Thunder-Fire Moxibustion for Cervical Spondylosis: A Systematic Review and Meta-Analysis

Background Cervical spondylosis (CS) refers to the degenerative changes in the cervical spinal column, which affect the majority of middle-aged and elderly people. Thunder-fire moxibustion originated from thunder-fire miraculous needle, which has been applied widely for treating pain syndromes for thousands of years. The aim of our research is to provide evidence to assess the efficacy and safety of thunder-fire moxibustion in treating CS. Methods and analysis. Retrieved literature databases included Cochrane Library, MEDLINE, Web of Science, EBSCO, EBASE, Springer, PubMed, WFDP, CNKI, VIP, and CBM. The period of retrieval was from the establishment of the database to December 2018. Randomized controlled trials which compared thunder-fire moxibustion and other therapies in CS were included. The quality of inclusive trials was accessed though a Cochrane risk of bias tool. According to the test results of heterogeneity, a random effect model or fixed effect model was used to analyze the data. Results Meta-analysis was conducted for the total effective rate of thunder-fire moxibustion, traditional Chinese medicine syndrome score, pain score, satisfaction score, and score of the symptoms and functional rehabilitation of cervical vertigo. The analysis results were as follows: compared with other therapies, the efficacy of thunder-fire moxibustion was statistically significant, total effective rate increased (OR = 2.48; 95% CI [1.80, 3.41]; P < 0.00001), traditional Chinese medicine syndrome score decreased (SMD = −3.05; 95% CI[−4.18, −1.93]; P < 0.00001), traditional Chinese medicine syndrome score decreased (SMD = −3.05; 95% CI[−4.18, −1.93]; P < 0.00001), traditional Chinese medicine syndrome score decreased (SMD = −3.05; 95% CI[−4.18, −1.93]; P < 0.00001), traditional Chinese medicine syndrome score decreased (SMD = −3.05; 95% CI[−4.18, −1.93]; P < 0.00001), traditional Chinese medicine syndrome score decreased (SMD = −3.05; 95% CI[−4.18, −1.93]; Conclusion Based on the existing evidence, the curative effect and safety of thunder-fire moxibustion on CS were statistically significant. We should interpret the results scrupulously because of the low evidence level. Large-scale, high-quality, rigorous RCTs with long-term follow-up should be performed in the future.


