Traditional Chinese Medicine for Neck Pain and Low Back Pain: A Systematic Review and Meta-Analysis

Background Neck pain (NP) and low back pain (LBP) are common symptoms bothering people in daily life. Traditional Chinese medicine (TCM) has been used to treat various symptoms and diseases in China and has been demonstrated to be effective. The objective of the present study was to review and analyze the existing data about pain and disability in TCM treatments for NP and LBP. Methods Studies were identified by a comprehensive search of databases, such as MEDLINE, EMBASE, and Cochrane Library, up to September 1, 2013. A meta-analysis was performed to evaluate the efficacy and safety of TCM in managing NP and LBP. Results Seventy five randomized controlled trials (n = 11077) were included. Almost all of the studies investigated individuals experiencing chronic NP (CNP) or chronic LBP (CLBP). We found moderate evidence that acupuncture was more effective than sham-acupuncture in reducing pain immediately post-treatment for CNP (visual analogue scale (VAS) 10 cm, mean difference (MD) = -0.58 (-0.94, -0.22), 95% confidence interval, p = 0.01), CLBP (standardized mean difference = -0.47 (-0.77, -0.17), p = 0.003), and acute LBP (VAS 10 cm, MD = -0.99 (-1.24, -0.73), p< 0.001). Cupping could be more effective than waitlist in VAS (100 mm) (MD = -19.10 (-27.61, -10.58), p < 0. 001) for CNP or medications (e.g. NSAID) for CLBP (MD = -5.4 (-8.9, -0.19), p = 0.003). No serious or life-threatening adverse effects were found. Conclusions Acupuncture, acupressure, and cupping could be efficacious in treating the pain and disability associated with CNP or CLBP in the immediate term. Gua sha, tai chi, qigong, and Chinese manipulation showed fair effects, but we were unable to draw any definite conclusions, and further research is still needed. The efficacy of tuina and moxibustion is unknown because no direct evidence was obtained. These TCM modalities are relatively safe.


Introduction
Neck pain (NP) and low back pain (LBP) are common symptoms bothering people in daily life. In developed countries, more than 70% of people experience LBP [1], whereas approximately two-thirds of people experience NP [2,3] at some point in their lifetimes. These two disorders occur most frequently among the middle-aged population, in which the proportion of females is higher than that of males [4]. A majority of the acute NP and LBP sufferers obtain spontaneous relief within days or weeks, although approximately 10% of acute NP [2] and 20% of acute LBP [5] suffers experience the condition as chronic or persistent. NP and LBP can be caused by specific pathological diseases, such as tumors, infection, fracture, and inflammation. However, the pain in most individuals (approximately 85% for LBP) [1] is non-specific, which indicates that the pain is not attributable to one of the definite above pathologies but instead to some ambiguous etiology. Conventional treatments, such as medications [6] and surgery [7], have demonstrated some efficacy. Nonetheless, these treatments were not always effective, and even had some serious adverse effects [7,8]. Consequently, to find some more effective therapeutic methods, many individuals have turned their attention to some other treatments, such as complementary and alternative medicine (CAM). Although CAM may also have some adverse effects of itself, given the numerous therapeutic methods of CAM and its positive effectiveness to some extent, a growing number of researchers have focused on various CAM therapies, such as acupuncture, massage, exercise, and hydrotherapy [9]. Most significantly, as a type of traditional Chinese medicine (TCM), acupuncture has shown respectable efficacy and is broadly accepted internationally [10][11][12].
TCM is based on the fundamental theory of balance among yin and yang, five basic elements, and a relationship between humans and nature [13,14]. TCM has been used to treat various diseases in China and even throughout East Asia for more than 2000 years, and it still remains the first choice of treatment for many people. However, the different models of thought that are the foundations of TCM and modern science are not compatible, hindering the spread of TCM worldwide. Nevertheless, there are many articles published in various scientific journals that have attempted to explain some phenomena and mechanisms of treatments in TCM from the perspective of modern medicine.
Although acupuncture is a typical representative of TCM, it is only one of the various general therapies for NP and LBP, such as acupressure, cupping, moxibustion, tuina, gua sha, tai chi, qigong, Chinese herbal medicine, and Chinese manipulation (for definitions, see S1 Table). The objective of the present study was to review and analyze the existing data about pain and disability associated with TCM treatments for NP and LBP. The question of our study is "whether TCM treatments are more effective in pain relief or disability improvement as compared with other treatments for people with NP or LBP?".
Retrieval System (TCMLARS) and China National Knowledge Infrastructure (CNKI) and the Wan Fang database. The search was conducted between the inception of each database and September 1, 2013, and updated on May 25, 2014 using disease-specific search string combinations, partly according to the strategy outlined by the Cochrane Back Review Group (CBRG) (S2 Table), with subject limitations within the English or Chinese language. There were no restrictions about publication status of the searched trials. The reference lists of identified studies were screened manually for more studies related. Experts in the representative fields were also contacted for unpublished trials. The search was conducted by a veteran librarian.

