Efficacy of meridian massage for motor function after a stroke: a systematic review and Meta-analysis

Abstract

OBJECTIVE:

To evaluate the efficacy of meridian massage on motor function after a stroke.

METHODS:

This systematic review and Meta-analysis of randomised (RCTs) and quasi-randomised controlled trials (qRCTs) was performed to determine the effects of meridian massage on motor dysfunction following a stroke. Articles published until December 2020 related to stroke and meridian massage from National Library of Medicine, Cochrane Central Register of Controlled Trials, Embase, KoreaMed, Japan science and Technology Agency, China National Knowledge Infrastructure Database, Wanfang Database, Chinese Biomedical Literature Database and China Science and Technology Journal Database were searched. The primary outcomes included Fugl-Mayer Assessment (FMA), modified Barthel Index (mBI) and clinical efficiency, while the secondary outcomes included modified Ashworth (mAS) and Berg Balance Scale (BBS). Two authors independently selected the literature, extracted the data and assessed the risk of bias, after which Meta-analysis with subgroups and sensitivity analysis was conducted.

RESULTS:

A total of 16 RCTs and 3 q-RCTs involved 1556 patients satisfied the inclusion. Meta-analysis showed that compared to conventional rehabilitation, meridian massage had significantly better FMA [mean difference (MD) = 1.16, 95% confidence interval (CI): 0.43, 1.89, P = 0.002], FMA-L (MD = 3.21, 95% CI: 1.97, 4.45, P = 0.000 01), mBI (MD = 8.87, 95% CI: 4.24, 13.49, P = 0.000 02), clinical efficiency [relative risk (RR) = 1.37,95% CI: 1.24, 1.51, P < 0.000 01], mAS (MD = -0.69, 95% CI: -1.26, -0.11, P = 0.02) and BBS (MD = 5.07, 95% CI: 2.22, 7.93, P = 0.0005). Subgroup analysis indicated no significant difference between meridian massage and control group when the intervention duration was >8 weeks (MD = 0.89, 95% CI: -0.64, 2.42, P = 0.25) but improving motor function in the meridian massage when the intervention duration was ≤ 8 weeks (< 4 weeks intervention duration MD = 0.84, 95% CI: 0.59, 1.10, P < 0.000 01; 4-8 weeks MD = 2.41, 95% CI: 0.80, 4.01, P = 0.003).

CONCLUSIONS:

Meridian massage as an alternative and complementary therapy seems to have short-term benefits in improving post-stroke motor function (particularly in the lower limbs), quality of life, clinical efficiency, and balance stability and muscle spasms. The high heterogeneity of the included studies and the low methodological quality certainly limits the confidence placed in this systematic review and Meta-analysis, warranting future well-designed studies.

1. INTRODUCTION

Stroke is a group of diseases characterised by the loss of local nerve function caused by cerebral blood circulation disorder, leading to high morbidity, high mortality, high disability and high recurrence worldwide.^1-3 Between 1990 and 2017, a dramatic increase in the of stroke had been noted, with an estimated 104.2 million cases in 2017, almost double that of 1990.⁴ while modern medicine and health care have increased the survival rate for patients with stroke, varying degrees of disability have been observed from survivors.^5-7 Unfortunately, the prevalence of stroke in China has not decreased, with the annual stroke mortality rate being estimated at approximately 115 per 100 000 individuals and over 2 million new stroke patients accompanied by the highest disability-adjusted life-years lost.^8-10 Individuals who suffer a stroke need their families and society to bear the financial and health care burden, which has continued to increase and is becoming excessive.¹¹ Given that approximately 50%-60% of stroke patients experience motor-related disabilities that decrease their quality of life, increased attention should be provided to related treatment and rehabilitation.^12,13 Therefore, there is an urgent need to identify efficient rehabilitation strategies to address this pressing concern.