Introduction
Cervical spondylosis (CS) is a chronic cervical spine disc degenerative pathology, which influences the vertebral bodies and cervical intervertebral discs. It may develop into bone spur production, protrusion of intervertebral disc, and spinal cord compression [1]. e most prevalent symptoms are headaches, neck stiffness, numbness, neck and/or arm pain, neck and/or arm tingling, and bladder function problems [2]. CS is a common disease in middle-aged and senile populations. About 50% of people over 40 suffer from cervical spondylosis and about 85% of people over 60 were diagnosed with cervical spondylosis by radiology [3]. Age plays a double role in cervical spondylosis development. e morbidity of cervical spondylosis grows with age before 50 years old, while it does the opposite after 50 [4]. As the spine exposed to increasing loads with aging, it may lead to hypertrophy of the facet joints, ligamentum flavum, and posterior longitudinal ligament. e diameter of the cervical canal decreases, and then spinal cord compression develops [5]. e etiological factors include age, occupation, behavior, heredity, and sports-related activities [6][7][8]. Many asymptomatic CS patients may be diagnosed lingeringly because it attacks insidiously. So, sufferers usually have serious symptoms and permanent disability when they seek medical treatment [9]. Treatment for CS abroad includes conservative and surgical therapy. Conservative therapy consists of rest, immobilization of the cervical spine, drug therapy, and physical therapy [10]. Surgery can cure the core symptoms of CS, but there are still marginal symptoms, such as Barre-Lieou syndrome, tinnitus, blurred vision, cervical angina, cervical migraine, etc., [11]. Surgery is expensive and not appropriate for every patient [12]. Although these conservative therapies have demonstrated some total effective rates, these treatments are not always effective and even have some side effects.
Moxibustion is an ancient Chinese medicine therapy with a history of more than four thousand years. Moxibustion is to place the moxa cones on some acupoints of the body, and then ignite the moxa cones. Generally, the moxa cone is removed before it burns out, and it is replaced with a new moxa cone. ere is another method of moxibustion, where the therapist holds a cigar-like moxa stick and applies moxibustion at the acupoints. Moxibustion points are usually selected according to the diseases to be treated. In traditional Chinese medicine, meridians and collaterals are channels to move qi and blood and connect viscera, body surface, and all parts of the body and are regulating systems of human functions. Moxibustion can promote the dredging of meridians and collaterals, improve local microcirculation, and accelerate the absorption of inflammatory media, so as to reduce local inflammatory response and relieve pain. In addition, moxibustion also has the effect of infrared resonance radiation. During the treatment process, it can promote the resonance of abnormal functional cells, make them undergo apoptosis, and at the same time, it can promote the activation and regeneration of tissue cells, so as to accelerate the local tissue metabolism and promote the functional recovery.
under-fire moxibustion originated from the thunderfire miraculous needle which had been used for thousands of years. It is a new moxibustion therapy ameliorated by Profession Zhao Shibi due to her experience of decades of medical practice [13]. under-fire moxibustion is widely used in China to treat diseases, such as otolaryngological diseases, osteoarthritis, and gynecological diseases, and pain caused by any disease [14][15][16][17][18][19]. Different from the traditional moxibustion stick, the thunder-fire moxibustion adds agallochum, eaglewood, banksia rose, dried ginger, virgate wormwood herb, and stem of capillary sagebrush to a moxa stick [20]. ese drugs play a role in warming meridians and dredging collaterals. When burning, thunder-fire moxibustion can produce strong firepower, thermal infrared radiation force, and medicine chemical factor [21]. e heat of thunder-fire moxibustion can improve blood circulation and stimulate related acupoints to activate meridians. e drug of thunder-fire moxibustion, therefore, penetrates and reaches the acupoints with a high drug concentration area in the human body surface. According to the theory of traditional Chinese medicine, the comprehensive effect of these factors can act on meridians and acupoints to raise local temperature, stimulate local blood circulation and lymph circulation, and accelerate metabolism so as to achieve the function of warming meridians to dredge collaterals, dispel coldness, eliminate wind-dampness, activate yang, transform qi, and adjust physical enginery [22,23]. At the same time, special treatment techniques, such as bird pecking, rotation, and array, are used to mend the curative effect.
Previous studies had shown the effect of thunder-fire moxibustion in CS. However, the effectiveness of thunderfire moxibustion in treating CS is still contentious because of inadequate sample size, low methodological quality, and lack of high-quality study in related clinical trials. As far as we know, there is no systematic review of thunder-fire moxibustion for CS. We, therefore, aim to evaluate the safety and effectiveness of thunder-fire moxibustion on CS and help clinicians to make recommendations and promote the research progress in thunder-fire moxibustion.

Study Registration.
is systematic review and metaanalysis were registered at the International Prospective Register of Systematic Reviews (CRD42018115686). e report is composited based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guide [24].

Eligibility Criteria
2.2.1. Types of Study. All randomized controlled trials (RCTs) that examine thunder-fire moxibustion efficacy in the treatment of CS were included.

Types of Participants.
Researches on adults, who had CS, were included. It had no limitation of gender, race, and type of cervical spondylosis. Trials containing patients with pregnancy, cancer, kidney disease, or liver disease were excluded.

Types of Interventions.
e intervention group adopts single thunder-fire moxibustion or thunder-fire moxibustion combined with other therapies, while the control group receives other therapies except thunder-fire moxibustion, such as usual care, acupuncture, moxibustion, medication or physical therapy, and so on.

