The efficacy of manual therapy for chronic obstructive pulmonary disease: A systematic review

Background Manual therapy (MT) can be beneficial in the management of chronic obstructive pulmonary disease (COPD). However, evidence of the efficacy of MT for COPD is not clear. Therefore, we aimed to review the effects of MT, including Chuna, in people diagnosed with COPD. Methods MEDLINE via PubMed, EMBASE, The Cochrane Central Register of Controlled Trials (CENTRAL), China National Knowledge Database (CNKI), KoreaMed, Korean Medical Database (KMbase), and Oriental Medicine Advanced Searching Integrated System (OASIS) were searched. Randomized controlled trials (RCTs) and crossover RCTs were included. The main inclusion criteria were COPD diagnosis (forced expiratory volume in the first second [FEV1]/forced vital capacity [FVC] < 0.70). The primary outcomes were lung function and exercise capacity. The secondary outcomes were symptoms, quality of life (QoL), and adverse event (AE)s. Studies reporting one or both of the primary outcomes were included. The Cochrane RoB 2.0 tool was used to assess the risk of bias. Data synthesis and analysis were conducted according to the trial design. Results Of the 2564 searched articles, 13 studies were included. For the primary outcomes, the effect of MT on pulmonary function and exercise capacity in COPD was partly significant but could not be confirmed due to the limited number of studies included in the subgroups. For the secondary outcomes, no definitive evidence regarding the improvement of symptoms and QoL was found, and some minor adverse effects were reported. Conclusions There is insufficient evidence to support the role of MT in the management of COPD. High-quality studies are needed to thoroughly evaluate the effect of MT on COPD.

Introduction Chronic obstructive pulmonary disease (COPD), characterized by persistent respiratory symptoms and airflow limitation caused by airway and alveolar pathologies, is a complex condition with various patterns of symptoms, progression, and associated comorbidities [1,2]. The Global Burden of Disease Study estimated that there are approximately 174 million patients with COPD [3]. Moreover, 3.2 million patients worldwide died in 2015 due to COPD [4].
Patients with chronic respiratory diseases experience dyspnea, fatigue, and exercise intolerance [5,6]. These conditions are associated with a reduced quality of life (QoL) due to a decrease in physical activity levels [7][8][9][10], and the clinical manifestations do not improve with appropriate pharmacological therapy [11,12]. Patients with COPD not only present with respiratory diseases but also extra-pulmonary manifestations, muscle weakness, and medical and mental comorbidities [13,14]. Therefore, non-pharmacological interventions to improve the QoL of patients with COPD are important [15,16], and pulmonary rehabilitation (PR) is recommended for these patients [17].
PR [5], which has been shown to improve QoL, is a multidimensional therapy that encompasses education, physical exercise, and mental assistance. Physical exercise, as the primary intervention, includes active exercises such as walking and stair exercises. Some researchers recommend Manual therapy (MT) as an additional treatment option in association with other interventions, such as physical exercise [18], and MT that targets the respiratory muscles would be particularly beneficial for patients with COPD to develop muscle strength and maintain muscle movement [19].
Although several systematic reviews (SRs) assessing the efficacy of MT have been published [20][21][22][23], it is difficult to reach a conclusion due to the contradictory results. Heneghan (2012) [20] concluded that MT relieves dyspnea and improves overall health conditions but reported only minimal improvement in pulmonary function. Galletti et al.'s study [22] suggested a link between MT and an improvement in physical ability level; however, no clinically significant improvements in QoL or pulmonary function were observed. Nevertheless, the existing systemic reviews have limitations since they included research on MTs published only in English and excluded non-English research.
In Asian countries, a technique called Chuna adds the traditional concept of meridian massage to MT [24]. In particular, Korea has developed a unique form of Chuna by combining it with traditional Korean medicine [25,26]. The effects of Chuna on musculoskeletal pain relief [27,28], as well as its efficacy in the treatment of internal diseases [29][30][31][32][33], have been reported. However, there has been no research on its effects in COPD treatment. Thus, this study intended to elucidate the efficacy of MT, including Chuna (which is used in Korea and China), in the COPD population.

