Effectiveness, Medication Patterns, and Adverse Events of Traditional Chinese Herbal Patches for Osteoarthritis: A Systematic Review

Objective. The aim of this study is to systematically evaluate the evidence whether traditional Chinese herbal patches (TCHPs) for osteoarthritis (OA) are effective and safe and analyze their medication patterns. Methods. A systematic literature search was performed using all the possible Medical Subject Headings (MeSH) and keywords from January 1979 to July 2013. Both randomized controlled trials (RCTs) and observational studies were included. Estimated effects were analyzed using mean difference (MD) or relative risk (RR) with 95% confidence intervals (CI) and meta-analysis. Results. 86 kinds of TCHPs were identified. RCTs and controlled clinical trials (CCTs) which were mostly of low quality favored TCHPs for local pain and dysfunction relief. TCHPs, compared with diclofenac ointment, had significant effects on global effectiveness rate (RR = 0.50; 95% CI (0.29, 0.87)). Components of formulae were mainly based on the compounds “Xiao Huo Luo Dan” (Minor collateral-freeing pill) and “Du Huo Ji Sheng Tang” (Angelicae Pubescentis and Loranthi decoction). Ten kinds of adverse events (AEs), mainly consisting of itching and/or local skin rashes, were identified after 3-4 weeks of follow-up. Conclusions. TCHPs have certain evidence in improving global effectiveness rate for OA; however, more rigorous studies are warranted to support their use.


Introduction
OA, which is manifested by joint pain, disability, stiffness, and/or swelling, is a common chronic disease in the elderly worldwide [1][2][3]. In Traditional Chinese Medicine (TCM), symptoms of OA are usually known as "Bi-arthralgia" or "flaccidity" [4].
Conventional therapies for the management of OA consist of exercises, weight loss, acetaminophen, and oral and topical nonsteroidal anti-inflammatory drugs (NSAIDs), as well as intra-articular injection and several other treatments [3]. Such treatments may prove ineffective in some OA patients, and NSAIDs often have gastrointestinal (GI) and cardiovascular adverse events (AEs) [3,5], so patients are turning increasingly to complementary and alternative medicine (CAM) as treatment options for OA [5]. Some reviews provide evidence for the effectiveness of herbal medicines for OA [5][6][7][8]. At the same time, external medications for the treatment of mild or moderate OA pain have been advocated by both the Chinese Orthopedic Association (COA) and the American College of Rheumatology (ACR), because it is deemed to have relatively less AEs and is more convenient for use [3,9]. In many topical herbal medications, the patch or plaster is most frequently prescribed [10]. TCHPs are a class of transdermal plasters that dissolve or mix different herbs with the adhesive matrix, which could then be made into a thin patch. When affixed to the injured area or acupoints, it would have a therapeutic effect locally or even systemically [4,11].
The biological mechanisms of TCHPs for OA are known to have the following characteristics. (1) Their herbs could reach the lesions with the help of the transdermal delivery 2 Evidence-Based Complementary and Alternative Medicine system, so they could continue to achieve an analgesic and anti-inflammatory effect [12,13]. Some studies reveal that serum prostaglandin E2 (PGE2), interleukin-1 (IL-1), and interleukin-6 (IL-6) were decreased, while -endorphin ( -EP) was increased in OA patients after using TCHPs [14,15]. (2) They have a slight fixation effect and could help patients overcome fear of pain as taping [4,16]. (3) The way of dispelling "cold evil, " removing "dampness evil, " and activating blood circulation might possibility have an impact on the immune and neurochemical systems to improve TCM syndrome [4].
At present, transdermal patches as TCHPs have been widely applied for patients with OA or chronic joint pain in China and worldwide and have accumulated abundant data in clinical practice [4,10,17]. To date, no comprehensive study has been documented for their effectiveness, medication patterns, and AEs, while such information would be of great value in guiding TCM practitioners or health care providers in the management of OA. Therefore, this systematic review is undertaken to investigate these important aspects of TCHPs for OA. RCTs and CCTs were chosen to evaluate the effectiveness, whereas both interventional and observational studies were included for analyzing medication patterns and AEs of TCHPs.