Types of Outcome Measures.
We analyzed the total effective rate of the thunder-fire moxibustion, traditional Chinese medicine syndrome score [25], pain score, satisfaction score, and score of the symptoms and functional rehabilitation of cervical vertigo. ree outcome measures were positive indicators, such as total effective rate, satisfaction score, and score of the symptoms and functional rehabilitation of cervical vertigo, and the other two were negative indicators, such as pain score and traditional Chinese medicine syndrome score. Among the positive indicators, the higher the score, the better the effect of the intervention. Among the negative indicators, the lower the score, the better the effect of the intervention. e definitions of "total effective rate" in each study were different, including improvement of symptoms and signs such as vertigo, headache, neck pain, the neck and shoulder movement function, and ability to work normally. Although the definition of "total efficiency" was not the same in each study, they all referred to the remission of symptoms, the recovery of daily work and life. e traditional Chinese medicine syndrome score was measured by the TCM syndrome score and Neck Pain Questionnaire. e pain scores were assessed by the Numeric Rating Scale (NRS) [26] or Visual Analog Scale (VAS) [27]. e satisfaction score was evaluated by the hospital self-made satisfaction questionnaire. e score of the symptoms and functional rehabilitation of cervical vertigo were the evaluated by cervical vertigo symptom and function assessment scale.

Search Strategy.
All researchers worked together to develop retrieval terms and strategies based on the Cochrane Manual. We used the following search terms: Cervical Spondylosis (e.g., Cervical Spondylosis, Spondylosis, Cervical, Spondylosis Deformans) and under-fire Moxibustion (e.g., under-fire Moxibustion, Leihou Moxibustion). We used the search terms with the same meaning to retrieve domestic databases. Table 1 shows us the search strategy of PubMed.

Research Selection.
Before searching studies, investigators had discussed and decided screening criteria in the group. Firstly, the retrieved literatures were imported into the literature management system of EndNote X7 in order to delete duplicate documents. Secondly, two of the researchers (HR and HY) read the titles and abstracts of included literatures to exclude unqualified studies. irdly, these two researchers obtained and read the full text, and discussed in the group to select qualified studies. One researcher (CX) contacted the author to obtain relevant information of the researches as there was lack of related data. Any branching between researchers was solved by discussion or consulting a senior reviewer (CX).

Data Extraction.
We had made a data collection form to record the data of selected studies such as the first author, published year, sample size, age, intervention characteristics of both groups, moxibustion method, duration of treatment outcome measures, and adverse events, before extracting the useful information. e data extraction was completed by two reviewers (HR and HY) independently, and the extracted information was rechecked after finished. e divergence of opinion was solved through consulting the senior researcher (CX). If related data were deficient, one reviewer (CX) got in touch with the writers of articles for lost information through telephone or email. If we could not obtain sufficient information after hard work, our researchers processed the existing information and then explained the result.

e Assessment for Risk of Bias.
e Cochrane's risk of bias (ROB) was used as a tool to evaluate the risk of deviation [28]. Two of the researchers (HR and HY) evaluated the methodological quality of the including literatures on their own. We evaluated the following projects: random sequence generation, allocation concealment, blinding of researchers, subject and outcome assessment, incomplete outcome information, selective result reporting, and other sources of bias.

Data Analysis and Synthesis.
e collected data were analyzed though Review Manager (Version 5.3). Based on the result, the total effective rate, traditional Chinese medicine syndrome score, pain score, satisfaction score, and score of the symptoms and functional rehabilitation of cervical vertigo were analyzed, respectively. We used odds ratio (OR) to deal with dichotomous variable and mean differences (MDs) to deal with continuous variable. At the same time, 95% CI was calculated. We used the forest plot to explain treatment effect. When there was potential heterogeneity (clinical and methodological), due to the weak power of the Q statistics to detect heterogeneity, we should avoid using I 2 to determine the model selection. If the number of studies involved was small, it may not be able to accurately estimate the variance between studies, and the fixed effect model is the first choice. P < 0.05 showed that there was a Leihuo moxibustion 07 Or 05-06 08 Randomized controlled trial 09 Randomized trial 10 Randomly 11 Trial 12 Or 08- 11 13 04 and 07 and 12 Evidence-Based Complementary and Alternative Medicine statistically significant difference. When substantial heterogeneity exited, we conducted subgroup analysis or sensitivity analysis to explore heterogeneity sources. We performed subgroup analysis according to the duration of treatment. If we could not find some resources, we reported a narrative description of the included researches, instead of conducting subgroup analysis. A sensitivity analysis through the leave-one-out method was carried out while substantive heterogeneity existed among trials. When there were more than ten included researches, we used the funnel plot to illustrate the reporting biases. e symmetry of the funnel plot reflected reporting bias. If the funnel plot was symmetrical, there was no bias, and vice versa.

