Clinical Efficacy and Safety of Acupressure on Low Back Pain: A Systematic Review and Meta-Analysis

Objectives To evaluate the effectiveness and safety of acupressure on low back pain (LBP). Methods We searched 7 electronic databases and 2 trial registries through December 28, 2020. Randomized controlled trials (RCTs) of acupressure on LBP were considered for meta-analysis with Revman 5.3 and Stata 15.0 software. Methodological quality was evaluated using the Cochrane Collaboration's tool. Trial sequential analysis (TSA) was used to quantify the statistical reliability. HETRED analysis and GRADE were used to determine the heterogeneity and quality of the results, respectively. Results Twenty-three RCTs representing 2400 participants were included. Acupressure was superior to tuina massage on response rate (RR 1.25; 95% CI, 1.16 to 1.35; P < 0.00001) and in the standardized mean difference (SMD) for pain reduction [SMD −1.92; 95% CI, −3.09 to −0.76; P=0.001]. Likewise, acupressure was superior to physical therapy [SMD, −0.88; 95% CI, −1.10 to −0.65; P < 0.00001] and to usual care [SMD, −0.32; 95% CI, −0.61 to −0.02; P=0.04] in pain reduction. The Oswestry Disability Index was significantly improved by acupressure compared with usual care [SMD, −0.55; 95% CI, −0.84 to −0.25; P=0.0003]. The combination of acupressure with either manual acupuncture or electro-acupuncture showed significant improvements over the adjuvant therapies alone in response rate [RR 1.19; 95% CI, 1.13 to 1.26; P < 0.00001], pain reduction, and the Japanese Orthopedic Association score (JOA). However, each study displayed substantial heterogeneity. Through subgroup sensitivity analysis and -HETRED analysis, the heterogeneity of acupressure compared with manual acupuncture decreased while the results maintained significance with respect to pain reduction [SMD −0.9; 95% CI, −1.21 to −0.6; P < 0.00001] and JOA [SMD, 0.66; 95% CI, 0.33 to 0.98; P < 0.00001]. Similar results were obtained comparing acupressure with electro-acupuncture with respect to pain [SMD, −1.07; 95% CI, −1.33 to −0.81; P < 0.00001] and JOA [SMD, 0.89; 95% CI, 0.51 to 1.27, P < 0.00001]. TSA demonstrated the effectiveness of acupressure as a standalone or as a combinative treatment (with manual acupuncture or electro-acupuncture) for LBP. Conclusion Acupressure is an effective treatment for LBP. However, GRADE assessments downgraded the evidence in the trials, indicating that additional investigations are needed to confirm these observations.


Introduction
Low back pain (LBP) affects 540 million people worldwide and is the leading cause of disability [1]. e number of years lived with LBP has increased by 54% between 1990 and 2015 at all income levels and age groups worldwide [1]. In the USA, annual healthcare costs of LBP are estimated to be $100 billion, imposing an economic burden on the healthcare system [2]. An underlying cause for LBP has not been identified for most patients. However, several pathological causes have been identified for some patients such as intervertebral disc damage, vertebral fracture, tumors, infection, and physical and mental comorbidities [3]. Many therapeutic options are currently recommended for LBP, but the emphasis is now on complementary medicine such as exercise and physical therapy rather than on pharmacological and surgical treatments [4].
In traditional Chinese medicine (TCM), LBP is closely associated with the obstruction of the meridians and collaterals of the lower back, which start at the head, descending parallel to the midline and the lateral of the back, and end in the feet. e treatment is based on dredging the obtrusion of these meridians [5]. Acupressure on the meridians and collaterals is a noninvasive and straightforward technique, embracing the principle of Yin and Yang and the close relationship between Qi and blood circulation [6]. Acupressure from fingers, palms, or elbows on the appropriate acupoints helps promote the circulation of Qi and blood in the meridians of the lower back, relaxing muscles, and alleviating LBP [7]. Previous studies have suggested that acupressure may increase the production of endogenous sedatives and analgesics by the stimulation of the autonomic nervous system [8]. e efficacy of acupressure on LBP has been assessed in several meta-analyses. However, these reviews discussed the effectiveness of auricular acupressure on LBP based on limited researches or the acupressure treatments from nontherapists or the use of instrumentation [9][10][11]. We performed this analysis because of the absence of a meta-analysis on the therapeutic effects and safety of traditional acupressure on the effectiveness of traditional acupressure treatment of LBP.

