The Effect of Different Traditional Chinese Exercises on Blood Lipid in Middle-Aged and Elderly Individuals: A Systematic Review and Network Meta-Analysis

Although the impact of physical exercise on blood lipids is well documented, less information is available regarding the effect of traditional Chinese exercises (TCEs), and it is unclear what the best TCE treatment for dyslipidemia in middle-aged and elderly individuals is. The aim of this study was to systematically assess the effects of TCEs (Taijiquan, TJQ; Wuqinxi, WQX; Baduanjin, BDJ; Liuzijue, LZJ; Yijinjing, YJJ; Dawu, DW) on blood lipids in middle-aged and elderly individuals. Chinese and English databases were searched, including PubMed, China National Knowledge Infrastructure, Wanfang Database, Chongqing VIP, and Web of Science. A total of 42 randomized controlled trials (RCTs) including 2977 subjects were analyzed. Outcome indicators include total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), triacylglyceride (TAG), and high-density lipoprotein cholesterol (HDL-C). Summary mean differences (MD) were calculated using pairwise and network meta-analysis with a random-effects model. The results of this study showed that compared to non-exercise intervention (NEI), all six kinds of TCE treatment had some kind of influence on blood lipid indicators, among which WQX and TJQ could improve all four blood lipid indicators, whereas BDJ was effective on three indicators but not on TC. The results of cumulative probability ranking showed that WQX (84.9%, 73.8%, 63.4%, 63.1% to TC, TAG, HDL-C, LDL-C, respectively) was at the top spot being the best intervention, followed by BDJ (55.6%, 83.7%, 68.4%, 56.1%) and TJQ (73.7%, 47.6%, 63.1%, 54.1%). The network meta-analysis of RCTs demonstrates that WQX may be the best TCE treatment for dyslipidemia in middle-aged and elderly individuals.


Introduction
Blood lipids are an important predictive factor for atherosclerosis and an independent risk factor for coronary heart disease and ischemic stroke, which means that dyslipidemia might lead to many ongoing pathologies in people's daily life, such as metabolic syndrome and cardiovascular disease [1]. Moreover, a close association between serum lipid levels and the incidence of coronary heart disease has been well proven in middle-aged and elderly people with a probable degeneration of metabolic functions [2]. Lipid profiles have been used to define dyslipidemia, and abnormal serum lipid profiles include changes in four indicators: high total cholesterol (TC), high triacylglyceride (TAG), low high-density lipoprotein cholesterol (HDL-C), and elevated low-density lipoprotein cholesterol (LDL-C) [3]. Therefore, improving the blood lipid indicators of middle-aged and elderly people has become a hot topic that is receiving increased attention at present.
The inclusion criteria strategy was defined according to participants, interventions, comparisons, outcomes, and study design (PICOS): (1) Subjects of both sexes, aged 40 years old and over, and with a clinical diagnosis of healthy status, dyslipidemia, or chronic disease were included; (2) the type of study included was RCT comparing the effects of six TCEs on lipids. Experimental groups adopted BDJ, WQX, YJJ, LZJ, DW, or TJQ. The control group was a simple control group with non-exercise intervention (NEI), or routine aerobic exercise (AE), such as walking and jogging, or other specific TCEs; (3) the outcomes included at least one of the four lipid indicators (TC, TAG, HDL, CLDL-C); (4) ethically approved RCTs were included.

Exclusion Criteria
The exclusion criteria included: (1) literature not published in English or Chinese; (2) repetitions of previously published literature; (3) theoretical and review literature; (4) literature with only abstracts but no full text; (5) literature designed for non-randomized controlled studies-for example, before-after studies on the same patient; (6) joint-intervention trials; (7) studies where the experimental subjects were not middle-aged or elderly; (8) studies where the experimental data were not clear, and it was not possible to calculate the average and standard deviation of the outcome indicator; (9) studies with control groups that did not meet our requirements, such as drug control; (10) studies with data errors or missing literature.