A massage can be described as the systematic and methodical manipulation of body tissues best performed with the hands to affect the nervous and muscular systems and the general circulation.¹⁴ As an alternative and complementary therapy, massages have been widely used in obstetrics, gynaecology, encephalopathy, cardiology, orthopaedics, oncology and other medical disciplines, which can be classified according to different countries and regions, such as Swedish massage, Thai massage, Traditional Chinese Medicine (TCM) massage, Malay massage, Ayurveda massage and so on.^{14, 15} Different from a normal massage, a meridian massage, which is an orthodox treatment based on the meridian theory and Yin-Yang theory in eastern philosophy combined with various manipulation techniques, such as palm-pulling, point-pushing, rou-kneading and others, has been practiced for over 5000 years in China. A meridian massage has been speculated to affect change through the external force produced by manipulation, acting on specific parts of skin and acupoint force produced by massage manipulation, which is transformed into ‘Qi’ and subsequently into ‘information’ that can be recognised by the human body, thereby stimulating the meridian system and regulating the function of the corresponding organs.^16,17 The meridian massage has been widely employed for neurological disorders diseases, possibly due to its ability to increase the circulation of ‘Qi’ and relieve Qi-blockage, dredge the meridians, regulate tendons, enhance joint flexibility, adjust the Yin-Yang balance and boost immune function, or other advantages.^18,19 Recently, several clinical studies^20-41 have shown that meridian massages can efficiently improve motor dysfunction of the affected limb and boost the quality of life after a stroke. However, the conclusions presented in the aforementioned studies have been inconsistent and controversial owing to the small sample size and uneven quality of the different studies. The current systematic review and Meta-analysis therefore aimed to evaluate the efficacy of meridian massage therapy for post-stroke motor function.

2. METHODS

This systematic review and Meta-analysis has been registered with PROSPERO (International prospective register of systematic reviews, CRD42020156584).

2.1. Search strategy

A literature search was performed on the following databases for article published from their inception up to December 2020 without language restrictions: three English medical databases (MEDLINE, Cochrane Central Register of Controlled Trials and Embase), a Korean medical database (KoreaMedwww.koreamed.org/), a Japanese medical database (Japan Science and Technology Agency, https://jglobal.jst.go.jp/en), and four Chinese medical databases (China National Knowledge Infrastructure Database, Wanfang Database, Chinese Biomedical Literature Database and China Science and Technology Journal Database). Taking PubMed as an example, the following key words were used for the literature search: ((((massage[MeSH Terms]) OR (tuina) OR (anmo) OR (acupressure)) AND ((Stroke[MeSH Terms]) OR (cerebrovascular disorders) OR (cerebrovascular disease) OR (cerebrovascular event) OR (cerebrovascular accident) OR (intracranial haemorrhage) OR (brain haemorrhage) OR (transient ischaemic attack)) AND ((randomised controlled trials[Publication Type]) OR (random*[Title/Abstract]) OR (placebo[Title/Abstract])))) NOT ((animals[MeSH Terms]) NOT ((animals [MeSH Terms]) AND humans[MeSH Terms]))

2.2. Inclusion criteria

Studies that satisfied all of the following inclusion criteria were included: (a) study design: randomised (RCTs) or quasi-randomised controlled trials (qRCTs) on meridian massage for post-stroke motor dysfunction; (b) participants: adults with motor dysfunction due to stroke; (c) interventions: the experimental group was treated with meridian along with or without conventional rehabilitation (CR); (d) ccomparison: the control group can be CR, sham or placebo; (e) no limit on meridian selection was established.

2.3. Exclusion criteria

Studies that satisfied the following exclusion criteria were excluded: (a) non-randomised controlled trials; (b) non-clinical trials; (c) repeated publications; (d) review or case reports; (e) unavailability of data after contacting the author.

2.4. Outcome measures

2.4.1 Primary outcomes

The primary outcomes included fugl-meyer assessment (FMA), modified barthel index (mBI), activities of daily living (ADLs), and effective rate.

