The aim of the current study was to systematically assess the effectiveness of herbal medicines in treating chronic obstructive pulmonary disease (COPD).
Randomised clinical trials (RCTs) testing herbal medicines against any type of control intervention in patients with COPD and assessing clinically relevant outcomes were included. The selection of studies, data extraction and validation were performed independently by at least two reviewers. Methodological quality was evaluated using the Jadad score. Effect sizes and their 95% confidence intervals were calculated.
Fourteen eligible RCTs, testing 14 different herbal medicines, were located. Herbal medicines were compared against placebo or no treatment in six trials. Significant intergroup differences for one or more outcome were reported for several herbal medicines including Panax ginseng and Salvia miltiorrhiza. In seven RCTs, which compared herbal medicines with other herbal medicines, the results were mixed. A single trial compared a herbal medicine (Hedera helix leaf extract) with a conventional treatment (ambroxol tablet) and reported no significant difference between groups. Due to the heterogeneity of the data, statistical pooling was not performed. The median methodological quality score was 2 out of a possible maximum 5.
The effectiveness of herbal medicines for treating chronic obstructive pulmonary disease is not established beyond reasonable doubt. Currently, the evidence from randomised clinical trials is scarce and often methodologically weak. Considering the popularity of herbal medicine among chronic obstructive pulmonary disease patients, rigorously designed studies seem warranted.
Chronic obstructive pulmonary disease (COPD) is a group of conditions characterised by airflow obstruction 1. Chronic bronchitis (CB) and emphysema are two major conditions within this group. COPD is a public health concern worldwide, and the prevalence of this disease is increasing. According to the World Health Organization report in 1998, COPD was the sixth leading cause of death and the twelfth most common cause of morbidity worldwide. Both the direct and indirect economic costs of COPD to the society are substantial. In the USA, 16 million people have symptomatic COPD incurring estimated total economic costs for COPD-related morbidity and mortality of US$23.9 billion 2. In the UK, about 1.5 million patients suffer from COPD with a total annual cost to the National Health Service of ∼£491 million for direct costs only and £982 million including indirect costs 1.
There is no cure for COPD 3. Current conventional treatment is aimed at relieving symptoms, preventing recurrent exacerbations, preserving optimal lung function and enhancing the quality of life 4. Smoking cessation is the only therapeutic intervention shown to reduce disease progression 5. Although the conventional management of COPD has been improved since the 1990s, the progress is slow 6. Unsatisfactory treatment outcomes from conventional drugs, and adverse effects associated with several classes of drugs, such as steroids and theophylline, contribute substantially to the increasing popularity of complementary and alternative medicine (CAM) and, in particular, herbal medicine 7.
There is a long history of using herbal remedies to treat COPD, particularly in China, India and other Asian countries. Herbal expectorants, based on extracts from Hedera helix (ivy) or Thymus vulgaris (thyme) also enjoy considerable popularity in many European countries 8. Despite the popularity of herbal medicine, there has been no comprehensive systematic review of herbal medicines for treating COPD. The objective of the current review, therefore, was to systematically review the existing evidence on the effectiveness of herbal medicines for the treatment of COPD.
All methods detailed below were according to a pre-defined, unpublished protocol.
Five electronic databases (AMED, MEDLINE, EMBASE, CINAHL and the Cochrane Library) were searched, from their respective inception to August 2005, without language restriction. Search terms were as follows: bronchitis, chronic obstructive pulmonary disease, COPD, acute exacerbation of chronic bronchitis (i.e. AECB), AECB, emphysema, herb*, botanic*, phyto*, Chinese medicine, plant extract*, plant preparation*, and individual common plant names and botanical names. Hand searches were performed in files and journals of the authors’ own library. Three manufacturers of relevant herbal remedies were contacted and asked to contribute further information, particularly unpublished data. The bibliographies of all included trials and other relevant reviews were searched to identify further potential trials.
The clinical trials included in this review had to be of herbal preparations administered systemically for CB, emphysema or COPD, in which patients of either sex and of any age were randomly assigned to receive either herbal medicines or control treatments (i.e. placebo, no treatment, conventional therapy or other herbal medicines). Only trials assessing clinical outcomes (e.g. forced expiratory volume in one second (FEV1), global clinical assessment of effectiveness, symptom scores, health-related quality of life, exacerbation severity and frequency) were included. Trials including asthma patients were excluded.
