Chinese herbal medicine (“3 medicines and 3 formulations”) for COVID‐19: rapid systematic review and meta‐analysis

Abstract Background To evaluate the evidence behind claims that Chinese Herbal Medicine, specifically “three medicines and three formulations” (3M3F, comprising Jinhua Qinggan, Lianhua Qingwen, Xuebijing, Qingfei Paidu, Huashi Baidu, and Xuanfei Baidu), is an effective treatment for COVID‐19. Methods We searched PubMed, MEDLINE and CNKI databases, preprint servers, clinical trial registries and supplementary sources for Chinese‐ or English‐language randomized trials or non‐randomized studies with comparator groups, which tested the constituents of 3M3F in the treatment of COVID‐19 up to September 2020. Primary outcome was change in disease severity. Secondary outcomes included various symptoms. Meta‐analysis (using generic inverse variance random effects model) was performed when there were two or more studies reporting on the same symptom. Results Of 607 articles identified, 13 primary studies (6 RCTs and 7 retrospective non‐randomized comparative studies) with 1467 participants met our final inclusion criteria. Studies were small and had significant methodological limitations, most notably potential bias in assessment of outcomes. No study convincingly demonstrated a statistically significant impact on change in disease severity. Eight studies reported sufficiently similar secondary outcomes to be included in a meta‐analysis. Some statistically significant impacts on symptoms, chest CT manifestations, laboratory variables and length of stay were demonstrated, but such findings were sparse and many remain unreplicated. Conclusions These findings neither support nor refute the claim that 3M3F alters the severity of COVID‐19 or alleviates symptoms. More rigorous studies are required to properly ascertain the potential role of Chinese Herbal Medicine in COVID‐19.


| INTRODUCTION
China was the first country to be seriously affected by COVID-19.
The first version of the Novel Coronavirus Pneumonia Treatment Plan was published on 16th January 2020, 1 and the Plan was soon revised into the seventh edition. 2 From the fourth revision, the Plan included Chinese herbal medicine (CHM) and recommended CHM to be effective to patients with all stages of disease from observation period to critical phase. 3,4 Six CHM recipes, known as the "3 Medicines and 3 Formulations" (3M3F, [三药三方]), were selected for use.
The The 3M3F were claimed to have significant efficacy after observation of population data, and the role of 3M3F in COVID-19 treatment was officially announced in a Chinese government press conference on 23 March 2020, with promotion as being able to relieve symptoms, and reduce the number of mild of moderate cases progressing to severe cases. 8 Specific claims included that the compound significantly improves immunological indicators for both mild and severe COVID-19; that one of the Medicines (LHQW) and the three formulations are effective in improving radiologically-assessed lung infiltrates; that one of the formulations (XFBD) improves lymphocyte count by 17% and cure rate by 22%; and that another of the formulations (HSBD) reduces the time for viral testing to turn negative and shortens hospital stay by 3 days. One Medicine (LHQW) was suggested to have antiviral and anti-inflammatory effects by inhibiting the SARS-COV-2 replication and reducing the pro-inflammatory cytokines production at the mRNA levels. 9 These claims were widely reported in the Chinese press and also in Chinese researchers' communication to the WHO that the traditional and complementary medicine unit of the WHO highly appraised the role of 3M3F. 10 Due to the concise nature of the official statement, the above findings were communicated in concise language with little detail of data supporting claims. However, despite the paucity of available data, 3M3F was readily and significantly implemented into COVID-19 treatment management. The South China Morning Post reported that over 90% of Chinese COVID-19 patients had been treated with CHM. 11 Large quantities of 3M3F were shipped as part of the Chinese government's aid package to other countries such as Italy, Iran and Iraq. 12,13 Despite unclear evidence of efficacy and some negative press in the West, 14 they have been distributed by local civic organizations such as the Red Cross and Chinese embassies. 15 These organizations were taking the lead because regulations limit the official inclusion of 3M3F in many settings outside of China. Nevertheless, the Chinese guidelines have informed national guidelines for traditional medicine use in COVID-19 in other countries such as Japan and South Korea, which have fully or partially incorporated 3M3F. [16][17][18] It is critically important to independently review the evidence base behind such claims considering such formally promotion in China and on international stages.
Whilst multiple reviews have reviewed the role of herbal medicine-and CHM specifically-for COVID-19, none of them look at 3M3F specifically. Independent review is essential to shed light on the debate around the effectiveness of CHM in the COVID-19 pandemic.
As such, our review is the first systematic review to evaluate whether 3M3F improves outcome in COVID-19 and test the specific efficacy claims outlined above.