Literature Screening and Data Extraction
According to the inclusion and exclusion criteria, a unified method and standardized search and selection were used, with the two authors (Y.G. and L.Y.) conducting searches in turn and independently. The figures and tables for information were produced after the two authors had verified their results. An independent arbitrator (H.H.) resolved any discrepancies in data extraction. The following data were collected: (1) basic information extracted, including the name of the first author and the year of publication; (2) the demographic characteristics of the subjects, such as sex and age; (3) information regarding study design, such as sample size, interventions, measurement parameters, follow-up duration, and information related to bias risk assessment.
All the measurement parameters of blood lipids were converted to mean differences (MD) on blood lipids following exercise intervention under the international standard system of units. If an included study reported outcomes of different follow-up times or had more than one trial arm, each different follow-up time and trial arm were treated as a separate trial. The Cochrane Handbook for Systematic Review of Interventions provides detailed measures for dealing with such situations-one way to overcome this is to perform a fixed-effect meta-analysis across comparisons within a study, and a random-effects metaanalysis across studies; in practice, the difference between different analyses is likely to be trivial [13].
Since the unit of a certain blood lipid parameter is uniform, the standardized mean difference was not chosen to illustrate the pooled effect. The software STATA ® 16 (StataCorp LLC, College Station, TX, USA) was used to analyze the combined effect.

Quality Assessment
The Cochrane Collaboration tool was used to assess the risk of bias. The study quality was assessed and graded independently by two authors (Y.G. and L.Y.) according to the criteria described in The Cochrane Handbook. The risk of bias graph and the risk of bias summary graph were produced using RevMan5.3. If the evaluation results were inconsistent, the issue was resolved following discussion with the third researcher (A.W.).

Statistical Analysis
A set of multivariate meta-analysis programs in STATA ® 16 (StataCorp LLC, College Station, TX, USA) software was used to deal with the statistical analysis, draw the net relation diagram of different interventions, and output the table of direct pairwise comparisons of different interventions as well as the forest plot to present the results of network metaanalysis visually. When there was a closed loop, the consistency between direct comparison and indirect comparison was judged by the node-splitting value, and inconsistency was considered to be significant when p < 0.05. When there was no closed-loop structure within the interventions, there was no need to make a consistency test. Continuous variables (TC, TAG, HDL-C, and LDL-C) were analyzed by mean differences (MDs) and 95% credible intervals (95%CI). The effectiveness of these six interventions can be ranked by the surface under the cumulative ranking curve (SUCRA). SUCRA values range from 0% to 100%. The higher the SUCRA value, and the closer to 100%, the higher the likelihood that therapy is in the top rank or one of the top ranks.

Literature Selection
In our initial search, we found a total of 1003 articles, of which 320 were duplicates and thus excluded. After deduplication and application of the exclusion criteria, 42 studies  out of the total 1003 studies were included for analysis. The flow diagram can be seen in Figure 1. Based on the information from all the included full texts, the results of data collection and a summary measure of each included study can be seen in Table 1. Figure 1 shows the selection process for the relevant studies.
As the intervention methods included in this study were all exercise therapy, most of the included RCTs did not adapt blind methods (patient, care provider, and outcome assessor). In addition, other biases were mainly low-risk bias, so the overall quality of the included literature was high.  As the intervention methods included in this study were all exercise therapy, most of the included RCTs did not adapt blind methods (patient, care provider, and outcome assessor). In addition, other biases were mainly low-risk bias, so the overall quality of the included literature was high.

Evidence Network Relationship
The edges connecting the net point indicate the presence of direct comparison evidence between networks. Numbers in addition to edges indicate the number of independent trials comparing the corresponding pair of treatments head-to-head. Comparisons with more trails have a wider edge. In the network diagram of the effects of eight interventions (including the control group) on the blood lipids of middle-aged and older people (Figure 3), the area of the circle represents the size of the corresponding intervention study sample size. Among all interventions, the order of the sample size was NEI > BDJ > TJQ > WQX > AE > YJJ > LZJ > DW (Figure 3).

TC
Using the classical frequency method with Stata software, traditional direct comparison and indirect comparison were combined at the same time to evaluate all the intervention measures [56]. Network meta-analysis summary plots are shown in Table 2. and Appendix A Figures A1-A4 (available online).
Life 2021, 11, 714 9 of 24 dence between networks. Numbers in addition to edges indicate the number of indepe ent trials comparing the corresponding pair of treatments head-to-head. Compari with more trails have a wider edge. In the network diagram of the effects of eight in ventions (including the control group) on the blood lipids of middle-aged and older ple (Figure 3), the area of the circle represents the size of the corresponding interven study sample size. Among all interventions, the order of the sample size was NEI > BDJ > TJQ > WQX > AE > YJJ > LZJ > DW (Figure 3).

TC
Using the classical frequency method with Stata software, traditional direct com ison and indirect comparison were combined at the same time to evaluate all the in vention measures [56]. Network meta-analysis summary plots are shown in Table 2.
As shown in Figure 3 (Table 2). Figure 4 and Appendix B Tables A1-A4 show the ranking probability of each intervention, Table 3 shows the cumulative probability data of each intervention, and Appendix B Figures A5-A8 show the surface under the cumulative ranking plots (available online). The ranking probability gram can help researchers to predict the best or worst intervention quickly, but interventions with higher ranking probability are not necessarily the most effective, and there are still many uncertain factors that can interfere with the ranking. If the optimal intervention cannot be obtained, the SUCRA probability gram can help in decision-making [57].