Study Selection Criteria
We included any randomized controlled trial (RCT) meeting all of the following criteria: (1) the work is published in the English or Chinese language; (2) the subjects included are men or women (age 17 years) with NP or LBP (with or without radiating pain) of any duration; (3) at least one of the therapies assessed pertains to TCM; (4) a comparison should be done between TCM and other treatment (e.g. TCM versus other treatment, TCM versus no treatment, TCM plus other treatment versus other treatment); (5) at least one of the following outcomes was evaluated: pain intensity or disability; (6) the principle summary measures should better be commonly used, such as pain intensity (e.g., visual analogue scale, VAS; numerical rating scale, NRS) and disability (e.g., Oswestry Disability Index, ODI; Neck Disability Index, NDI); (7) the duration of follow-up should be at least one day after all treatment sessions were concluded according to the study design of each corresponding trial.
We excluded trials of neck or back pain caused by trauma, infection, cauda equina syndrome, bone rarefaction, compression fracture of a vertebral body, tumor, or fibromyalgia.

Data Extraction
Two evaluators independently extracted the data from the studies or SRs, and discrepancies were resolved by negotiation or a third party.
Primary outcomes included pain intensity (e.g., visual analogue scale, VAS; numerical rating scale, NRS) and disability (e.g., Oswestry Disability Index, ODI 0-60 points). Additionally, side effects (including the names of adverse effects and the number or proportion of individuals experiencing them) were recorded.
The study, treatment, population, and outcome characteristics are summarized in tables.

Assessment of Study Quality and Reporting
Two independent assessors evaluated the quality of every trial included in our review. Discrepancies were resolved by negotiation or an authoritative third party. The quality of the individual trials was rated according to the criteria of the Cochrane Back Review Group (Table 1) [15]. Depending on the number of "Yes" responses (coded as "1") for four particular items (e.g., allocation concealment, baseline similarity, patient blinding, and number of or reason for dropouts), yielding a range of scores from 0-4, the quality of individual trials was graded according to the following three ranks: good (score = 4), fair (score = 2-3), or poor (score = 0-1). Because the number of "Yes" responses was recorded for 4 domains, in case of a single study, N was a whole number (0, 1, 2, 3, or 4); in case of multiple studies, N was the average number, which may have been either a whole number or a fraction (S3 Table).

Quantitative Synthesis
We grouped the results with respect to the interventions used (e.g., acupuncture), the general locus of the pain (e.g., neck or low back), the persistence of the pain (e.g., acute, sub-acute, or chronic), and the cause of pain (e.g., specific or non-specific).
The data abstracted were classified into continuous and dichotomous variables. Generally, fixed-effects models (inverse-variance method) were used in the meta-analysis. However, we also used random-effects models (DerSimonian-Laird method) to pool data if the statistical heterogeneity was high (I 2 50%). The source of heterogeneity was explored by fitting covariables (ie, intervention characteristics, mean age, baseline total symptom scores) one by one in the meta-regression. We analyzed the subgroups according to the source of heterogeneity if possible, and sensitivities if there were unaccountable sources of heterogeneity. We used contour-enhanced funnel plots and Egger test to examine publication bias if the number of pooled trials were near or above 10 [16,17]. All analyses were performed in STATA 12.0 (StataCorp LP, College Station, TX).
If the data permitted the assessment and there were statistically significant differences across the pooled data in pain relief or disability improvement, clinical importance was assessed according to Cohen's 3 levels (Table 2) [18].