The FMA, which includes five domains [i.e., motor function (in the upper and lower extremities), sensory function, balance (both standing and sitting), joint range of motion and joint pain], has been considered the gold standard for evaluating motor function recovery after stroke.^42,43 The mBI has been considered the best scale for assessing ADLs and motor function, with an internal consistency reliability coefficient of 0.90 for the modified scoring.⁴⁴ The effective rate based on the Neurological Impairment Scale (NIS): Cure (reduction in the NIS score by 91%-100%); marked effective rate (reduction in the NIS score by 46%-90%); improvement (reduction in the NIS score by 18%-45%); failure (reduction in the NIS score by < 18%).

2.4.2 Secondary outcomes

The second outcomes included modified ashworth scale (mAS) and the Berg Balance Scale (BBS).

The mAS had been used to measure muscle tone, grading spasticity from 0-4.^45-47 A score of 0 indicates no increase in muscle tone; a score of 1 indicates a slight increase in muscle tone as manifested by catch and release or by minimal resistance at the end of the range of motion (ROM) during flexion or extension of the affected part(s); a score of 1+ indicates a slight increase in muscle tone as manifested by a catch followed by minimal resistance throughout the remainder (less than half) of the ROM; a score of 2 indicates a more marked increase in muscle tone through most of ROM, but affected part(s) easily moved; a score of 3 indicates a considerable increase in muscle tone and passive movement difficultly; a score of 4 indicates rigidity during flexion or extension of the affected part(s). The BBS had been used to assess postural stability.⁴⁸ This scale consists of 14 items scored from 0 to 4, the sum of which provides a total score ranging from 0 to 56, with higher scores indicating better balance.

2.5. Data extraction

The literature management software ENDNOTE X7 was used to retrieve studies and exclude duplicates. Two authors (XYH and ZQ) independently reviewed the studies’ content and abstracts according to the precise search strategies, after which they initially screened the studies that satisfied the inclusion criteria. Two authors (XYH and ZQ) then read the full text to further determine suitability for inclusion. Any disputes were settled through discussions or consultation with a third author (GH).

2.6. Quality assessment

Two authors (XYH and ZQ) independently assessed the risk of bias among the included studies using the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.3.⁴⁹ Accordingly, studies were evaluated based on seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases. The same two authors then independently tagged low risk studies (low probability of bias) as ‘yes’, high-risk studies as ‘no’ and those that lacked sufficient information or were uncertain as ‘unclear’. Any differences were resolved through discussions or consultation with GH.

2.7. Data syntheses

This study utilised Rev Man 5.3 software for Meta-analysis. RR and 95% CIs were calculated for dichotomous data, whereas MD and 95% CIs were calculated for continuous data. The included studies were tested for heterogeneity before Meta-analysis. When P < 0.1 and I ² ≥ 50%, indicating significant heterogeneity, the random effects model (REM) would be selected for Meta-analysis; otherwise, the fixed effects model (FEM) was selected. Sources of heterogeneity were determined using sensitivity analysis and subgroup analysis. Descriptive analysis was performed when the source of heterogeneity could not be determined.

3. RESULTS

3.1. Study selection

Our search strategy yielded a total of 1267 related studies. After excluding 613 duplicate studies, 739 studies remained. After reading the titles and abstracts, 49 studies were included. After reading the full text, excluding case reports, reviews, non-RCTs, animal trials, those with incomplete data, those with an incomparable baseline, duplicate publications, 19 studies were ultimately analysed. The study screening process is outlined in Figure 1.

Figure 1. Study screening process.

JST: Japan Science and Technology Agency; CNKI: China National Knowledge Infrastructure Database; CBM: Chinese Biomedical Literature Database; VIP: China Science and Technology Journal Database

