A randomised controlled trial comparing the effectiveness of tai chi alongside usual care with usual care alone on the postural balance of community-dwelling people with dementia: protocol for the TACIT trial (TAi ChI for people with demenTia)

This is a randomised, assessor-blind, two-arm, parallel group, superiority trial, with an embedded intervention pilot study [21]. In this RCT, 150 dyads, each comprising a person with dementia and their informal carer, will be randomised to either the control group (usual care) or the intervention group (usual care plus the TACIT Tai Chi intervention) in a 1:1 ratio at each of the three treatment sites (see Fig. 1).

This study will be conducted in three locations across the South of England (see ClinicalTrials.Gov registration for list of study sites). Participants will be identified and recruited via various sources, including National Health Service (NHS) research / clinic databases, memory assessment services (where patients are provided with a dementia diagnosis), local charities, and self-referral. Baseline and follow-up appointments will be conducted in participants’ homes or local venues. Tai Chi classes will be provided in suitable venues (e.g. church halls) and home-based Tai Chi instruction delivered in participants’ homes.

The target group for this study is people with mild to moderate dementia and their informal carer. Both the PWD and the carer must consent to participate in the study for the dyad to be eligible for inclusion. Informal carers may be the spouse, close relative, friend, or neighbour who live with the PWD participant or visit at least twice a week. While there can be more than one informal carer to support the PWD during the trial, data will be collected from only one primary informal carer.

PWD must satisfy the following criteria to be enrolled in the study: Aged 18 or above, living at home, have a diagnosis of a dementia (indicated on their medical record held by the NHS or general practitioner [GP]), are physically able to do standing Tai Chi, and willing to attend weekly Tai Chi classes. The following exclusion criteria will be applied: Living in a care home; in receipt of palliative care; have severe dementia, a Lewy body dementia or dementia with Parkinson’s disease, or severe sensory impairment; are already currently practising or have been practising within the past 6 months Tai Chi or similar exercise (Qigong, yoga, or Pilates) on average once a week or more; are currently under the care of or have been referred to a falls clinic for assessment, or are currently attending a balance exercise programme (e.g. Otago classes); or lack mental capacity to provide informed consent. In regard to dementia severity, the Mini Addenbrooke’s Cognitive Examination (M-ACE) will be conducted at the initial visit after informed consent has been obtained [22]. Scores of 9 or less will be regarded as being indicative of severe dementia and grounds for exclusion from the trial. No ceiling cut-off will be used. In relation to types of dementia, the fall rates in those with either Lewy body dementia or dementia with Parkinson’s disease are significantly higher than other forms of dementias and so would require a different intervention (e.g. [23]).

Informal carers must satisfy the following criteria to be enrolled in the study: Able to commit to supporting the PWD by participating in data collection throughout the trial and in the intervention components, if allocated to the intervention group, physically able to do standing Tai Chi, and willing to attend weekly Tai Chi classes. The following exclusion criteria will be applied: Carers who have severe sensory impairment or lack mental capacity to provide informed consent.

Both trial groups will continue to receive usual care during the course of the trial. In the UK, those who are concerned about their memory or are suspected to have dementia by their GP are referred to a memory assessment service. The memory assessment service will diagnose a person with dementia, and depending on the type and severity of dementia, may prescribe medication. Otherwise, no further treatments are offered for the PWD, and in particular, no Tai Chi or other physical activity is prescribed. Patients under the care of one of the NHS Trusts that are recruiting to the trial are provided with an educational course to provide information, support, and advice to PWD and a similar course is provided separately for informal carers. In addition, for those living in some areas, a referral is made to a memory advisor who provides telephone contact to check on the wellbeing of the PWD and their informal carer. Under the two other trusts, there is support from the voluntary sector in the form of memory advisors for every patient and informal carer before and after diagnosis, in the form of advice, information, guidance, and signposting to local services.

Participants in the control group will be asked not to take up Tai Chi or similar exercise (Qigong, yoga, or Pilates) during the period of the project.

The Tai Chi intervention comprises 3 components: (1) Tai Chi classes, (2) home-based Tai Chi exercises, and (3) a behaviour change component (see logic model in Table 1). The intervention has been designed for participants to accrue 50 h or more of Tai Chi physical activity, because exercise-based interventions to prevent falls are more effective if: they have a higher dosage (50 h or more) and challenge balance (e.g. exercises are conducted while standing) [24]. To allow sufficient time for participants in the intervention condition to be allocated to a class and to complete the 20-week intervention, the post-intervention outcome assessments are scheduled 6 months post-baseline, to coincide with the end of the 20-week intervention period. Information about the Tai Chi exercise prescription and instructional methods are provided in Table 2.

