Effects of Movement-Based Mind-Body Interventions on Physical Fitness in Healthy Older Adults: A Meta-Analytical Review

Life expectancy has notably increased in the 21st century as a result of societal and health care changes [1]. To guarantee an independent living and a high quality of life in old age, the maintenance of healthy physiological functioning is considered of utmost importance. Aging induces declines of strength, flexibility, endurance, mobility, balance, and coordination performance and thus results in lower physical activity levels of older adults that hamper healthy aging with notable effects on morbidity and mortality [2]. This vicious circle can be cut through by physical activity and exercise training that specifically promotes the components of intact physical functioning [3, 4]. Innovative and integrative interventions based on exercises and tailored to the needs of older adults are required [5]. Those interventions should motivate older persons to engage in physical activity in the long term, considering the main types of training/strain (balance, endurance, strength, flexibility, and coordination) and to rather delay clinical limitations (e.g., diseases, falls, and hospital administration).

In this regard, mind-body interventions (MBIs) have already proven to be a suitable approach in improving physical functioning of older adults [4, 5]. MBIs have been described by the National Centers for Complementary and Integrative Health, as interventions which aim to unite the body and mind to positively influence physical function, psychological status, and symptoms [6]. This integrative framework mainly targets relaxation techniques, meditation, and breathing exercises with structured movements [6]. Prominent agents of these movement-based MBIs are Yoga, Qi Gong, Tai Chi [7], and Pilates [8]. Tai Chi and Qi Gong have become increasingly popular within the MBIs although the effects, including breathing, relaxation, and movement control, are minor [9]. The respective movements of the MBIs are mainly performed in easy to more complex exercise sequences [10]. Studies on Tai Chi observed positive effects on flexibility, balance, and gait in older adults [9]. Yoga studies revealed improvements in numerous physiological and functional parameters such as walking, balance, muscle strength, cardiovascular health, and blood pressure [11]. On the other hand, Pilates is a relatively new and underexplored mode of MBIs. It has been estimated that 80% of all available Pilates studies were conducted between 2010 and 2014 [4]. Pilates follows the key principles of muscle strengthening, especially the lumbar-pelvic region, dynamic stretching, body control, flow of movements, concentration, and breathing [12] and can therefore be classified as MBI [13]. The aspects of all MBIs can be considered integratively with multimodal impact on balance, certain strength aspects, and functional mobility [4, 9-11, 14].

Several studies have already been conducted on the effects of MBIs on health-related outcomes. However, compelling and comprehensive evidence pooling of available MBI findings and examining its effects on a variety of relevant physical dimensions in older adults are still lacking. Therefore, the present meta-analytical review focuses on different types of movement-based MBIs (Yoga, Qi Gong, Tai Chi, and Pilates) and their physiological effects compared to both active control (AC) and inactive control (IC) conditions in healthy older community dwellers. Our work may contribute to a better understanding of the occurrence and magnitude of MBI-induced effects on older adults’ physical functioning and thus to better tailor and adapt MBIs in order to improve a broad range of health-related physical function indices.

This meta-analytical review was performed in accordance with the PRISMA guidelines [15]. The literature search was independently conducted by two researchers (S.A.E. and L.M.) until 23 November 2019 in three health-related, biomedical, and sports-related databases (Web of Knowledge, PubMed, and SPORTDiscus). Relevant search terms (operators) were combined with Boolean conjunctions (OR/AND) and applied on the following three search levels (see Table 1). Citation tracking of the articles as well as hand searching (e.g., within all reference lists of the articles) for important primary articles and review articles were additionally carried out. After removing duplicates, the remaining studies underwent a manual screening process. Three search levels of screening (1) title, (2) abstract, and (3) full text were applied. Irrelevant articles, which had no title at all (e.g., numbers and symbols), remaining duplicates that have not been automatically detected, pilot or preliminary studies, and reviews or surveys were consensually excluded. According to the criteria mentioned below, two independent researchers (S.A.E. and L.M.) made a consensual final inclusion or exclusion decision. If a deviation between both examiners occurred, the title was rechecked and further abstract screening was conducted.