Registration.
e review was registered with the International Prospective Register of Systematic Reviews (PROS-PERO) database (registration number: CRD42020144586).

Data Source and Search Strategy.
e meta-analysis and systematic review were conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [12]. Literature searches were conducted through the following electronic databases from their inception to November 25, 2019: PubMed; Embase; the Cochrane Central Register of Controlled Trials (CENTRAL); Clinicaltrials.gov; the Chinese Scientific Journal Database (VIP); Wan-fang Data; the Chinese Biomedical Literature database (CBM); the China National Knowledge Infrastructure (CNKI); and the Chinese Clinical Trial Registry (ChiCTR). We searched MeSH term trees for "acupressure" and "low back pain" in PubMed, and the keywords searched included "acupressure", "shiatsu", "zhi ya", "low back pain", "lumbago", "low back ache", and "randomized controlled trial". e above keywords were translated into Chinese and searched in the abovementioned Chinese databases as well. Search terms were combined with the Boolean "AND" and "OR" terms in search strategies, e.g., "acupressure" OR "shiatsu" OR "zhi ya") AND ("low back pain" OR "lumbago" OR "low back ache") AND ("randomized controlled trial" ( Table 1 in the Supplement). Selected studies and reviews were screened for relevant trials. Journal editors and experts in the relevant fields were consulted for additional study sources. Finally, Open Grey was searched for qualifying studies.

Types of Study.
Randomized control trials (RCTs) that evaluated the effectiveness of acupressure on LBP were included. No restrictions were set on publication type, language, or status.

Types of Participants.
Patients (≥18 years old) who were diagnosed with LBP were eligible for inclusion without any restriction on sex, nationality, race, the period or nature of prior treatments, past or existing diseases, economic status, or inpatient or outpatient care.

Types of Interventions.
For interventions, acupressure only or adjunctive treatments, such as usual care, sham acupressure, acupuncture, or physical therapy, combined with acupressure was applied in the test group.
For control groups, usual care, sham acupressure, acupuncture, or physical therapy could be applied in the control group. If acupressure was applied with adjunctive treatments, the identical adjunctive treatment should be applied in the control groups as well.

Types of Outcome Evaluations.
Response rate, pain intensity, and functional ability were used as primary outcomes to evaluate the effectiveness of acupressure compared to other treatments. Response rate is defined as the proportion of participants who reported relief from pain or symptoms. Pain intensity was measured by the Visual Analogue Scale (VAS). Since VAS has either 0-10 cm or 0-100 mm range, the values are normalized up to a 0-10 cm range for comparability. For functional ability, we used the Oswestry disability index (ODI) and Japanese Orthopedic Association scores (JOA). e JOA has either 0-17 or 0-29 scores, and all 0-17 scales are normalized to 0-29 scores for comparability.
Adverse events during treatments and the usage count of each acupoint selected in the articles were analyzed as secondary outcomes.

Exclusion Criteria.
Studies applying auricular acupressure or acupressure on specific reflexology areas of the hands or feet were excluded to study the effectiveness of acupressure on the meridians and collaterals of the body. Investigations in which acupressure was undertaken by therapeutic devices or non-medical staff were excluded as well.

Data Extraction.
Data extraction was completed in duplicate and independently by two reviewers (TL and XL). Specific characteristics were extracted: author; publication year; sample size; the detail of the intervention and comparison; frequency; duration of treatments; follow-up intervals; and outcomes. We contacted the authors of the studies twice over six weeks via e-mail for missing or unclear data. If no response was received, the data were marked as unclear. A third reviewer (QT) assisted for consensus when disagreement occurred.

Assessment of Quality and the Level of Evidence.
Methodological quality was assessed with the Cochrane Collaboration's tool for assessing risks of bias from 7 domains: random sequence generation; blinding of patients; allocation concealment; selective reporting; incomplete outcome data; and other biases [13]. Each of these domains was rated as low, high, or unclear risk. Each study was classified at one of three levels: low, high, or unclear risk of bias based on the overall risk assessments of the 7 domains. e quality of the body of evidence for the different outcomes was assessed with the GRADE approach (GRADE pro, Version 3.6 for Windows, Grade Working group) [14].

