Acupoint Injection for Nonspecific Chronic Low Back Pain: A Systematic Review and Meta-Analysis of Randomized Controlled Studies

Introduction Nonspecific chronic low back pain (NCLBP) became a public health and economic problem. Acupoint injection was used widely for patients with NCLBP. However, there were inconsistent results on the efficacy for these people. Therefore, this review was performed to systematically assess the efficacy and safety of acupoint injection. Materials and Methods The literature sources were collected via EMBASE, Medline, CENTRAL, CINAHL, CNKI, VIP, Wanfang, and Sino-Med Database from their inception to October 13, 2019. Endnote X7, widely used document management software, was used to manage and screen the literature sources. Each record was screened according to the predetermined inclusion criteria by two review authors independently. Quality assessment tool, “Risk of table,” was used to assess the quality of the included studies according to the recommendation of the Cochrane Handbook for Systematic Reviews of Interventions. Data extraction was performed by one reviewer and verified by another reviewer. Any disagreement was addressed via consulting with a third reviewer in the abovementioned processes. All procedures were performed according to the Cochrane Handbook for Systematic Reviews of Interventions. Results This review included 13 studies involving 1381 patients with NCLBP. Quantitative analysis results indicated that there is no sufficient evidence that acupoint injection can improve the pain of patients with low back pain based on two trails: Visual Analogue Scale (VAS: MD = −1.33, 95% confidence interval (95% CI) −3.30 to 0.64, P=0.18, random-effect model). When assessing the effectiveness of acupoint injection therapy, the results indicated that acupoint injection can improve the effective rate for nonspecific chronic low back pain (OR = 3.64, 95% CI 2.4 to 5.21, P < 0.0001, fixed-effect model). Conclusion There is insufficient evidence to indicate that acupoint injection therapy could improve the pain for patients with NCLBP. However, the level of evidence was downgraded to “very low quality” because of the poor methodological quality and clinical heterogeneity. The results should be interpreted with caution. Higher quality RCTs with more appropriate comparison, more objective outcome instruments, and adequate follow-up periods are necessary to assess the efficacy of acupoint injection for NCLBP. The PROSPERO Research registration identifying number is CRD42019119158.


Introduction
e low back pain (LBP) refers to pain occurring between the 12th rib and the inferior gluteal fold [1,2]. Epidemiologic studies reported that the prevalence of LBP is 23% to 84% [3]. Most patients (about 62%) still have pain after one year of the first episode of LBP, with 11-12% of those being disabled [3]. It was reported that 16% of CLBP patients were absent from work due to disability [4]. It is estimated that the direct medical cost of evaluating and treating LBP in the United States exceeds $33 billion annually. If the indirect costs of lost work are taken into account, it is more than $100 billion. It is a cause of Years lived with disability (YLDs) reported by the Global Burden of Disease Study (GBD) [5]. LBP is not only a major individual health problem but also became a public health and economic problem worldwide [6].
Chronic low back pain (CLBP) is defined as LBP over 3 months. e main symptom of CLBP is pain, and only 42% of patients were pain free after one year [7]. e causes of CLBP include biophysical factors, comorbidities, and poor mental health [8]. However, no cause can be found in most cases with low back pain. at is, the LBP is "nonspecific" [8][9][10].
Many interventions were used to improve the pain of CLBP, including surgery, drugs, and exercise [11]. However, there is no intervention that is better than others [11,12]. erefore, it is necessary to explore an effective and safe intervention for them.
Acupuncture has been used to treat CLBP for thousands of years [13,14] and has been recommended by the American College of Physicians [15]. Based on the traditional acupuncture and injection therapy, acupoint injection was developed. It was a therapy that combined acupuncture and injection [16]. It was used to relieve pain for patients with CLBP widely in China and Korea [17]. Due to the usage of special forms and drugs, acupoint injection was also known as aqua acupuncture, herbal acupuncture, pharmacoacupuncture, water acupuncture, hydroacupuncture, or squirt cut. is therapy has the dual effects of acupuncture and medication. Specific substances injected into acupuncture points include vitamin [18][19][20][21][22], anesthetics [19,20,23,24], glucocorticoid [19][20][21], Chinese herbal extracts [23][24][25], or other material [26,27]. As with traditional acupuncture, acupoint injection stimulates acupoints to play a therapeutic role for CLBP [28]. Different from traditional acupuncture, acupoint injection also has therapeutic effects through substances, which was beneficial for patients with CLBP [17]. ese substances work as a neuroprotectant, analgetic, or antioxidant. Acupoint injection enhanced stimulation on acupoints [17,29]. It was recommended as an effective intervention that is also a simple, convenient, and time-saving treatment [30]. However, there are inconsistent results for the evidence for the effect of acupoint injection for patients with NCLBP [19,[21][22][23][24]26]. erefore, this systematic review was designed to assess the efficacy and safety of acupoint injection for NCLBP.

