3.1Search results
The flow diagram of the selection process is summarized in Fig.1. 349 potentially eligible reports or articles were founded through electronic searches.224 articles still remained by eliminating repeated records.We excluded 169articles of these based on the title, abstract. Of the 16 remaining articles, an additional 43 were excluded. The most common reasons for exclusion were a non-RCT design, unrelated outcomes,Non-clinical research,or patients in unstable phase of COPD . Eventually,16 RCTs were deemed eligible for inclusion and selected for the final analysis.
3.2 Study characteristics
A total 16 studies[18-33] involving 1039 individuals were selected for this meta-analysis, one studies[23] were published in English, and fifteen studies[18-22,24-33] were published in Chinese. All eligible studies were published from 2008 to 2019. There were 13 studies with a sample size≥40 participants,and the average age of participants is >60 years old. The quality of each study was assessed using the Jadad scale, Five studies were high-quality [22-23,29,31,33],Ten studies were low-quality[18-21,24-28,30,32].The basic characteristics of the included studies were shown in Table 1
3.3 Methodological Quality Assessment
The methodological quality of all included studies were evaluated according to the bias risk
assessment tools provided by the Cochrane Handbook. All of the included trials described randomized allocation,and they were low risk in the fields of randomized allocation.Twelve studies were classified as having an unclear risk in the fields of allocation concealment[18-21,24-28,30-32] and one was high risk[22].There was high risk of bias in the domain of blinding of participants and personnel,only three studies[23,29,33]used single-blind method,but no specific method of blinding was mentioned in these studies.Only five of them were shown to blind their outcome assessment[21,23-24,32-33].All trials reported methods with a low risk of incomplete outcome data .With regard to selective outcome reporting bias, ten studies were determined as low risk and the remaining were determined as unclear risk.All studies were graded as unclear risk of other bias.These results were summarized in Figure 2.
3.4 Effects of Liuzijue breathing exercise on primary outcomes
3.4.1 Effects of Liuzijue breathing exercise on pulmonary function
Pulmonary function test was the main objective index to judge airflow limitation. FEV1,FEV1% and FEV1/FVC(%) are important component of pulmonary function.Seven trials[19,21,23,26-29]used FEV1 to evaluate the therapeutic effect of Liuzijue on stable COPD patients.The fixed effects model was utilized to integrate the results(I2=7%).The results showed that,overall,Liuzijue breathing exercise significantly improved the FEV1 compared with the control group [MD=-0.16,95%CI (0.09,0.23), P<0.00001,Fig3], the efficacy on FEV1 were noticeable in both short term duration of study[MD=0.14, 95%CI (0.06, 0.23), P=0.001;Fig3] and medium-term duration of study [MD=0.19, 95%CI (0.07, 0.30), P=0.002;Fig3].Eleven studies[21,23-24,26-33]used FEV1% to assess the improvement of pulmonary function of COPD patients by Liuzijue breathing exercise.The fixed effects model was conducted to incorporate the results,they had high heterogeneity (I2 =63%).The effect size of the studies[21,23-24,26-33]showed that Liuzijue exercise could significantly improve the FEV1%[MD=9.71,95%CI(8.44,10.98), P<0.00001;Fig.4]. The treatment effect on FEV1% were also prominent in both short term [MD=11.02, 95%CI(9.39,12.66),P<0.00001;Fig.4] and medium-term duration of study [MD=7.71,95%CI (5.69,9.73), P<0.0001;Fig.4].Nine trials used FEV1/FVC(%) to evaluate the enhancement of lung function of COPD patients.The fixed effects model was conducted to incorporate the results (I2 =50%).The effect size of the studies[19,23,26-29,31-33]showed that Liuzijue breathing exercise could significantly improve the FEV1/FVC(%)[MD=4.81, 95%CI (2.12, 7.51),P=0.0005;Fig.5].The effect of FEV1/FVC(%) had significant amelioration for COPD in short term [MD=5.35, 95%CI (1.93,8.78),P=0.002;Fig.5] and with marginal in medium-term duration of study [MD=2.99, 95%CI (-1.11,7.09), P=0.15;Fig.5]
3.4.2 Effects of Liuzijue breathing training on exercise endurance
Nine studies used 6MWD to evaluate the effect of Liuzijue on exercise endurance of COPD patients.The random effects model was adopted to incorporate the results(I2=43%).The overall effect of the studies[18-25,33]found that Liuzijue breathing exercise was associated with significantly increase the length of 6MWD compared with the control group [MD=21.89,95%CI(14.67, 29.11),P<0.00001,Fig6],The effect on 6MWD were remarkable in both short term duration of study[MD=16.78, 95%CI (14.18,19.38),P<0.0001,Fig6] and medium-term duration of study [MD=42.72, 95%CI (28.36, 57.08),P<0.00001,Fig6],but there was no significant effect in the studies with a long-term duration of study [MD=23.11, 95%CI (-3.35,49.57), P=0.09;Fig6].
3.5 Effects of Liuzijue exercise on secondary outcomes
3.5.1 Effects of Liuzijue exercise on health-related quality of life
Five of 7 trials used CAT to evaluate the effect of Liuzijue exercise on health-related quality of life of COPD patients while the other 2 articles used SGRQ .There was high heterogeneity among the included studies(I2=65%), and a random effect model was used for merge the results.The overall effect of the studies[20-24,29-31,33]showed significantly improvement the health-related quality of life compared with the control group [SMD==-0.84,95%CI(-1.12,-0.55),P<0.00001,Fig7], The effect on health-related quality were outstanding in short term duration of study[SMD=-0.70, 95%CI(-1.01,-0.38),P<0.001,Fig7],medium-term duration of study [SMD=-1.10,95%CI (-1.67, -0.52),P=0.0002,Fig7],and long-term duration of study [SMD=-0.57,95%CI(-0.91,-0.23), P=0.001; Fig7].
3.5.2 Effects of Liuzijue exercise on dyspnea
Only three studies[20,22,33]used the mMRC to evaluate dyspnea of patients with COPD .The random effects analysis was managed to merge the results (I2 = 42%).The results showed that Liuzijue breathing exercise significantly lowered the mMRC compared with the control group [MD= -0.73,95%CI (-0.96,-0.50), P<0.00001,Fig8]
3.6.Adverse events
None of the articles informed adverse events. Thence,this information could not be searched from the RCTs analyzed.
3.7.Sensitivity analysis
Some results of this study had high heterogeneity eg.FEV1(%) ,FEV1/FVC(%) and health-related quality, by removing single studies i.e. the study[26] of FEV1(%) ,the study[32]of FEV1/FVC(%) ,study[31] of health-related quality,the sensitivity analyses showed obvious changes in the statistical significance of outcomes.
3.8.Publication bias
As shown from the Egger’s asymmetry tests,there was little indication of publication bias on the these outcomes(p=0.764 for FEV1;p=0.436 for FEV1%;p=0.076 for 6MWD;p=0.348 for health-related quality,respectively),only the outcome of FEV1/FVC%(p=0.048)was showed publication bias.