Statistical Analysis.
We analyzed all outcomes with Stata 15.0 (Stata Corp, College Station, TX, USA). Studies were arranged according to the outcome measures and the type of intervention (acupressure monotherapy or in combination with acupuncture). Pooled dichotomous outcomes were presented as risk ratios (RRs) and 95% confidence intervals (CIs) while pooled continuous outcomes were presented as weight mean differences (WMDs) with 95% CIs or standardized mean differences (SMDs) with 95% CIs when using multiple measurements. e fixed-effect model was applied if heterogeneity was not significant (I 2 ≤ 50% or P ≥ 0.10 in the Q test). Otherwise, subgroup analysis was performed to identify the source(s) of the heterogeneity. Planned subgroup analyses included the types of adjunctive interventions (manual acupuncture or electro-acupuncture) and the duration of treatments (≤2 weeks or > 2 weeks). Sensitivity analysis and HETRED analysis were conducted to further identify the trials resulting in significant heterogeneity [15]. If the source(s) of the heterogeneity was still unidentifiable, a random effect model was utilized for data synthesis.

Publication Bias.
Publication bias was assessed using funnel plots and Egger's test quantitatively when at least 10 trials were included. Significant publication bias was considered when P < 0.05.

Trial Sequential Analysis.
Trial sequential analysis (TSA) was conducted to detect the robustness of the results of the meta-analysis by calculating the required information size (RIS) with TSA 0.9.5.5 software (Copenhagen Trial Unit, Centre for Clinical Intervention Research, Rigshospitalet, Copenhagen, Denmark) [16]. We took the sample size as RIS.
e required indexes for TSA were type I error (α � 0.05) and type II error (β � 0.2).

Literature Search Results.
e Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram ( Figure 1) shows the entire review process from the original search to the final selection of studies. Twenty-three studies [8, with a total of 2400 participants were included. e characteristics of the included RCTs are presented in Table 1. Table 1 summarizes the characteristics of 23 RCTs in this analysis. In seven studies [20,23,28,33,34,36,38], acupressure was compared with tuina massage (TM). In these studies, acupressure was distinguished from TM by the methods recorded, and no acupressure techniques were involved in TM in the control group. In three studies [18,19,26], acupressure was compared with physical therapy (PT). In two of these trials [18,19], sample size, patient characteristics, and evaluation of outcomes differed but otherwise used similar research designs and interventions. is indicated that these studies were not duplicates. In seven studies, acupressure combined with electro-acupuncture was compared to electroacupuncture alone [21,25,[29][30][31][32]37]. In four studies, acupressure combined with manual acupuncture was compared to manual acupuncture alone [22,24,27,35]. In two of these studies [24,31], the frequency and duration of the acupressure treatments differed between trials. In two studies [8,17], acupressure was compared with usual care (UC), which included conventional medication or health education. In 17 studies [8, 20-23, 25-27, 29, 30, 32-38], acupressure was performed at least once a day. In three studies [31,35,37], acupressure combined with acupuncture was performed once every other day. In two studies [17,28], acupressure was performed once a week. e duration of treatment periods among trials ranged from 2 weeks to 5 weeks (Table 1).
Four studies [28,33,34,38] assessed the analgesic effect of acupressure compared with tuina massage. e pooled data suggested a greater pain reduction, although with substantial heterogeneity (I 2 � 97%). Owing to the similar clinical characteristics in these studies, we were unable to perform subgroup analysis. Sensitivity and HETRED analysis detected no potential source of heterogeneity. After confirming the data accuracy with the studies' authors, we adopted the random effect model and the subsequent analysis demonstrated favorable effects of acupressure [SMD, −1.92; 95% CI, −3.09 to −0.76; P � 0.001] ( Figure 6).   Evidence-Based Complementary and Alternative Medicine Six studies [21,22,24,25,30,35] reported the effect of combination therapy of acupressure and acupuncture compared with acupuncture alone on pain intensity. Less pain intensity was indicated when using the combined treatment of acupressure and acupuncture [SMD, −1.13; 95% CI, −1.31 to −0.94; P < 0.00001]. However, since significant heterogeneity existed among these trials (I 2 � 68%), a subgroup analysis was performed ( Table 2). In the subgroup, the pooled result favored the combinative therapy of acupressure and electro-acupuncture with low heterogeneity [SMD, −1.07; 95% CI, −1.33 to −0.81; P < 0.00001; heterogeneity, I 2 � 0%; P � 0.37] (Figure 8(a)). In the subgroup for acupressure with manual acupuncture, the heterogeneity remained considerable (85%). Sensitivity and HETRED analyses were therefore performed (Figure 7). After removing the study from Zeng and Zhao [21], the I 2 value decreased (85% to 2%) (Figure 8(b)).