| METHODS
This rapid systematic review is reported following the PRISMA checklist. We largely followed Cochrane Interim Rapid Reviews Guidance produced specifically for the COVID-19 pandemic, 19 except for tailoring our search to Chinese bibliographic database. Our team included bilingual authors experienced undertaking systematic review tasks in English and Chinese and familiar with both health systems.

| Search strategy and selection criteria
In early May 2020, we searched PubMed, MEDLINE and CNKI (China National Knowledge Infrastructure) databases with date restrictions (2019-2020). We used keywords and MeSH terms in domains of COVID (e.g., "COVID-19", "Coronavirus"), Chinese and herbal medicine (e.g., "Herbal medicine", "Traditional Chinese Medicine"), official terms for the 6 Medicines, (e.g., "Lianhua Qingwen") and Chinese, English and botanical terms for individual ingredients associated with the 3 formulations (e.g., "Ma Huang"). Using the same or similar key- We included all Chinese-and English-language comparative studies of 3M3F, including randomized controlled trials (RCTs) or nonrandomized studies of interventions. We included any of the 3M3F used separately or together, and alone or in conjunction with other medicines. To be included, a study of any of the three formulations had to report reasonable details of the formulation which were consistent with guidelines from the State Administration of Traditional Chinese Medicine. We included any study on confirmed COVID-19 patients, including those suspected initially and diagnosed retrospectively. We placed no limitation on age, disease severity or ethnicity (in practice, most participants would have been Chinese).

| Quality appraisal of studies
We used the version 2 of the Cochrane Risk of Bias for randomized trials (RoB 2) 23 and the Newcastle-Ottawa Scale for non-randomized studies. 24 One reviewer extracted data and critically appraised the studies (Y. W., J. W.). A second reviewer double checked (reviewer 4, reviewer 5). Disagreements were resolved by a third reviewer.

| Data extraction
Data were extracted by Y. W., reviewer 4, and reviewer 5. for Chinese-language sources, and by J. W. and reviewer 4 for English-language sources. We charted the following fields onto a data extraction sheet: geographic location of recruitment, care setting, inclusion criteria including participants' starting disease severity category, age, gender, proportion of immuno-depression, pre-existing conditions, and pregnancy status.

| Outcome measurements
We predefined a primary outcome domain ("change in disease severity category at the end of treatment"), since this was a major claim at the government press conference. We sought clearly-defined categories (preferably from guidelines) and used clinically in the study settings.
China standardized definitions of disease severity early in the COVID-19 outbreak. The Chinese national guideline categorizes disease severity into mild, moderate, severe and critical; the Chinese Center for Disease Control and Prevention has mild, severe and critical categories. 25 This definition is cited in the US CDC guideline. 26 In other countries, "triage category" is used in regional or local settings. 27 Usually, these categorisations take many clinical characteristics into consideration, including vital signs, symptoms, laboratory, and radiographic findings. We did not include "disposition" (e.g., home care or hospital admission) on its own as a marker of disease severity unless the triage criteria were clearly stated. We included categories "dead" and "cured" if the definition of "cured" was clear, and we did not apply time limitations for disease progression or treatment. Only categorisations from studies using the same definition were eligible for meta-analysis.
We took an emergent approach to secondary outcomes, adjusting our data extraction sheet to reflect outcomes reported in primary studies. Although a disease severity category is already a composite measure, we analysed changes in symptoms separately as secondary outcomes, because of official claims that 3M3F could relieve symptoms. We extracted treatment outcomes of the symptoms reported in COVID-19 patients.
At the time of this review, there was no international consensus on the outcomes that should be reported when studying COVID-19, so we extracted non-symptom outcomes if they were reported in the primary studies; these included laboratory, radiology and healthcare utilization measures. All these outcomes were mentioned in the press conference. 8