3.2. Characteristics of the included studies

Table 1 summarises the characteristics of the 19 included studies, among which 16 were RCTs^{21-26, 28-39} and 3 were quasi-RCTs.^{20, 27, 41} The included studies were published from 2002 to 2018, among which two were conducted in Korea, ^{20, 27} while the remaining 18 were conducted in China. Among the included studies, stroke duration varied from 10 days to 21 years, while the intervention duration ranged from 2 to 12 weeks. Among the nine studies that reported stroke types, ischaemic strokes outnumbered haemorrhagic strokes. Moreover, 17 studies had more males than females. Two studies utilised placebo massage in the control group, ^{20, 39} while the rest used conventional rehabilitation treatment or routine care. Three studies^{34, 39, 41} provided follow-up for 2-3months, while four studies^{26, 27, 39, 41} had clear sample size calculation formula. Three studies mentioned no adverse effects, ^{31, 34, 39} while the rest provided no detailed information on the same. Four studies had reported attrition,^32,34,39,41 two had no attrition,^32,34 and reported no such information in detail.The details of selected acupoint or meridians, manipulations, sample size estimation, baseline comparison, adverse effects, drop out, and fund (Table 1).

Table 1.

Characteristics of the included studies

No.		Study		Country		Participates															Treatment							Control				Duration		Outcome
						Total No. of random participates			Sex (M/F)		Type (Ischaemic stroke/ Haemorrhagic stroke/ Mixed stroke)				Age		Disease duration			Main content			Length of session/ frequency					Main content		Length of session/ frequency
1		Di HY et al 201721		China		150			T: 48/27 C: 46/29		T: 42/33 C: 39/36				T: 61.4 ± 5.2 C: 61.7 ± 5.3		< 3 m			Acupoint and meridian massage+ CR			2*15 min/d					CR		1*20 min/d		3 w		FMA, mAS, mBI, TCM Syndrome Score, Total efficacy
2		Lin JZ et al 201728		China		60			T: 10/20 C: 12/18		-			T: 62.04 ± 9.31 C: 60.86 ± 9.72			1-21 y			Acupoint and meridian massage + CR			5*40 min/w					CR		5*40 min/w		8 w		FMA, mBI, Total efficacy
3		Tang JL et al 201135		China		111			56/55		T: 41/15 C: 37/18			T: 52.4 ± 2.3 C: 53.1 ± 1.7			T: 13.6 ± 5.7(d) C: 14.5 ± 4.8(d)			Acupoint and meridian massage + CR			2*10 min/d					CR		-		3 w		mBI, Total efficacy
4		Wu XL et al 201837		China		70			35/35		-			T: 5074 C: 51-75			T: 2-12 w C: 2-11 w			Acupoint and meridian massage + CR			2*30 min/d					CR		2*30 min/d		4 w		mAS (data lack), Total efficacy
5		Sun PP et al 201534		China		60			T: 21/9 C: 25/5		60/0			45-75			1-5 m			Acupoint and meridian massage + CR			5*20 min/w					CR		5*20 min/w		3 w 90 d follow-up		FMA-U, FMA-L, mBI, Rankin
6		Yang YJ et al 201639		China		90			T: 34/11 C: 27/18		-			T: 59.93 ± 16.87 C: 62.73 ± 11.22			T: 3 ± 5 m C: 3 ± 7 m			Tuina therapy + CR			5*40-50 min/w					Placebo-Tui Na therapy (gentle rubbing) + CR		5*40-50 min/w		4 w, 3 m follow-up		FMA-U, FMA-L, mBI, mAS
7		Yue SJ et al 201241		China		69			T: 19/16 C: 18/16		-			T: 68.1 ± 11.7 C: 68.8 ± 10.3			T: 27.8 ± 6.8 d C: 30 ± 5.4 d			Acupoint and meridian massage + CR			30mins/d					CR		-		1 m-hospital, 2 w-home, 2 m-follow-up		FMA, mBI
8		Wang XH et al 201536		China		104			T: 40/12 C: 38/14		-			T: 67.54 ± 11.35 C: 67.27 ± 11.23			-			Acupoint and meridian massage + CR			2*40-60 min/d					CR		-		10 d		FMA