Rating the Strength of Evidence
The overall strength of evidence was evaluated with the aid of the grading system outlined in the Methods Guide prepared by the AHRQ Evidence-based Practice Center (EPC) program

Rating Range
Small a weighted mean difference (WMD) less than 10% of the scale (e.g., <10 mm on a 100 mm VAS); a standardized mean difference (SMD) or "d" score <0.5; a relative risk of <1.25 or >0.8 (depending on whether the report referred to the risk of benefit or the risk of harm, respectively) Medium a WMD from 10-20% of the scale; an SMD or "d" score from 0.5 to 0.8; a relative risk between 1.25 and 2.0 or between 0.5 and 0.8 (depending on the factor described above) [19]. The rating scheme focuses on 4 major domains: risk of bias (high, medium, low), consistency, directness, and precision. Consistency indicated that 75% of the trials showed that effects were in the same direction (positive or negative) or that heterogeneity was low (i.e., I 2 < 50%).
The strength of evidence was classified into one of four levels: high, moderate, low, or insufficient (no evidence) ( Table 3, S4 Table) [19]. The level was lowered one stage if any of the aforementioned domains was not met and was directly lowered two stages if the trial had a high risk of bias. Additionally, the level was lowered one stage if the sample size was smaller than 40 patients per group (to enable adequate power) [20].

Results
After searching the databases rigorously and systematically, 658 unique records were identified, and the titles and abstracts were screened. The full-text of 243 articles were assessed for eligibility, 75 studies  were included in the systematic review ( Fig. 1). Of these studies, 12 studies [35,36,59,[78][79][80][81][90][91][92][93] were in Chinese (2 unpublished trials about Chinese herbal medicine), others in English. The kappa value for agreement between the reviewers (YQL and LL) was 0.90 which indicated an excellent agreement. We found that most of the included studies were principally about acupuncture, acupressure, and cupping (Table 4). In contrast, the number of studies on the other seven treatments was less than 3 (most = 1) for each treatment. The treatment sessions and treatment durations of interventions also were shown in Table 4. The specific results of our meta-analysis were shown in S5 Table.

Study characteristics
Seventy five studies involving 11077 subjects ranging from 17 to 90 years were included. A majority of the participants were females (> 60%) with chronic neck pain (CNP) or chronic low back pain (CLBP). The basic characteristics of the trials were presented in S6 Table. For the findings of our risk-of-bias assessment were shown in Fig. 2 (and S7 Table). The median and interquartile range (IQR) of the quality score of the studies was 6 (4.5 to 8), which meant that the overall quality were of higher-quality. Most of the studies didn't provide adequate information on outcome assessor blinding, co-intervention and compliance. Given the characteristics of some interventions, the blinding of the care provider was unapplicable. The strength of evidence and its clinical importance was presented in S8 Table.
Acupuncture versus sham treatments (inactive treatments) There were three trials (272 subjects) identified that hold comparisons between acupuncture and sham transcutaneous electrical nerve stimulation (TENS) [25,26,31]. Compared with sham TENS for pain, acupuncture didn't display any differences in pain relief ( Fig. 6) and disability improvement (Fig. 7) for CNP at immediate term and even at short term (p > 0.10), and these results were still robust in sensitivity analysis [25,26,31]. One trial (108 subjects) compared acupuncture and sham laser [29]. Similarly, no difference was found between acupuncture and sham laser in pain relief immediately post-treatment (p = 0.202) (S5 Table) [29].
Acupuncture versus waitlist (no treatment) Only one trial (30 subjects) showed a significant difference in pain for CNP immediately post-treatment on VAS 10 cm, with an odds ratio of 26.00 (3.69 to 183.42, p = 0.001) [37].
Side effects   There were some complaints reported in the trials, such as local bleeding, numbness, pain, and fainting when needles were inserted in some subjects. All these side effects were transient and mild. No life-threatening side effects were reported [21,25,26,29].
Summary Several studies showed that there was moderate evidence with small clinical importance that acupuncture was more effective than sham-acupuncture in reducing pain and disability associated with CNP in the immediate term and at the one-month follow-up.
Acupuncture versus TENS Two studies [56,57] (n = 70) compared acupuncture with TENS and showed no significant differences between groups with respect to pain (Fig. 16), not only in the immediate term (p = 0.81) but also at short term follow-up (p = 0.33). However, functional status was not assessed.