Design and Registration.
e Research registration number of this study is CRD42019119158 (https://www.crd. york.ac.uk/PROSPERO/), and the protocol of this study was published previously [31]. For the literature research of this study, ethical authorization is not required. tabase. e English terms such as "acupuncture point injection," "acupoint injection," "pharmacoacupuncture," "hydroacupuncture," "pharmacoacupuncture," "LBP," "low back pain," and "chronic low back pain" were used individually or combined in English databases. e terms including "shu xue zhu she (acupoint injection)," "shui zhen (acupoint injection)," and "xue wei zhu she (acupoint injection)" were used in Chinese databases. e search strategy for Medline via PubMed is shown in Table 1 or acquired from the protocol of this study [31]. e full search strategy used in this study was submitted as supplementary material file 1.

e Inclusion Criteria and Exclusion Criteria of is
Review.
e studies were screened according to the inclusion criteria as follows: (1) patients: ≥18 years of age and suffering from nonspecific CLBP (persists for longer than 12 weeks); (2) intervention: acupuncture point injection therapy without limitation in drug and acupuncture point; (3) comparison: placebo-controlled or other nonacupoint injection therapy; (4) outcomes: the outcomes reported were not limited. e pain intensity was considered as the primary outcome such as the visual analogue scale (VAS) and proportion of pain-free patients; and (5) publication type: randomized controlled trials (RCTs). Exclusion was defined as follows: (1) patients: without nonspecific chronic low back pain; (2) intervention: nonacupoint injection was used, or acupoint injection was one part of a treatment package; (3) comparison: just different acupoint injection intervention methods, or treatment package in which the acupoint injection was used as one part of package acupoint injection; and (4) publication type: quasi-RCTs, non-RCTs such as case reports, systematic reviews, and letters to editors. e duplicate records were removed according to the title, author, journal, and publication date. After that, two reviewer authors screened the records independently. e irrelevant literature was excluded according to the inclusion criteria and exclusion criteria via browsing the title and abstract of the article. Two independent review authors evaluate whether the study can be included according to predetermined selection criteria via reading the full text. In order to extract data accuracy and reliability, only full-text literature in English or Chinese were included in this study. e original author was contacted to obtain additional information when it is needed to make a judgment. Any inconsistencies were resolved with a third reviewer via discussion and consultation. e details of literature

Data Extraction and Management.
A predesigned, verified, and formatted table was used to extract information. e extracted information includes five parts for each study. e information of the title, author, journal, and publishing year was extracted as basic information. e characteristics of participants of each study, including gender, age, inclusion, and exclusion criteria, were collected as the second part. e details of intervention used in the experimental and control groups such as the frequency, intervention time, follow-up period, injection content, dosage, and volume of injection at each point were extracted as the third part. e design information of each study, including the random sequence generation, randomization method, allocation concealment, blinding of participants and personnel, and blinding of outcome assessment was also extracted.
e outcomes reported in each study such as outcome indicators, assessment tools, adverse events, and dropouts were extracted as the fifth part. e extraction of information was performed by two reviewers independently. One of them extracted the data and another checked the data. For the missing data, we have contacted the original author. For the studies that had more than two intervention Evidence-Based Complementary and Alternative Medicine groups, the data of intervention groups related to this review were extracted. For the studies that have two or more experimental groups, we combined the groups to create a single pairwise comparison.