9		Lee JS et al 201033		Korea		40			T: 10/10 C: 9/11		32/6/2			<50:3/1 50-59: 10/17 60-69: 3/2 >70:6/8			<14 d: 11/10 15-30 d: 11/13 31-56 d: 6/5			Acupoint and meridian massage + CR			40 min/d					Placebo massage + CR		30 min/d		2 w		Index finger circumstance, mBI, depression scale
No.		Study		Country		Participates														Treatment							Control					Duration		Outcome
						Total No. of random participates			Sex (M/F)		Type (Ischaemic stroke/ Haemorrhagic stroke/ Mixed stroke)			Age			Disease duration		Main content				Length of session/ frequency				Main content			Length of session/ frequency
10		Hyun SK et al 200828		Korea		56			T: 14/14 C: 17/11		T: 16/12 C: 18/10			<50:5/3 51-60: 8/8 61-70: 10/12 >71: 5/5			<6 m: 6/7 6-12 m: 10/10 >12 m: 4/3		Acupoint and meridian massage + CR				10 min/d		CR					No detailed description		2 w		Grip power, Self-reported pain score ROM, index finger circumstance, mBI, depression scale
11		Yang M et al 201538		China		80		T: 25/15 C: 26/14			-		T: 62.9 ± 3.9 C: 63.0 ± 3.8				T: 2-13 m C: 3-13 m		Acupoint and meridian massage + CR				No detailed description		CR					3*10 min/d, 4d/w		3 m		FMA, mBI, Total efficacy
12		Zhang MN et al 201430		China		80		22/18			-		T: 53.6 ± 10.7 C: 48.5 ± 12.3				2 w-3 m		Acupoint and meridian massage + CR				2*40 min/d,5 d/w		CR					No detailed description		12 w		FMA-L, MT
13		Xu JJ et al 200926		China		166		T: 43/44 C: 49/30			T: 71/16 C: 72/7		T: 67.08 ± 9.42 C: 68.08 ± 9.43				2 w-6 m		Acupoint and meridian massage + CR				No detailed description		CR					No detailed description		4 w		FMA, mBI, CNS, SF-36, MMT, HAMA, HAMD
14		Wu PX et al 201032		China		100		T: 29/21 C: 37/13			T: 43/7 C: 42/8		T: 65.54 ± 11.03 C: 61.7 ± 9. 82				T: 11.9 ± 3.14 (d) C: 11.88 ± 4.24 (d)		Acupoint and meridian massage + CR				30 min/d		CR					Gradually increase time		3 m/6 m		NDS, mBI, SF-36 (modified)
15		Yang GP et al 201525		China		60		T: 19/11 C: 20/10			-		T: 52.13 ± 6. 95 C: 54.28 ± 7. 01				-		Acupoint and meridian massage + CR				2*40 min/d, 5 d/w		CR:					2*40 min/d, 5 d/w		8 w		FMA-L, MWS
16		Liu J et al 201229		China		60		38/22			38/22		57.3 ± 5.8				-		Acupoint and meridian massage + CR				30-50 min/d		CR					30-50 min/d		4 w		Brunnstrom, BBS, mBI
17		Wang YF et al 200240		China		27		T: 12/6 C: 7/2			-		T: 54.778 ± 12.436 C: 54.222 ± 13.517				< 6 m		Acupoint and meridian massage + CR				3*45 min/w		CR:					3*45 min/w		4 w		FMA, mAS, BI, Brunnstrom
18		Pang HP et al 201131		China		80		T: 28/12 C: 26/14			-		T: 55.31 ± 7.54 C: 54.68 ± 8.21						Tui Na therapy + CR				2*40 min * 5 d/w		CR					2*40 min*5 d/w		8 w		FMA-L, MWS, gait analysis
19		Geng XD et al 201824		China		120		T: 29/31 C: 32/28			-		T: 58.2 ± 2.4 C: 57.8 ± 2.1				3-12 w		Tui Na therapy + CR				2*45 min * 5 d/w		CR					2*45 min*5 d/w		8 w		FMA-L, 6MWS, gait analysis

Notes: T: treatment group; C: control group; CR: conventional rehabilitation; d: days; w: weeks; m: months; -: not reported; FMA(U/L): Fugl-Mayer Assessment-(Upper/Lower); mBI: modified Barthel Index; TCM: Traditional Chinese Medicinel; mAS: modified Ashworth; ROM: range of motion; SF-36: the 36-item shot-form health status survey; MMT: manual muscle testing; HAMA: Hamilton Anxiety Scale; HAMD: Hamilton Depression Scale; 6MWS:6 min walking test;

