Meta-Analysis on the Chinese Herbal Formula Xiaoer-Feike Granules as a Complementary Therapy for Children With Acute Lower Respiratory Infections

Background Over the past five years the Chinese herbal formula (CHF) medicine, Xiaoer-Feike granules (XFG), has become a widely used adjuvant therapy for acute lower respiratory infections (ALRI). Considering the rapid popularization and application of XFG, and the lack of systematic evidence evaluating its effectiveness and safety in treating ALRI, it is necessary to conduct a meta-analysis to determine its benefits for patients. Methods This study systematically identified randomized controlled trials (RCTs) of XFG treatments for ALRI through July 2019 using four English-databases (PubMed, Cochrane Library, Ovid, and Web of Science) and four Chinese-databases (Sino-med database, China National Knowledge Infrastructure (CNKI), VIP database, and the WANFANG database). We then performed a quality assessment and data analysis with Review Manager 5.3.5 and Stata 15.1. Results Twenty-one RCTs involving 3425 patients were randomly divided into an XFG group and a conventional medicine (CM) group. The results showed that the clinical efficacy rate (CER) of the XFG group was significantly higher than that of the CM group (RR=1.17, 95% CI =1.13-1.22, P< 0.00001). In comparison with the CM group, the XFG group had strikingly shortened: resolution time of cough (RTC) (MD = -1.92; 95% CI =-2.33, -1.51, P<0.00001); resolution time of rale (RTR) (MD = -1.68; 95% CI =-2.27, -1.10, P<0.00001); resolution time of fever (RTF) (MD = -1.46; 95% CI =-1.92, -1.00, P<0.00001); resolution time of inflammatory lesions (RTIL) (MD = -2.43, 95% CI =-2.94, -1.93, P< 0.00001); and hospital stays (HS) (MD = -2.26, 95% CI =-3.03, -1.49, P< 0.00001). At the cellular and molecular level, the CD4, CD8, CD4/CD8, IL-6, TNF-α, and CRP levels were significantly improved when CM was complemented with XFG. In addition, no significant difference was observed between the XFG and CM groups in terms of the adverse events (AE) (RR =0.97, 95% CI= 0.61-1.54, P= 0.89). Conclusions The findings of this meta-analysis support the use of XFG in the treatment of ALRI. However, these results should be treated with caution due to the significant heterogeneity and publication bias of existing data. Further well-designed and high-quality RCTs are needed to interrogate the efficacy and safety of XFG.


Article Inclusion and Data Extraction
This systematic review was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analyses Statement (PRISMA). We selected eligible studies based on the following inclusion criteria: (1) the experimental group was treated with the XFG included in the Chinese Pharmacopoeia (edition 2015) regardless of the pharmaceutical company, (2) the control group was treated with conventional medicines alone, and (3) the clinical efficacy rate (CER), resolution time of cough (RTC), resolution time of rale (RTR), resolution time of fever (RTF) and HS were used as the primary outcomes by referring to guidelines or consensus views or the evaluation criteria. (4) In addition to the CER, at least one other outcome was employed. The exclusion criteria were: (1) non-RCTs or only one outcome; (2) the XFG group received the combination treatment of XFG and acupuncture therapy or other TCM; (3) studies that did not have control groups, or studies in which control subjects received TCM treatment including herbal medicine, acupuncture, or acupoint injection therapy; (4) the XFG that was not recorded in the Chinese Pharmacopoeia; and (5) systematic review, important data reports, and case reports were addressed. Two reviewers independently checked for relevant literature, including the first author's last name, year of publication, sample size, gender distribution, age, disease duration, intervention, duration of treatment, dose, outcomes and adverse events, and clinical diagnosis.

Quality Assessment
Two reviewers independently evaluated the methodological qualities of the trials according to the Cochrane manual. We identified a list of seven categories for risk of bias: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Each item was classified into low bias risk, high bias risk, and unclear bias risk. Disagreements between the reviewers were settled through discussion.

Data Synthesis and Analysis
In this review, statistical analyses were conducted by a reviewer manager (version 5.3.5), and we used OR with 95% CI for the analyses of dichotomous data, whereas the continuous data were presented as MD or sMD with 95% CI. The data were merged according to the Mantel-Haenszel (fixed-effects) model and the DerSimonian and Laird (random-effects) model. The heterogeneity between studies was determined by the chisquare test. With the I 2 statistic, an I 2 < 25% indicating that heterogeneity may not be important, values between 25% and 50% represent moderate inconsistency, and I 2 > 50% suggest severe heterogeneity. We defined P≥ 0.1 and I 2 < 50 as indications that the results are consistent and that the fixedeffects model would be used, while I 2 > 50% was an indicator of significant heterogeneity among trials. Then, a random-effects model was used to pool the results to minimize the influence of potential clinical heterogeneity. We evaluated the robustness of publication bias and sensitivity analyses to evaluate the robustness of the merged results with Stata 15.1. All the statistical tests were two tailed, and the differences were statistically significant at P< 0.05.

