Background

Dementia is a syndrome marked by cognitive and functional decline, associated with over 100 diseases [1]. Approximately 55 million people live with dementia worldwide, and each year the number of new cases increases by nearly 10 million [2]. Mild cognitive impairment (MCI) is conceptualised as the prodromal phase for dementia, with ~ 10–15% of individuals with MCI converting to dementia annually [3]. There is no cure for dementia and limited symptomatic relief in the short-term, thus, delaying deterioration and supporting wellbeing is imperative.

Yoga is a system of mind-body practices that includes gentle movements or postures (asanas), breathing (pranayama) and relaxation techniques, reciting mantras, visualisations, and meditations, all of which can be adapted to suit practitioner ability [4]. For example, chair-based yoga has been successfully applied in geriatric [5] and palliative care settings [6]. Systematic reviews of yoga-based interventions in cognitively healthy older adults reported improvements in muscle strength [7], balance and mobility [6, 8], cardiovascular health [9], sleep quality [10], mental wellbeing [11], and quality of life [7, 12]. Yoga’s potential to support the functional independence and psychosocial wellbeing in older people has resulted in its application in long-term care settings [13] and its expansion to older people with cognitive deficits [14]. A scoping review of yoga and mindfulness-based interventions identified in PubMed highlighted benefits for people in the early stages of cognitive decline, but as noted by the authors, this literature base was limited [15]. Following our own scoping review across five major databases to inform the present study (see Eligibility Criteria in Methods), we found that the evidence for yoga studies involving people with dementia has not been reviewed systematically. We thus aimed to fill this literature gap.

As interest regarding yoga’s benefits for people with neurocognitive disorders continues to grow [16,17,18,19], an evaluation of its efficacy and safety is required. This systematic review aimed to examine the research question “what are the study characteristics, cognitive, mental, and physical health effects, and safety of yoga-based interventions in people with MCI or dementia”. The findings can inform future interventions and provide guidance to practitioners to maximise their use in clinical care settings.

Methods

This systematic review was registered with the PROSPERO international database on 22 September 2021 (#CRD42021217969) and follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for transparent and objective reporting [20].

Eligibility criteria

A scoping review assessed the existing literature in Cochrane Central Register of Controlled Trials (CENTRAL), EBSCOHost (indexing CINAHL Plus, PsycINFO, Psychology and Behavioural Sciences Collection), and PubMed and identified the population, intervention, comparisons, outcome, and study design principles used to define the eligibility criteria.

  • Population: People with MCI or dementia.

  • Intervention: Yoga as primary intervention.

  • Comparisons: Outcome changes over time (pre-post intervention) in the same sample of participants or between-group changes relative to a control.

  • Outcomes: Cognitive, mental, or physical health or safety.

  • Study designs: Repeated measures design of ≥4 weeks in duration.

Journal articles published in English were eligible for inclusion without limitation on publication year. As quantitative data collection may not be feasible in the later stages of dementia [21], studies using qualitative assessments were also included. Articles that did not meet the inclusion criteria were excluded. Multi-modal and mindfulness-based interventions were also excluded as they had been reviewed elsewhere [15].

Systematic search strategy

Literature searches were performed in CENTRAL, EBSCOHost (indexing CINAHL Plus, PsycINFO, Psychology and Behavioural Sciences Collection), and PubMed using terms related to yoga and dementia (see Additional file 1 for PubMed strategy which was adapted for other databases). Searches were conducted from the databases’ date of inception through to 4 September 2020, after which alerts were activated to capture articles published until article submission on 31 March 2022. Additional studies were identified through secondary citation searching of the reference lists of relevant articles and reviews.

Study selection and data extraction

Two researchers (DK, MdM) independently screened titles/abstracts before obtaining full texts. Study data were independently extracted from included articles by two researchers (DK, TT): authors, location, study population, mean age, female representation, sample sizes, demographics, study design, assessment timepoints, intervention characteristics (type, frequency, duration, supervision), comparison characteristics (if relevant), outcome measures and description of effects. Any disagreement about study selection or data extraction was resolved by discussion with another researcher (DC).

Data synthesis

Study characteristics and primary findings were extracted and presented in relevant tables (see Tables 1 and 2) as the substantial heterogeneity across the included studies (in population, intervention characteristics, and study quality) precluded meta-analysis.