Database and search strategies
Electronic searches and other sources. This study was performed based on a previously described method [37]. Searches were conducted independently in online electronic databases, including MEDLINE via PubMed, EMBASE, the Cochrane Central Register of Controlled Trials (CENTRAL), and the China National Knowledge Infrastructure database, to retrieve research relevant to the use of MT in COPD. Additionally, the following three Korean medical databases were searched: KoreaMed, Korean Medical Database (KMbase), and Oriental Medicine Advanced Searching Integrated System. The reference lists of the retrieved articles and relevant SRs were searched manually. Ongoing RCT registers, such as Clinicaltrials.gov and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), were also reviewed. The study authors were contacted to confirm any unconfirmed data whenever possible and necessary.
Search strategy. The method used in this study has been previously described [37]. Articles were searched using Related Medical Subject Heading terms and synonyms in various combinations in each database from inception through January 31, 2021, without language restrictions. The search strategies consisted of relevant disease-and intervention-level word combinations. The terms relevant to the disease included "COPD," "emphysema," and "bronchitis." The terms relevant to interventions included "manual therapy," "manipulation," "mobilization," "massage," and specific words for treatment modalities such as "physiotherapy," "osteopathy," "chiropractic," and "Chuna (Tuina)." The detailed process is shown in the S2 File. Only disease-and intervention-level words were used as electronic search terms. After searching the databases, two researchers evaluated whether the studies should be excluded according to the eligibility criteria of the review.

Study selection
According to the pre-defined PICOS criteria, two reviewers (JAR and KIK) independently assessed the titles and abstracts of the search results. With the same set of inclusion criteria, the full text of the articles was reviewed for further inclusion by the two reviewers separately. All included studies were uploaded to EndNote X9.3.3, (Clarivate Analytics, Philadelphia, PA) for bibliographic management and review. Any disagreement was resolved by discussion with a third researcher (HJJ).

Data extraction and management
A standard extraction form (a pre-designed Excel file) was used for data extraction by the two reviewers (JAR and KIK). Detailed raw data, such as publication year, first author, country, title, study design, group allocation, method of randomization, number of groups, blinding procedure, number of withdrawals and dropouts, inclusion/exclusion criteria, diagnostic criteria for COPD, number of participants, patient's age, sex, COPD stage, patterns of symptoms based on traditional Korean medicine or traditional Chinese medicine (syndrome differentiation/pattern identification), interventions (the type of the intervention, number of treatments, duration of individual treatment, total number of treatments, type of practitioner), comparators, outcomes (primary and secondary outcomes), and AEs, were extracted.

Assessment of risk of bias
The Cochrane RoB 2.0 tool was used to perform a literature analysis for the SR and meta-analysis of RCTs. The current version of RoB was separately used to assess the risk of bias in individually randomized parallel-group trials (2019), and the RoB 2.0 tool was used to assess the risk of bias in individually randomized crossover trials, wherein each trial was categorized into the one of the three groups: (1) high risk of bias; (2) some concerns; and (3) low risk of bias [49,50].
The risk of bias consists of the following five domains: (1) bias arising from the randomization process, (2) bias due to deviations from the intended interventions, (3) bias due to missing outcome data, (4) bias in the measurement of the outcome, and (5) bias in the selection of the reported result. These five evaluation domains differed from the seven domains stated in our protocol [37].
Two reviewers (JAR and KIK) independently evaluated the risk of bias in the included studies, in accordance with the tool's recommendations. The risk of potential biases was considered for studies in which data were missing or the process of analysis was unclear. Disagreements were resolved by consulting with another researcher (HJJ).

Data synthesis and analysis
In addition to the previously described protocol [37], we further attempted to analyze the RCTs according to their research designs in this review. The analysis protocol was divided into four categories, as follows:  [51]. Fixed-effects modeling was applied to data with substantial homogeneity (I 2 value <50%), and randomeffects modeling was applied to data with heterogeneity (I 2 value �50%, P-value <0. 10).
For studies with missing data or unclear methodology, the authors were contacted via email to clarify the information. If there was no reply or insufficient data after contact, only the available data were used for analysis.
To consider heterogeneities other than statistical heterogeneity that can influence the results, rather than pooling all studies, we pooled a group of studies that had the same trial design and outcomes into a meta-analysis. Multi-arm RCTs were excluded from the metaanalysis. However, whenever two subgroups were reported for an intervention, we combined the two reported subgroups into a single group and pooled the mean (or mean change) and standard deviation (SD) of the combined group into the meta-analysis [52]. While the RoB results did not influence the pooling in the meta-analysis, they influenced the interpretation of the results.
The review manager software (RevMan, Version 5.4.1 for Windows; Copenhagen, The Nordic Cochrane Center, The Cochrane Collaboration, updated in September 2020) was used for data analysis. If the data were adequately homogeneous for analysis, a meta-analysis was performed using fixed-or random-effects models. Random-effects models were used to consider the heterogeneity between interventions in individual clinical research. The mean difference (MD), confidence interval (CI), and P values were used for analysis. We used the MD and SD of post-intervention values in the meta-analysis. If necessary, mean change scores and SD obtained pre-and post-intervention were used [52]. Weighted MDs with 95% CIs were calculated for continuous data obtained using the same measurement scale. In case measurements were made on a different scale, a standardized MD was used. Odds ratio, risk ratios with 95% CIs, and P values were used for dichotomous outcomes. When quantitative synthesis was not appropriate due to heterogeneity, we summarized the study characteristics and outcome measures and conducted a narrative synthesis. Since only a few studies with the same trial design were included in the meta-analysis, analyses with forest plots were not possible for all studies. Moreover, funnel plot, subgroup, and sensitivity analyses could not be performed as planned.
Grading the quality of evidence. The levels of evidence for outcomes and recommendation strengths were assessed according to the Grading of Recommendations, Assessment, Development, and Evaluation [53].