Selection Criteria.
Given that chronic joint pain is the major symptom of OA and that a large number of TCHPs have listed chronic joint pain, rather than OA as their indications, it is necessary to index both OA and chronic joint pain during the search process.
Data has been pooled from the 2010 version of China Pharmacopoeia (one) and electronic databases from past decades, as they both provide clear evidence for TCHPs in the treatment of OA. When retrieving data from China Pharmacopoeia (one), TCHPs were required to show the indication of OA or chronic joint pain.
In the electronic searches, relevant articles published in English or Chinese were included if all the following criteria were met: treating OA or chronic joint pain, RCTs or observational studies; the case number enrolled into the treatment group of at least 15, and describing the main traditional Chinese herbs (commercially available or exclusively applied in the hospital). A study was excluded if, it was treating rheumatoid arthritis, gouty arthritis, or psoriatic arthritis; TCHPs were employed as one method in a combined therapy, and/or the unbalanced baseline before interventions, because it is not possible to identify the effect; it was a review or experimental articles or if there was no clinical data and/or details of herbs provided.

Study Selection.
Titles and abstracts of all records were initially checked to find relevant studies. If this information was insufficient, whole articles were retrieved to check whether the article had been missed in the initial search. Full text articles were retrieved for final analysis. The two reviewers (XZW and SPW) independently conducted study selection and assessed articles by the strategy of the established criteria.

Data Extraction.
All articles were read and data was extracted, based on predefined standardized forms. This data mainly included first author, year of publication, title of study, simple size, types of trial, treatment and control group, methodological quality, eligibility criteria, outcome measures, name and components of TCHPs, descriptions of effectiveness, details of AEs, and follow-up period for each study.
A classical textbook was referred to, to standardize the herbal name involved in all TCHPs [18]. Synonyms of herbs were merged and different herbs were distinguished. Matrices such as honey, rosin, and licorice were excluded, because they act as processing materials, usually with no detailed dosage available.

Quality
Assurance. All authors worked together to develop relevant MeSH, keywords of each database, and screening methods of citations. All works were applied independently by two authors to screen the full texts of articles. In case of disagreement, the two reviewers tried to discuss and achieve a consensus. When a consensus could not be reached, a third reviewer (YLC) was consulted to make the final decision.

Analysis Plan.
According to the unique philosophical and methodological characteristics of TCM [19], evidence of effectiveness, medication patterns, and related AEs of TCHPs have been synthesized, respectively.
The qualities of the reports of RCTs and CCTs were assessed by the Cochrane Collaboration's tool for assessing risk of bias to address the following domains: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other bias. Judgments were categorized as low risk of bias, high risk of bias, or unclear risk of bias. If insufficient information was prevented to make judgment, trials were categorized into high risk of bias; if adequate reporting was provided, trials were categorized to low risk of bias, and the rest were recorded unclear risk of bias. The effectiveness of TCHPs for OA was defined as a significant improvement compared to the placebo, NSAIDs, or other therapeutic interventions (e.g., infrared therapy) in outcomes of OA pain, dysfunction, or global effectiveness rate (TCM syndrome). Statistically, the difference between the intervention group and the control group was considered to be an improvement ( < 0.05 or < 0.01); noninferiority results of TCHPs group compared to NSAIDs were also included.
Data was pooled using MD with 95% CI for continuous outcomes or RR with 95% CI for binary outcomes through Revman 5.2 software. Meta-analysis would be done if RCTs had a good homogeneity and the funnel plot would explore publication bias if enough trials were identified. When the 2 was greater than 50%, higher levels of statistical heterogeneity were existed and random effects model was used. When 2 was less than 50%, a fixed effects model would be more appropriate. RCTs and observational studies were included for analyzing the medication patterns and AEs in all included TCHPs.   [20][21][22][23][24] and 7 studies reached noninferiority effect [25][26][27][28][29][30][31], including 3 studies using diclofenac ointment and 1 using diclofenac [25][26][27][28]. The characteristics of TCHPs compared with diclofenac ointment or placebo were listed in Table 1 and 686 participants were involved. The duration of treatment ranged from 7 to 42 days  [4,28]. 18 studies provided information on patients' syndrome differentiations (TCM-Zheng) as the basis of effectiveness for using TCHPs [4, 12, 21, 22, 25-27, 29, 32-41]. For example, when applied for knee OA with syndrome of kidney deficiency and blood stasis, the "Huo Xue Hua Yu" patch (Gao) significantly improved total scores of TCM syndrome and OA symptoms compared with diclofenac ointment [21].   1607 participants were included. The duration of treatment ranged from 9 days to 5 weeks, even with 1-year follow-ups [47]. A three-arm study concerning knee pain and range of motion (ROM) about "Gu Ci" patch versus control patch was obtained [43]. One reported using "Gu Bi" patch for the chronic joint pain of knee, ankle, and shoulder [47]. In these studies, the outcome measure was unclear and stated rare in 4 trials [42,43,48,49].