Search Results and Study Characteristics
A total of 272 papers were obtained from a database search of the following sources: Web of Science (n=0), Cochrane Library (n=2), Ovid (n=0), PubMed (n=0), CNKI (n=53), Sino-med (n=48), VIP (n=67), WANFANG (n=102), from which 129 duplicated publications were removed. Fifty-nine citations of irrelevant topics were excluded after reading the titles and abstracts (irrelevant studies (n=52), review studies (n=5), and animal test studies (n=2)), and 63 studies were ruled out following a screening of the full text (irrelevant interventions (n=39), not RCT studies (n=15), and conference papers (n=9). Finally, 21 RCTs published between 2014 and 2018 involving 3425 patients with ALRI were eligible according to the inclusion criteria. Two reviewers extracted data from the literature, including information on general trial characteristics (first author's last name, publication date); baseline patient and disease data (number of patients in each group, age and disease course); interventions (XFG, conventional medicine, treatment duration, and dose), outcome definitions, detailed adverse reactions, and clinical diagnosis. The sample size was 60 to 600 with significant age differences. The treatment duration ranged from 5 to 14 days. The XFG was provided by two Chinese pharmaceutical companies, Tian-sheng Pharmaceutical Group co., LTD, and Chang-chun People Pharmaceutical g=Group co., LTD. Twelve trials reported a specific number of adverse events. Of the 21 trials, 14 included patients diagnosed with acute bronchitis, and the other 7 studies selected pneumonia patients. Generally, the basic characteristics of the two groups were consistent, and no significant difference was found before the intervention. The procedure of the literature search is shown in Figure 1, and the general characteristics of the selected studies are listed in Table 1.

Methodological Quality Assessment
The specific randomized methods were detailed in 11 studies by random number tables in the assessments of selection bias, and we considered them as low-risk (Wang and Guo, 2014;Du, 2016;Ye and Zhen, 2016;Ma and Lu, 2017;Chu, 2018;Liao and Lou, 2018;Qian and Li, 2018;Wei and Li, 2018;Zhang and Guo, 2018;. One study (Liu, 2015) was high-risk due to its use of the registration order to generate a random sequence. The remaining 9 articles did not offer any specific information regarding the generation of random sequences. Almost all the studies failed to give the specific allocation concealment, performance bias, and detection bias. The reporting bias was at high risk in 3 documents (Liu, 2015;Kong, 2016;Ye and Zhen, 2016) because they failed to report the pre-listed outcomes. On the whole, the 21 studies fell in the middle because of the high proportion of the unclear risk of biases in the majority of studies. The specific results of the bias assessment are summarized in Figure 2.
XFG and CM, and 1642 patients in the control groups were treated with CM alone. The heterogeneity test suggested that the fixed-effects model was more suitable (P=0.008, I 2 = 48%). The result showed a statistically significant difference between the two groups, which indicated that adding XFG provided more benefits to ALRI patients. (RR=1.17, 95% CI =1.13-1.22, and P< 0.00001), and no significant difference was found between acute bronchitis and pneumonia groups (P= 0.37, I 2 = 0%), as shown in Figure 3.

Subgroup Analysis
Subgroup analyses of the treatment durations were introduced to determine the subgroup differences according to the primary outcomes, including the CER, RTC, RTR, and RTF. For the CER, RTC, and RTR, the results across the various subgroups were highly consistent, and the benefits of XFG were significant. However, a significant difference was observed between the "5-8 days" and "14 days" groups in terms of the RTF. Generally, the "14 day" group showed a lower effect size value and a narrower confidence interval than that of the "5-8 days" group, which  indicated that "14 days" of treatment was more likely to bring the maximum therapeutic effect to the ALRI patients in terms of improving the RTF. Despite the significant difference favoring the 14-day group, it is unreasonable to conclude that 14 days of XFG was more effective than 5-8 days because the number of studies between subgroups varied widely, and there are some differences between bronchitis and pneumonia. In terms of the dose, most studies followed the packaging instructions for the   drug provided by the two pharmaceutical companies (ages: 0-1 (2g); 1-4 (3g); 5-8 (6g), tid, PO). Therefore, considering the large age difference between the studies and the ALRI severity difference, it is reasonable to adjust the medication according to different conditions, as shown in Table 2.