Level of Evidence.
We assessed the evidence level with the help of Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) [29]. e level of evidence from low to high was classified into 4 grades: very low, low, moderate, and high. Evidence level was assessed based on eight factors. Risk of bias, imprecision, indirectness, inconsistency, and publication bias were downgrading factors. e remaining three were upgrading factors, which were dose-response gradient, large effect, and plausible confounding, which would change the effect.

Study Characteristics.
e table shows the characteristics of the researches. All of the trials included had been conducted in China and published from 2003 to 2018. A total of 12 articles included 1304 patients. Seven trials [30-33, 36, 39, 41] had a treatment duration less than or equal to 2 weeks, and 5 trials [34,35,37,38,40] had a treatment duration more than 2 weeks. e baseline data between the two groups in every research were comparable.
Ten researches [30][31][32][33][35][36][37][38][39][40] [32,33] measured satisfaction score, and 2 researches [34,41] measured the score of the symptoms and functional rehabilitation of cervical vertigo. Table 2 provides a detailed summary of the inclusive researches. [30,34,38,40] using the random number table were identified as low risk of bias and the remaining 8 trials [31-33, 35-37, 39, 41] were identified as unclear risk of bias. All trials did not indicate allocation concealment. Due to the particularity of intervention, all trials were unable to implement blinding, so the blinding of the researchers and subjects was assessed as high risk. e data in all studies were inadequate to assess the blindness of the outcome assessors, so we rated it as high risk. We considered all researches to be of low attrition bias. Since there was insufficient information to determine the selective reporting, we assessed all researches at unclear risk of bias. We thought that all trials have a low bias risk in other sources of bias (Figures 2 and 3).

Traditional Chinese Medicine Syndrome Score.
e score of TCM syndrome was assessed in 4 trials [31,32,36,38], which was scored according to the severity of symptoms. e result showed that SMD � − 3.58 (P < 0.00001), indicating that thunder-fire moxibustion was more effective than the control group; however, there was substantial heterogeneity (I 2 � 53%) ( Figure 6). With the purpose of finding the source of heterogeneity, sensitivity analysis was conducted. e heterogeneity dropped to 0% after removing Yang Yang research. After comparison, we speculated that the heterogeneity may be caused by the single treatment time. e single treatment time of this study was 10 min, while the single treatment time of remaining studies was more than or equal to 20 min. We performed subgroup analysis according to the treatment days. We divided the trials into two groups, one is less than or equal to two weeks, the other is longer than two weeks. ree studies [31,32,36] in the subgroup of "less than or equal to two weeks" revealed a statistical significance (SMD � − 3.68; 95% CI [− 4.05, − 3.30]; P < 0.00001; I 2 � 0%).

Pain
Score. Four trials [30,31,38,40] reported pain sore, two [38,40] of them used the NRS and the other two [30,31] used the VAS. e higher the score was, the more serious the condition was. e result showed that SMD � − 0.78 (P < 0.00001), indicating that thunder-fire moxibustion was more effective than the control group; however, there was substantial heterogeneity (I 2 � 96%) (Figure 7). We performed sensitivity analysis to find sources of heterogeneity. e heterogeneity dropped to 10% after removing Zhang huajun research. We conjectured that herb was used in both groups in this research, while other researches did not use the herb. Two trials [38,40] assigned to the group less than or equal to two weeks, which were the same trials evaluated by the NRS, revealed a statistical significance (SMD � − 0.37; 95% CI [− 0.58, − 0.15]; P � 0.0009; I 2 � 52%). Two more trials [30,31] assigned to the group more than two weeks ago, which were the same trials evaluated by the VAS, revealed no statistical significance (SMD � − 1.24; 95% CI [− 1.47, − 1.01]; P < 0.00001; I 2 � 98%).