Assessments of Risk of Bias of Included RCTs.
RCTs were generally of poor methodological quality or were poorly reported (supplementary Appendix 3). The randomized allocation of participants was declared in all included RCTs. However, only 12 trials mentioned methods for sequence generation, such as random number table [21,22,30,35,36,41,51,52] or computer software [4,10,23,25]. Three trials were of single-blind design [40,51,53] and 2 were of double [4,31]. Two trials were assessed as having concealment and obtained research ethics approval [4,36]. Nearly, all the trials provided baseline data for the comparability among groups. Four trials reported information on withdrawal/dropout [20,25,28,35]. Majority of studies lacked the information for dropouts and outcome measures were quite varied. Most studies have conducted follow-ups of 3 to 4 weeks. Risk of bias summary of TCHPs versus diclofenac ointment or placebo was shown in Figure 2.

Effect Estimates of RCTs.
All the RCTs demonstrated a positive effect, favoring TCHPs for OA, except one study protocol [10]. Pain relief was the most frequently reported positive benefit of TCHPs [21,24,31,35,36,38,40,41,[53][54][55][56]. The onset time of reducing pain was from 4.02 hours to 15.40 hours [56,57], and the effect could be maintained from 2.30 days to 6.77 days [53]. The reported outcome measures included local pain relief, dysfunction relief, and overall effectiveness rate. According to the analysis plan and established outcome measures, firstly, TCHPs were compared with diclofenac ointment or placebo. Effect estimates of TCHPs compared with diclofenac ointment or placebo for OA were shown in Table 3.
For the global effectiveness rate, as most studies have used diclofenac ointment as the control group, the meta-analysis has been applied. Concerning specific outcomes of local pain or dysfunction relief, there was high heterogeneity in the aspects of control group and methodology design, so data was only synthesized using MD or RR with 95% CI rather than the meta-approach.   [8,21] or placebo [4] in the aspect of improving dysfunction of knee OA. The "Huo Xue Hua Yu" patch was documented as giving an improved effect on ROM and flexion deformity over a 14-day treatment as compared to diclofenac ointment [MD 0.06, 95% CI −0.29 to 0.41; MD 0.06, 95% CI −0.40 to 0.55; 1 trial; = 60] [21]. One demonstrated similar effect compared with diclofenac ointment for the treatment over 42 days [MD −1.30, 95% CI −6.46 to 3.86; = 60] [28], but one showed no significant improvement in the aspect of stiffness and physical function compared with placebo in the visit of the 6th day [4].

Meta-Analysis for Global Effectiveness Rate.
Six trials with 6 comparisons including 417 participants demonstrated the effectiveness rate of TCHPs versus diclofenac ointment [20,22,23,25,26,58] (Figure 3). As reported, 4 trials showed that TCHPs had better effect on the global effectiveness rate [20,22,23,58]. Two trials demonstrated noninferiority results favoring TCHPs [25,26]. As 2 was less than 50%, lower levels of statistical heterogeneity were denoted and the fixed effects model was used for meta-analysis. Collectively, the results showed that TCHPs had significant effects on the global effectiveness rate [RR = 0.50, 95% CI 0.29 to 0.87; 6 trials; = 417]. However, the funnel plot of comparison of TCHPs versus diclofenac ointment for OA demonstrated asymmetry, suggesting the possibility of publication bias (Figure 4).
Compared to other RCTs interventions, TCHPs could reduce local pain by the 10th day [36], local pain, joint swelling, and locomotor disability by the 14th day [31], local pain by the 21st day [55], locomotor disability by the 14th/28th day [27,41], local pain, stiffness, and dysfunction by the 28th day [35], local pain, locomotor disability, and burning sensation [24], local pain by the 60th day. They could also improve walking distance and ROM by the 60th day [54]. Table 4. Global effectiveness rate from 75% to 97.91% was the main outcome [46,50]. Kuang   Data was synthesized using MD with 95% CI for continuous outcomes or RR with 95% CI for binary outcomes; * there was no statistically significant difference between the intervention and control group in score reduction or global effectiveness rate ( > 0.05); # noninferiority results.  One was a class of dispelling cold-damp, promoting blood circulation and strengthening analgesic efficacy to treat syndrome of cold-damp stasis blockage (wind-cold-damp Biarthralgia or tendons-muscular stasis) with local joint pain, swelling or effusion, feeling of heaviness, and functional impairment. All these symptoms are most likely to become worse on cloudy and rainy days, as well as a preference for warmth and pressing, unchangeable skin color, thick tongue, thin or greasy tongue coating, thin or string pulse condition, and so forth. Such targeted TCHPs were the "Gou Pi" patch, the "Fu Fang Nan Xing Zhi Tong" patch, and the "Hei Yao" patch [4,26,60].