Data extraction and methodological quality assessment
Titles and abstracts of identified articles were screened and full-text articles of potentially relevant trials were obtained. Articles in English and Chinese were read by R. Guo and articles in German by M.H. Pittler. These authors then discussed the articles and made decisions whether to include or exclude and, if needed, the third co-author (E. Ernst) was consulted. Data concerning the details of study design, participants, interventions, outcomes and adverse events were extracted from all included articles (R. Guo and M.H. Pittler). Missing data from one trial 9 were requested by contacting the author of the report. To date, a response has not been received.
The methodological quality of the included trials was assessed using the five-point scale developed by Jadad et al. 10. This was done either independently by two authors (R. Guo and M.H. Pittler) for randomised clinical trails (RCTs) published in English or by discussion of the same two authors for RCTs published in German or Chinese.
The included RCTs were categorised according to the type of control intervention and the following comparisons were made. 1) Herbal medicine versus placebo or no treatment; 2) herbal medicine versus conventional therapy; and 3) herbal medicine versus herbal medicine.
Due to the inadequate reporting, the overall results were unclear in some trials. The effect sizes and the 95% confidence intervals (CI) of primary outcomes of each of the studies were therefore calculated (R. Guo). Most trials reported FEV1 (in absolute volume or percentage of predicted value) as primary outcomes. Symptom scores and exacerbation frequency were also reported in some trials. Several Chinese trials reported the results of global assessment of effectiveness as the primary outcome measure. This was based on the “Chinese National criteria for the clinical diagnosis and treatment evaluation of chronic bronchitis” 11, where responses are categorised into four levels (excellent, good, moderate and no effect). The responder rates for the levels “good” and “excellent” were assessed. This was defined as the proportion of patients who met the following criteria: at least two of four symptoms (cough, sputum, dyspnoea and rale) disappeared and the other symptoms significantly improved, or at least three of the four symptoms significantly improved and the other symptom improved. Results for continuous data (FEV1, vital capacity (VC), symptom score, and exacerbation frequency and severity) were calculated as weighted mean differences and results for dichotomous data (responder rate) were calculated as risk ratio (the proportion of the responders in the test group divided by the proportion of responders in the control group). These results are presented in table 1⇓ and are summarised descriptively. Pooled results were not calculated because only single trials were located for each herbal intervention.
The current authors located a total of 529 potentially relevant titles and abstracts. Of these, 48 articles were obtained for detailed evaluation and 19 RCTs 9, 12–30 were subsequently identified for further analysis. No unpublished RCTs were located. Of the 19 RCTs identified for further analysis, five were excluded because they did not report clinically relevant outcomes 26, 27, did not test a herbal remedy 29, did not use a systemic route of administration 28, or included asthmatic patients 30. Figure 1⇓ provides a flowchart of all included and excluded trials.
Fourteen RCTs 9, 12–24 met all inclusion criteria and were assessed. The main characteristics of these studies are summarised in table 2⇓. They originate from four countries (China, n = 10; Germany, n = 2; Switzerland, n = 1; and Italy, n = 1) and were published between 1991 and 2004 in three languages (Chinese, n = 8; English, n = 4; and German, n = 2). A total of 1,359 subjects were randomised and 1,314 were analysed by the original investigators. Most of the trials studied herbal mixtures whereas four trials tested herbal monopreparations. Eleven studies included only adult subjects (three trials specified their subjects as senile), two included both children and adults, and one included children only. Patients were diagnosed as having COPD in five RCTs and CB in nine RCTs. In one RCT, patients were specified as being in remission whereas three other RCTs included patients in acute exacerbation.
The methodological quality of the trials was generally low with a median Jadad score of 2, ranging from 1 to 5 (tables 2⇓ and 3⇓). Of the 14 included RCTs, only three 9, 16, 24 described the methods for randomisation and five 9, 15, 17, 21, 22 mentioned double blinding, of which only one 9 described the method of blinding appropriately. Details of dropouts and withdrawals were described in only five trials 9, 15, 17, 21, 24. Only two trials 9, 15 were placebo controlled.
