The Effectiveness of Massage Therapy for Improving Sequelae in Post-Stroke Survivors. A Systematic Review and Meta-Analysis

Objective: To assess the effect of therapeutic massage for improving sequelae in stroke survivors. Methods: A systematic review of the nine medical databases from January 1961 to December 2020 was carried out. The bibliography was screened to identify randomized controlled clinical trials (RCTs). Two reviewers independently screened references, selected relevant studies, extracted data and assessed the risk of bias using the PEDro scale. The primary outcome was upper and lower limb motor function and spasticity. Results: A total of 3196 studies were identified and 18 RCT were finally included (1989 individuals). A meta-analysis of RCTs in the comparison of Chinese massage (Tuina) plus conventional physiotherapy versus conventional physiotherapy was performed. The mean difference (MD) in the subacute stage on upper limb motor-function using the Fugl Meyer Assessment was 2.75; (95% confidence interval (CI) from 0.97 to 4.53, p = 0.002, I2 = 36%). The MD on upper limb spasticity using modified Ashworth scale was −0.15; (95% CI from −0.24 to −0.06, p < 0.02, I2 = 0%).The MD on lower limb spasticity was −0.59; (95% CI from −0.78 to −0.40, p < 0.001, I2 = 0%) in the endpoint. Conclusions: Therapeutic massage, especially Tuina, in addition to conventional therapy is effective for improving motor function and for reducing spasticity in stroke survivors.


Introduction
Stroke is a leading cause of long-term adult disability. The individuals that have suffered from a stroke present various upper motor neuron syndrome, signs and symptoms, including weakness, spasticity, lack of coordination and agonist antagonist co-contraction, with up to 50% of survivors being chronically disabled [1]. They involve together in impairments and functional problems that can lead to costly complications. [2]. Physical therapy may contribute to the improvement of disabilities and quality of life in these individuals [3].
Manual therapeutic massage is the most applied type of passive physical therapy and it is one of the oldest forms of medicine known to humanity, having been practiced worldwide since ancient times [4]. All massage manipulations introduce mechanical forces into the soft tissues by means of "mechanotransduction" [5]. Massage can increase muscle mass temperature and blood flow, and this might help to increase muscle compliance and minimize muscle stiffness [6]. There are several kinds of therapeutic massage. The most

Protocol and Registration
We conducted a systematic review of the scientific literature to assess the effect of therapeutic massage interventions in stroke survivors. The PROSPERO (register of systematic reviews) number was CRD42020178942. The guidelines in the Cochrane Handbook for Systematic Reviews of Interventions Version 6 [18] and PRISMA [19] statement were followed.

Eligibility Criteria
Studies in Chinese, Spanish, French, Italian, Portuguese and English published between January 1961 and December 2020. Since only RCTs were to be analysed all other forms of literature were excluded as comments, reviews, observational studies, books, poster/oral abstract communications, case reports, non-randomized studies, protocols, systematic reviews and practice guidelines.
Therapeutic massage was compared to no treatment, sham treatment or active treatment. We defined therapeutic massage in this revision as "a patterned and purposeful soft-tissue manipulation accomplished by the use of digits, hands, forearms, elbows, knees and/or feet, with or without the use of emollients, liniments, heat and cold with the objective of therapeutic change" [20].

Data Items
The search strategy was designed using the PICO framework. (P) Adult post-stroke survivors, (I) receiving therapeutic massage alone, or combined with another rehabilitation approach or conventional physiotherapy, (C) compared to patients receiving another rehabilitation approach or conventional physiotherapy and (O) changes in the analysis of upper/lower limbs motor function, spasticity, activities of daily living, anxiety, pain, balance, gait, stroke disability and quality of life with or without a follow-up assessment.
The primary outcomes were upper/lower motor function and spasticity. The secondary outcomes were activities of daily living, pain, anxiety, balance, gait, stroke disability and quality of life.