Assessment of Methodological Quality of Included
Studies. e Cochrane Collaboration's tool, "Risk of bias," was used to assess the quality of included studies by two reviewers independently [32]. Review authors' judgments involve answering a specific question for each entry. ere are seven questions for reviewers to make a judgment. ey are as follows: (1) Was the allocation sequence adequately generated? (2) Was the allocation adequately concealed? (3) Was knowledge of the allocated interventions adequately prevented during the study? (4) Were incomplete outcome data adequately addressed? (5) Are reports of the study of the suggestion of selective outcome reporting? (6) Was the study apparently of other problems that could put it at risk of bias? For each trial, the abovementioned questions were answered using "No," "Unclear," and "Yes." For each question of each study, an answer "Yes" indicates a low risk of bias in this domain and an answer "No" indicates a high risk of bias in this domain. If there were very few details available to make a judgment of "low" or "high" risk, a judgment of "unclear" was made. Any inconsistencies were resolved with a third reviewer by discussion and consultation.

Processing of Missing Data.
For the missing data that were not reported in the manuscript, the original author was contacted to request additional data via e-mail. If the acquisition fails, only the available information and data were analyzed.

Data
Analysis. Data analysis was performed by the Cochrane Review Manager (RevMan 5.3). e results were shown using mean difference (MD) with 95% confidence intervals (95% CIs) for continuous data. e results were shown using Odds Ratio (OR) with 95% CI for the enumeration data. e pooled effect was estimated using the fixed-effect model when the results indicate there is no significant heterogeneity between the included studies. If not, the pooled effect was estimated using the random-effect model. e chi-square test (P < 0.1) was used to assess the statistical heterogeneity, and I 2 statistic was used to be quantified. If the I 2 value is greater than 75%, the significant heterogeneity between the included studies was demonstrated according to the Cochrane Handbook for Systematic Reviews of Interventions [32]. e overall effect was synthesized using meta-analysis [33].

Assessment of Publication
Bias. e reporting bias was assessed using funnel plots for the outcomes, which were reported by more than ten studies [33].

Grading the Quality of Evidence.
e quality of evidence for outcomes was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system, which rates the quality as very low, low, moderate, or high levels [34].

Literature Screening.
A total of 7,479 records were obtained from online databases in this study. Among them, 724 records were deleted because they were identified as duplicate records. Other 6755 records were excluded after screening the abstracts and titles by two reviewers. Finally, twelve full texts were included for this review according to our inclusion/exclusion criteria. e details of the literature screening are described in Figure 1. Table 2 shows the characteristics of the included studies. All twelve studies were designed as single-center studies. Nine studies were the pairwise comparison of intervention groups' design [18, 20-22, 25-27, 35]. e other three trials were multicomponent group intervention (three groups) design [23,24,36]. Five trials were followed up after intervention [18,20,26,27,35]. Also, the follow-up ranged from 12 weeks to 1 year. A total of 1381 participants with nonspecific chronic low back pain were included in this review. e number of participants in each trial ranged from 30 to 208. One of the twelve studies was from South Korea, while the others were from China. Only two trials reported the sources of patients [27,36]. All included trials reported the subject selection criteria. e points of acupoint used in each trial were diverse. e number of selected acupoints varies from 3 to 9. e Ashi acupuncture point in the low back was selected as the site of operation in nine of the twelve studies [19-24, 26, 35, 36]. Ten trials choose the points of the Taiyang Bladder Meridian of Foot (back Shu points) and five trials choose the points of the Governor Meridian (DU30). e BL23 was selected in seven trials. e BL25 was selected in seven trials. e BL40 was also selected in six trials [19-21, 23, 25, 26]. e GV03 was selected in five trials. Other points were selected including BL24, BL26, BL28, BL32, BL36, BL54, GB30, GB34, GB37, EX-B12, EX-B05, and SP9. e substance used for acupoint injection was diverse from each other. A single substance or a mixture of substances was used in the studies. ree trials used single substance including Angelica injection [24,36], Danshen injection [23], and Lidocaine injection [23]. e constituent substances of the mixture included Angelica injection [18,22], Lidocaine injection, Oxygen-ozone [24,36], Tong xi tong [22], Vitamin B1 [22], Vitamin B12 [18][19][20][21][22], Triamcinolone Acetonide Acetate [19][20][21], and Cervus and Cucumis Polypeptide injection [26]. Each acupoint was injected with 0.1 ml to 2 ml injectant. Each experimental group of included studies was treated once a day for four days [22], once a day for ten times [24,26,36], five times a week for 2 weeks [35], once a week for 2-4 weeks [20,21], the next day at a time for ten days [25], the next day at a time for twenty days [18], twice a week for 4 weeks [27], and every three days for 20 days [23]. e intervention of the control group in each trials included electroacupuncture [19-21, 24, 36], acupuncture [23,26,35], oral administration of traditional Chinese medicine decoction [22], and placebo control [27]. e visual analogue scale (VAS) was used to assess the effect of acupoint injection for nonspecific chronic low back pain in three trials [18,25,27]. e MOS item short form health survey (SF-36), the Oswestry Disability Questionnaire (ODQ), and the straight leg raising test (SLRT) were used to assess the effect of acupoint injection for nonspecific chronic low back pain in one trial [27]. e subjective cure rate was used to assess the effect of acupoint injection in eleven trials [18-26, 35, 36].