The following inclusion criteria were applied:

• Full-text article published in English in a peer-reviewed journal.

• Randomized controlled trial (RCT) intervention study with pre-post testing.

• At least one control group that either did not receive any active physical and functional exercise-based intervention (IC) and/or received an alternative exercise-based training program (AC).

• Yoga, Tai Chi, Qi Gong, and Pilates served as target interventions.

• Studies that included healthy community-dwelling older adults without neurological, orthopedic, and/or cardiac conditions.

• Exercise intervention program focused on improvements of physical fitness indices.

• Participants were at least 65 years of age (mean age of the given study sample was considered).

• Intervention duration of at least 4 weeks and/or 2 training sessions per week, and the studies must be published after 2000 so that the current study situation can be considered.

The methodological quality of the included RCTs was rated using the PEDro (Physiotherapy Evidence Database) scale [16]. This scale consists of 11 dichotomous items (either yes or no). The first item (indicating inclusion and exclusion criteria) is not included in the sum score. Studies were rated by two reviewers independently (S.A.E. and L.M.). After completing the evaluation, both examiners came to a consensus on every item by discussion. The raters were not blinded to study authors, place of publication, or results.

Physical fitness indices were divided into the following categories (subgroups): strength (lower and upper body), static balance, functional mobility, gait, endurance, and flexibility. Data were extracted by two researchers (S.A.E. and L.M.). Data were then transferred to an excel spreadsheet. Relevant study information regarding author, year, population, number and age of participants, AC condition and IC condition, intervention design, and training conditions was extracted. All outcome measures and the PEDro score of each study were listed (see Table 2).

Standardized mean differences (SMD, with 90% confidence intervals) were computed separately for each study. Therefore, the difference of the target outcome between the intervention and the respective IC and AC condition including the pooled standard deviations was computed for both each category and subgroup. Standard errors were transformed into standard deviations by multiplying standard errors by the square root of the sample size. Therefore, Hedges g adjusted for small samples has been calculated. If studies only presented means and standard deviations in figures, the corresponding authors were contacted to provide the relevant means and standard deviations. If not available, the WebPlotDigitizer Version 4 (Free Software Foundation, Boston, MA, USA) was used to extract means with standard deviations. Negative effects were symbolized with a minus sign. The Cochrane Review Manager Software (RevMan 5.3; Cochrane Collaboration, Oxford, UK) was used to compute the effect sizes using an inverse-variance approach [46] with a random-effects model [47]. Forest plots were generated for the respective outcome measures. Thereby, the magnitude of SMD was classified according to the following scale: 0–0.19 = negligible effect, 0.20–0.49 = small effect, 0.50–0.79 = moderate effect, and 0.80 = large effect [48]. To examine a potential publication bias, a funnel plot was created and qualitatively evaluated. Statistical analyses were performed using the software Cochrane Review Manager Software (RevMan 5.3; Cochrane Collaboration, Oxford, UK). A p value <0.05 was seen as statistically significant.

In total, 2,597 potentially relevant articles were found (see Fig. 1). After removing duplicates, 2,055 article titles remained for a careful title screening. The abstracts of the remaining 608 potentially relevant articles were thoroughly studied. One-hundred nine full texts were further perused of which 79 did not meet inclusion criteria. Thirty studies were finally included in the quantitative meta-analysis. All studies were published after 2000.