Search
A computerized search strategy of the following databases was performed: Medline/PubMed, Cochrane Central Register of Controlled Trials (CENTRAL), Physiotherapy Evidence Database (PEDro), Scielo, Tripdatabase, Web of Science, Scopus, CINHAL and Epistemonikos. A manual search was also performed. We used the Medline search strategy, and adapted it to other databases (Appendix A).
Searches for eligible articles and data extraction were conducted independently by three authors (VG-R/JC-S/RC-V). Proquest Refworks discarded duplicate articles, and the remaining studies were analyzed for their appropriateness. Selection was initially based on the title or abstract, and subsequently on the full text of the articles. They were thoroughly checked to confirm the selection criteria. We analyzed whether the studies included followed the template for the intervention's description and the replication checklist (TIDieR) [21]. The following data were extracted: patient stroke characteristics, number of participants in each group, type of massage, co-interventions, zone of massage application, number of sessions, comparisons, outcome measures and tool used, follow up and main results. Conventional physiotherapy interventions were described.

Risk of Bias in Individual Studies
The risk of bias assessment was assessed by two authors (J.A.-K. and P.S.-L.) using the PEDro scale [22]. In cases of doubt or disagreement, a discussion took place with a third reviewer (R.C.-V.) until a consensus was reached. The PEDro is an 11-item scale, in which the first item relates to external validity (not used to calculate the total score) and the other 10 items assess the internal validity of a study. The higher the score, the greater the study's risk of bias as assessed by the following cut-points: 9-10: excellent; 6-8: good; 4-5: fair; <4: poor [23].

Synthesis of Results
Data from the studies were summarized narratively using text and tables. Studies were grouped by comparator. A meta-analysis was performed whenever possible. The treatment effect sizes were calculated using the Revman 5.3 software package [24], based on the mean scores and standard deviations of the studies. Post-intervention effects were analyzed by calculating the change between the baseline and the immediate post-intervention assessment and persisting effects by computing the change between the baseline and the final follow-up assessment. These changes were compared between groups. When the outcomes were continuous and measured in the same unit, a mean difference was used; otherwise, a standardized mean difference was used. The effect size was categorized as 0.2, 0.5, 0.8 and 1.3, which were considered small, medium, large and very large, respectively. Funnel plots were used to illustrate the risk of publication bias [25].
The heterogeneity was assessed visually by means of forest plots and by reporting the I 2 statistic. The I 2 statistic describes the percentage of total variation across studies that are attributable to heterogeneity rather than chance. A value greater than 25% is considered to reflect low heterogeneity, 50% moderate, and 75% high heterogeneity [26]. The fixed effect model was applied by default, and the random-effect model was used in cases of substantial heterogeneity [27]. When there was insufficient data for quantitative analysis, the review only represents and summarizes the evidence. Missing data was requested by contacting the corresponding author.

Risk of Bias within Studies
The mean PEDro score assessing the risk of bias was 6.3 points (range 4-9) from 10 criteria (Supplemental Table S1), indicating a good score. Only one study [35] with 9 points was found, because is difficult to blind the therapist and patients in a physical therapy intervention.

Risk of Bias within Studies
The mean PEDro score assessing the risk of bias was 6.3 points (range 4-9) from 10 criteria (Supplemental Table S1), indicating a good score. Only one study [35] with 9 points was found, because is difficult to blind the therapist and patients in a physical therapy intervention.