Effect Estimates of TCHPs for OA about CCTs. Results of favored TCHPs for OA about CCTs were shown in
Another small portion was a class for clearing heat and damp, cooling blood, and relieving pain to treat syndrome of wind-damp-heat Bi-arthralgia with pain or tingling, increased skin temperature, effusion or swelling, functional impairment, associated with local burning, thirst or lack of thirst, bitter mouth, dry stool and yellow urine, red tongue, thin yellow or yellow greasy tongue coating, slippery-quick or string pulse condition, and so forth. Targeted TCHPs were the "San Huang" patch, the "Huang Bo Wu Wei" patch, and the "Xi Tong Ning" patch [24,44,61].
It is clear that the top 7 herbs are suited for the syndrome of cold-damp stasis blockage: Chuan Wu and Cao Wu are used to dispel the evil of wind, damp, and cold as well as relieve pain; Mo Yao and Ru Xiang promote blood circulation to achieve analgesic effect; Chuan Xiong accelerates blood circulation and qi and relieves pain; the effects of Dang Gui enrich the blood and promote blood circulation; and the emitting of Bing Pian stimulates drug absorption, so they have the effect of promoting blood circulation and dredging meridians, eliminating swelling and pain with the help of a transdermal delivery system [12,13,50]. After categorizing the different herbs according to their actions, the most used ones are those which have the effect of dispelling cold-damp and promoting blood circulation [18] (Table 6). If we investigated the formula, it is clear that they are mainly based on "Xiao Huo Luo Dan" (minor collateral-freeing pill) and "Du Huo Ji Sheng Tang" (Angelicae Pubescentis and Loranthi decoction).

Adverse Events.
For AEs of patches in 80 literature sources, 38.75% studies (31 of 80) did not mention whether they had monitored AEs or not, 32.50% studies described AEs, whereas the remaining 28.25% reported no incidence of AEs. Among all the reports, detailed information of AEs was identified (Tables 1 and 7). Apart from a special therapy of blistering [41], the incidence of AEs ranged from 0.66% to 12.24% [20,36]. When we compared the incidence of withdrawal and AEs of TCHPs with diclofenac ointment or placebo, it was found that the TCHPs group was more than the placebo [7.56% (9/119) versus 0.00% (0/30)] [4]    Values are the number (frequency). Herbal name presented only when the value is above 1.