2,792 healthy community dwellers with a mean age of 71.2 ± 4.7 years have been enrolled in the 30 included trials. Eleven studies included only female participants [17-26, 49]. The mean sample size was 96 ± 128 ranging from 24 [27] to 684 [28] including older participants. Trials comprised various study arms: MBIs (n = 1,177), IC condition (total number of participants, n = 1,163), and AC condition which included Pilates plus inspiratory muscle training (n = 11), traditional balance training (n = 43), low-impact exercise (n = 28), unstable support surface exercises (n = 20), traditional exercises (n = 11), Tai Yoga (n = 26), Otago exercises (n = 24), Tai Chi 1 time per week (n = 233), Western exercises (n = 39), and Huber training (n = 17). In total, we included 11 Pilates and 19 Tai Chi study interventions. All studies applied a standard randomization procedure for group assignment. Seven out of the 30 studies used a 3-armed design [9, 18, 28-32]. The study quality (PEDro score) ranged from 3 [20] to 9 [17, 33-35, 49] (see Table 3). Thirteen studies blinded the testing personnel [9, 17, 22, 27, 28, 32-37, 49]. Five studies with a PEDro score between 3 and 7 did not report inclusion and exclusion criteria [19, 20, 22, 38, 39]. Training intervention duration ranged from 8 [19, 20, 22, 25, 26, 29, 33, 39] to 48 weeks [35], with most interventions lasting 8 or 12 weeks [17, 19-27, 29-33, 36, 38-41, 49] (n = 22). Training frequencies varied between 2 and 7 times per week. More than half of the trials used a training frequency of 2 times per week (n = 21). Training session duration was between 30 and 90 min. Roughly, half of the trials conducted a 60-min workout (n = 17). When Tai Chi was indicated, Tai Chi Chuan, Sun style, and Yang style were subsumed under the umbrella term “Tai Chi” [23, 26, 32, 33, 38, 39, 42, 49].

The majority of the comparisons did not reveal a clear or trending funnel shape (see Fig. 2, 3) indicating that small and large studies with different effect and sample sizes are missing among all trials and outcomes. Thus, a tendency of publication bias can be assumed across the majority of all computed comparisons. Only balance, functional mobility, and endurance depict a slight funnel shape for the IC comparison (Fig. 2b, c, e) and strength, balance, and functional mobility in the AC comparison (Fig. 3a–c). Moreover, few data points are available in the majority of the AC comparisons.

Highly significant and large overall effects were observed for total strength (p = 0.001, SMD: 0.87 [90% CI: 0.43, 1.30], I² = 94%; Fig. 4). A moderate to large and significant effect was found for upper body strength (p = 0.001, SMD: 0.79 [90% CI: 0.39, 1.19], I² = 82%; Fig. 4) in favor of MBI compared to IC. With respect to lower body strength, a large and significant effect (p = 0.04, SMD: 0.91 [90% CI: 0.18, 1.64], I² = 96%; Fig. 4) in favor of MBI compared to IC was observed.

Small overall but significant effects with comparatively narrow confidence limits but large heterogeneity were found for static balance performance (p = 0.02, SMD: 0.37 [90% CI: 0.10, 0.63], I² = 79%; Fig. 5) in favor of MBI compared to IC.

Moderate and significant overall effects with narrow confidence limits and large heterogeneity were observed for functional mobility (p = 0.009, SMD: 0.55 [90% CI: 0.20, 0.89], I² = 83%; Fig. 6) in favor of MBI compared to IC.

With very large heterogeneity, moderate but insignificant effects were found for gait performance (p = 0.22, Fig. 7) favoring MBI compared to IC.

Small to moderate but significant and very homogenous overall effects with narrow confidence limits were found for endurance performance (p < 0.001, SMD: 0.44 [90% CI: 0.25, 0.62], I² = 0%; Fig. 8) in favor of MBI compared to IC.

Relevantly small to moderate and mildly heterogeneous overall effects with narrow confidence limits were found for flexibility performance (p = 0.003, SMD: 0.46 [90% CI: 0.21, 0.72], I² = 54%; Fig. 9) in favor of MBI compared to IC.

The overall effect relevantly favored AC at borderline significance level with, however, trivial effect sizes. Trivial and insignificant effects with acceptable heterogeneity were found for the upper body and lower body strength performance indices (p = 0.08, Fig. 10) favoring AC compared to MBI.

