reminders of class time and encouraging messages. The text response rate was lower among participants who reported higher levels of inattention (r = −0.66, p = 0.014), so in future work we intend to follow up in case of no response by text, email, and phone. Finally, to improve home practice we may apply social cognitive theory and the habituation-intention framework [44].

The Tai Chi and Active Control interventions were generally acceptable to the participants. Satisfaction scores for class assignment, content, teacher, and motivation were similar between the Tai Chi and Active Control arms. Thus, the “Box Blast” class is a practical choice for the Active Control. Some participants in both arms complained of minor muscle or joint soreness. These and other potential adverse reactions should continue to be monitored in future work.

In addition to addressing logistical issues such as recruitment and retention, we gauged the feasibility of obtaining suitable clinical outcome measures. Our observed baseline value of the primary clinical outcome measure (CAARS-S:L DSM-IV Inattentive, 68+/−11) is in good agreement with that reported elsewhere for college students with ADHD [84], [85]. Test-retest reliability in the Inactive Control and Active Control participants was very good (r = 0.87) and comparable to that reported for a subset of the CAARS scales (r = 0.80-0.91) [60]. Internal validity of the CAARS-S:L item responses as gauged by the Inconsistency Index was good and agreed with that reported elsewhere [85].

Several secondary measures exhibited good test-retest in the Inactive Control and Active Control participants and tended to be consistent with expectations and existing literature. The FFMQ acting with awareness subscale correlated negatively with inattention at baseline. This is notable especially in light of a reported improvement in response to Tai Chi training by typical undergraduates in FFMQ “Total Mindfulness”, which also correlated with inattention in our sample (r = −0.68, p = 0.001, n = 20, not shown) [86]. Inattention correlated as well with the PSQI Daytime Dysfunction subscale at baseline, as has been observed by others in a large typical college student sample [87]. There were negative correlations between inattention and the card sort reaction time mean and standard deviation, which might appear paradoxical, but to our knowledge this relation has not been examined previously in adults with ADHD.

Limitations

This feasibility trial was subject to several limitations. The participants were likely open to the benefits of Tai Chi and might be prone to providing positively biased self-report data. Expectations of improvement, however, were equivalent for Tai Chi and the Active Control among all participants at baseline. As addressed elsewhere in the Discussion, the enrollment target was not met, participants were not allocated by a completely random process, and adherence in the Tai Chi arm was variable. To increase sample size, Active Control participants were included in the test-retest evaluations. Class attendance was not recorded in the Active Control arm. Our experience with these issues will inform the design of subsequent trials.

Future Directions

The results of this feasibility trial argue for an exploratory clinical trial with the goals of improving recruitment and adherence by means including those suggested above. Such a study should systematically address various other aspects as well, including trial design, participants, interventions, logistical feasibility, primary and secondary outcome measures, sample size, randomization, blinding, and analysis methods. We have developed a detailed preliminary protocol for such an exploratory clinical trial. If such work is successful, a full-scale high-quality efficacy RCT could follow. This work might also be extended to other populations, such as technical college students, young adults generally, and high school students. To estimate the sample size needed for fully powered trials in these populations, we would aim for sensitivity to approximately half the effect obtained with medication [5], [7]. We therefore target Cohen’s d = 0.25, which requires n = 253 per group for α = 0.05 and 1 - β = 0.8 [88], [89]. This effect would amount to a decrease of 2.8 points in the primary clinical outcome measure.

To help guide the development of combined motor and attention training interventions, it would be useful to explore potential mechanisms. Future work might examine the neurobiological underpinnings of a potential effect of Tai Chi training for ADHD. We propose that the combined motor and attention training inherent in Tai Chi can improve catecholaminergic signaling to ameliorate core deficits and reduce symptom levels. Figure 3 illustrates a model, in which the midbrain groups of dopamine cell bodies form a locus of synergy underlying the hypothesized reinforcing effects of motor and attention activity [90], [91]. This model might be tested with human neuroimaging as well as animal studies [92].

Figure 3. Neurobiological model of a potential synergistic effect of the motor and attention training inherent in Tai Chi.

Concurrent firing of neurons in the mesostriatal, mesocortical, and mesolimbic pathways may engage synaptic plasticity in midbrain to enhance dopaminergic response to salient stimuli. This model predicts increases in midbrain dopaminergic function in response to combined motor and attention training greater than the sum of the individual effects.

Conclusion

We have performed a feasibility trial to help guide the design of an efficacy RCT of Tai Chi as therapy for ADHD in undergraduates. Participant retention was very good as was the quality of the data gathered, including the primary clinical outcome measure. We identified barriers, especially related to recruitment and adherence, and we have developed a strategy to address these critical aspects of clinical trial implementation in future work.

Highlights.

• A feasibility trial of Tai Chi training as therapy for ADHD was performed

• 19 college students were followed through a 7-week intervention period

• The primary clinical outcome measure of self-reported inattention symptoms was reliable

• Steps were identified to enhance recruitment, randomization, and adherence

• These results will inform the design of a fully powered randomized controlled trial (RCT)