Satisfaction
Score. Two studies [32,33] assessed satisfaction score of patients. e higher the score was, the better the treatment effect was. e result showed that SMD � 5.35 (P < 0.0001), indicating that thunder-fire moxibustion was more effective than the control group; however, there was substantial heterogeneity (I 2 � 95%) (Figure 8).

Score of the Symptoms and Functional Rehabilitation of
Cervical Vertigo. Two studies [34,41] assessed the score of the symptoms and functional rehabilitation of cervical vertigo. e higher the score was, the better the treatment effect was. e result showed that SMD � 4.12 (P < 0.00001), indicating that thunder-fire moxibustion was more effective than the control group; however, there was substantial heterogeneity (I 2 � 75%) (Figure 9).

Adverse Events.
Two trials [30,36] reported adverse reactions. Huang [30] reported one case of burning pain but no scalds and two cases of mild cough which relieved after proper ventilation. Wang [36] reported no discomfort of the gastrointestinal tract, peripheral blood vessels, and tissues in the experiment group. e routine examination of blood, urine, stool, and liver and kidney functions was normal.  (Table 2).

Publication Bias.
We used RevMan to perform publication bias only on the total effective rate, which included 10 studies. ese 10 studies were performed in China, and the sample sizes were small. e results showed that the funnel plot was asymmetrical, illustrating that there may be publication bias in the efficacy of thunder-fire moxibustion in the treatment of CS (Figure 10). For a long time, the asymmetry of the funnel plot has been regarded as publication bias, but the funnel plot should be regarded as a general means to show the effect of small-scale research. In addition to publication bias, there are other factors that may lead to small-study effect, such as the exaggeration of results in small researches due to the low quality of methodology, real heterogeneity, and accidental factors.

Level of Evidence.
We assessed the evidence quality with the aid of GRADE. e quality of evidence for all comparisons was very low, limiting our recommendation of research findings. ere were five factors that can reduce the level of evidence: risk of bias, imprecision, indirectness, inconsistency, and publication bias. e risk of bias was assessed to be high. e inconsistency was low in all researches. All studies compared treatment outcomes straight, so indirectness did not downgrade. Several comparisons displayed high heterogeneity to downgrade the evidence level. Small sample sizes in most studies led to imprecision and publication bias. Of the three escalation factors, none of the studies achieved.
ere upgrading factors, which were dose-response gradient, large effect, and plausible confounding, were not achieved in all studies (Table 4). Due to the above 8 factors, the quality of evidence in all comparisons was very low. Evidence-Based Complementary and Alternative Medicine 7

Discussion
We want to appraise the curative effect and safety of thunder-fire moxibustion on CS. We included 12 RCTs for meta-analysis after searching and screening the major domestic and foreign database by evidence-based medicine. e result showed that thunder-fire moxibustion was more effective than the control group. under-fire moxibustion can improve the total effective rate of treatment and relieve pain, numbness, and other symptoms of cervical spondylosis. We recommend the effect of thunder-fire moxibustion in CS finitely because of the very low level of evidence. Traditional Chinese medicine syndrome score, pain score, satisfaction score, and score of the symptoms and function Heterogeneity: chi 2 = 2.14, df = 3 (P = 0.54); I 2 = 0% Test for overall effect: Z = 2.36 (P = 0.02) Figure 5: Forest plots of total effective rate (more than two weeks).

