Different Training Durations and Frequencies of Tai Chi for Bone Mineral Density Improvement: A Systematic Review and Meta-Analysis

Objective Tai Chi shows potential as a safe and cost-effective intervention to improve bone mineral density (BMD). However, the various effects caused by different training durations and frequencies have not been evaluated. This updated systematic review aims to explore the effectiveness of Tai Chi in attenuating bone mineral density loss based on different training durations and frequencies. Methods We conducted an extensive database search in Cochrane Central Register of Controlled Trials, PubMed, EMBASE, Web of Science, Chinese Biomedical Literature Database, China Knowledge Resource Integrated Database, Wanfang Data, and China Science and Technology Journal Database on randomized controlled trials that examined Tai Chi for BMD improvement. Two reviewers independently performed data screening and extraction. Study quality was evaluated using the Cochrane Handbook for Systematic Reviews of Interventions. Results A total of 23 randomized controlled trials involving 1582 patients were identified. The aggregated results have shown significant benefits in favor of Tai Chi on BMD improvement in the lumbar spine (SMD = 0.36, 95% [0.13, 0.59], P=0.002), femoral neck (SMD = 0.40, 95% [0.16, 0.63], P=0.0009), femoral trochanter (SMD = 0.43, 95% CI [0.20, 0.66], P=0.0002), and Ward's triangle (SMD = 0.31, 95% [0.15, 0.48], P=0.002). Such favorable benefits in Tai Chi can only be seen when compared with the nonexercise group, and Tai Chi showed no significant improvement in BMD change when compared with other exercises group. Subgroup analyses showed various effects of BMD improvement based on different training durations and frequencies of Tai Chi. Tai Chi is effective in attenuating BMD loss with an intervention frequency of >4 days/week in the lumbar spine, with an intervention frequency of >4 days/week or an intervention duration of >10 months in the femoral neck, and with an intervention duration of >10 months or a frequency of ≤4 days/week in Ward's triangle. Conclusions The results demonstrated that Tai Chi may have benefits in attenuating BMD loss. Different training durations and frequencies may result in variable effectiveness. Researchers should focus more on the training durations and frequencies of Tai Chi so that a more definitive claim can be made regarding the beneficial effects for BMD improvement.


Introduction
e decrease in bone mineral density (BMD) is closely related to age [1]. Osteoporosis caused by age-related bone loss puts a heavy burden on families and society [2,3]. Currently, there is no safe and effective radical cure to restore osteoporotic bones, making early prevention of osteoporosis particularly important [4,5]. e essential elements in preventing osteoporosis are achieving normal peak bone mass and attenuating age-related bone mass reduction [6].
In recent years, in addition to pharmaceutical treatments, the positive effect of exercise on bone density has attracted more and more attention [7,8]. Studies have shown that exercise can prevent and treat osteoporosis in the following ways: (1) it can stimulate osteoblasts and bone marrow stem cells to produce biological effects through mechanical stress [9,10], (2) it can up-regulate the expression of key factors in signaling pathways including WNT, BMP, OPG/RANK/ RANKL [11,12], and (3) it can regulate the endocrine system and increase the estrogen level in the body [13,14]. At present, the positive effect of exercise on osteoporosis has become a research hotspot.
Tai Chi is a mind-body, low-impact, weight-bearing exercise that is growing in popularity worldwide and shows potential as a safe and cost-effective intervention to improve BMD. It is an enjoyable and gentle activity that involves the entire body with a high adherence [15]. A handful of studies have evaluated the effectiveness of Tai Chi on BMD [16][17][18][19], and more and more systematic reviews (SRs) in this field are emerging [20][21][22][23]. However, the retrieval time of the most recent SR is until 2017, and new randomized controlled trials (RCTs) have been published in the past two years. Besides, no SR focused on the BMD change in different body parts, as well as different BMD improvements caused by various intervention durations and frequencies of Tai Chi. erefore, we conducted an updated SR to comprehensively and systematically evaluate the effects of Tai Chi on BMD change based on different training durations and frequencies and provided evidence-based recommendations to clinicians.