Study Characteristics
An overview of the studies included and the patients' characteristics was provided ( Table 1). The total population analyzed included 1989 individuals, of whom 1273 were male and 683 were female. Ahmed (2015) did not specify the gender distribution, 1057 had suffered from an ischemic stroke and 511 were hemorrhagic, and five studies did not report this information. There were 439 patients with left side hemiparesis, 420 had right side hemiparesis and it was bilateral in 63 individuals and not specified in eleven studies. The age of the participants ranged from 32 to 86 years old and the average of most studies was 60 years. Four studies were multicenter. Most of the studies performed their intervention in the subacute stroke phase (≤3 months), while only three studies recruited a mixture of chronic and subacute phase. Three authors did not specify which phase.
Regarding outcomes ( Table 2) eleven studies assessed motor function and ten of them used the Fugl Meyer Assessment [46]. Six authors evaluated spasticity, and all authors used the modified Ashworth Scale [47]. Most of the studies performed their intervention in the subacute stroke phase (≤3 months), while only four studies [30,32,36,38] recruited a mixture of chronic and subacute phase. Three authors did not specify which phase. Only four studies [35,36,42,45] performed a follow-up of 3 months, one had a follow-up of 6 months, and one had a follow-up of 3 days. Wang (2019) was the study that recruited the most patients, with 397.    As regards as therapeutic massage only four studies [28,30,33,41] performed massage alone in the experimental group. Twelve studies performed Tuina massage, six of them combined it with conventional physiotherapy, three performed Tuina with acupuncture, two performed Tuina with acupuncture plus conventional physiotherapy, one performed Tuina in combination with medicinal herbs. Three studies performed Dalk massage, one performed slow-stroke back massage; another performed Thai massage and one performed Swedish massage in combination with footbath and conventional physiotherapy. The number of massage sessions ranged from 7 to 40, the frequency was usually once a day and the duration of treatment ranged from 1 to 8 weeks, with a massage time ranging from 10 to 60 min. Most of the studies involved daily sessions except for Pan (2011) and Han (2015), which involved 2 sessions per day lasting 40-50 min per massage session (Supplemental Table S2).

Efficacy of Therapeutic Massage
Two comparisons were performed: (1) Tuina massage plus conventional physiotherapy versus conventional physiotherapy; (2) Tuina massage plus acupuncture versus conventional physiotherapy. A meta-analysis was performed for motor function, spasticity, and activities of daily living.

Activities of Daily Living
The first comparison was performed. The MD for activities of daily living at the endpoint evaluated by the Barthel index was 1.91; (95% CI from −0.98 to 4.80, p = 0.20) and in the long-term it was 0.26 (95% CI from −3.19 to 3.71, p = 0.88) (Figure 7).

Activities of Daily Living
The first comparison was performed. The MD for activities of daily living at the endpoint evaluated by the Barthel index was 1.91; (95% CI from −0.98 to 4.80, p = 0.20) and in the long-term it was 0.26 (95% CI from −3.19 to 3.71, p = 0.88) (Figure 7).

Activities of Daily Living
The first comparison was performed. The MD for activities of daily living at the endpoint evaluated by the Barthel index was 1.91; (95% CI from −0.98 to 4.80, p = 0.20) and in the long-term it was 0.26 (95% CI from −3.19 to 3.71, p = 0.88) (Figure 7).

Anxiety and Stress
Anxiety and stress were evaluated only in three studies with different interventions, involving Swedish, Thai and slow-stroke back massage, and all had positive results.

Pain
Tuina massage was effective at reducing shoulder-hand pain. One study Yang (2011) also performed it in combination with conventional physiotherapy, while another Li (2012) performed it in combination with electrical acupuncture.

Gait and Balance
Regarding gait, Pan HP (2011) found that Tuina massage improves it and Zarnigar (2012) improves it with Unani exercises. Wu (2013) observed that balance with Tuina massage plus balance training was statistical significant better than in the control group.

Adverse Events
No adverse events were reported during therapeutic massage treatment. All adverse events were considered unrelated to the rehabilitation techniques. The drop-out and loss of patients during the follow-up were unrelated to the study intervention.

Risk of Bias Publication
All the funnel plots were symmetrical, and as such, publication bias is low, the following are available online (Supplemental Figure S1-5).

Discussion
The main findings of this systematic review and meta-analysis are that therapeutic Chinese massage (Tuina) combined with conventional physiotherapy is an effective method to improve motor function and to reduce spasticity in stroke survivors, especially in subacute stage. The combination of Tuina massage plus acupuncture also improves the symptoms. The results of this review are important, because the therapeutic massage intervention was mostly performed in the subacute stage of the stroke, with positive results in upper/lower limbs motor function. Recovery of upper limbs function remains a major scientific, clinical and patient priority [48].
Surprisingly this review only found one study that used the Swedish massage as an intervention. It was used to decrease anxiety. This was unexpected to us, because Swedish

Anxiety and Stress
Anxiety and stress were evaluated only in three studies with different interventions, involving Swedish, Thai and slow-stroke back massage, and all had positive results.