Total (95% CI)
Heterogeneity: chi 2 = 6.42, df = 3 (P = 0.09); I 2 = 53% Test for overall effect: Z = 18.97 (P < 0.00001) Test for subgroup differences: chi 2 = 6.14, df = 1(P = 0.01), I 2 = 83.7%   Evidence-Based Complementary and Alternative Medicine rehabilitation of cervical vertigo were statistically significant with substantial heterogeneity. Although subgroup analysis was conducted, heterogeneity still existed in these comparisons. e variety of acupoint selection scheme and therapeutic manipulation may have caused the unresolved heterogeneity. e evidence quality was very low. Due to the low methodology quality of the included literature and the small sample size and the fact that they were all carried out in one country, the conclusion of the composite outcome was not very reliable.
Only one trial reported adverse reactions such as burning pain and mild cough. Routine examination of blood, urine, stool, and liver and kidney functions were normal. We can not draw a conclusion on the safety of    Evidence-Based Complementary and Alternative Medicine thunder-fire moxibustion due to the insufficient number of studies included. In this research, we have seen that thunder-fire moxibustion alone or associated with other therapies had a similar curative effect. Due to the following limitations, we were incapable of reaching an exact finding regarding the efficacy of thunder-fire moxibustion. e methodological quality of inclusive researches was low. Despite the comprehensive database search, only studies performed in China were included, leading to publication bias. Only few   studies had reported the generation of random sequences. e sample size of most researches was small. ere was no multicentre study, and the outcome indicators were subjective. All studies had not performed allocation concealment, blinding of participants, subjects, and outcome measurers, which would lead to high risks of selection bias and performance bias, along with detection bias. ese factors exaggerate the efficacy of thunder-fire moxibustion in CS [42,43]. Only one research reported the results of one-month follow-up after treatment. One research reported that one case of thermalgia but not scalds and two cases of slight cough. e potential mechanism of thunder-fire moxibustion for CS is not yet distinct, but it does have a positive therapeutic effect in CS. Chen [44] found that thunder-fire moxibustion effectively reduced the contents of TNF-α and IL-1β in serum in the rat model of knee osteoarthritis, thereby inhibiting inflammatory response and alleviating symptoms.
under-fire moxibustion is widely used in China in treating any disease, including pain caused by a variety of reasons [45][46][47]. In general, the normal thunder-fire moxibustion operation will not cause invasive injury to patients. Noninvasive treatment can increase participant compliance. Compliance to a certain extent affects the successful implementation of the experimental scheme, the clinical research quality, as well as the dependability of test results [48,49]. e mechanism, efficacy, and safety of thunder-fire moxibustion deserve intensive study.
ere are some limitations and shortcomings in the study. Only few trials of low methodological quality were found, and all were performed in China. e follow-up data were deficient to assess the long-term efficacy. Owing to the diversity of interventions, there were a fewer trials comparing each intervention, and sample sizes were also reduced for each comparison. Meanwhile, this research has not analyzed the different categories of acupoint selection schemes, moxibustion method, or duration, which may influence the findings. Less reporting of adverse reactions can also lead to bias. It was difficult to draw an exact conclusion because of the lack of trials.
Researchers need to consider the following factors for future research. Randomized controlled trials designed rigorously with a large sample size should be performed. e objective index and follow-up information of thunder-fire moxibustion are still insufficient, so we advise that long-term clinical trials with objective index should be performed in future to access the efficacy of thunder-fire moxibustion. Strict protocol is required before performing research, and researchers need to report on the basis of systematic reporting guidelines, for example, Consolidated Standards of Reporting Trials (CONSORT) and Extension for Herbal Interventions 2006 [50]. At the same time, researchers should use CONSORT in Harms Data Recommendations to detect and report adverse reactions in future studies. As the mechanism of thunderfire moxibustion in relieving CS has not yet been established, researchers should pay attention to this aspect in future research.

Conclusion
Based on the existing evidence, the curative effect of thunder-fire moxibustion in CS is weak. under-fire moxibustion has a fewer side effects and is relatively safe. In addition, thunder-fire moxibustion is noninvasive and comfortable and can improve patient compliance. e methodological quality of the relevant researches is low due to small sample size and rough experimental design. erefore, future researchers should perform large-sample, well-designed, long-term RCTs to assess the efficacy and safety and the mechanism of thunder-fire moxibustion.

CS:
cervical spondylosis PRISMA: preferred reporting items for systematic reviews and meta-analyses RCTs: randomized controlled trials NRS: numeric rating scale VAS: visual analog scale WFDP: Wanfang digital periodicals database CNKI: China national knowledge infrastructure CBM: Chinese biomedical literature database TCM: traditional Chinese medicine