Sensitivity Analysis
Stata 15.1 was employed for sensitivity analysis of the primary outcomes including the CEF, RTC, RTR, and RTF. The centralized small circles in the sensitivity analysis plots suggested that the results were not reversed by removing any study for each outcome, indicating that the results were reliable, as shown in Figure 10.

Publication Bias
We used Stata15.1 software to detect possible publication bias according to primary outcomes, and the trim and filling method was used to cope with the striking publication bias if the P<0.05.
The result of Egger's test suggested that there was no publication bias in terms of the RTC (P >| t |=0.542, 95% CI= −2.6 to 4.79) and RTF (P >| t |=0.48, 95% CI= -4.17 to 8.21). However, significant publication bias was observed in the CER (P >| t |= 0.0001, 95% CI= 1.61-2.84) and the RTR (P >| t |=0.04, 95% CI= 0.35 to 12.56). The trim and filling method was then used to evaluate the reliability of results affected by significant publication bias. After two iterations, no potential missing study was filled. The result suggested that the MD and 95% CI   .0001) before and after applying the trim and filling method was consistent. Generally, the publication bias induced by CER and RTR had no significant effect on the pooled results, as shown in Figure 11.

DISCUSSION
This meta-analysis assessed the evidence from 21 RCTs. A total of 3425 ALRI patients were randomly assigned to receive either an XFG addition with CM or CM alone. The findings of this study suggested that XFG supplementation of the ALRI treatment was effective and safe, and it was associated with significant improvements in the RTC, RTR, RTF, RTIL, CD4, CD8, IL-6, TNF-a, and CRP. Both the symptoms and immunity-related outcomes of the patients were significantly improved when CM was complemented with XFG, which indirectly reduced the use of antibiotics. However, substantial heterogeneity was present in the three primary outcomes (RTC, RTR, and RTF). According to the sensitivity analysis, two studies Yan, 2017) significantly increased the overall heterogeneity. A possible explanation for this finding is that Li, 2015 had a larger sample size compared with the other 20 included studies, while Yan, 2017 had a small sample size. In addition, differences in age and the severity of the disease appear to be another important source of heterogeneity. XFG is a patented preparation. Two Chinese pharmaceutical companies, Tian-sheng Pharmaceutical Group Co., LTD, and Chang-chun People Pharmaceutical Group Co., LTD, produce XFG according to the guidelines outlined in the latest edition of the Chinese pharmacopoeia, which consists of Ginseng radix et rhizoma, Poria, Atractylodis macrocephalae rhizoma, Citri reticulatae pericarpium, Rhei radix et rhizome, Lycii cortex, Glehniae radix, Glycyrrhizae radix et rhizome, Artemisiae annuae herba, Ophiopogonis radix, Cinnamomic ramulus, Zingiberis rhizoma, Aconiti radix, Trichosanthis fructus, Farfarae flos, Asteris radix et rhizome, Mori cortex, Astragali radix, Lycii fructu, Arisaema cum Bile, Galli gigerii endothelium corneum, Trionycis carapax, and saccharose. The procedure for making this preparation is as follows. First, ten herbs are mixed, including Astragali radix, Lycii cortex, Glehniae radix, Glehniae radix, Glycyrrhizae radix et rhizome, Artemisiae annuae herba, Cinnamomic ramulus, Trichosanthis fructus, Asteris radix et rhizome, and Mori cortex. They are decocted with water twice, for 2 hours per decoction, and the decoctions are combined,   filtered, and concentrated to form a thin extract with a relative density of 1.26-1.30 (80°C). Second, the other twelve ingredients are pulverized into a powder. Third, in addition to the above thin extract, the powder and saccharose are mixed, dried, and made to reach 1000g. This preparation method for XFG is stable and repeatable ( Figure 12). The primary pharmacological effects of the 22 ingredients include lung injury protection, enhancing immune function, antimicrobial, anti-inflammatory, and anti-asthmatic activities, neuroprotection, and antitussive and expectorant effects. The majority of the 22 ingredients have strong anti-microbial and anti-inflammatory properties, which significantly enhanced resistance to pathogenic microorganisms in ALRI patients. Seven herbs, namely, Citri reticulatae pericarpium, Glycyrrhizae radix et rhizome, Glycyrrhizae radix et rhizome, Trichosanthis fructus, Farfarae flos, Trichosanthis fructus, and Mori cortex, are widely used for antitussive, expectorant, and anti-asthmatic effects by TCM physicians for thousands of years. Modern pharmacology has also confirmed these effects on respiratory disease (Ruan and Yue, 2004;Shi et al., 2009;Sun et al., 2012;Kim et al., 2017;Singh et al., 2018;Fu et al., 2019;Li et al., 2019). Seven ingredients (Ginseng radix et rhizoma, Poria, Atractylodis macrocephalae rhizoma, Ophiopogonis radix, Astragali radix, Astragali radix, and Trionycis carapax) showed biological activity in immune enhancement, especially Ginseng radix et rhizome, which significantly increased the viability and proliferation of spleen cells by promoting proliferation from CD19(+) B cells and increased the surface expression of CD25, CD69, and interleukin-2, which significantly enhanced the immune system of ALRI patients. Two herbs (Lycii cortex and Cinnamomic ramulus) showed striking neuroprotection effects, with kukoamine A (Zhang et al., 2016) from Cortex lycii reducing the malondialdehyde level and increasing the glutathione level, superoxide dismutase, catalase (CAT) activities, and the expression of BDNF. Three herbs significantly improved acute lung injury in mice (Asteris radix et rhizome, Astragali radix, and Lycii fructus) due to mechanisms that enhanced the activation of Nrf2, inducing human cathelicidin antimicrobial peptide LL-37 in respiratory epithelial cells, promoting the repair of vascular endothelial cells, and inhibiting NF-kB signaling pathway. On  the whole, the TCM efficacy of XFG was consistent with its pharmacological effects, and XFG plays a therapeutic role through coordinated cooperation among the 22 ingredients. Therefore, based on TCM theory, through a multiple-target pathway, multi-level and holistic therapy, XFG was found to be a reliably effective and safe complementary therapy in ALRI treatment, as shown in Tables 3, 4.