Methodological Quality of Study.
e risk of bias summary is shown in Figure 2 for the included studies. All trials reported random allocation implemented during the study. e random sequence generation methods reported included a random number generator in three studies [18,24,36] and Minitab software [27], for which the risk of bias of random number generation were "low risk." One trial reported that the random number was generated according to the intervention in which the risk of bias of random number generation was "high risk." Other seven trials did not describe the methods of random generation in which the risk of bias was unclear. One study reported allocation concealment [27]. e risk of performance bias, the risk of detection bias, the risk of attrition bias, and the risk of reporting bias were judged as "unclear risk" of eleven trials because there is insufficient information about these domains to permit the judgment of "Yes" or "No." One study was judged as "low risk" in terms of performance bias, detection bias, attrition bias, and reporting bias [27]. Five trials were judged as "low risk" in terms of other bias because a comparison of baseline data before treatment was reported [18-20, 25, 36], while the other seven trials were judged as "unclear risk" because there is no enough information to permit the judgment of "Yes" or "No" [21-24, 26, 27, 35].

e Effectiveness of Acupoint Injection in Terms of Pain for Patients with NCLBP.
ree of twelve trials reported the effects of acupoint injection in terms of pain for patients with NCLBP using the VAS [18,25,27], while we only obtain the Evidence-Based Complementary and Alternative Medicine Evidence-Based Complementary and Alternative Medicine original data of the VAS from two studies [18,25]. us, only two studies were included in the meta-analysis for the primary outcomes [18,25]. e results demonstrated that there is not enough evidence to indicate that acupoint injection can improve the pain of patients with nonspecific CLBP using the VAS (n � 243, MD � −1.33, 95% CI −3.30 to 0.64, P � 0.18, the random-effect model), with a substantial heterogeneity (I 2 � 98%) as shown in Figure 3. e effectiveness of acupoint injection in terms of subjective effective rate for patients with NCLBP.
Eleven trials reported the effect of acupoint injection therapy using a subjective effective rate for NCLBP [18-26, 35, 36]. e results demonstrated that acupoint injection can improve the symptoms of low back pain using a subjective effective rate (n � 1261, OR � 3.64, 95% CI 2.54 to 5.21, P < 0.00001, the fixed-effect model), with homogeneity (I 2 � 0%) as shown in Figure 4.

Dropout.
e number and reasons for dropout were reported in one study [27].
e intension-to-treat (ITT) analysis and per-protocol (PP) analysis were carried out in one trial [27]. ere was no information of dropout reported in the other eleven studies [18-26, 35, 36].

Adverse Effects.
e adverse reactions were reported in two trials which reported that skin flair, edema, and skin rash have occurred in the experimental group and skin rash, headache, hand-foot tingling occurred in the placebo control group [23,27]. e original author also reported that "all adverse reactions disappeared after the study without the use of antihistamines or medical interventions" [27]. Other studies did not report there were adverse events.

Publication Bias.
e publication bias was performed for the subjective effective rate reported by eleven trials in this review [37]. Although the funnel plot is symmetrical (supplementary material file 2), we also considered that there is reporting bias due to language bias and poor methodological quality.