Summary
Several studies showed that there was low evidence that acupuncture was more effective than sham-acupuncture, waitlist care, or usual care in reducing pain and disability for CLBP immediately post-treatment. Moreover, three small studies showed that acupuncture may be more effective than sham-acupuncture in reducing pain in the immediate term for acute or sub-acute LBP (moderate evidence). Acupressure in NP We found one study (32 individuals), which was of fair quality but had a small sample size [70].
Acupressure (+ acupoint stimulation + CT) versus conventional therapy (CT) (sub-acute LBP) Compared with the control group, reduction in the pain intensity in the intervention group was statistically significant at the 1-month follow-up (approximately a 23% reduction in the VAS, p = 0.02, effect size = 0.43) [70].
Side effects None reported. Summary The sample size was so small that we were unable to draw any definite conclusions.
Summary Two large studies showed that there was moderate evidence that acupressure could be more effective than physical therapy for pain and disability associated with CLBP in the immediate and intermediate term.

Cupping in NP
We identified 5 RCTs (241 subjects) for cupping therapy, all of which were of fair quality and had small sample sizes [73,74,77,82,95]. Almost all of the trials principally targeted patients with chronic non-specific neck pain (CNSNP).
Cupping versus progressive muscle relaxation One trial [73] (n = 61) was found. However, no significant differences were reported in pain (VAS 100 mm, MD, -0. Side effects Three trials found some adverse events, which were minor and transient [73,74,82]. Summary Two small (number per group < 40) studies showed that cupping could be more effective than waitlist with respect to pain and disability in the immediate term for CNP (moderate evidence).

Cupping in LBP
Six RCTs pertaining to cupping therapy were identified [76,[78][79][80][81]83]. Four studies were in Chinese which compared cupping with medications, and all these studies were of poor quality [78][79][80][81]. The other two fair-quality studies were in English.  Cupping versus medications Four studies [78][79][80][81] with seven trials (n = 430) compared cupping and medications (e.g. NSAID) for the outcome of pain or disability scores. With respect to pain, all the seven trials [78][79][80][81] were pooled in the random-effects model (Fig. 25), and the MD was-0.54 (95% CI, . That is, the mean pain scores were 0.54 units on the VAS (10 cm) scales significantly lower in the cupping groups than in the medications groups. However, this was highly heterogeneous across studies (I 2 = 81.7%). No evidence of publication bias was found (Egger test, P = 0.138). Meta-regression identified that the source of heterogeneity was the types of the cupping (Adjusted R-squared = 96.34%). Subgroup analysis was performed by the types of the cupping. We found statistical significance between cupping (balance\moving\wet) and medications, but no difference about cupping with retention. With respect to disability, three studies [78][79][80] with six trials (n = 360) were pooled with a MD of-3.77 (95% CI, -5.85 to-1.69, random, I 2 = 83.8%) (Fig. 26). This indicated that cupping could decrease mean disability score of 3.77 units on ODI (0-50) scale. The Egger test suggested no evidence of publication bias (P = 0.166). Meta-regression suggest that the types of the cupping was the source of heterogeneity (Adjusted R-squared = 100.00%). We performed subgroup analysis according to the types of the cupping. Significant differences were still found between cupping (retention\balance\wet) and medications.
Cupping versus waitlist One RCT (n = 32) was identified [76]. There were no significant differences between the two groups in pain and disability (S5 Table).
Cupping versus usual care Usual care indicated that participants received no study-related care-just the care, if any, that they and their physicians chose: mostly massage and physical therapy visits and continued use of medications (mostly NSAIDs). One trial (n = 98) was found [83]. Upon a 3-month Side effects Three individuals experienced fainting (vaso-vagal shock) [83]. However, no adverse events were found in another study [76].
Summary Several small and lower-quality studies showed that there was low evidence with small clinical significance that cupping was more effective than medications (e.g. NSAID) in reducing pain and disability at immediate term for CLBP. One large study showed that there was moderate evidence with a large clinical significance that cupping was more effective than usual care in treating pain and disability in the short term for CLBP.

