Meta-analysis of clinical trials on traditional Chinese herbal medicine for treatment of persistent allergic rhinitis


  • Edited by: Wytske Fokkens


Yu Fan and Wei Qian, Institute of Molecular Biology & Translational Medicine, The Affiliated People's Hospital, Jiangsu University, Zhenjiang, China.

Tel.: +86 511 88915233

Fax: +86 511 85234387




Chinese herbal medicine (CHM) has been used for the prevention and treatment of persistent allergic rhinitis (PAR), but results are still equivocal. This study was to assess the clinical effectiveness of CHM in patients with PAR.

Materials and methods

Databases searched included articles published in the Cochrane library, MEDLINE, EMBASE, China National Knowledge Infrastructure, and Wanfang database from 1999 to 2011. The studies included were randomized controlled trials (RCTs) comparing CHM to placebo if they included patients with PAR. The main outcomes were the changes in the standardized mean difference (SMD) of nasal symptom scores and total serum IgE level. Methodological quality was assessed by the modified Jadad's scale.


Seven RCTs with 533 patients were identified and analyzed. In the meta-analysis, CHM reduced the total nasal symptom scores compared to placebo (SMD, −1.82; 95% confidence interval [CI], −3.03 to −0.62; = 0.003). The effect estimate was in favor of the CHM intervention (SMD, −1.09; 95% CI, −2.74 to 0.55) in reducing the total serum IgE level, although this was not significant (= 0.19).


CHM interventions appear to have beneficial effects in patients with PAR. However, the published efficacy studies are too small to draw firm conclusion.

Allergic rhinitis (AR) is one of the most common allergic diseases encountered in clinical practice, and its prevalence has exhibited a marked increase in Western industrialized countries. Between 10% and 20% of the population suffer from AR [1]. The incidence of AR is estimated to be as high as 36% in the general population of the United States [2]. Of 11 cities in China, the prevalence of AR varies from 8% to 24.1% [3].

According to ARIA guidelines [4], AR can be divided into intermittent or persistent based on the duration of symptoms. Persistent allergic rhinitis (PAR) comprises approximately 29% of all incidences of AR in Europe [5]. It is defined by the presence of the following cardinal symptoms: sneezing, pruritus (itching), runny nose, and nasal congestion [6]. Although AR is not life-threatening, it can significantly impair quality of life [7], mainly during allergen exposure [8], and there is epidemiologic evidence that AR is a risk factor for asthma [9]. The therapeutic strategy of PAR mainly includes avoidance of exposure to allergens, treatment with corticosteroids and/or antihistamines, and immunotherapy. However, these measures do not completely control AR symptoms [9, 10]. Long-term treatment is usually required, and side-effects of the drugs are common [11]. In these circumstances, seeking complementary therapies for AR, and using them to alleviate allergic symptoms, may be a feasible therapeutic strategy.

The use of complementary or alternative medicine is widespread in the world. Chinese herbal medicine (CHM) has been used for centuries to treat allergic conditions, such as nasal symptoms similar to AR. There are numerous herbal remedies for PAR. Based on clinical evidence, a well-written review of herbal medicines for treating PAR has been published [12]. Some Chinese herbs have been found to have anti-allergic, anti-inflammatory, or immune modulation effects and might be beneficial for the symptomatic treatment of AR. A recent systematic review, mainly focused on seasonal (intermittent) AR, has provided some suggestive evidence that CHM may be beneficial for AR treatment [13]. In the present review, we evaluated findings from recent randomized controlled trials (RCTs) on the efficacy and safety of CHM used for treating PAR to determine whether CHM is beneficial to patients with PAR. The findings were then summarized in the present paper. The current review focused only on the patients with PAR to avoid the seasonal variation of nasal symptoms.