The 14 RCTs were categorised into three groups according to the type of control interventions. Detailed results are summarised in table 1⇓ and described below. Information on the preparations, such as manufactures and compositions, can be found in the Appendix.
Herbal medicines versus placebo or no treatment
The current authors located a total of six RCTs that compared herbal medicines with placebo or no treatment. Two RCTs 9, 15 compared herbal medicines with placebo and the other four 12–14, 16 compared herbal medicines with no treatment. In all of these RCTs, both the experimental and the control groups received the same adjunctive treatment regimen with conventional therapies (table 2⇓). Significant differences for pulmonary function, symptom score and/or responder rate were seen in several herbal remedies.
In one double-blind, placebo-controlled RCT 9, Panax ginseng (Ginsana®; Pharmaton SA, Basel, Switzerland) was found to be significantly superior to placebo in improving pulmonary functions, including FEV1. However, the available data were insufficient for effect size calculations. Another RCT 14 tested Shen Mai injection (Affiliated Pharmaceuticals of West China University, Chengdu, Sichuan, People’s Republic of China), which is a combination preparation containing three herbs including P. ginseng as the main ingredient (see Appendix), against no treatment. Significant improvements were found in all outcomes, including pulmonary function tests (FEV1, VC), global clinical assessment of effectiveness (responder rate) 11 and the Borg scale symptom score.
One RCT 12 tested Salvia miltiorrhiza injection (Danshen injection; Sichuan Ya-an Pharmaceuticals, Ya-an, Sichuan, People’s Republic of China) against no treatment and found a significant improvement in FEV1. No significant difference in responder rate was found between treatment regiments.
Traditional Chinese herbal medicine
Two RCTs 13, 16 tested two traditional Chinese herbal medicine (TCHM) decoctions, Jinshui Liujin and Jiawei Yupingfeng (Appendix), against no treatment. In the Jiawei Yupingfeng trial 16, the responder rate in the TCHM group was found to be significantly higher compared with the no treatment group. In the Jinshui Liujin trial 13, the author reported significant improvement of all outcomes in both groups from baseline (p<0.05). Calculations of effect size and 95% CI showed no significant difference in any of the outcomes, including responder rate, FEV1 and VC, between groups.
One double-blind, placebo-controlled RCT 15 studied the effectiveness of Esberitox® N (Schaper and Brümmer GmbH & Co., Salzgitter, Germany), a liquid extract made from three herbs (Appendix), including Echinacea, as a supportive medication. Significant improvements in FEV1 were reported.
Herbal medicine versus conventional therapy
Only one RCT 17 was located that compared a herbal remedy with a conventional therapy. VC was assessed as the primary outcome. No statistically significant difference was seen between H. helix leaf extract (Prospan®) and ambroxol tablets (table 1⇓).
Herbal medicine versus herbal medicine
Seven RCTs 18–24 compared one herbal medicine with another (table 1⇓). One RCT 21 compared two forms of H. helix extract (Prospan® cough syrup and Prospan® herbal drops). No significant differences were found between groups in either FEV1 or VC. The other six RCTs tested one TCHM remedy against another. Five of these RCTs reported the results for the responder rate. Four trials 18, 19, 20, 23 showed positive results favouring test groups and one 22 showed no significant difference between groups. Only one trial 18 reported data for the pulmonary function test (FEV1), which showed no significant difference between groups. Two trials 24, 20 reported the results of symptom scores and both showed positive results favouring the test group. One trial 23 reported the assessment of exacerbation severity and frequency. The results suggested that both severity and frequency of exacerbation were significantly reduced in the test group compared with the control group.
Only five out of 14 trials included information on adverse events (AE). Two of these RCTs reported no AE in either group. The other studies reported mild-to-moderate AE with no further details (table 1⇓).
The present study has identified several herbal remedies with the potential to improve pulmonary function, to relieve symptoms or to reduce exacerbation severity and frequency in the treatment of COPD. Studies on herbal products containing P. ginseng, H. helix, S. miltiorrhiza and some TCHM decoctions generated encouraging results. However, interpretation and extrapolation of these results are difficult for a number of reasons.