Data Sources and Search Strategies.
e following databases were searched using a computer: Cochrane Central e primary search terms used were "Taiji," "Bone mineral density," "Bone health," "Bone metabolism," "Osteoporosis." e retrieval time was from the establishment of the database to May 2020, and only studies published in English or Chinese were included. Relevant systematic reviews and the references to the included articles were also searched to supplement other potentially relevant literature.

Criteria for Considering Studies for
is Review. Studies should meet the following inclusion criteria (PICO format): (1) Participants: individuals without serious complications. e age, gender, case source, nationality, disease duration, or race of subjects was not restricted. (2) Interventions: the experimental group was given Tai Chi Chuan, and the style, the intervention duration, and the intervention frequency were not restricted. Other Tai Chi exercises like Tai Chi push hands or Tai Chi ball were excluded. (3) Control: any type of control group, including usual care, conventional Western medicine, no exercise, and any kind of exercise, was acceptable. (4) Outcomes: the primary outcomes were dual-energy X-ray absorptiometry measures of BMD of the spine, the femoral neck, the femoral tuberosity, and Ward's triangle. (5) Study type: RCTs.

Literature Screening.
Two investigators reviewed the titles and abstracts independently according to the preset inclusion criteria and excluded unrelated literature. en full-text screening of the remained studies was carried out, and the two investigators identified the final included research independently according to the inclusion criteria. e results were cross-checked, and the investigators' differences were resolved by consensus with a third investigator.

Data Extraction.
Two reviewers extracted the data while blinded to each other's review according to the prepiloted, standardized forms. e original author(s) was contacted in case of incomplete information provided in the article. e data extraction included the following aspects: (1) general information: first author, publication year, literature topics; (2) research characteristics: baseline comparability, sample size, sex ratio, country, intervention measures, treatment course, follow-up; (3) outcome indicators; and (4) relevant factors for evaluating the risk of bias. A third reviewer was consulted if there was disagreement during cross-correction.

Assessment of Risk of Bias in Included Studies
. Two reviewers independently assessed the risk of bias in accordance with the assessment tool suggested in the Cochrane Handbook for Systematic Reviews of Interventions [24] and then conducted cross-checking. e following aspects were included: random sequence generation (selection bias); allocation concealment (selection bias); blinding of participants and personnel (performance bias); blinding of outcome assessment (detection bias); incomplete outcome data (attrition bias); selective reporting (reporting bias); and other sources of bias [24]. Any disagreement during this procedure was resolved by consensus with a third investigator. RevMan 5.3 provided by the Cochrane Collaboration was used to created plots demonstrating the risks of bias.
2.6. Statistical Analysis. Revman 5.3 was also used for statistical analysis. All outcomes were continuous variables and were calculated as standard mean difference (SMD) and 95% confidence interval (CI). We extracted the mean and standard deviation of the change from baseline and transformed it into a standard format to make sure that it was implemented successfully in our analysis. e chi-square test and I 2 statistic were used to check the heterogeneity of the results. P ≥ 0.1 or I 2 < 50% was considered to indicate low heterogeneity, and a fixed-effects model was established for statistical analysis; otherwise, subgroup analyses according to control type, intervention duration, and intervention frequency were performed, as well as sensitivity analyses if necessary. Publication bias was estimated with a funnel plot.