Gua sha in CNP
There were only two studies (n = 69) that had researched the efficacy of gua sha for NP [84,85], both of which were fair-quality RCTs.
Gua sha versus waitlist/no treatment There was only one RCT (n = 21) found [84]. Seven days after treatment, significant differences favoring gua sha were found for CNP with respect to pain (VAS 10 cm, MD-1.6 (-3.0, -0.1)) (S5 Table). The study did not report disability.
Side effects One trial [85] reported that there may be petechiae, slight muscle aches, and soreness in the application area, with none of these effects being serious.
Summary Small studies showed that there was low evidence with medium to large clinical importance that gua sha was more effective than waitlist and thermal therapy for pain in the immediate term for CNP.

Gua sha in CLBP
Only one study was identified [84].
Side effects None reported. Summary A very small study provided low evidence with medium clinical importance that gua sha was more effective than waitlist in treating pain in the immediate term for CNP.

Qigong in CNP
There were three fair-quality RCTs associated with the use of qigong therapy for NP [86][87][88].
Summary Two small studies showed that qigong may be superior compared with waitlist for CNP sufferers (moderate evidence), but no differences were found compared with other exercises.

Qigong in LBP
There were no RCTs identified.

Tai chi in NP
There were no RCTs identified.

Tai chi in LBP
Only one trial of tai chi for LBP was found. This was a fair-quality RCT with a large sample size (n = 160) [96]. All of the subjects experienced CNSLBP.
Side effects None reported. Summary A large study showed that tai chi could be more effective than waitlist for CLBP (moderate evidence). Because the findings were derived from only one RCT, the validity of the evidence above should be treated cautiously.

Chinese Herbal Medicine in NP
There were 3 studies identified. One study [90] retrieved from database, and two other studies (H. Wang, personal communication, 2004; H. Wang, personal communication, 2005) were identified through consultation with experts. All these studies were in Chinese and 2 of these studies were unpublished. All 3 included trials assessed pain severity. Functional status was not assessed in all 3 trials. There were no continuous outcomes reported. The primary outcomes were expressed with an ordinal scale. So we re-expressed the result as continuous data using NRS (0-3 points, the lower indicates the better) (Fig. 29).
Side effects All of the trials included reported side effects, such as watery stool, abdominal pain, stomachache, pruritus, reddish skin, and small blisters on the skin. However, no life-threatening complications were reported in any of the trials. Summary A large trial showed that there was moderate evidence with large clinical importance that the Chinese herbal medicine Qishe may be more effective in treating pain for CNP in the short term than placebo. In view of the limitations of the included trials, such as a lack of allocation concealment, blinding, and intention-to-treat analysis, the results above should be treated with caution.

Chinese Herbal Medicine in LBP
No RCTs or SRs were identified.
Chinese manipulation versus cervical traction There was only one study [93] (213 subjects, higher quality) that reported pain. This study showed that CM was more effective in pain relief immediately post-intervention (VAS 10 cm, MD, -1.06 (-1.37, -0.75)). Disability was not reported.
Side effects Two studies mentioned adverse events; none were observed in either study [91,92]. Summary Two small studies showed that there is moderate evidence with large clinical importance that Chinese manipulation is more effective than traditional Chinese massage in relieving pain associated with CNP in the immediate term. Additionally, one large study also showed this favorable effect compared with traction.