Small to moderate but significant overall effects with narrow confidence limits and moderate to large heterogeneity were observed for static balance performance (p = 0.01, SMD: 0.46 [90% CI: 0.16, 0.76], I² = 73%; Fig. 11) in favor of MBI compared to AC.

No effects in favor of either training arm with large heterogeneity were found for functional mobility performance (p = 0.69, Fig. 12).

No effects in favor of either training condition were found for gait performance (p = 0.89, Fig. 13). Large underlying heterogeneity was observed.

No effects were found for endurance performance (p = 0.81, Fig. 14).

No effects were found for flexibility performance (p = 0.85, Fig. 15).

This is the first meta-analytical review that comparatively investigated the effects of movement-based MBIs compared to both AC and IC conditions on different dimensions of physical functioning in healthy community-dwelling older adults. This review mainly provides the evaluation of two different forms of MBIs (Pilates and Tai Chi) as only these studies met the inclusion criteria. Our meta-analytical review revealed pronounced effects in favor of MBIs for upper and lower body strength, static balance, functional mobility, endurance, and flexibility compared to the ICs and for static balance compared to the ACs. Only regarding strength indices, the AC conditions were slightly favored compared to MBI. Compared to being inactive, it can be concluded that MBIs can be regarded as a multimodal, time-saving, integrative, and functional training strategy that trains relevant domains of physical functioning such as strength, balance, mobility, endurance, or flexibility.

Significantly small to large overall effects in favor of MBI for the majority of major physical fitness components, such as upper and lower body strength, static balance, functional mobility, endurance, and flexibility, have been observed. Performance improvements of upper and lower body strength in favor of MBI are plausible as exercise-based MBIs include various movement patterns that require adequate lower and upper body strength levels [50]. These findings are in line with previous studies on multimodal resistance training compared to IC conditions [51, 52]. Static balance performance showed only small overall effects compared to moderate strength performance effects. This finding seems reasonable as balance training adaptations are considered highly task specific, and mere static balance tasks are not well targeted and reflected in MBI approaches [53]. In contrast to that, a previous meta-analysis showed that MBIs with a main focus on Pilates can reveal positive effects on static balance in older subjects, when balance is incorporated, and seniors have less initial balance performance [54]. Two further meta-analyses revealed positive effects on static balance performance following Tai Chi interventions [43, 55], which comprise a variety of different static balance conditions. Functional mobility performance showed overall moderate effects with large effects in some selected studies. Thus, MBI has a huge potential to improve functional mobility performance. In this regard, the multimodal character of MBIs proves to be useful and beneficial to elicit relevant improvement of health-related physical domains in general and strength and functional mobility in particular. It seems furthermore reasonable to assume that an increase in lower body strength supports the effects on functional mobility. This assumption is underpinned by findings of a study conducted by Chen et al. [56]. Although MBIs are not primarily designed for improvements in endurance performance, we found at least small but relevant effects in our meta-analysis. These relevant endurance effects can already be with low-impact exercises, as already shown by Hopkins and colleagues [57]. In our meta-analysis, flexibility performance has also slightly benefitted from MBI and showed small but meaningful overall effects in favor of MBI. As Pilates training reflects stretching components [58], this finding can be attributed to the task-specific application of some Pilates exercises [13, 59]. This also applies for Tai Chi: besides the postural aspects when performing different poses or during a change between poses, large functional strength and flexibility components are considered. Accordingly, the targeted tensing of muscles and relaxing, the control of the target muscles, and the correct alignment of body segments seem crucial [60]. Therefore, it is assumed that especially the task-specific components in the Pilates interventions [13] and the flowing changes between different movements and poses of Tai Chi exercises [41, 61] mainly account for the positive effects on flexibility. Gait performance was not affected, as gait is considered a robust pattern [62] that less likely benefits from MBI. A recent study showed that either a multicomponent training program or interval-walking training can improve gait parameters over a short time (Two weeks). These results also indicate that the type of training is essential in terms of intended adaptations [63]. Accordingly, the lack of gait adaptations can be at least partly explained by the fact that Pilates and Tai Chi are less gait-specific training approaches. Overall, MBIs provide multimodal neuromuscular and cardiocirculatory training stimuli that improve the majority of health-related physical fitness parameters. Therefore, MBI serves as a less time-consuming training approach in older adults that covers a wide spectrum of training adaptations.