3.3. Assessment of bias risk

Figure 2(a, b) detail the results of our bias risk assessment. With regard to randomisation, seven studies used the random digital table method,^{20,25,27,29,30,34,36} one used computer software to generate random numbers,³⁸ and one used draw,⁴¹ all of which were assessed as ‘low bias risk’. Moreover, three studies were randomly assigned in the order of admission, with a ‘high bias risk’.^31,32,35 With regard to allocation concealment, three studies utilised a random allocation scheme through sealed envelopes,^30,38,39 all of which were ‘low bias risk’. Moreover, three studies were considered ‘high bias risk’ according to admission order.^31,32,35 Only two studies described participant and personnel blinding, both of which were considered ‘low bias risk’.^27,38 With regard to incomplete outcome data, three studies reported attrition and adopted intention-to-treatment (ITT) analysis,^20,38,39 two reported no attrition^{25, 31} and others provided no such description. All 21 studies were assessed to have ‘uncertain risks’ in selective reporting and other biases due to insufficient evidence to determine them.

Figure 2. Risk of bias graph.

A: risk of bias in all included studies; B: risk of bias summary.

3.4. Meta-analysis of the results

3.4.1 Primary outcomes

FMA: as shown in Figure 3, seven studies^{21,26,28,36,38,39,41} utilised FMA to evaluate the ability of meridian massage for improving motor dysfunction after a stroke. Among such studies, two^21,36 had an intervention duration of less than 4 weeks, two^26,28 had an intervention duration of 4 weeks and another two^39,41 assessed FMA at 4 and 12 weeks. Our Meta-analysis showed that the massage group exhibited greater improvement in motor function compared to those received the conventional treatments [MD = 1.16, 95% CI: (0.43-1.89), P = 0.002, I² = 96%, REM, n = 663]. Subgroup analysis showed no significant difference between the massage and control group when the intervention duration was more than 8 weeks but better limb motor function in the massage group than in control group when the intervention duration was less than 8 weeks [intervention duration less than 4 weeks: MD = 0.84, 95% CI: (0.59, 1.10), P < 0.000 01; 4 to 8 weeks: MD = 2.41, 95% CI: (0.80, 4.01), P = 0.003; more than 8 weeks MD = 0.89, 95% CI: (-0.64, 2.42), P = 0.25]. Sensitivity analysis indicated that removing the Lin JZ study²⁸ decreased the heterogeneity from 97% to 39%.

Figure 3. Forest plot for the Meta-analysis on the Fugl-Mayer assessment.

As shown in Figure 4, three studies^34,39,40 used the FMA-Upper for evaluating upper limb motor function. Our Meta-analysis showed no significant clinical difference in upper limb motor dysfunction improvement between the massage and control group [MD = -0.13, 95% CI: (-3.61, 3.35), P = 0.94, I ² = 38%, FEM, n = 177].

Figure 4. Forest plot for the Meta-analysis on the Fugl-Mayer assessment-upper.

As shown in Figure 5, seven studies^{24,25,30,31,34,39,40} utilised the FMA-lower to assess lower limb motor function. Our Meta-analysis showed that the massage group exhibited better improvement in lower extremity motor dysfunction compared to the conventional treatments for motor dysfunction after stroke [MD = 3.21, 95% CI: (1.97, 4.45), P = 0.000 01, I ² = 74%, REM, n = 516]. Sensitive analysis show that I² declined from 74% to 8% after removing the Yang et al’ s study.³⁹

Figure 5. Forest plot for the Meta-analysis on the Fugl-Mayer assessment-lower.