Chinese Manipulation in LBP
No RCTs were found.

Moxibustion in NP
No RCTs were found.

Moxibustion in LBP
No study reported directly about pain and function, which were of principal concern in our study. Summary There was insufficient evidence supporting moxibustion because the outcomes reported in the review were not directly about pain and function.

Tuina in NP
No RCTs were found.

Tuina in LBP
No studies directly compared tuina alone with one of the other treatments; however, tuina was combined with other traditional Chinese therapies (tuina-focused integrative Chinese medical therapy, TICMT), and TICMT was then compared with other traditional Chinese therapies. And no trial reported pain and disability.
Side effects None reported. Summary There was insufficient evidence regarding tuina because there were no direct comparisons between tuina alone and other treatments.

Discussion
There were a total of 10 methods of treatment for NP or LBP. The included studies concentrated primarily on three therapies (i.e., acupuncture, acupressure, and cupping), with an emphasis on acupuncture, whereas the studies on the other seven treatments were fewer in number. This finding might suggest that these treatments were not well known in western countries and were not standard of care.
All the treatments showed positive effectiveness compared with baseline measurements. Compared with sham acupuncture (SA), acupuncture may be more effective in reducing pain and disability in the immediate and one-month term for individuals with CNP, which was consistent with the former SRs [97,98]. However, inconsistency was found with another SR, which reported no significant differences between these two groups [9]. We determined that this inconsistency arises from three subsequent RCTs with larger sample sizes and positive results favoring acupuncture. Similarly, these differences in immediate-term and short-term outcomes about pain also existed for individuals with CLBP, but no difference about disability. Moreover, the difference was evident for acute LBP in the immediate term. Nevertheless, the difference in clinical importance between acupuncture and SA was small. The SA group was used to estimate the specificity of the acupuncture points and of the technique itself. However, a standardized SA has not yet been established. Therefore, it has been a challenge for researchers to choose the correct acupoints for the SA group. As a result, the effect of true acupuncture will be underestimated. Thus, various degrees of efficacy were observed in different studies. For example, the trial by Vas et al. (2012) [49] reported the effects of acupuncture in subjects with acute nonspecific LBP. In its protocol [99], the authors had chosen some acupoints for SA groups, such as Neiguan (PC-6) and Kongzui (LU-6), which were proven to be effective in previous studies. In 1989, Geng [100] found that 51 patients with acute LBP (pain duration < 10 days) were treated only by PC-6. Thirty eight patients got complete pain relief after just one treatment session. Meanwhile, a case report by Xu [101] reported that the acute LBP of a 37-year-old man was relieved by only using LU-6. A review [102] reported that up to forty one acupoints had been found to be efficacious for individuals with acute LBP, including the PC-6 and LU-6. Thus, it is particularly important to establish a unified standard with which to choose the correct acupoints for SA. We found that acupressure may be more effective than physical therapy but sham-acupressure for CLBP. Additionally, small studies (number per group < 40) showed that CNP patients treated with cupping experienced more pain relief than waitlist controls with CNP and that cupping was more effective at reducing CLBP than medications (e.g. NSAID). The studies associated with these two treatments were relatively higher in number and quality, such as the trial on acupressure by Hsieh [71]. We found that the Chinese herbal medicine Qishe was more effective for reducing CNP in the short term compared with placebo. Small studies showed that gua sha may be more effective than thermal therapy or waitlist control. However, the sample sizes were very small (number per group < 10), and it was difficult to draw any conclusions. Chinese manipulation may be effective for CNP, with results that are consistent with the results of studies on conventional Western manipulation [103,104]. As specific types of mind-body exercises, tai chi and qigong show superior effects for CNP or CLBP compared with effects observed in the waitlist control groups; this finding is consistent with the previous reviews regarding other exercises [105,106], especially yoga, which is also a mindbody exercise. However, the number of studies was so sparse that we were unable to draw conclusions regarding these two treatments. There was insufficient evidence supporting tuina and moxibustion because there were no direct comparisons between tuina and other treatments, and no direct outcomes on pain or disability were reported for moxibustion.