Consistency Analysis
Consistency can be evaluated by node splitting, with each direct comparison being excluded from the network and then estimating the difference between this direct evidence and the indirect evidence from the network. If there are differences, it means that there are inconsistencies, which need to be fitted with an inconsistent model. If there is no difference, it means that there is no inconsistency, and the consistency model is used to fit it [57].
As can be seen from Table 4, all p values except AB to TC were greater than 0.05, indicating that there was no obvious inconsistency.

Discussion
As the General Administration of Sport of China recommends, more and more people are choosing TCEs to exercise at home. This manuscript has been written based on the hypothesis that TECs have a positive effect on the blood lipid parameters of middle-aged and older people. This systematic review included 42 RCTs, with 6 TCEs with 2977 subjects aged 55 to 60 years, providing high-quality evidence of the effect of six kinds of TCE on blood lipids. Maoxing Pan et al. (2019) [58] studied the effect of Qigong on blood lipids in middle-aged and elderly people through a network meta-analysis, but TJQ was not included. Taijiquan is one of the main forms of TCEs, and it is also very popular in China. To the best of our knowledge, at present, this review is the first network meta-analysis comparing the effects of six TCEs.
People's blood lipid parameters are becoming worse with the development of society. Around the world, over 4000 people die of cardiovascular diseases caused by dyslipidemia every day [59]. Medicine treatments are underused in clinical practice because most doctors worry about their hepatotoxicity and nephrotoxicity [58]. According to a survey of physical examiners in Beijing [59], lifestyle changes play a great role in preventing and treating hypercholesterolemia. TCEs are mind-body exercises that focus on posture, coordination of breathing patterns, and meditation. Movements in TCE are smooth and slow, making this type of exercise safe for middle-aged and older people. In terms of metabolic types, oxidative metabolism, in which fat is oxidized by the body to provide energy, is dominant in TCEs. The positive effects of aerobic exercise on blood lipid parameters have already been verified in many trials enrolling subjects with dyslipidemia [60]. Improvements in lipid levels include lowering the levels of TC, LDL-C, and TAG and increasing the level of HDL-C. Some studies have found that during aerobic exercise, total energy consumption and exercise intensity will affect the improvement of blood lipids and have a positive effect on the improvement of HDL-C and LDL-C [61]. Moderate-intensity aerobic exercise can cause a significant increase in HDL-C, and improvement and reduction in LDL-C may require more intensive aerobic exercise. For middle-aged and elderly individuals, adhering to moderate physical activities can not only delay the senile degenerative changes of various system organs but also maintain relatively high physiological function [62]. Therefore, TCEs can help to improve both physical and psychological conditions of middle-aged and older people [63][64][65].
A total of 42 RCTs were included in this study, including six kinds of TCE and two general controls. Network meta-analysis of direct and indirect evidence showed that six kinds of TCE can effectively improve some blood lipid indicators in middle-aged and elderly individuals. Compared with the NEI group, WQX and TJQ had a significant effect on the decrease in TC, TAG, and LDL and an increase in HDL in middle-aged and elderly people. BDJ had a significant effect on the decrease in TAG and LDL and an increase in HDL in middle-aged and elderly people. YJJ had an obvious effect on the decrease in TAG and LDL in middle-aged and elderly people. LZJ had an obvious effect on the reduction in LDL and can increase the level of HDL in middle-aged and elderly people, while DW had an obvious effect on the reduction in blood LDL in middle-aged and elderly people. Compared with the YJJ group, the level of TC in the WQX group decreased more significantly. We can conclude that all six TCEs are effective in improving LDL-C and are partially effective for other lipid indicators. There are no significant differences in the consistency analysis, which reveals that the statistical model of indirect comparisons is reliable. This may be due to the similar exercise program and outcome indicators among the included studies. Therefore, it can be suggested that patients can choose an appropriate TCE according to their preferences and physical conditions to improve blood lipid parameters.