The Tai Chi instructor will keep a record of dyads’ weekly class attendance and a record of all home visits, along with a checklist to confirm all materials have been provided to each dyad. For home practice of Tai Chi, dyads will be asked to record their practice each week in a diary and return this to the instructor at the next class.

Qualitative data will be collected and reported separately from the main trial outcome paper. A researcher will observe 10% of the classes and make qualitative observations in relation to class-based adherence. At these observations a checklist will be used to confirm the Tai Chi instructors’ fidelity to the intervention protocol, and informal feedback from dyads and instructors will be sought at the end of each observation session and recorded using field notes. At around week 16 of the Tai Chi intervention, joint interviews will be conducted with a purposive sample of around 15 dyads in their homes and will focus on adherence to the intervention and in particular the home-based Tai Chi exercises.

The majority of measures will be taken at baseline and six months post-baseline (see Table 3). Weekly telephone calls with PWD for data collection will also serve as weekly contact to encourage retention. If one member of a dyad wishes to withdraw from the study, or is advised to withdraw by their general practitioner, they will be asked whether they are willing for data gathered up to the point of withdrawal to be retained for intention-to-treat analysis.

Continuing participation of the other member of the dyad will be sought. Participants will not be replaced. The choice of key outcome measures was informed by a search conducted on The COMET database (Core Outcome Measures in Effectiveness Trials; www.​comet-initiative.​org). Both dynamic and static balance have been shown to be important modifiable risk factors for falling among community-dwelling older PWD [35], and can be measured with valid and reliable tools in this patient group [36, 37]. Descriptions of the primary and secondary outcome measures are presented in Table 4.

Falls measurements will be collected prospectively from baseline until six months post-baseline. These include the number of falls by PWD (count data), the proportion of PWD who have fallen (binary outcome measure), and number of injurious falls (count data). The collection of falls data is in accordance with recommendations made by the Prevention of Falls Network Europe (ProFaNE) [38]. In our trial, the definition of a fall is, “an unexpected event in which the participants come to rest on the ground, floor or lower level” [38]. Falls will be recorded by dyads daily using prospective monthly calendars and returned on a monthly basis by post for data entry. Where falls are reported by dyads or calendars are not returned, a researcher will conduct a telephone interview to collect further information about the fall/clarify missing data. Telephone calls will be conducted weekly with the PWD as recommended for falls data collection with PWD [39]. In addition, to ascertain the accuracy of different recall periods, telephone calls about fall incidents by the PWD will be made monthly with the PWD and every 3 months with the carer. Fall injury will be recorded by telephone interview when recording falls as described above, using an existing set of definitions for severity of injury ([11, 40], p.).

Within the intervention group only, data will be collected half-way through the intervention on a 5-item questionnaire to ask about the importance of features of the intervention such as enjoyment of the classes and confidence in being able to continue to do the home practice. At the end of the final home visit, all dyads will be asked to complete a brief structured exit interview that asks about their hypothetical willingness to pay for the Tai Chi intervention should it be offered in the future as part of routine NHS care. The researcher will also ask participants if there have been any significant changes during the trial to their medical status (e.g. medication usage or elective surgery), which may influence performance on the primary outcome.

A formal sample size calculation was undertaken based on detecting a difference in mean timed up and go (TUG) test scores between the intervention and control arms of the study. We found no studies looking at minimum clinically important difference, but did find two studies that had estimated smallest detectable change (i.e. the smallest change that we can be reasonably sure is not measurement error). The two values were 4.09 [41] and 5.88 [37], and we used a conservative value of 4. To find an estimate of standard deviation (SD) for TUG, we found several relevant papers but the values were sensitive to the mix of age and severity of dementia, with one study that seemed to most closely match our population and be based on a reasonable sample size (n = 58) [37]. SDs from two different time points were presented (9.74 and 9.01) and we used the average of these (9.38) [37]. In our analysis we will enter baseline TUG as a covariate and assume a correlation of 0.7; test-retest reliability of TUG is excellent [37, 42] but we acknowledge the duration between time points in our study will be longer than prior studies and so we have used a more modest correlation.

Therefore, using a mean difference in TUG of 4, an SD of 9.38, a correlation of 0.7 and a 2-sided 5% significance level, the study will have 90% power when the sample size is 120 (60 per group). Allowing for up to 20% withdrawal / non-completion of outcome measures, we will recruit 150 dyads into the trial (75 per group). These calculations were conducted using nQuery Advisor 7.0. The intervention partly takes place in classes, and there is a possibility of clustering effects. However for the purposes of the sample size calculation we have assumed that these will be negligible, because (a) much of the intervention occurs outside the class environment, and (b) interaction between PWD is not a specific purpose of the intervention’s effect on the primary outcome (only to enhance adherence to the intervention). Clustering effects will be considered in the statistical analysis.