Discussion
This is the first study to systematically investigate the evidence of effectiveness and AEs and analyze medication patterns of   TCHPs for OA. As there is no current data to support a particular group of patches possessing overwhelming efficacy in the treatment of OA, and since there was no meta-analysis available, we therefore comprehensively sourced all the evidence from both clinical studies and China Pharmacopoeia (one). The review showed that TCHPs, which were mostly of low quality, could obviously improve global effectiveness rate, reduce local pain, and/or raise function comparing with diclofenac ointment or placebo. The result of metaanalysis showed a statistically significant improvement to global effectiveness rate of OA participants [RR = 0.50, 95% CI 0.29 to 0.87; 6 trials; = 417]. Formulae of TCHPs were mainly based on Xiao Huo Luo Dan and Du Huo Ji Sheng Tang. The incidence of AEs was less than diclofenac ointment group. Ten AEs mainly concerning itching and/or rashes of local skin were identified.
The efficacy of Chinese herbal medicine for OA was found to be better than or similar to conventional therapies [8,75]. Consistent with our study, a previous review which detected the external use of Chinese herbal medicine has also documented a good efficacy and safety for OA. Apart from TCHPs, it has included other intervention methods and the control group was also of diversity (NSAIDs, Cox-2 inhibitors, sodium hyaluronate intra-articular injection, and pain spot blocking), so results were combined and the incidence of AEs was therefore smaller than that of our result (1.87% versus 2.82%) [8].
Although the review demonstrates that TCHPs could ease OA symptoms, it may be affected by low methodological quality of included RCTs and potential publication bias indicated by asymmetry funnel plot. It is known that low methodological studies indicated greater differences between test and control group than those well conducted [76]. Therefore, further trials with more rigorous design and unpublished studies are needed in this area.
This study has documented those herbs with the effect of dispelling cold-damp, promoting blood circulation, and relieving pain, such as Chuan Wu, Cao Wu, Mo Yao, Ru Xiang, and Dang Gui which were the major components of TCHPs. Furthermore, results of analyzing both the herbs' frequencies and their effects were consistent. Formulae of TCHPs were mainly based on Xiao Huo Luo Dan and Du Huo Ji Sheng Tang. Xiao Huo Luo Dan was documented in formulae by the Taiping Pharmaceutical Bureau for Benevolence to relive pain, so that the wind-cold-damp evil might be relieved [77]. Pharmacological studies have confirmed its anti-inflammatory, analgesic, and immunosuppressive role [78], so it has a good therapeutic effect on the early and mid-OA [79]. Du Huo Ji Sheng Tang is derived from "Bei Ji Qian Jin Yao Fang. " Topical use is mainly for removing wind-damp evil to warm and dredge meridians and cure Bi-arthralgia [4,80]. Whether via oral administration or topical use, its efficacy in the treatment of knee OA has been confirmed [81][82][83].
It is commonly believed that the AEs of TCHPs should be less and TCHPs is convenient for topical use, but there are still 10 kinds of self-reported AEs identified by this study. On the one hand, a large part of these studies have no description of AEs, indicating insufficient information about monitoring and reporting of AEs, and, on the other hand, over half of these studies (53.06%; 26 of 49) have demonstrated AEs. Given that OA is a chronic progressive disease, results from current relatively short term studies (mostly 3-4 weeks) seem to have underestimated the incidence of AEs. Nevertheless, for the incidence of AEs, the TCHPs group was less than diclofenac ointment (2.82% versus 8.99%) [25,28,58]. AEs reported were mild to modest, as the majority of allergic reactions were local skin itching and/or rashes.
TCHPs are warm and dry in nature and drastic in potency, so it is appropriate for patients with a strong constitution and should be applied with caution for those with heat syndrome due to damp obstruction, Yin deficiency and/or for pregnant women. As Chuan Wu, Cao Wu, or Ma Qian Zi has potential kidney or liver toxicity, it should be used with caution when administered in high-dose or for long-term use [77]. If blister, itching, and/or rash occurs in the local skin, it should be stopped immediately.
In the included articles, all the prescriptions of TCHPs were based on the medication patterns, and only 18 studies provided information on patients' syndrome differentiation as the basis of effectiveness for using TCHPs [4, 12, 21, 22, 25-27, 29, 32-41]. It is recommended to use TCHPs based on both OA symptoms and TCM syndrome if applicable. Otherwise, the effectiveness might be decreased and the incidence of AEs might be increased theoretically, although there are currently no reports concerning this issue.
This study has had several limitations. Firstly, diagnosis/grade of OA was not clear in most of the trials for insufficient reporting of either ACR or COA criteria and Kellgren-Lawrence scale, so findings would limit the generalization to OA population. It is noteworthy that OA is a chronic longer-term condition; it therefore remains presenting symptoms and signs of disease-stage with the majority of OA participants. Secondly, no data was available to support which Chinese herbs were superior to others. Hence, the top 7 most common herbs and components of formulae were listed so as to provide a broader reference to future study. Thirdly, the funnel plot showed asymmetry, which indicated the possibility of publication bias existed in this review. Although this might also be induced by language bias as we only retrieved literatures in Chinese and English, factors from relatively small sample sizes of these studies could not be excluded. Finally, the included RCTs were low in quality and a meta-analysis could be only applied in the aspect of global effectiveness rate rather than in specific outcome measures such as pain and dysfunction. However these have been synthesized through standard RR or MD. Moreover, the initial evidence for the effectiveness and clue of AEs as well as specific components of formulae may provide beneficial references to practitioners when using TCHPs.

Conclusions
In summary, this review suggests that TCHPs have certain evidence for OA in improving global effectiveness rate. Components of formulae were mostly based on Xiao Huo Luo Dan and Du Huo Ji Sheng Tang. The main AEs were itching and/or rashes of local skin, but further studies concerning AEs, effectiveness, and medication patterns are warranted to support their use.