Search strategy

Systematic literature searches were conducted in the following electronic databases: the Cochrane library, MEDLINE database (1999 to January 2011), EMBASE, China National Knowledge Infrastructure (CNKI) database (1999 to February 2011), Wanfang database (1999 to January 2011). All databases were searched without language limitations. Both journal articles and unpublished reports were included in the review. Potentially relevant trials included the word ‘allergic rhinitis’ plus at least one of the following terms related to traditional Chinese medicine (TCM) modalities: TCM, CHM, and herbs, and we retrieved the full texts to assess studies for inclusion. Non-English publications satisfying the inclusion criteria were translated into English and then assessed. Only data available in the full papers were reviewed. We searched the existing meta-analysis and scanned the cited references in published studies to identify relevant trials and reviews. We hand-searched the following journals published in Chinese: the Chinese Journal of Otorhinolaryngology of Integrated Traditional and Western Medicine (1999 onward), the Chinese Journal of Chinese with Materia Medica (1999 onward), Chinese Traditional and Herbal Drugs, Chinese Traditional patent Medicine (1999 onward), the Chinese Journal of Integrated Traditional and Western Medicine (1999 onward). All data were put into Review Manager Software 5.0 by one author and checked by another. Any disagreement was resolved by discussion between the two reviewers and by seeking the opinion of a third reviewer when necessary.

Inclusion criteria

Studies satisfying the following criteria were included in the review: (i) RCTs with testing herbal preparation; (ii) RCTs investigating patients with CHM compared to placebo (or inactive CHM treatment); (iii) patients, men or women of any age, with a diagnosis of PAR made by a qualified physician and assessed clinically relevant outcomes; (iv) patients who had typical symptoms of AR including runny nose, sneezing, nasal obstruction, and itching in nose and eyes for more than 1 year; (v) patients who were not using any traditional CHM therapy for PAR 1 month before entering the study; (vi) allergy must be confirmed using an objective test such as positive skin prick test (SPT) or a serum IgE test. For the included randomized crossover trials, only data from the first phase were used.

Exclusion criteria

Subjects were excluded if they had received (i) allergen injections in the previous 2 years; (ii) regular medications for AR or other allergic disorders; (iii) systematic corticosteroid within the last 3 months or nasal corticosteroid within the last 15 days; they were also excluded if presented (iv) seasonal AR; (v) vasomotor rhinitis and rhinitis medicamentosa; (vi) nasal structure deformities; (vii) other active respiratory disorders; and (viii) active medical disorders such as cancer, infection, hematologic, renal, hepatic, cardiovascular, metabolic, or gastrointestinal diseases; (ix) pregnant women and women at risk of conception were also excluded; finally, (x) studies based on single chemical extracts or on combination preparation containing synthetic chemicals were excluded. CHM studies including Western medicine as intervention were also excluded.

Review methods

Data extraction

The two reviewers extracted the data according to predefined criteria using a ‘TCM for PAR extraction form’ developed specifically for the review. Items on the form included (i) baseline demographics (author, year of publication, country of study); (ii) participants (sample size, age); (iii) CHM intervention; (iv) control intervention; (v) duration of intervention; (vi) outcome measures; (vii) summary of results; (viii) adverse effects; and (ix) discontinuation. Where discrepancies were identified, the reviewers resolved these by discussion.

Outcome measurements

Trials had to report at least one of the following outcome measures: (i) the total nasal symptom scores including runny nose, nasal obstruction, sneezing, itchy nose, and itchy eyes before and after treatment; (ii) total serum IgE; (iii) withdrawal or dropout rate. Adverse effects were also extracted.

Quality assessment

The methodological quality was assessed using the modified Jadad's scale by two reviewers independently [14]. The effect analysis was performed to explore the difference between the two groups. The modified Jadad's scale is an eight-item scale designed to assess randomization, blinding, withdrawals/dropouts, inclusion/exclusion criteria, adverse effects, and statistical analysis. The maximum number of points available was 8. Scores of 4–8 represent good to excellent (high quality) and 0–3 poor or low quality.

Statistical method

Meta-analysis was used the Revman 5.0 software from the Cochrane Collaboration (version 5.0.18 2009, Cochrane Collaboration and Updated Software). Included trials were categorized by the treatment interventions, and the following comparisons were made as CHM versus controls. Dichotomous data were calculated as the relative risk (RR) with 95% confidence interval (CI). Continuous data were calculated as mean difference (MD) or standardized mean difference (SMD) with 95% CI. Heterogeneity of effect sizes was evaluated by the I2 statistic [15]. P>0.05 or I2<50% was taken as an indicator in the same scale of outcomes using a fixed-effect model. P ≤ 0.05 or I2>50% was taken as an indicator in different scales of outcomes, and an SMD with 95% confidence intervals using a random-effect model. A funnel plot analysis was carried out to assess the potential publication bias, if the number of trials was sufficient. A P-value<0.05 was considered as statistically significant.