Description of the included studies
A total of 2562 articles were obtained from the initial search of seven databases, and two studies were additionally obtained from the bibliographic information of the retrieved articles. After excluding 393 duplicates from the 2564 articles, the titles and abstracts of 2171 articles were screened for eligibility. Further, 2099 articles were discarded because of the issues related to participants, interventions, or study type. After complete review, 59 articles were further excluded from the remaining 72 articles. Finally, among the 2564 retrieved articles, we included 13 RCTs that satisfied the eligibility criteria. All included RCTs [54][55][56][57][58][59][60][61][62][63][64][65][66] were used for qualitative analysis, and seven studies [57-60, 63, 64, 66] were included for additional quantitative analysis. A flow chart is presented in Fig 1. The general characteristics of the included studies are summarized in Table 1. All 13 included studies were published in journal articles. Among the 13 included RCTs, 3 were crossover RCTs [55][56][57]. The year of publication of the included articles ranged from 1975 to 2019. Three of these studies were conducted in the United States [54,55,66]; two in Australia [57,63]; two in Brazil [60,64]; two in Italy [58,61]; two in Poland [56,59]; and one each in China [65] and Pakistan [62].
A total of 394 patients were assessed across the 13 studies, and 9 of the RCTs recruited outpatients [54-58, 60, 62, 64, 65]. The age of the participants ranged from 57 to 71 years. No study evaluated the syndrome/pattern differentiation for symptoms and signs in patients with COPD based on traditional Korean and/or Chinese medicine (TKM and TCM, respectively).
Based on the GOLD report, five studies recruited individuals with the moderate-to-severe stage of COPD [59,61,[63][64][65], and three studies recruited those with a severe stage of COPD [56,58,60]. In the other five studies, severity was classified as "unknown" since the severity of COPD could not be determined due to insufficient baseline characteristics [54,55,57,62,66].
The analyses results for the articles included in the study designs are as follows:

PLOS ONE
Efficacy of manual therapy for COPD

Study quality and risk of bias
RoB 2.0 was used to evaluate the effect of assigned intervention for the three crossover trials [55][56][57]. The results of the RoB analysis are presented in Fig 2 and the S3 File. Additionally, three studies [55][56][57] showed a low risk of bias for "measurement of the outcome" and "selection of the reported result" categories. The overall biases of the crossover studies were high [57], some concern [55], and low [56], because of the influence of randomization (domain 1) and performance biases (domain 2). ROB 2.0 was used to evaluate the effect of assigned intervention for 10 studies [54,[58][59][60][61][62][63][64][65][66]. The results of the RoB analysis are presented in Fig 3 and the S4 File. Four studies mentioned randomization without describing the detailed methods of allocation and/or concealment [54,59,61,65]. Of the studies, seven did not have adequate blinding between participants and healthcare providers [54,59,61,62,[64][65][66]. In the domain of selection of the reported result, no study had a low risk of bias. Therefore, regarding the overall bias, no study had a low risk of bias.