Table 1 Study characteristics of yoga interventions involving people with MCI and dementia
Full size table
Table 2 Pre/post-intervention outcome measures and results of yoga interventions involving people with MCI and dementia. Within-group findings are narratively discussed first, followed by between-group outcomes (if relevant)
Full size table

Risk of bias

An 11-item tool, based on the Cochrane Handbook [36], was developed to assess risk of bias (RoB) in the included studies (see Additional file 2). The tool captured bias in sampling, random sequence generation, blinding, intervention description, incomplete outcome handling, selective reporting, adjustment for confounders, contamination, validity and reliability of outcome measures, statistical power, and protocol compliance [36]. Each item was rated as yes and assigned a score of 1 or rated as no/unsure and scored as 0. Higher scores indicated lower RoB; total scores ≥10 were considered to have low RoB. Multi-paper studies were reviewed as a single unit, such that all related publications were assessed for each criterion and a yes rating was provided if any associated paper met a particular criterion. Two researchers (DCM, DK) independently evaluated RoB and any discrepancies were resolved with a third researcher (AS). Evidence strength was evaluated qualitatively based on the RoB for included studies.

Results

Study selection

Figure 1 depicts the study search and selection process. Of the 1, 431 articles identified, fourteen met the review inclusion criteria, representing 10 unique studies. Reasons for article exclusion are provided in Fig. 1.

Fig. 1
figure 1

Flow diagram of the study selection process

Full size image

Study characteristics

Table 1 details the characteristics of the 10 included studies. There were three papers with different outcomes from the one trial [22,23,24], four articles that reported quantitative and qualitative intervention data [30, 31, 33, 34], and the remaining seven articles reported outcomes from unique interventions [25,26,27,28,29, 32, 35]. The multi-paper studies were grouped together in the reporting of study characteristics and outcomes in Tables 1 and 2, respectively. Publication years ranged 2010–2020.

Five studies used a randomised controlled design [22,23,24,25, 27, 29, 33, 34]. Of these, one trial compared yoga to an active control group with cognitive training [22,23,24], two trials used passive control groups of music listening [27] or continuing with usual care [29], and another two studies involved two intervention groups (yoga and structured exercise) and a passive control group with music listening [33, 34] or refraining from any complementary exercise [25]. An additional intervention involving cognitive activities was implemented across all participant groups in one study [25]. The remining studies included one nonrandomised controlled trial [28] and four nonrandomised pre/post-intervention trials [26, 30,31,32, 35].

Study participants

The total sample size across the 10 unique studies was 421. Individual sample sizes ranged 9–81 participants. The mean age of participants was 70.6 ± 6.9 years, although three publications [30,31,32] reported age range only; these were excluded from the calculation.

Interventions involving people with dementia were most common (n = 5), while two studies recruited MCI participants only [22,23,24,25], another two involved mixed cohorts of people with MCI or subjective cognitive decline (SCD) [27], or their caregivers [26], and one study recruited people with MCI, SCD, and Alzheimer’s Disease (AD) [28]. The five dementia studies [29,30,31,32,33,34,35] examined 199 participants, whilst the two MCI only studies analysed 130 participants [22,23,24,25]. Participant characteristics of the remaining studies involving mixed cohorts can be viewed in Table 1.

Intervention design

Intervention length ranged 8–16 weeks, with 12 weeks (n = 5) being the most common, and frequency ranged from 1 to 7 sessions per week with two sessions being the most common (n = 5). The style of yoga varied with Kundalini Yoga (KY) [22,23,24, 26,27,28] and chair-based yoga (CY) [30,31,32,33,34] equally represented in four unique studies each and Hatha yoga applied in two studies [29, 35].

Yoga’s impact on cognitive, mental, and physical health

Table 2 displays each study’s outcome measures and intervention effects. Cognition was assessed in 7 studies using a variety of quantitative measures [22,23,24,25,26, 28, 30, 35]; the MMSE was most used [22,23,24,25, 28, 35], followed by the Trails B test [22, 28]. Four publications reported qualitative assessments of cognition with self-reports by the participants [27] or caregivers [34] only, a combination of researcher and caregiver observations [31], and both participant and caregiver reports during an exit interview [35]. In terms of yoga’s effect on cognition, three interventions reported improvements in samples of people with MCI [22, 23], their caregivers [26], and people with MCI, SCD or AD [28]; namely on tests of memory function. Three qualitative assessments reported cognition-related benefits in cohorts involving people with SCD and MCI [27], and participants with dementia [34, 35]. The remaining four publications reported non-significant effects [24, 25, 30, 31].