After the application of search strategy, a total of 266 potentially relevant publications were identified. After reviewing the full texts, 259 studies were excluded and 7 studies met the inclusion criteria. In total, 533 patients were included and treated. The trial duration ranged from 2 weeks to 3 months. All included studies were RCTs. Overall, the methodological quality of the included studies was low to moderate. Two studies were performed in Hong Kong [16, 17] and Taiwan [18, 19], and one in China [20], Australia [21], and Korea [22] (Fig. 1). Therefore, seven clinical trials of CHM treating PAR were included in the meta-analysis. The included trials were reviewed by two independent investigators (W SJ and T QF).

Figure 1.

Flow chart of trial selection process for meta-analysis.

Description of studies

The characteristics of included trials, along with details on adopted herbal preparations, are listed in Table 1.

Table 1. Summary of clinical studies included in meta-analysis
StudyMethodSample SizeExperimental HerbControlDurationOutcomesResult summary
  1. QoL, Quality of life; PMN, polymorphonuclear neutrophils; PGE2, prostaglandin E2; LTC4, leukotriene C4; sICAM-1, soluble intercellular adhesion molecule 1; IL-8, interleukin 8; IL-10, interleukin 10.

Chui et al. (2009)Randomized, double-blinded controlled trial20/20Herba Centipedae, 23%; Herba Menthae, 16%; Radix Paeoniae Alba, 16%; Radix Scutellariae, 10%; Radix Glycyrrhizae, 6%; Radix Platcodi, 6%; Floz Lonicerae, 5%; Fructus Zizyphi Jujubae, 5%; Rhizoma Coptidis, 4%; Radix Ledebouriellas, 5%; Pericarpium Citri Reticuulatae, 4%. (Allergic rhinitis nose drops,2 sprays per nostril, 5 times a day)Radix Condonopsis, Pilosulae, and Radix Rehmanniae (which have no specific treatment value or any known adverse effect on AR.)2 weeksNasal symptom scores QoLDecreased nasal symptom score; significant improvements in complexion and sleep (P < 0.05)
Zhao et al. ([17])Randomized double-blind control trial63/63Angelica dahurica (Fisch. ex Hoffm.) Benth, 20 g; Saposhnikovia divaricata (Turcz.) Schischk, 7.5 g; Magnolia biondii Pamp, 5 g; Asteraceae, Fructus, 7.5 g; Gentiana scabra Bunge, 5 g; Verbena officinalis L, 5 g.Placebo contained brown-colored starch resembling the SBL powder4 weeks

Nasal symptom scores

Serum IgE levelVisual analog scaleSkin prick testsQoL

Decreased nasal symptom score; visual analog scale; the changes in serum IgE were not significant; improved the nose condition and enhanced some domains of QoL when compared to placebo (< 0.05).
Yang et al. ([17])Randomized control trial36/24Astragalus mongholicus Bunge, 6.0; Panax ginseng C.A.Mey, 4.0; Citrus reticulata Blanco, 2.0; Atractylodes macrocephala Koidz, 2.0; Angelica dah-rica Fisch. ex Hoffm, 2.0; Cimicifuga foetida, L.1.0; Bupleurum chinense DC., 1.0; Zingiber officinale Rosc, 3.0; Glycyrrhiza uralensis Fisch, 4.0; Ziziphus jujuba Mill. var. inermis Rehd, 2.0. (Bu-zhong-yi-qi-tang, every 12 g of the water extract is derived from 27 g of the raw materials)Atractylodes lancea Thunb, 4.0; Magnolia officinalis Rehd. Et Wils., 1.5; Citrus reticulata Blanco, 1.0; Zingiber officinale Rosc, 0.5; Ziziphus jujuba Mill. var. inermis Rehd, 0.5; Glycyrrhiza uralensis Fisch, 1.0. (Ping-wei-san, every 4.5 g of the water extract are derived from 8.5 g of the above raw materials)3 months