Modified Barthel Index: as shown in Figure 6, 13 studies utilised mBI to assess ADL and motor function.^{20,21,26-29,32,34,35,37-41} Among such studies, five^{20,21,27,34,35} had an intervention duration of less than 4 weeks, four^26,28,29,40 had an intervention duration of 4 weeks and two^39,41 were assessed mBI at 4 and 12 weeks. Our Meta-analysis showed that the massage group showed better improvement daily life ability compared to those treated conventional rehabilitation [MD = 8.87, 95% CI: (4.24-13.49), I² = 94%, REM, n = 1069)]. Subgroup analysis found that the massage group had better improvement in daily life ability compared to the control group at any intervention duration [less than 4 weeks, MD = 8.16, 95% CI: (5.07, 11.24), P < 0.0001; 4 to 8 weeks, MD = 6.38, 95% CI: (2.89, 9.87), P = 0.0003; more than 8 weeks, MD = 14.29, 95% CI: (2.85, 25.72), P = 0.01]. Sensitive analysis showed no change after individually removing the studies.

Figure 6. Forest plot for the Meta-analysis on the modified Barthel index.

Clinical efficiency: As shown in Figure 7, five studies^{20,28,35,37,38} reported on the clinical efficiency of meridian massage treatment. Our Meta-analysis showed that the massage group had better overall clinical effectively compared to the control group [RR = 1.37, 95% CI: (1.24, 1.51), P < 0.000 01, I ² = 29%, FEM, n = 471].

Figure 7. Forest plot for the Meta-analysis with clinical efficiency.

3.4.2 Secondary outcomes

Modified Ashworth Scale: as shown in Figure 8, three studies^20,39,40 utilised the mAS to measure muscle tone. Our Meta-analysis showed that the massage group exhibited better improvement in limb spasm in contrast to conventional rehabilitation [MD = -0.69, 95% CI: (-1.26, -0.11), P = 0.02, I ² = 81%, REM, n = 247].

Figure 8. Forest plot of the Meta-analysis on the modified Ashworth Scale.

Berg Balance Scale: as shown in Figure 9, two studies^28,29 used the BBS to assess postural stability. By contrast, the meridian massage displayed better improvement in balance than the conventional rehabilitation [MD = 5.07, 95% CI: (2.22, 7.93), I ² = 0.0005, I ² = 38%, FEM, n = 120].

Figure 9. Forest plot for the Meta-analysis on the Berg Balance Scale.

3.4.3 Publication bias assessment

More than ten studies utilised only FMA and mBI, for which funnel plots were presented to determine publication bias. Accordingly, Figures 10 and 11 show asymmetric funnel plots, indicting some publication biases. Funnel plots are shown in supplementary figures 10 and 11, respectively.

3.4.4 Safety evaluation

A total of three studies^{31, 34, 39} reported on safety evaluation, whereas 16 provided no such data. Yang et al’ s study³⁹ reported no adverse reaction in any of the cases mentioned at all study sites. The drop out and loss of patients during the follow-up were unrelated to the study intervention. Meanwhile, Pang’s study³¹ showed that none of the cases developed any adverse reactions during treatment, suggesting the considerable safety and reliability of the treatment methods in each group. Sun Per Protocol’s (PP) study provided a brief description of the research process without adverse reactions.

4. DISCUSSION

4.1. Summary of studies

4.1.1 Efficiency summary

A total of 19 studies were included in this systematic review and Meta-analysis, wherein different types of meridian massage were compared to conventional rehabilitation in 1664 patients. Consequently, our Meta-analysis revealed that meridian massage was more effective than conventional rehabilitation in terms of improving motor function (particularly lower limb function), activities of daily life, total clinical efficiency, limb spasm, and balance ability. Moreover, our findings suggested that meridian massage did not have sufficient ability to improve upper limb motor function, which may be related to the innervation pattern of the distal hand and the nature of hand function itself, thereby leading to relatively slow.⁵⁰ Our subgroup analysis indicated that meridian massage was more effective compared to conventional rehabilitation when the treatment duration was less than 8 weeks.