Assess of Risk of Bias.
e risk of bias assessment of all included studies is shown in Figures 2 and 3. Four trials reported the method of randomization in terms of a computer-generated random sequence [16,17,37]; 6 trials used a random number table [30,32,35,36]; 1 trial applied the method of drawing lots [27]; the remaining trials lacked descriptions of particular random sequence generation method. Four trials used sealed opaque envelopes to perform the allocation concealment [16,26,37], while the remaining trails failed to provide sufficient details about allocation concealment. Due to the apparent difference, whether to use Tai Chi or not, between the two groups, a blind method could not be used in the participants or administrators in these studies. Four trials were shown to blind its outcome assessment [16,17,37]. Five trials reported a high but nonrandom drop-out rate [17,27,30]. All studies reported all outcomes listed in their methods section. e data necessary for judging the risk of other biases in all trials were insufficient.
en, we conducted subgroup analyses according to intervention durations (>10 months and ≤10 months) and intervention frequencies (>4 days/week and ≤4 days/week) in the nonexercise group (Table 2).

Publication Bias.
e funnel plot was drawn based on RCTs that included the outcome of BMD change in the lumbar spine (L2-L4). e funnel plot was obviously asymmetrical, suggesting that publication bias might exist ( Figure 8). Four RCTs, which were significantly a skewed in the graph, showed no difference in methodology and other aspects [25,31,36,37].

Safety Monitoring.
ree RCTs reported that there were no serious adverse events [16,35,36], while the remaining trials did not mention adverse events.

Discussion
Currently, there are several meta-analyses published on the same topic, as presented in Table 6. e highlights of our work are as follows: (1) firstly, our analyses included new RCTs that were published in the past two years. Since the            Evidence-Based Complementary and Alternative Medicine results of previous systematic reviews were not uniform, the inclusion of further RCTs can lead to more accurate conclusions; (2) secondly, we added some results, including BMD change in the femoral trochanter and Ward's triangle, to study the effect of Tai Chi on different parts of the body; (3) lastly, we conducted subgroup analyses on control type, intervention duration, and intervention frequency. In this way, the effects caused by different training durations and frequencies were evaluated, and we found that different training durations and frequencies of Tai Chi could result in variable effectiveness.

Main Results and Analysis.
is systematic review and meta-analysis, which was based on 23 RCTs involving 1582 participants, found that Tai Chi may have a positive effect on BMD improvement in the lumbar spine, the femoral neck, the femoral trochanter, and Ward's triangle. Our study showed benefits in improving BMD in the four parts mentioned above in favor of Tai Chi compared with the nonexercise group. However, no significant improvement was found between the Tai Chi and the other exercises group. Besides, different intervention durations and frequencies of Tai Chi may lead to various effects. For BMD in the lumbar spine, only Tai Chi with an intervention frequency of >4 days/week was shown to have a beneficial effect, while no significant improvement was found with an intervention frequency of ≤4 days/week. For BMD in the femoral neck, the pooled result showed that significant improvement could only be found in subgroups of an     Tai Chi is a moderate-intensity aerobic exercise, characterized by a high demand for neuromuscular coordination, low velocity of muscle contraction, and no jumping. Tai Chi affects bone health through the two following aspects: one is the mechanical regulation system, that is, the beneficial effect on bone comes from the mechanical stress generated during the Tai Chi exercise; the other is the hormone regulation system, and one study found that the serum bone-specific alkaline phosphatase (BAP) concentration and the ratio of BAP to pyridinoline (BAP/PYD) increased significantly after six weeks of Tai Chi exercise, indicating the bone-promoting effect in Tai Chi [39]. In addition, Tai Chi also plays a positive role in psychological adjustment [40]. Although showing no significant BMD improvement compared to other exercises, including resistance exercise, rapid walking, dancing, and rope skipping, Tai Chi is still recommended as a suitable exercise for older patients with rheumatoid arthritis and osteoporotic conditions for smooth, slow, calm, and conscious-based movements, which are barely seen in other strengthening exercises [41]. Till now, some studies have focused on the effectiveness of different training frequencies and durations in Tai Chi. One study has found that Tai Chi was shown to have a significant improvement that is positively related to the intervention duration in functional gait assessed by timed up-and-go tests [42]. Another study has found that a brief high-impact exercise every day is more beneficial to BMD improvement than an exercise of 4 days per week [43]. According to our review, different training durations and frequencies could result in variable effectiveness. However, the interpretation of these results should be treated with caution. In this meta-analysis, high heterogeneity, a