A range of mental health domains were examined across 10 publications [22, 25,26,27, 29,30,31, 33,34,35]. The most common quantitative measures used were the GDS for mood [22, 35] and the PSQI for sleep [25, 33]. Qualitative assessments were reported in three publications using caregiver observations [31, 34] and participant and caregiver exit interviews [35]. Regarding yoga’s effects, four quantitative studies reported mental health improvements in cohorts involving people with MCI [22], their caregivers [26], or people with dementia [29, 30], while two studies reported non-significant effects in women with MCI [25] and people with dementia [35]. One study reported a worsening in agitation in people with dementia [33]. Qualitative assessments identified mental health benefits in cohorts of people with MCI or SCD [27], and dementia [34], while one AD study did not observe mental health-related changes [35].

Various physical health domains were assessed across 6 studies using a range of measures. Balance was most examined with the BBS [30, 32], one-leg standing test [29], and 8-ft up & go test [25]. Blood pressure [17, 32], cardiopulmonary fitness through the 2-minute step test [25, 29], and body composition through BMI [29, 33] were also commonly measured. Qualitative physical health assessments were reported in two publications through researcher and caregiver observations [31] or with caregivers only [34]. Yoga’s effects on physical health included balance enhancements in women with MCI [25] and people with dementia [29, 34], but non-significant effects in two dementia studies [30,31,32]. Blood pressure improved in people with MCI and their caregivers [26], and people with dementia [29]. Cardiopulmonary fitness also improved in people with dementia [29] and women with MCI [25], but this latter finding was apparent in both intervention groups. Body composition effects in dementia studies were mixed with one study reporting improvements [29] and another identifying non-significant changes [33]. One qualitative dementia study observed flexibility and strength benefits [31].

Safety, withdrawal, and compliance

Four studies assessed adverse effects [22,23,24, 28, 29, 33] and most reported zero safety events [28, 29, 33]. One reported yoga-related dizziness that led to a participant’s withdrawal [22]. Most studies (7/10) adequately reported on withdrawals and provided reasons [22, 26,27,28, 30, 32, 33]. The remaining studies reported drop-out rates but did not elaborate on reasons. Compliance was satisfactorily measured in four studies [27, 28, 33, 35] with intervention adherence rates ranging 73.5–93.0%. Two publications reported on the handling of protocol non-compliant participants [30, 31], and the remaining studies did not report on compliance at all.

Study quality

Table 3 depicts the RoB judgements for each study, noting that none had low RoB (i.e., ≥ 10). Three unique interventions received a moderate rating [16, 29, 30, 33, 37, 38] and the remaining studies received high RoB ratings. All articles had valid and reliable outcome measures (item 9). Although all were free of suggestion of selective outcome reporting (item 6), only one intervention protocol was pre-registered [16, 29, 30]. Most studies sufficiently described the yoga intervention to allow identification and replication of the key components (item 4; 8/10 studies), and clearly described participants and eligibility criteria (item 1; 6/10 interventions). Where relevant, about half of the articles appropriately adjusted for confounders and outliers (item 7; 4/7 interventions), and adequately described randomisation methods (item 2; 3/6 interventions). Most studies reported on participants’ compliance to the protocol (item 11; 6/10 studies). Where applicable, only 4 studies had blinded outcome assessments and analyses (item 3) or conducted intention to treat and/or sensitivity analyses (item 5). No studies were adequately protected against contamination from other interventions (item 8) or were adequately powered to detect hypothesised changes (item 10).

Table 3 Risk of bias (RoB) ratings for included studies
Full size table

Discussion

Yoga-based intervention studies involving people with MCI or dementia were summarised and critically evaluated for effects on cognitive, mental, and physical health here. The fourteen included articles were published from 2010 onwards, highlighting the emerging scientific interest in this field. Most studies focused on participants with dementia [17, 18, 35, 37,38,39,40], used Kundalini Yoga [16, 29, 30, 32,33,34] or chair-based yoga [18, 31, 35, 37,38,39] in their intervention, were 12 weeks in duration [16, 17, 31, 33, 37], assessed cognition with the MMSE [30, 31, 34, 40], used the GDS to assess depression [29, 32, 40], and examined physical heath in the context of balance [17, 18, 31, 39]. The proceeding section summarises the effects and safety of these yoga-based interventions, the quality of these studies, recommendations for future research, and implications for practitioners in clinical settings.

Cognitive health

In line with another scoping review [15], we found most studies reported cognitive benefits in people with MCI or dementia [16, 29, 32,33,34, 38, 40], but only three of these employed a RCT design [16, 29, 33, 38] – the gold standard for effectiveness research. Three of the four studies using the MMSE reported non-significant changes with yoga [30, 31, 40]. This is important, as global cognition improvements have been reported with physical activity interventions [39].