Outcomes
By considering the heterogeneity of the studies, a separate analysis was performed on each outcome depending on the study design. If the same variable was observed in each study design, a quantitative analysis was performed. Otherwise, a qualitative description was provided. The main outcomes of this review are summarized in Tables 2 and 3.  [54][55][56][57]60], and the short-term effects of 1-2 treatment on lung function were evaluated in four studies [54][55][56][57].
Primary outcomes (PFT, 6MWD). In terms of primary outcomes, a meta-analysis could only be performed for VC among PFT parameters. FEV 1, FVC, FEV 1 /FVC, RV, and TLC were all reported on in three articles, with only one treatment [54][55][56].
(1) PFT 1) FEV 1 : FEV 1 decreased by 3.3% (-0.04 L, SD 0.636) after the intervention compared to the pre-intervention value (mean±SD, 1.22±0.65 L) in the integrative osteopathic MT (OMT) intervention group, which was not a significant difference [54]. This was also the case in a study using the same integrative OMT [56]. Myofascial release also resulted in a decrease in FEV 1 by 2.6% (P = 0.03) after a single OMT intervention [55].
2) FVC: FVC decreased by 5.6% (-0.14 L, SD 0.935) after the intervention compared to the pre-intervention value (mean±SD, 2.50±0.94 L) in the integrative OMT intervention group, which was not a significant difference [54]. However, in the study using the same integrative OMT [56], FVC increased by 0.3 L in the intervention group, which was also not statistically different. Thoracic lymphatic pump (TLP) with activation also decreased FVC by 4.9% after a single OMT intervention (P<0.05) [55].
Secondary outcomes (symptoms, QoL, AE). (1) Symptoms: Dichotomous questions were used in two studies to describe the results of patient-perceived symptomatic improvements after OMT [54,55], and one study reported perceived dyspnea using the Borg scale [57], while yet another study used a VAS for dyspnea [56]. The percentage reflected for dichotomous results, instead of the odds ratio, due to the multi-arm crossover RCT design. Positive response rates of symptoms after treatments were 82% [54] and 50-78% [55]; however, those of controls were 68% [54] and 44% [55], respectively. No changes were observed in the pre-or post-intervention results for either the Borg scale or dyspnea VAS [56,57].
(2) QoL: No study in this group reported QoL; hence, the evaluation was impossible.
(3) AE: AEs after one-session treatment were found using patient-reported outcomes in two studies [54,55]. The following types of musculoskeletal pain were the most frequently