Japanese Orthopedic Association Score (JOA).
e effects of acupressure plus acupuncture compared to acupuncture on LBP using JOA in 6 studies [21,22,24,25,30,35] were examined. Because substantial heterogeneity was found in these studies (I 2 � 84%), we performed a subgroup analysis based on the duration of treatments or types of adjunctive interventions ( Table 2). In the subgroup for the combination of acupressure with manual acupuncture, the pooled data in fixed model demonstrated a significant increase of JOA scores [SMD, 0.66; 95% CI, 0.33 to 0.98); P < 0.0001; heterogeneity, I 2 � 0%; P � 0.70] (Table 2 and Figure 9(b)). In the subgroup of acupressure with electro-acupuncture compared with electro-acupuncture alone, the heterogeneity was still considerable (I 2 � 83%). e following sensitivity and HETRED analysis indicated that the inclusion of two studies [21,25] might be the cause of the high heterogeneity ( Figure 10). After removing these studies from the analysis, the pooled results in the fixed model showed a significant improvement in the test group [SMD, 0.89, 95% CI, 0.51 to 1.27; P < 0.00001; heterogeneity, I 2 � 0%; P � 0.36] (Figure 9(a)).

Adverse Events.
Adverse events reported in studies were sparse. Six studies [17,19,22,32,33,35] [17,32] reported the total number of the symptoms and the solutions to adverse events. e incidence of adverse reactions was 0.42% (5/1200) in the test groups compared with 0.25% (3/1200) in the control groups. e primary adverse reactions observed in the treatment group included muscle pain or headache while those in the control group included dizziness, urticaria, and abdominal pain. e adverse events of the two groups were tolerable and did not require specific interventions.

Usage Counts of the Acupoints.
e usage counts of each acupoint selected for treatment were calculated. e most frequently-used acupoints were Weizhong (BL40), Huantiao (GB30), Chengshan (BL57), Dachangshu (BL25), Ashi points, Yanglingquan (GB34), Kunlun (BL60), Zhibiao (BL54), Yinmen (BL37), and Shenshu (BL23) (Figure 12). (TSA). TSA was undertaken with the data from 2 meta-analyses. A TSA on the comparison of the response rate of acupressure with that of tuina       Figure 11: Forest plot of ODI massage revealed that the cumulative Z-curve crossed the traditional boundary of 5% significance (horizontal line) and the monitoring boundaries (inward sloping curves) (Figure 13(a)). After the inclusion of Wang 2010, the significance of 5% was reached every time a new trial was added to the meta-analysis. e results support the conclusion that acupressure improves the response rate over that of tuina massage with RR � 1.25.

Trial Sequential Analysis
In the comparison between acupressure plus acupuncture versus acupuncture, TSA revealed that the cumulative Z-curve crossed the traditional boundary of 5% significance and the monitoring boundaries (Figure 13(b)), indicating that the sample size achieved the required 150 participants and confirmed that the combinative treatment could improve the response rate on LBP (RR � 1.19).

GRADE.
e grade of the evidence obtained for the response rate using JOA, VAS, or ODI for acupressure was low or very-low (Table 3). ese results were similar to the grade of evidence for the combinative treatment of acupressure and acupuncture. Only the trials comparing acupressure with physical therapy provided moderate-grade evidence (Table 3). e reasons for downgrading the evidence were the poor methodological quality, high heterogeneity, wide confidence intervals, and insufficient sample size among relevant trials. Wide confidence intervals and small sample sizes also contributed to the downgrading of the evidence. No serious indirectness was identified. e summary of findings and evidence profile is presented in Table 3.