Yoga may be associated with domain specific improvements in cognition. Here, we found that yoga interventions enhanced executive function [16, 34], visual [29] and semantic memory [34]. These domain specific improvements align with physical activity interventions applied in MCI and dementia, also showing enhancements in executive function (and other domains including processing speed) using a range of neuropsychometric measures [37, 38]. Other studies assessing the efficacy of components of yoga, such as meditation, have also shown improvements in attention and verbal fluency [40]. Potential mechanisms underpinning these domain-specific improvements may be neuroplastic changes in the hippocampus and widespread executive function networks including prefrontal hubs [41].

Mental health

Previous systematic reviews of yoga-based interventions in cognitively healthy older adults reported improvements in sleep quality [10], mental wellbeing [11], and quality of life [7, 12]. Similarly, the most common mental health improvements here from yoga-based interventions for individuals with MCI, dementia, or their caregivers were mood [29, 32, 35, 38, 40] and sleep quality [31, 37]; though two studies reported non-significant effects [31, 40] and one reported worsening agitation [37]. Further research is needed to resolve inconsistencies. While well-validated mental health outcome measures were used, studies relied heavily on self-reports that may be biased or less accurate in populations where cognitive capacity is reduced. Future studies might complement self-report with clinical interviews and/or objective measures such as polysomnography or actigraphy for sleep. Furthermore, future studies should compare yoga-based interventions to active control, evidence-based, first-line mental health interventions like cognitive behavioural therapy to determine their efficacy relative to existing therapies.

Physical health

Yoga has previously been reported to improve an array of physical outcomes including muscular strength, cardiorespiratory fitness, balance, and flexibility [42]. While certain yoga studies included here demonstrated that populations with MCI or dementia may experience improvements across various domains of physical health [25, 29, 30], the findings were not uniform as non-significant changes were also reported [31,32,33]. Furthermore, of the three studies that compared the effects of yoga to a control group [25, 29, 33], only two reported significant balance, muscular strength and cardiorespiratory fitness improvements [25, 29]. Whether yoga-based interventions are superior to other more established lifestyle therapies such as aerobic or resistance training is unclear [43]. Studies incorporating adequate comparator and control groups, sufficient sample size, and gold-standard measurements of physical health and fitness are required to determine the efficacy of yoga-based interventions in populations with MCI or dementia.

Safety of yoga

Across the seven interventions where adverse effects and withdrawal reasons were reported, yoga was considered relatively safe. Only one instance of yoga-related dizziness was identified [16], flagging considerations for falls and/or injury risk in a population that experiences neurological issues.

Study quality

Despite the domain-specific cognitive, mental, and physical health benefits identified, the RoB across these studies was high. These outcomes must therefore be treated with caution. Inadequate powering and insufficient reporting regarding contamination from other interventions, withdrawals, and compliance mars the safety and effectiveness data. These issues can be overcome by registering or publishing trial protocols, a practice that is essential in pharmacological trials and is increasingly being recommended for nonpharmacological interventions. Lastly, caution must be applied to the interpretation of effects in three studies that did not account for the variability in disease severity [33, 34] or cognitive status [32]. Future studies involving a mix of participants should adequately control for these factors. Although these methodological limitations hinder evidence certainty, they serve as key recommendations to improve study conduct and reporting quality.

Review limitations

Our comprehensive approach to reviewing quantitative and qualitative yoga studies limited our ability for quantitative data synthesis. There was a high degree of variability in the yoga intervention protocols, populations studied, and outcome measures used, as well as small sample sizes. The development of a RoB tool, although based on the Cochrane Handbook, also limited quality assessments. However, with varying study designs included, a single tool to evaluate these was deemed more efficient than the use of specific tools for each design (e.g., RoB v2 for RCTs and ROBINS-I for non-randomised trials). Further, the RoB criteria were appropriately modified to judge intervention qualities (e.g., see RoB items 4, 7 and 8 in Additional file 2). As research into the benefits of yoga for people with MCI and dementia continues, more thorough reviews will be required.

Implications and conclusion

In an emerging field of interest, these preliminary studies show that yoga may be safe and beneficial for the wellbeing of people with MCI or dementia. From a clinical perspective, it is recommended that yoga practitioners seeking to apply or recommend this complementary therapy follow the protocols described in these studies and undertake dementia awareness or competency training to appropriately facilitate sessions. This is especially important when applying person-centred care and adapting the exercises to suit and meet the needs of the person living with cognitive decline. Health professionals may also advise patients to engage in yoga with qualified practitioners to manage their wellbeing and ensure their safety throughout the classes. From a research perspective, the scientific rigour of this field must improve with more high quality RCTs that are designed to minimise bias [36] and reported according to the Consolidated Standards of Reporting Trials guidelines [44].