Nasal symptom scores

Serum IgE levelLTC4, PGE2 productionCOX-2 mRNA

Improved nasal symptom scores; total serum IgE, the IL-4-stimulated production of PGE2 and LTC4 by PMN were significantly suppressed in active treatment group. Suppressed COX-2 mRNA expression in IL-4-stimulated PMN
Yang et al. ([19])Randomized, double-blinded control trial62/38Magnolia liliflora desr, Asarum heterotropoieds, Saposhnikovia divaricata schischk, Angelica dahu-rica Benth. et Hook, Liqusticum sinense Oliv, Liqusticum Wallichi Franch, Cimicifuga foetida L, Akebia quinata Decne, and Glycyrrhiza uralensis Fisch. (Each capsule contained 5 g three times a day)Placebos were prepared by packing starch in the same-colored and same-sized capsules3 months

Nasal symptom scores

Nasal airflow resistanceNostril dissection areaSerum IgE antibodies

Attenuated nasal symptoms (sneezing and rhinorrhea) and nasal congestion; reduced nasal airflow resistance; increased nostril dissection areas; suppression of serum IgE and increased production of IL-10, sICAM-1, and IL-8 compared to control group.
Zhang H (2004)Randomized control trial60/30

Radix Astragalus seu Hedysari, 20g; Radix Ledebouriellae, 10g; Rhizoma Atractylodes Macrocephalae, 10g; Herba Asari, 3g; Ramulus Cinnamomi, 6g; Fructus Schisandrae, 6g; Radix Paeoniae Alba, 6g; Herba Ephedrae, 3g; Fructus Mume, 6g; Fructus Chebulae, 6g; Radix Glycyrrhizae, 6g. (Above herb decocte into 100 ml, two times a day.)

(Each BMP tablet contained 0.4 g of dried herbal extract)

Radix Astragalus seu Hedysari, Radix Ledebouriellae, Rhizoma Atractylodes Macrocephalae, Rhizoma Zingiberis Recens. (Ping-wei-san, every 5.0 g, three times a day)12 weeks

Nasal symptom scores

Serum IgE level

Decreased nasal symptom scores and serum IgE level.

There was significant difference between active and placebo treatments (< 0.05).

Hu et al. ([21])Randomized, Double-blind, controlled trial26/32Rehmannia glutinosa, 0.46g;Scutellaria baicalensis, 0.46g; Polygonatum sibiricum, 0.368g;Ginkgo biloba, 4.6g; Epimedium sagittatum, 0.46g;Psoralen corylofolia, 0.46g; Schisandra chinensis, 0.368;pulp of prunus mume, 0.184g; Ledebouriella divaricata, 0.46g;Angelica dahurica, 0.368g; Astragalus membranaceus, 0.552g. (Each Biminne capsule contained 460mg of dried herbal extract)Placebo contained brown-colored starch resembling the Biminne12 weeks

Symptom diary

Visual analog scaleQoLSerum IgE level

Significant improvement in some of the symptoms; QoL and visual analog scale in active group; serum IgE level was reduced after the herbal treatment (< 0.001).
Jung et al. ([22])Randomized, double-blind, placebo-controlled trial30/29Fermented red ginseng soft capsules (250 mg fermented red ginseng/capsule), three capsules two times dailyPlacebo capsules were filled with starch4 weeks

Nasal symptom scores

Rhinitis QoLSkin prick testSerum IgE level

No significant difference in nasal symptom scores; for nasal congestion, ginseng was significantly effective (< 0.005); improved rhinitis QOL in ginseng group (< 0.05); reduced skin reactivity to allergens (< 0.05); control group showed a significant increase in serum IgE, whereas the experimental group did not.

Methodological quality

A quality assessment of the primary studies is summarized in Table 2. The Jadad's scores ranged from 2 to 6 points. All of the trials were described as randomized but only three of the 7 trials gave details of the method used.