Discussions
e systematic review and meta-analysis included 23 RCTs with 2400 participants with LBP. Consistent with previous systematic reviews [6,40], moderate-quality evidence revealed an association between acupressure and greater pain relief compared with physical therapy. Although rated as very-low to low, poor quality evidence suggested that acupressure, with or without combinative acupuncture therapy, contributed to a greater amelioration of pain and functional disability from LBP compared with usual care, tuina massage, or acupuncture. TSA results revealed that adequate studies supported the significance of the clinical response rate of acupressure, with or without combinative acupuncture therapy, compared to other treatments.
Although positive results suggesting acupressure as a standalone or as a combinative treatment for LBP, highquality evidence was insufficient to make an informed decision. e methodological limitations of the included RCTs may have impacted the accuracy and reliability of the evidence. Owing to the complex and variable characteristics of therapies like acupressure, acupuncture, and massage, most studies used a nonblinded pragmatic trial to study the efficacy of the therapies, which may have increased their performance bias. Rigorous research methods, such as sham-controlled double-blinded designs, are necessary to minimize bias in evaluating the effectiveness of acupressure. Most outcome measures were dependent on patient-reported scales or questionnaires (JOA, ODI, and VAS), which are vulnerable to the subjective conditions of the patient and/or assessor. Physiological outcome indicators, including the range of motion, pain threshold, and muscle tone in the lower back, are encouraged to be included in future planned research protocols. Only 3 RCTs conducted mid-to-longterm follow-up indicating that the long-term efficacy of acupressure remains to be considered. e most frequently-used acupoints were located on the bladder and the gallbladder meridians. In TCM theory, these acupoints are closely correlated with the liver and kidney, which govern the function of bone and tendon. Acupressure over these points could promote the Qi and blood circulation of these specific meridians to improve the body's ability to recover the proper functions of muscles, tendons, and bones in the lower back [19]. Modern medical studies have suggested stimulation of acupoints could enable the production of endogenous opioids (endorphin) and certain peptides, which act as both analgesics and sedatives and alleviate LBP [41][42][43].
When acupressure was compared solely or as an adjunctive with acupuncture, substantial heterogeneity was found among the study's results. We conducted subgroup analyses on the duration of treatment and adjunctive interventions and found that adjunctive interventions might be contributing to the heterogeneity. However, considerable heterogeneity remained unexplained in the pooled results. In the comparison between acupressure and Tuina massage on pain intensity, the heterogeneity was substantial and unexplained. Acupressure is sometimes considered a subapplication of tuina massage since both are noninvasive techniques. However, the locations of interventions and operational techniques between the two forms are different. Tuina massage includes different techniques such as rolling, pushing, shaking, kneading, scrubbing, or stretching. We speculate that details of the tuina techniques, such as types, Evidence-Based Complementary and Alternative Medicine strength, angle, and duration of each movement, as well as details of the treatments, such as frequency and duration of the treatment, might result in the heterogeneity found in our analysis. Recording these details could improve assessments of the effectiveness of acupressure, but few studies have collected such information. Sparse and mild adverse events were reported in the RCTs, consistent with findings in the previous reviews [6,40,44]. However, the unclear methodological protocols for assessing adverse events, the unclear predictors of these events, and the limited number of RCTs reporting them contribute to a low overall apparent incidence of adverse events. erefore, therapists need to fully inform patients of the potential risks of adverse events and to pay close attention to their potential occurrence. e most important limitation in this systematic review was the limited number of studies with a low risk of bias. Only 5 eligible RCTs (22%) were rated as low risk of bias in the overall assessment. Most of the studies included in this review were evaluated as "high" or "unclear" risk of bias. We did not conduct a funnel plot and Egger's test to detect publication bias since there were fewer than 10 studies in each comparison. Although analysis of combinative MA and EA individually suggested that each was superior to acupuncture alone, the combinative therapies that included acupressure combined with acupuncture in test groups had high heterogeneity, making it difficult to determine the efficacy of acupressure. e outcome measures could have had high heterogeneity because most of the RCTs used patient-reported scales or questionnaires, which could be influenced by memory or emotional bias, undermining the reliability of the results. Moreover, there is no study investigating the effectiveness of acupressure on acute low back pain. Most studies included patients with chronic low back pain, and no clear distinction of patients' histories was made in the remaining studies.

Conclusion
We demonstrated that acupressure could provide clinical benefits to LBP conditions and had a significant short-term response rate in LBP management. However, the overall reliability of this conclusion is limited by the methodological quality of the included trials. Better designed large-scale RCTs using reliable quantitative methods are necessary to confirm the efficacy of acupressure for LBP.

Data Availability
e data supporting this systematic review are from previous studies and datasets, which have been cited. e processed data are available from the corresponding author upon request. Note: (1) e basis for the assumed risk (e.g., the median control group risk across studies) is provided in footnotes. e corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). (2) GRADE Working Group grades of evidence: high quality: further research is very unlikely to change our confidence in the estimate of effect; moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate; low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; very low quality: we are very uncertain about the estimate. (3) a: randomization and blinding are not adequate or appropriate; b: high heterogeneity; c: confidence intervals are too wide; d: insufficient sample size.

Evidence-Based Complementary and Alternative Medicine 13
Disclosure e funding bodies play no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

Conflicts of Interest
e authors declare that they have no conflicts of interest.