Implications for Further Study
This review has evaluated the curative effects of XFG, but most of the included studies used CER, RTC, RTR, and RTF as the primary outcomes. A few studies reported on the immune cells (CD4, CD8, and WBC) and cytokine levels (TNF-a, INF-g, IL-4, IL-6, IL-8, IL-10, and CRP, etc.), Du, 2016;Ma and Lu, 2017;Zhao and Chen, 2017;Hu et al., 2018;Zhang and Guo, 2018;. This indicates that the therapeutic mechanisms of XFG are associated with immune enhancement and inflammation reduction. However, this explanation is far from conclusive, because these cellular and molecular parameters were only reported in three studies. The therapeutic effects of XFG are multiple-target, multi-level, and holistic, and therefore, at the cellular and molecular levels, the mechanisms of action of XFG warrant further investigation. Although the evidence analyzed in this study indicated that XFG is an effective and safe adjuvant therapy in treating ALRI, we found significant heterogeneity and publication bias. Therefore, high-quality, multicenter, double-blind RCTs are needed to provide reliable evidence that will guide the rational use of XFG in clinical settings.

Limitations
Several limitations should be highlighted in our meta-analysis. First, we searched only the primary English and Chinese databases. Therefore, some studies that meet our inclusion criteria in other languages or databases may have been excluded. All the included trials declared randomizations, but only eleven studies described a specific randomization method.
Blinded assessments were not detailed in all the included documents, which may have exerted a potential impact on the objectivity of the ALRI outcomes. Second, the inclusion criteria of these studies had small sample sizes with low-quality  (2)Poria cocos (Schw) Wolf., 20g; (3)Dried rhizomes of Atractylodes macrocephala Koidz., 8g; (4)Aged fruit peel of Citrus reticulata Blanco., 20g., (5) Dried rhizomes of Rheum palmatum L., 12g; (6) Dried root peels of Lycium chinense Mill., 23g., (7)      designs, which may have led them to overvalue the benefits of XFG. Additionally, there may be a degree of selective reporting bias because the majority of studies were not officially registered. Third, only a few eligible studies reported the HS, RTIL, cellular, and cytokine outcomes. Fourth, six papers did not mention any information on adverse reactions. Thus, insights as to the safety of using XFG to treat ALRI are limited. Fifth, although the RTC and RTR were widely used to evaluate the curative effect of XFG for ALRI in hospitalized patients, this was still somewhat subjective. Although these limitations may undermine the quality of the evidence, the trials are nevertheless comparable, and the documents were selected using strict inclusion criteria. Since the patients of the selected studies were primarily from China, the conclusion of this meta-analysis might not apply to other ethnic groups. Therefore, large sample trials of a higher-quality, with well-designed considerations of different ethnic groups should be conducted in the future to provide more reliable evidence regarding the efficacy and safety of using XFG to treat ALRI.

CONCLUSION
The findings of this meta-analysis support the use of XFG in the treatment of ALRI. However, the results should be treated with caution due to the significant heterogeneity and publication bias of

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

AUTHOR CONTRIBUTIONS
QY and LL conceived this review and completed the manuscript. DL performed the literature searches electronically and manually. DY and LC performed the study selection and data extraction and assessed the risk of bias. Y-pL and H-pC critically revised the paper. All authors contributed to the article and approved the submitted version.