With regard to clinical importance, we found that most of the study sizes were small, others were medium, and only a few were large. This result may be due to the small clinical significances between these treatments themselves, as well as to the small sample sizes, which may give rise to random errors. With respect to the strength of evidence, we found that most of the comparisons had low strengths of evidence, others were moderate, and none were high. We believe that the causes for this variation mainly derive from the following reasons: 1) the quality of the individual studies was lowered because most of the trials failed to include a blinded control group, which was due to the characteristics of the interventions themselves (e.g., tai chi, qigong, tuina, gua sha, cupping, and Chinese manipulation); 2) the sample sizes of some trials were smaller than 40 individuals in each arm, which would lower the strength of the evidence directly by one level; and 3) inconsistency (i.e., different directions of effects or high heterogeneity (I 2 50%)) exists between trials. Obviously, inconsistency of study results in a meta-analysis reduces the confidence of recommendations regarding treatment [107].
Our review has several main limitations, which were due to the studies included. First, we found that the number of studies was small (even zero for some treatments, such as qigong for NP). Thus, further studies in these areas are warranted. Second, the strength of the evidence was low or moderate rather than high, which means that the results may change through further research. Third, the analysis was limited to studies published in English and Chinese when numerous studies had been carried out in various countries. However, the outcome measures in Chinese studies were usually cure rates or efficacy rates of overall improvements, and we were unable to obtain data on pain and disability. Although there were usually larger sample sizes in Chinese studies compared with English studies, there were definite flaws in the designs of Chinese studies and higher risks of bias, such as a lack of allocation concealment [78][79][80][81]90]. Forth, the studies on moxibustion did not report the outcomes we pre-established, and the studies on tuina even compared tuina with itself. As a consequence, we were unable to obtain any direct findings regarding tuina and moxibustion.
The clinical heterogeneities of some of our meta-analyses might limit the translations of our results. First, the selections of acupoints for LBP varied from study to study, and even from patient to patient. However, these heterogeneities were inevitable because the selections of acupoints should be individualized and disease-specific according to TCM theory, but the meridians of these points were homogeneous in some extent. Second, the heterogeneities might be due to different population, such as the wide range of ages (17-90 years) and different peoples from different countries. Third, the underlying differential TCM-diagnoses in patients suffering from NP or LBP might be responsible for the heterogeneity to some extent. Lastly, the variances of treatments methods in different trials might also contribute to the heterogeneity, such as the numbers of treatment sessions or durations, and the frequencies or intervals of treatments. For these reasons, a standardized treatment for each group of a sub-diagnosis is needed in order to gain a higher homogeneity for future studies.

Conclusions
Although there were many deficiencies in our review, the major strength of our study was the coverage of all aspects of TCM in individuals with NP or LBP. Our review demonstrated that acupuncture, acupressure, and cupping could be efficacious in pain and disability for CNP or CLBP in the immediate term. Some treatments, such as gua sha, tai chi, qigong, and manipulation, showed fair effects, but we were unable to draw definite conclusions, and further studies on these treatments are needed. Currently, we do not know much about the efficacy of tuina and moxibustion because there was no direct evidence in any of the studies evaluated. No serious or life-threatening adverse effects were found, indicating that TCM treatments are safe for patients. Nevertheless, considering the sparse studies and inferior methodological quality of the individual trials, we propose that TCM could be used as a supplement to occupational therapies for people with CNP or CLBP. Moreover, sham therapies need to be designed to improve the strength of evidence. The studies in Chinese should report the outcomes of pain and disability definitely, which will benefit the evaluations of these outcomes used internationally. In summary, many more studies with higher quality and longer-term follow-ups are warranted. For future studies, a standardized treatment for each group of a sub-diagnosis is needed in order to gain a higher homogeneity.