Pain
Tuina massage was effective at reducing shoulder-hand pain. One study Yang (2011) also performed it in combination with conventional physiotherapy, while another Li (2012) performed it in combination with electrical acupuncture.

Gait and Balance
Regarding gait, Pan HP (2011) found that Tuina massage improves it and Zarnigar (2012) improves it with Unani exercises. Wu (2013) observed that balance with Tuina massage plus balance training was statistical significant better than in the control group.

Adverse Events
No adverse events were reported during therapeutic massage treatment. All adverse events were considered unrelated to the rehabilitation techniques. The drop-out and loss of patients during the follow-up were unrelated to the study intervention.

Risk of Bias Publication
All the funnel plots were symmetrical, and as such, publication bias is low, the following are available online (Supplemental Figures S1-S5).

Discussion
The main findings of this systematic review and meta-analysis are that therapeutic Chinese massage (Tuina) combined with conventional physiotherapy is an effective method to improve motor function and to reduce spasticity in stroke survivors, especially in subacute stage. The combination of Tuina massage plus acupuncture also improves the symptoms. The results of this review are important, because the therapeutic massage intervention was mostly performed in the subacute stage of the stroke, with positive results in upper/lower limbs motor function. Recovery of upper limbs function remains a major scientific, clinical and patient priority [48].
Surprisingly this review only found one study that used the Swedish massage as an intervention. It was used to decrease anxiety. This was unexpected to us, because Swedish massage is today the most popular and best-known type of massage in the Western world [49]. We expected to find more studies as other authors had used Swedish massage to improve spasticity and motor function in multiple sclerosis [50,51] and cerebral palsy [52,53]. Scientific publications in Europe, America, Africa or Australia were not found, they were only found in Asian countries, and especially in China.
The upper and lower limbs motor function was the outcome most evaluated, followed by spasticity the two outcomes are linked [54]. Motor impairments in stroke survivors can be described by a cycle of overactivity-contracture-overactivity evolving in parallel with the continuum of paresis-disuse-paresis. Both cycles must be disrupted to optimized motor recovery and function [55]. In fact, a more complete restoration of motor function is achieved when spasticity is absent [56]. It is important to reduce spasticity before the patient performs the voluntary movement in order to obtain a movement with some quality since this will influence the neuroplasticity of the individuals and their recovery [57]. According to several authors [54,58,59] the sensory system has an important role in spasticity mitigation and is the most important predictor for severe spasticity.
The mechanism behind elastic modulus changes in spastic muscle in stroke survivors is still under discussion. One possible hypothesis might be related to structural alterations in the muscle after a stroke. Shortened muscle fascicle length in the upper limb [60] and lower limb [61] has been observed. These results suggest that altered muscle morphology of the paretic muscle may contribute to abnormal muscle elastic properties during passive stretching [62].
As a result of damages to the motor cortex and its descending pathways and the subsequent unmasking of inhibition, there is evidence of upregulation of reticulospinal tract projections excitability on the contralesional side in stroke survivors [63]. Reticular nuclei receive sensory input from the periphery and neck proprioceptors. In addition to sensorimotor integration, the reticular formation also seems to play a role in preparation for a voluntary movement [64]. Therapeutic massage increases blood flow and parasympathetic activity, releases relaxation and stress hormones, and inhibits muscle tension, and neuromuscular excitability [65]. It could reduce the hyperexcitability of the reticulospinal tracts. The various types of therapeutic massage modalities could be most useful for the therapist to reduce muscle overactivity to enable other therapeutic interventions.
The results obtained for daily living activities, gait, balance, quality of life and stroke severity were inconclusive. The trend is positive when Tuina is used in addition to the conventional therapy or acupuncture. Regarding pain, our results are consistent with the literature available [66]. There is growing evidence to support the concept of an interactive network between the cutaneous nerves, the neuroendocrine axis and the immune system [67]. Therapeutic massage is reported to have several beneficial effects, including activation of the relaxation and growth response has been suggested to be mediated by oxytocin [68]. Stroke patients suffer from anxiety and massage could be helpful for relaxation to ease the patients' suffering [69]. The results of this review show positive effects, as it reduced anxiety in stroke survivors. Surprisingly, no study evaluated the range of motion as an important aspect to take into account when reducing spasticity [70].