Grading the Quality of Evidence.
e quality of evidence for outcomes was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system, which rates the quality as very low, low, moderate, or high levels [34]. For this review, the quality of

Discussion
is study assessed the efficacy of acupoint injection therapy on pain for patients with NCLBP collating evidence from RCTs. A total of 12 RCTs were included in this review involving 1381 patients. Based on our meta-analysis, there was not enough evidence to indicate that acupoint injection can improve pain for patients with nonspecific CLBP. However, acupuncture point injection therapy can improve the subjective effective rate for patients with nonspecific CLBP. e adverse reactions include skin flair, edema, and skin rash. All adverse reactions disappeared after the study without the use of antihistamines or medical interventions.
ere was no reason to upgrade the strength of evidence. However, due to the limitations in the design and implementation inherent imprecision (low quality of included studies) and publication bias (there were less than ten studies reported the primary outcome), the level of evidence was downgraded. Finally, the quality of the evidence was downgraded to "very low quality" for the primary outcomes of pain.

Strengths and Limitations.
is review and meta-analysis was performed according to the Cochrane Handbook for Systematic Reviews of Intervention. e formulation of the search strategy for each online database is helpful for the comprehensiveness and extensiveness of literature collection. e study protocol predetermined was helpful to select studies, extract data, and synthesize data. According to the study protocol, two independent review authors complete the literature screening, data extraction, quality assessment of the included trials. When encountering a contradiction, we discuss with a third reviewer to make a final decision. ese methods reduced bias and transcription errors. At the same time, there are some limitations we should note. e different interventions, different acupuncture points, intervention frequency, and duration lead to the high clinical heterogeneity between included studies. Another source of clinical heterogeneity is the diversification of the substance injected into points. e Angelica injection, Lidocaine injection, Oxygen-ozone, Tong xi tong, Vitamin B 1 , Vitamin B 12 , Triamcinolone Acetonide Acetate, and Cervus and Cucumis Polypeptide injection were used in different trials. e type of acupuncture point injection was different between trials including single acupoint injection and combined with acupuncture, electroacupuncture, or functional training. All these factors make it difficult to confirm a distinct relationship between acupuncture point injection and the improvement of pain. Although eleven of the included twelve trials reported the effectiveness of acupoint injection for the patients with NCLBP, the subjective instrument used in trials challenged the reliability and validity of the results. Only full-text records in English or Chinese  were included, and unpublished materials were not retrieved in this study, which will also bring limitations to the results.

Clinical Relevance and Future Directions.
ere is no sufficient evidence to indicate that acupuncture point injection can improve the pain of nonspecific CLBP. Acupuncture can improve the subjective effective rate for patients with nonspecific CLBP. But, the results have been challenged due to the subjective instruments used in the trials.
In future studies, the following issues need to be noted. To explore the optimal intervention intensity, frequency, and dosage of substances for acupoint injection may be the future research direction. e intervention of the control group should be set according to the purpose of research. For the studies whose aim is to demonstrate the effect of the injected drugs, the control groups should be placebo control. For the studies whose aim is to demonstrate the effect of the acupuncture points, the control groups should be nonacupuncture point control. e selection of instruments to assess the clinical effect of acupoint injection therapy is important. Assessment tools that are more objective and sensitive should be used in future studies. e long-term outcomes should be of concern, and other outcomes such as quality of life and recurrence of low back pain also should be of concern. e factors that influence the effects such as profession and daily activities, should be considered in future trials. Finally, and most importantly, the CONSORT guidelines should be followed to report the trials to allow a better evaluation of the methodological quality. Higher quality RCTs with more appropriate comparison, more objective outcome instruments, and adequate follow-up periods are necessary to assess the efficacy of acupoint injection for NCLBP.

Conclusions
In conclusion, there is no sufficient evidence to indicate that acupuncture point injection can improve pain for patients with nonspecific CLBP using the VAS. e acupuncture point injection may have a positive effect for the subjective effective rate. e adverse events reported included skin flair, edema, and skin rash. ese adverse events disappeared spontaneously without intervention. However, due to the differences in the acupoints, drugs, injection doses, frequencies, durations of acupuncture point injection, and the poor quality of the included studies, caution is needed when interpreting these results. Higher quality RCTs with more appropriate comparison, more objective outcome instruments, and adequate follow-up periods are necessary to assess the efficacy of acupoint injection for NCLBP. Data Availability e datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Disclosure
Research registration unique identifying number: CRD42019119158. Jianglong Liao and Ying Guo are the guarantors of the article.