The ranking results for TC were as follows: WQX > DW > TJQ > BDJ > LZJ > NEI > YJJ > AE. Compared with the NEI group, WQX and TJQ had significant effectiveness, and WQX had the best effect in terms of improving TC. The ranking results for TAG were as follows: BDJ > WQX > AE > YJJ > TJQ > DW > LZJ > NEI. Compared with the NEI group, BDJ, WQX, YJJ, and TJQ had significant effectiveness, and BDJ had the best effect in terms of improving TAG. The ranking results for HDL were as follows: LZJ > BDJ > WQX > TJQ > DW > YJJ > AE > NEI. Compared with the NEI group, LZJ, BDJ, WQX, and TJQ had significant effectiveness, and LZJ had the best effect in terms of improving HDL. The ranking results for LDL were as follows: DW > YJJ > LZJ > WQX > BDJ > TJQ > NEI > AE. Compared with the NEI group, DW, YJJ, LZJ, WQX, BDJ, and TJQ had significant effectiveness, and DW had the best effect in terms of improving LDL. Additionally, statistical factors, such as study design, number of controls, and number of subjects, and many other factors, may have contributed to the difference in outcomes. Although the dominant metabolic patterns of different types of TCE are similar, their activity patterns and training principles are different.
To sum up, based on the comparison with NEI and AE groups, it was found that WQX, TJQ, and BDJ were more effective in improving blood lipids in middle-aged and elderly people. WQX and TJQ had a significant effect on all four indicators of blood lipid indicators. However, BDJ was effective on three indicators of blood lipids, but not TC. Therefore, WQX and TJQ are recommended as appropriate TCE methods for blood lipid control. The results of cumulative probability ranking showed that the SUCRA value of WQX was the highest, while the PrBest value, followed by BDJ and TJQ, indicated that WQX may be the most effective TCE to improve the blood lipid status of middle-aged and elderly people. Therefore, two methods (WQX and TJQ) are recommended as TCE methods for blood lipid control, of which WQX is recommended as the best choice.
The influential mechanism of WQX on blood lipids may be that it emphasizes the control of respiration and regulation of the mind during exercise [66], which not only improves the function of the respiratory system but also accelerates blood circulation [67]. By controlling breathing, the balance of the autonomic nerve can be regulated [68], and the level of substance metabolism can be regulated, and thus, the disorder of lipid metabolism can be improved [39]. The essence of mind regulation is to dominate and exercise the motor nervous system and autonomic nervous system, regulate hormone-sensitive lipase by increasing the activity of the sympathetic nervous system, and finally regulate lipid metabolism [39]. The action mechanism of TJQ may be that its action plays a natural "massage" effect on vascular smooth muscle, promoting rhythmic contraction and relaxation of blood vessels [21], gradually scouring and eliminating the deposition of cholesterol and cholesterol esters on the blood vessel wall; furthermore, TJQ exercise helps to eliminate the anxiety and tension of practitioners, reduce the tension of sympathetic vasoconstrictor nerves, reduce the release of norepinephrine in its terminals [69], and improve the excitability of sympathetic vasoconstrictor nerves. However, the specific mechanism of these three kinds of TCE on improving blood lipids is not clear, with a noted lack of high-quality research at the cellular and molecular level, and thus, further research is needed.
There were some limitations in this network meta-analysis. Small sample sizes and large sex differences existed in some RCTs. There was a different number of RCTs included in different interventions, which may have had a certain impact on the evaluation of the intervention. Most of the RCTs were conducted in China and published in Chinese, which may affect the results. The differences between analyzed populations (middle-aged and elderly) were not reported. Most RCTs did not clearly report whether a random, doubleblind method was used. The research did not discuss the adaptation time, size, and actual needs of each TCE.
In follow-up studies, after the emergence of more multicenter, standardized, and high-quality research using larger samples, we can make use of the advantages of network meta-analysis to further study the impact of TCEs on blood lipids among middle-aged and elderly people from all aspects. More research and attention should be paid to the time for adaptation, magnitude, and practical needs of each TCE, as well as the specific mechanism and the best training program for each TCE.

Conclusions
Our network meta-analysis suggested that six kinds of TCE (WQX, BDJ, TJQ, YJJ, LZJ, and DW) are all effective in partially improving blood lipid indicators among middle-aged and elderly people, while WQX and TJQ can be effective for all four blood lipid indicators and seem to be recommended as the most appropriate way for the elderly to exercise. WQX, BDJ, LZJ, and DW might be the most effective, respectively, for improving TC, TAG, HDL-C, and LDL-C among the six kinds of TCE. According to a comprehensive ranking, WQX had the best effect on improving blood lipids. However, due to the limitations of this study, follow-up studies need to classify and explore the disease status of middle-aged and elderly people and further study the effects of exercise duration and frequency on blood lipid parameters. Furthermore, high-quality clinical trials are needed in the future to strengthen the supportive evidence.                Tables A1-A4 Different ranking probabilities of Interventions.