Table 2. Quality assessment of trials included in meta-analysis
StudyRandomization (appropriate description?)Concealment (appropriate description?)Blind (appropriate description?)Withdraw/dropoutSimilar at baseline?Jadad's score (max 8)
Hu ([21])1 (1)01115
Zhang H (2004)1 (1)00002
Yang ([17])100012
Chui (2009)101114
Zhao ([17])1 (1)11116
Yang ([19])101103
Jung ([22])101114

Total nasal symptom scores

Among included studies, 6 [16-20, 22] evaluated nasal symptom score: of 453 patients, 263 were assigned to the treatment groups, whereas 190 were assigned to the control groups. Their total nasal symptom scores were evaluated. The heterogeneity was significant (χ2 = 137.98, I2 = 96%), so we used the random-effect model method. The results showed that CHM reduced the total nasal symptom scores compared to placebo (SMD, −1.82; 95% CI, −3.03 to −0.62; P = 0.003). The follow-up studies showed that the nasal symptoms remained controlled after the completion of CHM treatment, whereas the symptoms did not improve in the placebo group [16, 19] (Fig. 2).

Figure 2.

Forest plot showing SMD (with 95% CI) for total nasal symptom scores of eligible studies comparing CHM with control in a random effect model.

Total serum IgE level

Three included studies [18, 20, 22] reported total serum IgE in such a way that the data could be used in meta-analysis. In another study [21], the data could not be incorporated because it reported the total serum IgE with means and SD values, but only reported the outcomes as medians. The study by Zhao et al. [17] did not report the raw data on total serum IgE and reported that between-group changes were not significant. Hence, meta-analysis of the total serum IgE level was performed in the three included studies. In 3 of the 7 included studies comprising 209 patients, 126 were assigned to the treatment group, whereas 83 were assigned to the control group. The heterogeneity was significant (χ2 = 52.20, I2 = 96%); hence, we used the random-effect model method. The effect estimate was in favor of the CHM intervention (SMD, −1.09; 95% CI, −2.74 to 0.55) but was not significant (P = 0.19) (Fig. 3).

Figure 3.

Forest plot showing SMD (with 95% CI) for the total serum IgE of eligible studies comparing CHM with control in a random effect model.

Adverse effects and discontinuation

Three of the 7 trials describing administration of CHM to patients included data on adverse events [17, 21, 22]. The adverse effects included nausea, bloating, headache, flatus, sore nose or eyes, diarrhea, vomiting, dizziness, and rashes on the eyelids. Most of these events were moderate in severity and had short duration. No serious adverse effects were reported in any of the RCTs. No significant changes in the results of the blood and urine tests were observed in the two groups during the study period. Furthermore, serious hepatic, renal, or hematologic abnormalities were not exhibited. The fixed-effect model method was used because the heterogeneity was not significant (P = 0.74, I2 = 0). The RR for adverse rate was 1.14, and there was no statistical significance between the two groups (95% CI, 0.60 to 2.18; Z = 0.41; P = 0.68).

Three included studies [17, 19, 21] described dropout or withdrawals. Fixed-effect model method was used because the heterogeneity was also not significant (P = 0.16, I2 = 46%). The RR for discontinuation rate was 0.89, and there was no statistical significance between the two groups (95% CI, 0.44 to 1.79; Z = 0.34; P = 0.73) (Fig. 4).

Figure 4.

Forest plot showing RR (with 95% CI) for adverse rate and discontinuation of eligible studies comparing CHM with control in a random effect model.

Publication bias

All trials claimed to have randomly assigned participants into different groups, but there were few records of clear randomization and concealment method. The CHM approaches were heterogeneous because of the different treatments used. Hence, most of the trials included in this current review had a risk of bias.

Sensitivity analysis

In interpreting the sensitivity analysis, the six studies reporting total nasal symptom scores were used because this measure has some clinical significance for the primary outcomes in the review, but because of differences in the treatment duration, one trial (treatment duration<2 weeks; not by oral administration) was excluded [16] and meta-analysis repeated. The results of sensitivity analysis (SMD, −2.34; 95% CI, −3.77 to −0.90; Z = 3.19; P = 0.001) were similar to the main analysis (with no considerable qualitative differences among them). The findings supported the stability of the analysis (Fig. 5).

Figure 5.

Sensitivity analysis was performed by omitting one study based on the treatment duration and administration methods.