Functional magnetic resonance imaging data have suggested that moderate pressure massage with movement is represented in several brain regions, including the amygdala, the hypothalamus and the anterior cingulate cortex, which are all areas involved in stress and emotion regulation [13,71]. Findings from the whole-brain meta-analysis of right-hand tactile stimulation highlight the importance of taking bilateral activation into consideration, particularly in the secondary somatosensory cortex [72].
Most of the articles in this systematic review used Tuina massage for improving outcomes. This is one of the four main branches of traditional Chinese medicine. However, although its roots in China are ancient, it is still relatively new in the West [9]. Tuina massage was originated from China over 5000 years ago and is commonly known today as "the grandfather of all therapeutic massage therapies". It follows the meridian theory and works on the organs, energy channels in muscle groups and points on the body using the same principles as acupuncture, except hands and fingers are used instead of needles [73].
It is combined with anatomical and pathological diagnosis in order to achieve dredging meridian, removes pathogenic factors and has a curative effect of a harmonic balance of Yin-Yang [74]. Tuina can act on the subcutaneous muscular layer; enhance local blood circulation, lymph circulation, tissue metabolism of the skin, can regulate physiological and pathological states, unblock meridians, and harmonize Qi (total life energy). In Chinese medicine, Qi disorder and Yin-Yang imbalance account for balance disturbances following stroke. It corrects an imbalance in the yin and yang and qi (energy) which when translated into Western medical terminology, can also been understood as the "modulation of the imbalance between parasympathetic and sympathetic activity". Yin-Yang imbalance also contributes to upper and lower limb spasticity following stroke, manifesting as 'flaccidity of Yang and spasm of Yin' [75].
Tuina massage encompasses techniques as grasping, pressing, rolling, round rubbing, holding-twisting, rub rolling, pushing, kneading, rotating, shaking, wiping, vibrating, digital striking, knocking, chapping, pressing, acupressure, myofascial release, reflexology, stretching techniques and joint mobilizations applied to specific body points [76]. Tuina is a functional massage and it can input substantial proprioceptive sensory impulses to the central nervous system through muscle, tendon and joint motion [77].
No study reports adverse events, but massage therapies are not totally devoid of risks. The incidence of adverse events is unknown, but is probably low [78]. The massage itself does not increase a person's risk of stroke, but some precautions need to be taken with certain individuals. If the individual has blood clots, there is a small chance they could be dislodged by massage. Individuals on blood thinner medication bruise more easily, so deep tissue massage should be avoided. Care should be taken around the neck area in the region of the carotid artery, but this should not be an issue with an experienced massage therapist.
Further investigations are required at both the experimental and clinical levels to compare therapeutic Chinese massage (Tuina) versus Swedish massage in stroke survivors. Surprisingly, therapeutic massage is not on the list of recommendations in Western stroke management guides. Rehabilitation therapy based on integrated Chinese and Western medicine could be effective for stroke survivors [79]. Finally, touch a patient has a therapeutic value and it has many benefits [80].
The current review has several limitations. First, the majority of patients in this review are Asian people. Second, only one Swedish massage article was found for stroke survivors. We do not know if Tuina massage will be effective in Western population.

Conclusions
The results of this systematic review suggest that therapeutic Chinese massage (Tuina) in addition to conventional physiotherapy is an effective non-invasive treatment for improving upper/lower limbs motor function and for reducing spasticity especially in the subacute stage of stroke.

Acknowledgments:
We would like to thank to Xinyi Li for translating several articles from Chinese to Spanish language and Selma Peláez Hervás for revising English language.

Conflicts of Interest:
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.