Recruitment of participants will utilise multiple pathways, including:

All potential participants will be asked if they are willing for their contact details to be provided to a member of the TACIT research team.

A member of the TACIT research team will contact potential participants to arrange an initial visit, typically at the potential participant’s home. The researcher will wait a minimum of 48 h before contacting potential participants to ensure they have had sufficient time to consider the information in the Participant Information Sheet. Both PWD and carer must be present at the initial visit and both must provide written informed consent. Researchers will use process consent with PWD, which is established in the evidence base [43‐45], in that consent will not be assumed for the duration of the visit or for the project after provision of initial informed consent. Informed consent will be re-confirmed verbally throughout the project.

After consent has been obtained, the researcher will administer the M-ACE to the PWD to confirm that they meet the entry criterion in relation to dementia severity. As version A is used in routine care, version B will be used at baseline and version C at follow-up (to reduce potential bias due to practice effects). Participants recruited from outside NHS / GP services will be required to sign a consent form to enable the research team to verify the PWD’s dementia diagnosis (type and when diagnosed) from their memory assessment service or provide written evidence of their diagnosis.

Baseline measures will be recorded at the PWD’s home before randomisation is performed. After the home visit, dyads will be informed in writing of their treatment arm allocation by staff at the CTU. Given the nature of the intervention, it will not be possible to keep PWD, carers or Tai Chi instructors blind to treatment arm allocation. The researcher who collects baseline data will be kept blind to treatment arm and, in particular, when they conduct the six-month post-baseline follow-up home visit. Participants will be asked at the time of arranging the follow-up visit and at the beginning of the follow-up home visit not to disclose their treatment condition, and to conceal any evidence of participating in class or home-based Tai Chi. After data collection is completed for each follow-up visit, the researcher will make a note of their guess of the treatment arm for the dyad and if dyads accidentally disclose their treatment condition.

Copies of the completed case report forms (CRFs), labelled with a unique trial number for each dyad member, will be sent to the CTU for double data entry into a web-based trial database designed and maintained by the CTU. The data will be exported for the statistician with treatment condition unblinded since information on class cohort is required for the analysis. Once the analysis is complete, the results will be revealed to the rest of the research team with trial arm identity concealed. After the results have been discussed and interpreted, the results will be unblinded for the research team to complete their interpretation and to begin dissemination of the findings.

A detailed statistical analysis plan will be finalised separately before follow-up data collection is complete. Participants will be analysed in the group they are randomised to, and (with the consent of participants) we will attempt to collect complete data on everyone and use those data in the analyses. Descriptive statistics will be presented for all baseline outcome data for each trial arm separately. Descriptive statistics will be presented on the following for the PWD and their carer: gender, age, relationship status, current living arrangements, education level, ethnicity, previous experience of Tai Chi and treatment site. Additional descriptive statistics will be presented in relation to PWD: type of dementia, time since diagnosis, other long term conditions, existing injuries, use of a walking aid, number of medications currently taken, falls history (in past 12 months and past month), and current level of physical activity (moderate and vigorous).

Mean TUG scores at 6 months, the primary outcome, will be compared between the two trial arms using a mixed (multi-level) model approach to take into account clustering resulting from the group-based nature of the Tai Chi intervention. Baseline TUG, treatment site, and falls history in the past 12 months (Y/N) will also be included in the model. We will assume that the missing data mechanism is “Missing at Random” (MAR). No imputation methods will be used for the main analysis but will be explored in additional analyses. The method of analysis will be similar for other secondary outcomes for the PWD and carer. In addition, falls count data and the proportion of participants who fell will be analysed using negative binomial and logistic models respectively, taking into account falls history and treatment site. We will also conduct a per protocol analysis that will exclude participants from the Tai Chi group if they received fewer than 50 h (the intended minimum dose of the intervention).

One subgroup analysis is pre-planned for the primary outcome variable (TUG). Fall history at baseline (had a fall or not in the past 12 months), which is also the variable used for minimization in addition to treatment condition, will be entered as an interaction term in a statistical model (intervention effect x effect modifier). Departures from the proposed analysis, for example due to poor recruitment, will be explained and justified in the main trial paper. Feasibility objectives will be addressed by estimating recruitment and attrition rates (with 95% confidence intervals) and falls data parameters (falls rate, proportion of fallers, and completeness of falls data) for a future sample size calculation. In addition, we will compare the frequency of falls reported by each of the individual methods of data collection and their combination to determine if the increased resource in collecting falls data weekly provides more accurate data.