CHM is widely used in treating PAR in China, and some clinical trials have been conducted to explore its efficacy. However, most clinical studies published in Chinese medical journals show lack of appropriate randomization, blinding, and control. Evidence cannot be provided because of the poor quality of the methodology, and there were still no relevant large-sample multicenter, randomized, controlled trials. Only seven trials met the inclusion criteria. The data from the analyzed RCTs showed that CHM produces statistically significant clinical efficacy in patients with PAR. Symptom reporting is the most common end point for the study of PAR. All the included studies used symptoms for evaluating efficacy, such as ‘nasal symptom score’ or ‘clinical efficacy’ also based on symptom diaries. CHM had beneficial effects in reducing symptoms, including runny nose, nasal obstruction, and sneezing. A follow-up assessment performed 1 year after the end of one of the trials showed long-term benefits produced by the treatment [21].

Some of the studies were designed to evaluate possible improvements in quality of life of the subjects using quality of life (QoL) questionnaires [16, 17, 21, 22]. Thus, CHM management might also have improved some aspects of the QoL of the subjects with PAR, although QoL outcomes were poorly reported in the evaluated studies or not reported at all. The current results are consistent with the Cochrane data indicating an advantage for CHM in PAR [13].

However, the exclusive use of subjective outcomes may limit the validity of results [23]. Various objective outcomes were measured in 6 of the 7 RCTs, which included nasal airflow resistance and dissection area of nostrils [19], total serum IgE [17, 18, 20-22], and serum-specific IgE antibodies [19]. Objective outcome measures should include specific modern parameters such as blood tests (blood eosinophil count and total serum IgE level), nasal cytology (scrapings of secretions and nasal mucosa cells), allergy testing (SPT, intradermal dilution testing), rhinomanometry, or other objective measurements of nasal airflow.

Although the mechanism of CHM in relieving the symptoms on PAR is not yet fully understood, several physiological processes have been considered. First, CHM could lower serum IgE [18, 19, 24]. Second, CHM could modulate the function of lymphocytes and neutrophils [24, 25], also inhibiting the inflammatory mediators involved in AR [18, 19, 26, 27].

Treatment based on pattern differentiation is a characteristic of TCM. According to clinical manifestations, PAR is similar to ‘BiQiu’ in TCM. In the included RCT studies, patients with PAR received CHM treatment, but the information for TCM pattern classification was not taken into consideration. TCM diagnosis includes tongue examination, and evaluation of pulse and symptoms based on TCM pattern differentiation. The signs and symptoms are classified into different patterns, and treatment strategies are made according to the different patterns. Moreover, each pattern has different treatment principles and herbal medicines. For these reasons, pattern classification could be a matter of heterogeneity and bias among the evaluated papers: in fact, because the evaluated trials did not take into account the pattern differentiation, this may have reduced the efficacy of the adopted treatments. We believe that the inclusion of pattern differentiation into the diagnostic process could improve the efficacy of TCM interventions.

The previous RCTs on the efficacy of CHM in PAR have a variety of methodological limitations. Adopting one herbal formula is based on a specific pattern in TCM, and not on the kind of disease; thus, stratifying PAR into several patterns may be a feasible method in clinical trials. The study design should be guided by TCM theory, which is a good method to re-evaluate the efficacy of herbal formulae. Pattern classification could help to evaluate the true clinical efficacy in clinical practice with TCM treatment.

The meta-analysis used in current study had several limitations. First, the quality of the individual-controlled trials was not particularly high. Second, for most of the studies, treatment duration was not long enough to correctly evaluate the efficacy of the adopted treatment for PAR. Most of the literature did not describe any follow-up evaluation, or this was not sufficient for evaluating long-term treatment effects; moreover, long-term follow-up studies might yield different results. Third, there is no assurance that all relevant RCTs were included in the search, which is generally limited at evaluating systematic reviews. Fourth, because CHM approach may be a mixture of different treatments, it may result heterogeneous.

In conclusion, the meta-analysis suggests that CHM treatments have beneficial effects in patients with PAR. However, the evidence remains weak owing to methodological flaws, making it is difficult to draw any firm conclusion. More well-designed studies with large samples and multicenter studies are needed to access the real effects of CHM in treating AR.


We gratefully thank GuangLi Du for his assistance in preparing for the manuscript.

Authors' Contributions

Professor Fan contributed to conception and design of the manuscript and drafted the article. Professor Qian revised the manuscript and corrected grammar and sentence construction. Dr Wang Shijun and Tang Qiaofei collected data and performed analysis.

Conflict of interest

There was no conflict of interest for each named author.