Reference FEV1 (L) FVC (L) FEV1/FVC (%) VC (L) RV (L) TLC (L) Extra PFT 6MWT (m) (Before (B) and After (A)) (Before (B) and After (A)) (Before (B) and After (A)) (Before (B) and After (A)) (Before (B) and After (A)) (Before (B) and After (A)) items (Before (B) and After (A))
Wada et al.  reported: muscle soreness (in the treatment-involved regions, such as the chest, back, and neck) and cardiovascular symptoms (elevated blood pressure, palpitation, and headache). G2. MT + CT versus CT alone. The combination of MT and CT was compared to CT alone in two studies [61,65] through a qualitative analysis of patients with moderate-to-severe COPD.
Primary outcomes (PFT, 6MWD). (1) PFT A dynamic volume was used for the PFT. After a total of 40 Chuna sessions combined with CT over an eight week period, PFTs (FEV 1 , FVC, and FEV 1 /FVC) significantly improved compared to CT alone (P<0.05) [65]. The post-intervention values of FEV 1 , FVC, and FEV 1 /FVC The arrows signify statistically significant differences within or between groups with P values. � Significantly different between groups at the P<0.01.
(2) 6MWD In terms of exercise capacity, only routine pharmacologic therapy (CT alone) increased the walking distance by 5.4% (17.86 m,SD 70.199), whereas the addition of MT (Chuna) to the routine pharmacologic therapy increased the walking distance by 17.9% (59.22 m, SD 69.811; P < 0.05) [65]. After the integrative OMT intervention, the intervention group's 6MWD significantly increased at both 4 (P < 0.0038) and 10 weeks (P < 0.05), but the raw data for these results were not reported [61].
Secondary outcomes (symptoms, QoL, AE). The mMRC score was recalculated according to its own grading scale for improvement when interpreting the mMRC dyspnea scale data of the study in China [65]. More significant improvement in symptoms was observed in patients who had undergone both routine pharmacologic therapy and MT than in those who had undergone routine pharmacologic therapy alone (P<0.01) [65]. Buscemi et al.'s [61] study showed that dyspnea and fatigue improved in the integrative OMT group, but raw data and Pvalues were not reported.
In a study that evaluated QoL using CAT, the change-from-baseline CAT score significantly improved in the treatment group at 4 weeks (treatment group: P< 0.0005; control group: P < 0.188). Additionally, this study reported that the change-from-baseline CAT score significantly improved in the treatment group at 10 weeks (P < 0.05). However, only P-values, no raw data, were not provided [61].
No AE was reported in Buscemi et al.'s [61] study, but after three integrative OMT sessions, there were minor side effects that did not require treatment (e.g., muscle and maxilla pain).
G3. MT versus deep breathing exercise. The study by Shakil-ur-Rehman et al. [62] was included in this category of interventions. The third category was classified separately since the control intervention group had different characteristics than the sham and PR subgroups.
Primary outcomes (PFT, 6MWD). The study [62] reported that there was a higher increase in FEV 1 /FVC after rib cage mobilization than that in the control group (P-value, raw data, and baseline characteristics were not given).
Secondary outcomes (symptoms, QoL, AE). Although the dyspnea index was used to assess symptoms, the results were not reported. In this group, QoL and AEs were not studied as outcome variables [62].
G4. MT + PR versus sham + PR. Two studies were included in this category: one study [64] included patients with moderate-to-severe COPD, while the other [58] included patients with severe COPD. Since the number of studies that reported each outcome variable was small, a qualitative analysis was performed. A meta-analysis was performed only using 6MWD findings from on two studies [58,64].
(2) 6MWD The meta-analysis showed that adding MT to exercise treatment or PR has a particularly beneficial effect on 6MWD (for 58, 64; MD 34.83, 95% CI 22.08 to 47.58, I 2 = 0%; Fig 5) Secondary outcomes (symptoms, QoL, AE). When dyspnea was assessed using the modified Borg scale [58,64], one study [58] described their results narratively without data and P-values, and the other study [64] reported that symptoms showed significant improvement based on the difference in mean post-intervention values (P<0.001), without presenting the pre-treatment data. No study in this group reported QoL. Only one study [58] reported AEs descriptively, and the number of such AEs was reported as "zero" after treatment once a week for 4 weeks.
G5. MT + PR versus PR alone. Three studies were included in this category [59,63,66]. Two studies [59,63] conducted clinical research on patients with moderate-to-severe COPD. Miller [66] did not specify the intervention schedule or severity of disease of participants. A study by Engel et al. [63] reported results at 16 and 24 weeks after starting the clinical research. Considering that MT is terminated at the 12 th week, we used the results of the 16 th week for an analysis of primary outcomes in this review. In addition, this study [63] had limitations in pooling due to its multi-arm RCT research design. Considering that both soft tissue therapy (ST) and spinal manipulative therapy (SM) belong to MT, we combined the results from ST and SM groups and used them for meta-analysis. Using the change-from-baseline value score [63] and post-intervention value score [59,66], we performed a meta-analysis on FEV 1 and 6MWD. The remaining outcome variables were analyzed qualitatively.
Secondary outcomes (symptoms, QoL, AE). Only one study [59] reported symptoms in this group. After six interventions of MT plus PR, the MRC scale score decreased further, by 0.9 (SD 0.802); intervention with PR alone decreased the MRC scale score by 0.3 (SD 0.624). However, none of the groups showed a significant difference. One study [63] evaluated QoL using the SGRQ score. In this study [63], all three groups showed a reduction in SGRQ scores after multiple-session treatments, without statistical significance. Only one study [63] documented two mild AEs using patient-reported outcomes after 131 sessions of ST + PR groups, and no moderate or severe AEs were reported after 136 sessions of ST + SM + PR.

Quality of evidence
The quality of evidence was graded as "very low" to "moderate" (Table 4). No "high" quality of evidence was identified for any parameter, as inconsistencies in the analysis of outcomes, high risk of bias, and small sample sizes were observed in the included RCTs.