However, no significant differences between meridian massage and conventional rehabilitation had been observed when the treatment duration exceeded 8 weeks, suggesting that meridian massage may be more effective in the short term. The FMA sensitive analysis found the high heterogeneity due to Lin JZ study (patients’ disease duration was 1-21 years), which may have been caused by inappropriate sequelae stage rehabilitation of stroke.⁵¹

4.1.2 Safety summary

Among the studies included in this systematic review and Meta-analysis, 3 reported no adverse effects in the massage group and control group, while 16 provided no data regarding adverse effects. A non-invasive treatment, meridian massage can be considered relatively safe. However, given that none of the included studies mentioned adverse effects during follow-up, its long-term safety remains unclear.

4.1.3 Studies quality summary

Most of the included studies were not of particularly high quality, with the shortcomings being summarised in the following discussion. First, incomplete information made analysis difficult. Among the studies included herein, 17 were completed in China, of which 15 were obtained from Chinese literature databases. Moreover, most of the studies were less than 2 pages, with low integrity of information, resulting in some studies deign determine to have ‘unclear risk’ of bias. Second, the sample sizes were considerably small. Notably, only four studies described sample size calculation. In all included studies, the smallest sample size was only 27 patients, with 3 studies having fewer than 60 patients. The small sample size reduced the generalizability of the conclusions. Third, the characteristics of included patients varied greatly. Different types of strokes and disease durations may alter the effects of meridian massage on motor dysfunction after a stroke, resulting in higher heterogeneity. Fourth, diverse intervention methods had been employed. Notably, selected meridians, manipulation, intervention time and frequency were not consistent among the included studies, resulting in higher heterogeneity. Fifth, the methodological quality was generally low. During the random sequence generation and concealment, three studies used admission order for randomisation and subsequently assessed as ‘high-risk’. Another seven studies merely described their cohort as ‘random’, the accuracy of which could not be determined. The rest of the studies did not describe the concealment of random allocation schemes and were assessed to have a higher risk of selection bias. Secondly, the level of blinding was insufficient. Although blinding among rehabilitation therapists is difficult to achieve, blinding methods for patients and outcome assessors and placebo-controlled methods should be recommended to minimise biases associated with the study design.

Generally, the aforementioned shortcomings reduce the effect of meridian massage on post-stroke limb dysfunction, making it difficult to draw accurate and definite conclusions.

4.2. Limitations and strengths

Our Meta-analysis showed that meridian massage might be a beneficial complementary therapy for improving motor function among patients with stroke. The current study also reported PRISMA Checklist (Table 2) and AMSTAR2 (Table 3). Some limitations of this study need to be addressed. First, although data from different databases had been utilised, several other databases, including clinical trials on meridian massage, exist across Southeast Asia and West Asia, which had not been included herein. Second, despite our attempts at contacting the corresponding authors of the clinical trials by mail for more detailed data and information, none of them replied. Third, the overall methodological quality of included studies is quite poor, thereby increasing the risk of bias. Fourth, the number studies included in this analysis was insufficient, especially for subgroup analysis. For better analysis, future studies should employ the following: (a) the clinical study design should utilise placebo treatment as the control group; (b) double-blind should be used to enhance the reliability of results; (c) adequate confidentiality of random distribution schemes should be ensured; (d) sample size estimates should be carried out prior to testing, and multiple research centres should be included for complete patient recruitment; (e) ITT analysis and PP analysis should be used when analysing the results of a trial involving patients with abscission; (f) rigorous evaluation and reporting adverse events are needed; and (g) long-term follow-up should be adopted to determine the long-term effects and other possible adverse effects of meridian massage therapy.

In conclusion, existing evidence suggest that meridian massage therapy is beneficial for improving motor function (especially that of the lower limbs), daily life ability, limb spasm and balance function of patients suffering from limb dysfunction after a stroke. Moreover, given that no serious adverse effects had so far been reported, meridian massage can be considered a relatively safe non-invasive therapy. However, the long-term efficacy and safety of meridian massage have yet to be determined, while most of the included studies exhibited low methodology quality, small sample sizes, and inappropriate study design. Therefore, more well-designed clinical studies on meridian massage therapy are very much warranted.

We would like to thank Dr. Zhang Lin at the Guizhou University of TCM Affiliate Second hospital for his assistance.