For MBI versus an AC condition, only static balance performance showed small overall beneficial effects in favor of MBI. In contrast, trivial overall effects in favor of the AC condition for upper and lower body strength were observed. No effects have been found in either direction for functional mobility, gait, endurance, and flexibility performance. Thus, no clear direction of an adaptation of the physical domains can be observed. This finding is mainly due to the large heterogeneity of the AC conditions reflecting a broad range of different task-related exercise contents. A further explanation could be that Tai Chi and Pilates entail a higher proportion of task-specific exercises, and thus neuromuscular adaptations were at least slightly higher than in the AC groups [64]. The lack of effects on upper and lower body strength, functional mobility, gait, endurance, and flexibility can mainly be explained by the fact that the AC conditions partly contained those components [64].

The reporting of this systematic meta-analytical review was conducted according to the PRISMA statement [15]. This is the first meta-analysis that provides pooled effect sizes of different MBI approaches in comparison with both AC and IC conditions in healthy community-dwelling older adults. Despite notable heterogeneity of the underlying studies in terms of sample sizes, study characteristics, training intensity, duration, and diverse alternative exercise programs, our meta-analyses provides a comprehensive, structured, and quantitative view on current scientific evidence on the effects of multimodal MBI (mainly Pilates and Tai Chi) on a variety of physical fitness measures, such as strength, gait, static balance, functional mobility, endurance, and flexibility outcomes. The 30 included studies with nearly 3,000 participants build a solid and valid data pool for robust conclusions on overall and subgroup level. However, the AC programs present notable heterogeneity with some overlap to certain components of MBIs. Nevertheless, AC conditions did never meet the criteria for being categorized as MBI by our underlying definition. Therefore, it would be interesting for future reviews or studies to focus on specific assessment tests and to ensure the quality and homogeneity of the results. Due to several lacking explanations of the interventions, it was not always possible to determine which form of Tai Chi training was involved. Therefore, it was not possible to make a statement about which form might be more effective for healthy, older adults. This was also the case in some explanations of the Pilates exercises. It can be assumed that the Pilates interventions included task-specific exercises that affected various target parameters.

MBI can serve as an integrative and multimodal training approach for the improvement of several physical fitness domains in older adults. This holds particularly true for Tai Chi and Pilates. Compared to traditional exercise recommendations for older adults that separately propose strength, balance, endurance, and flexibility, MBI might be employed as a time-saving functional training alternative. Gait and static balance should thereby be specifically tackled as those aspects are underrepresented in MBI approaches. However, the applied training modalities showed a wide range of variability which made it difficult to deliver consistent and valid training recommendations, particularly when compared to alternative training programs. Furthermore, high variability and heterogeneity of the test and training procedure impede reliable conclusions on the content, volume, and intensity of potentially superior MBI or alternative training approaches when the training parameters notably differ between studies. Future studies should therefore provide more information and data on the exact training purpose, intensity, volume, and mode of exercises. This information would facilitate a valid comparison or meta-regression in which intensity and volume could be further investigated as modulators of the occurrence and magnitude of specific training effects.

The authors have no ethical conflicts to disclose.

The authors have no conflicts of interest to declare.

No sources of funding were used to assist in the preparation of this article.

L.D., S.A.E. and L.M. developed the theoretical formalism, performed the analytic calculations, and performed the numerical simulations. M.M., S.H., T.M., and L.D. contributed to the final version of the manuscript. L.D. supervised the project.