4.2.
Limitations. e overall quality of the included studies was not high. Only 11 RCTs reported appropriate randomization methods [16,17,27,30,32,[35][36][37], and 4 RCTs mentioned allocation concealment [16,26,37]. Meanwhile, performance bias could not be ruled out because participants cannot be blinded to the Tai Chi exercise. In addition, four trials mentioned blinding of outcome assessment [16,17,37], and 5 RCTs reported a high but nonrandom drop-out rate [17,27,30]. At the same time, publication bias also existed. Four RCTs were significantly asymmetrical according to the funnel plot [25,31,36,37]. However, differences in the design and methodology of these RCTs cannot be found, which may be caused by the low quality and inadequate information provided in these studies. e heterogeneity among the studies was significant. Subgroup analyses with respect to control type, intervention duration, and intervention frequency were subsequently carried out, and we found that these three aspects might partly be the source of heterogeneity. However, the above-mentioned three categories can only explain the heterogeneity to some extent and should be interpreted with great caution. ere remained substantial unexplained heterogeneity in this review, including study design and study quality. For example, the detailed included participants in each study were slightly different: some were menopausal women, some were elderly women, some were elderly men, and some had hyperlipidemia. Besides, the training style and daily practice time in each study were also different. Various training styles of Tai Chi involving Yang-style, Sun-style, and 24-form were included in this review, and the daily practice time was not the same, ranging from 30 to 90 minutes per day. All these differences may result in high heterogeneity. e distribution of the included studies is uneven. Most studies were conducted in Asian countries, and only 1 study was from the USA. erefore, ethnic-based subgroup analysis could not be carried out. Besides, we only performed a search for English and Chinese studies, but other languages should be searched to expand the scope. In the absence of enough data, this study only focused on the efficacy of Tai Chi on BMD improvement, but other aspects, including muscle strength, functional mobility, and fracture incidence, may also be assessed.

Practical Implications.
We synthesized current data of Tai Chi on BMD improvement and found that Tai Chi seemed to be an effective exercise therapy to improve BMD in the spine, the femoral neck, the femoral trochanter, and Ward's triangle. e current finding, which should be interpreted with caution on account of the low methodological quality and the high heterogeneity, is still a promising reference for future clinical trials. Future studies should focus more on the effectiveness of various training types, frequencies, and durations of Tai Chi, and create a set of standard Tai Chi exercises with specific moves, practice frequency, and training duration for global promotion.

Conclusions
According to our review, Tai Chi may be effective for BMD improvement and can be promoted as cost-effective exercise therapy. e training time and frequency required for various parts are different. However, due to the limitations of the included studies, large-sample, multi-center, well-designed clinical trials are still required to verify this conclusion. Moreover, future studies should focus more on the relationship between the training time and frequency of Tai Chi and different body parts to create a set of standard Tai Chi exercises with specific moves, practice frequency, and training duration.

BMD:
Bone mineral density SR: Systematic review RCT: Randomized controlled trial CENTRAL: Cochrane Central Register of Controlled Trials CBM: Chinese Biomedical Literature Database CNKI: China Knowledge Resource Integrated Database VIP: China science and technology journal database SMD: Standard mean difference CI: Confidence interval.

Data Availability
All data generated or analyzed during this study are included within this article.

Disclosure
Funding bodies played an economic support role in research design.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
Y Z, Z-H Z, X-H F, and W-H L contributed to the conception of the study. e manuscript protocol was drafted by Y Z and Z-H Z and was revised by X-H F and W-H L. e search strategy was developed by all the authors and performed by Z-H Z and Y Z, who also independently screened the potential studies, extracted data from the included Evidence-Based Complementary and Alternative Medicine studies, assessed the risk of bias, and completed the data synthesis. Z C arbitrated in cases of disagreement and ensured the absence of errors. All authors have read and approved the manuscript.