First, insufficient data are currently available for any specific herbal remedy. Only one trial was identified for each of the tested herbal interventions and most of these trials were of small sample size and none reported formal sample size or power calculations.
Secondly, the overall methodological quality of the studies is low (table 3⇓). Blinding and randomisation are two essential features for minimising bias 31. However, in this review, only five out of 14 RCTs were double blinded. Part of the reason for the lack of blinding may be the difficulty in finding a credible placebo that is indistinguishable in colour, taste and smell, and also pharmacologically inert in the conditions under investigation. This may be particularly problematic for herbal decoctions. Similarly, four trials compared herbal medicines against “no treatment”, which does not account for placebo effects and is a source of bias. The method for randomisation was rarely described. In two RCTs 19, 20, there was a substantial difference between the numbers of patients in test and control groups. This could suggest that inappropriate methods of randomisation were used. These weaknesses, and the lack of descriptions of dropouts and withdrawals, contribute to the often low scores on the Jadad scale. The validity of some RCTs was further limited by failings to report the concealment of treatment allocation and details of statistical analysis, or inappropriate analyses.
Thirdly, seven RCTs compared a herbal medicine with another herbal medicine. Despite some positive results shown in these trials, interpretation is difficult due to the unknown effect of the control intervention.
Furthermore, quality control of the herbal extracts is important to ensure the reproducibility of the study. Of the 12 herbal preparations (excluding herbal decoctions) listed in the Appendix, data on extract standardisation are only available for one preparation. Details on the extract solvent were also only reported for one preparation.
Finally, the European Agency for the Evaluation of Medicinal Products recommended using both FEV1 and a measure of symptomatic benefit as a combined primary outcome, and states that the end-point for symptomatic benefit should be justified by referencing published data that support its validity 32. However, only one trial 14 used a referenced symptomatic benefit measure, in spite of the availability of a number of validated, well accepted and widely applied instruments, such as the St George's Respiratory Questionnaire 33 and Chronic Respiratory Questionnaire 34.
The increasing popularity and, particularly, the easy access to herbal remedies raises safety issues. The common assumption among consumers that herbal medicines are safe is dangerous. Herbal medicines might be linked to serious AE and herb–drug interaction may have serious clinical consequences 35. Several relevant herbs, such as Salvia milhiorrhiza, Glycyrrhiza uralensis, P. ginseng and Angelica sinensis were reported to interact with a range of conventional medicines 36–38. Unfortunately, in most of the included RCTs, AE were not monitored and herb–drug interactions were considered in none of the reports. The included studies involved a large number of individual herbs in various combinations (see Appendix). The AE and herb–drug interactions of most of these herbs and preparations are not well documented. Therefore, further studies on the safety of these remedies are required.
Limitations of the current review and, indeed, systematic reviews in general pertain to the potential incompleteness of the reviewed evidence. The distorting effects arising from publication bias and location bias are well documented 39–43. In the current review, 10 out of 14 included RCTs were published in China, a country which has been previously shown to produce large numbers of positive CAM studies 44. For the current study, databases were searched with a focus on the American and European literature and those which specialise in complementary medicine, and included hand-searches. There were no restrictions in terms of publication language. The current authors are, therefore, confident that their search strategy has located most of the published trials on the subject. However, whether all unpublished trials were identified remains uncertain. A further limitation of this review is (as pointed out previously) the paucity and limited quality of the primary data.
The effectiveness of herbal medicines for treating chronic obstructive disease is not established beyond reasonable doubt. Currently, the evidence from randomised clinical trials is scarce and often methodologically weak. Considering the popularity of herbal medicine among chronic obstructive pulmonary disease patients, rigorously designed studies seems to be warranted.
APPENDIX: COMPOSITION AND CHARACTERISTICS OF MEDICINAL HERBAL PREPARATIONS USED IN THE INCLUDED RANDOMISED CLINICAL TRIALS
- Received October 13, 2005.
- Accepted February 17, 2006.
- © ERS Journals Ltd