Descriptive data from the structured exit interviews will be presented on hypothetical willingness to pay for the intervention and perceived changes to health (PWD and carer). Data on number of Tai Chi classes attended and adherence to home practice will be presented. To assess adherence to the intervention in the Tai Chi arm we will calculate: adherence to classes (i.e. x attended out of y possible) and adherence to home exercise (i.e. x minutes exercise out of y recommended). The following data will not be reported in the main trial outcome paper but will be included in further secondary analyses in relation to visual spatial cognitive functioning (statue task); comparison of the different methods of data collection on incidence of falls; qualitative data; and uptake and adherence to the intervention (whether the PWD or carer had done Tai Chi before, mean (SD) intention and perceived behavioural control to do Tai Chi, and intervention group responses to the questionnaire delivered half way through the intervention).

Given the short-term (6-month) follow-up, low-risk intervention being tested, and no planned interim analyses, there will be no data monitoring committee for the trial. The trial will only be stopped early if the intervention is considered to be directly causing undue harm to patients as independently considered by the trial steering committee and/or NHS Research Ethics Committee.

We will assess the feasibility of collecting the data required for a health economic analysis. This will include detailed descriptive statistics on completion of the health service use telephone interviews, which will include social care costs and costs incurred by dyads. Health service use costs will include both those from falls (both intervention arms) and in relation to injury while carrying out Tai Chi. The total cost of providing the intervention to each patient will be estimated from weekly records collected from the Tai Chi instructor and in relation to: (a) the average cost of hire of the building for the classes (taken from the costs incurred in this study), and (b) the instructor’s time spent delivering the intervention as a function of hourly rate. Health service use on falls will be calculated in relation to presentation to the GP, out of hours, or ED. We will also explore cost effectiveness results for PWD and their carers between arms in terms of incremental cost per change in quality of life. All health economic analyses will be presented descriptively in terms of completion (missing data) and outcomes (e.g. mean and SD in relation to baseline and follow-up), as the study has not been powered to conduct a full health economic analysis, which is reserved for the subsequent definitive trial.

The monitoring of participants’ safety will be achieved primarily through weekly telephone contacts between unblinded researchers and dyads and dyads’ weekly record keeping. Participation in the trial is considered to be of relatively low risk and participants are not expected to be harmed as a result of taking part. As such, the recording and reporting of adverse events will be limited to adverse events related to the Tai Chi intervention and balance test assessments. No treatment will be provided to participants in addition to NHS usual care.

The TSC will oversee the conduct and safety of the trial, ensuring that milestones are achieved and general scientific probity is maintained. The TSC will meet by teleconference on a bi-annual basis. The TSC will review progress against key milestones and reports of aggregate data (blinded) with a particular focus on adverse events. The TSC will be independently chaired and include independent members from the trial that represent expertise in statistics, clinical care of older people, relevant research knowledge and experience, and representation of older people with dementia and their informal carers.

Anonymised electronic records and paper copies of records will be open to inspection and monitoring from a recognised representative from either the Sponsor, Bournemouth University, the CTU, or the funder (National Institute for Health Research). The trial management group will have access to the full dataset. Other interested parties may make a formal request to access the electronic dataset, which will be approved / declined by the CI in accordance with the Data Management Plan that will detail management of access, sharing, and preservation of the data. Any use of the electronic data set must comply with the dissemination policy (see below) and be requested via Bournemouth University Library (bordar@bournemouth.ac.uk) who will collaborate with the CI with regards to access. Non-digital data supporting this study will be stored by the corresponding author at Bournemouth University. Only electronic data will be shared with bona fide researchers intending to use the data for non-commercial research purposes, after an embargo period of approximately 24 months. Access to the following will be restricted to researchers who sign a confidentiality agreement and confirm their intention to use the data is for secondary data analysis for non-commercial research purposes using a Creative Commons licence: statistical analysis plan; where applicable, statistical code (for final analysis of primary outcome measure); and anonymised participant-level dataset and data documentation.

We intend to notify participants involved in the trial of the study findings through our annual public engagement events and a newsletter on study completion. We will disseminate to researchers and healthcare professionals through conference presentations and scientific journal publications, and the general public through our webpage for the trial (www.​bournemouth.​ac.​uk/​tai-chi). In all dissemination, no professional writers will be used. To establish authorship eligibility, we will use the uniform requirements for manuscripts submitted to biomedical journals established by the International Committee of Medical Journal Editors (ICMJE; www.​icmje.​org).