Summary of the systematic review
Aims and objectives. The present SR aimed to investigate the efficacy of MT in individuals with COPD. Existing PR has shown advantages in a moderate-to-severe COPD population [17]. Based on these results, we inferred that patients at various stages of COPD may benefit similarly from MT.
The primary outcomes. (1) G1 and G2: The effects of Short-term MT treatment (G1) could not be determined. Long-term MT treatment (G2) improved FEV 1 , FVC, FEV 1 /FVC, and 6MWD in patients with moderate-to-severe COPD. Compared to the results of G1, MT may be effective for long-term and multi-session treatments.
(2) G4 and G5: Compared to sham plus PR, MT added to PR reduced RV and significantly improved 6MWD in patients with moderate-to-severe COPD (G4). MT added to PR improved FEV 1 and 6MWD, but the effects were not significant compared to the solely PR group (G5).
The secondary outcomes.
(1) Dyspnea: The effects of MT on dyspnea were confirmed in four subgroups (G1-2 and G4-5). However, only subgroup G5 did not show a significant  difference. The overall effects of MT on symptoms were inconclusive, likely due to heterogeneity in the measurement of symptoms, recalculation of results [65], and incomplete selective reporting [58,64].
(2) QoL: There is insufficient evidence to state the effects of MT on QoL, although one study reported SGRQ [63] and another reported CAT [61].
(3) AE: The risk ratio of AE increased in the G1 subgroup [54,55], and two AEs were reported in G5 [63]. However, the AEs described in the present review were mild.
After one MT treatment, FEV 1 and FVC decreased and RV increased, which was unbeneficial for PFT. This is thought to be due to reduced soft tissue elasticity in older patients, which prevents them from responding to MT [67]. Previous research has recommended extending the treatment period and/or adding PR for these patients [56]. Although the effects of shortterm MT were not confirmed (G1), we found that lung function and exercise capacity were improved in patients who had more than eight weeks of MT (G2). MT added to PR significantly improved exercise capacity in subgroup G4, and addressed symptoms of dyspnea in both G4 and G5. Therefore, further research is needed to investigate the effects of MT in improving exercise capacity and alleviating shortness of breath. However, a definite conclusion could not be established in the present review, due to the increased heterogeneity in trial designs and interventions evaluated, as well as the high risk of bias for the available evidence. Clinicians may refer to the present study on whether MT should be used for personalized treatment.

Assessment of the risk of bias in manual therapy research
The importance of deviations from intended interventions (performance bias) can vary according to trial design. Concerning participant blinding, a stricter judgment must be made in the MT vs. sham-controlled design (G1, G4), whereas the MT vs. non-sham design (G2, G3, G5) requires less strict judgments. Such a difference in the importance of blinding is not reflected in the indiscriminating judgment made by domain two of the RoB 2.0 tool. Among the seven studies belonging to the MT vs. sham control design group (G1, G4), there was no study with a high risk of bias (0/4/3; high/some concerns/low). Among the six studies belonging to the MT group vs. the non-sham design group (G2, G3, G5), there were three studies with a high risk of bias (4/1/1; high/some concerns/low).

Study limitations
The heterogeneity in the study design of the included trials must be considered in the interpretation of the results. The included studies differed in terms of disease severity of the participants, treatment techniques, length and intensity of treatment, and the total duration of treatment. Moreover, none complied with the Consolidated Standards of Reporting Trials (CONSORT) extension for non-pharmacologic treatment (NPT) [68] for reproducibility and transparency of RCTs [54][55][56][57][58][59][60][61][62][63][64][65][66].

Study strengths
This review provided updated information based on available journal articles that investigated the effect of MT in patients with COPD. Due to the lack of language barriers in East Asian databases, we could include new research on Chuna treatments for COPD [65]. Although the heterogeneity of MT was a significant limitation for researchers, the present review addressed this limitation for the first time by classifying the included articles based on their trial design. MT, accompanied by CT, indicated that multiple-session MT may possibly improve exercise capacity and lung function in patients with COPD. The possibility of alleviating symptoms was also observed when MT was added to either PR or CT. Moreover, the present review has the advantage of evaluating associated AEs, whereas previous SRs did not [20,22].

Study implications for practice and further research
To create a high level of evidence regarding the efficacy of MT via SR, future studies that comply with reporting guidelines are required so that the methodological quality can be improved and more robust comparisons of different MT interventions be conducted. Given the inherent challenges of blinding in trials for studying MT [68], pragmatic trial designs may be worth considering. There is a need to recruit clinical trial participants by considering the severity of the disease according to official reports (e.g., GOLD 2020, [17]) for subgroup analysis. Given that QoL is an important measure to analyze the comprehensive health of patients with COPD [69,70], more research on QoL is needed. Additionally, evidence for the follow-up results of MT is needed, since these were obtained in two studies [61,63].

Conclusion
This review showed that there is insufficient evidence to support the role of MT in the management of individuals with COPD. This is because the included studies in each subset were too small, resulting in a smaller number of studies available for meta-analysis. Furthermore, the studies were of poor quality, with some concerns regarding a high risk of bias. In the future, high-quality studies designed to evaluate the effect of MT thoroughly should be conducted. In addition to this review, practitioners need to use clinical judgment for the utilization of MT.