| Data analysis
When there were two or more studies reporting on the same outcome measures, we conducted meta-analysis using RevMan [v5.4]. For continuous variables, because of variability in diagnostic and inclusion criteria, interventions, and length of treatments and follow-up, a generic inverse variance random effects model was utilized to pool the mean difference (MD) with 95% confidence interval (CI) to incorporate heterogeneity. 28 When the units of the outcome measures used across studies were not consistent, the effects as standardized mean differences (SMD) were reported. For dichotomous variables, we compared groups using risk ratio (RR) with 95% CI. Heterogeneity was judged moderate when I 2 > 30%, substantial when I 2 > 50%, and considerable when I 2 > 75%. 28 Potential sources were investigated in a sensitivity analysis if appropriate when interpreting the findings.

| Description of dataset
The study flowchart is shown in Figure 1. Thirteen studies -Six ran- The key characteristics of the included studies are given in In all studies except one arm in Ref. 39, CHM were used in conjunction with usual care (as recommended in the current version of the Chinese national guideline), and compared with usual care alone.
"Usual care" in all the studies included three main approaches: nutrition and supportive treatment, symptomatic treatment and antiviral and antibacterial treatment.

| Quality appraisal of included studies
The results of quality appraisal of the included studies are shown in Figure 2.
The quality assessment results of the RCTs are shown in Figure 2 (A). There were various forms of concerns for all six trails or they were considered to be at high risk of bias. When evaluating the randomisation process, three trials 34 Of the non-randomized studies (all of which were retrospective cohort studies), three studies 35,36,38 were found to be of fair quality, while the other four studies 30,32,33,37 were of outstanding quality ( Figure 2(B)). There were extensive exclusion criteria for major diseases (including renal disease, cancer and immunodeficiency) in all studies, and comorbid respiratory diseases were excluded in all but one study 41 . Though the presence of these comorbidities is low for Chinese COVID-19 patients, the population is likely to be representative of patients with COVID-19. 42 The exposed and non-exposed cohort were from the same community. Two studies 35,38 failed to be comparable on the basis of study design, and age or disease severity of patients was normally controlled in other studies. 30,32,33,36,37 All the studies were completed, but only two were considered to be of enough follow-up length: one study 30

| Primary outcome
Our primary outcome measure (change in disease severity category according to clinical guidelines) was adequately reported in only one (non-randomized) study. One study 36 reported that there was a significantly lower proportion of patients becoming severe in the treatment group compared to the comparator group, as judged by a p value less than 0.05 (see Table 2 for numbers). However, it was based on a small sample size with very few events in some of the cells on the 2 Â 2 One randomized controlled trial 40 reported changes in disease severity but we choose not to include these findings because the definition of category used as treatment outcome was not clear. There was also inconsistency in the numbers presented in this study (see Supplementary material S4). Moreover, the study included both mild and moderate patients, but only presented data on progression to severe or dead, missing progression from mild to moderate and progression to critical. We wrote to the corresponding author for clarification, but received no response.
One retrospective analysis 37 of QFPD decoction showed no significant difference in the numbers of patients being cured (as defined by the Chinese national guideline).

| Secondary outcomes Improvement in symptoms
Primary studies measured symptom resolution differently. Fever resolution, for example, was measured in three ways: time taken for fever to resolve, whether fever was resolved after at the end of treatment, and change in symptom score. Assigning a score to a symptom is a common practice in CHM studies, although it has been criticized for systematic errors, non-standardized use in each study and statistical inappropriateness. 43 As a result, we will not report on the Traditional Chinese Medicine (TCM) scoring of symptoms, but have included additional information in Supplementary material S5.     T A B L E 3 Impact on symptoms: findings from analyses not amenable to meta-analysis    concerns in all studies, with especially high risk of bias in outcomes assessment in the four RCTs. Missing and wrong protocol registration information intensifies our concern over the integrity of these studies.
Of the six remedies making up 3M3F, four had been tested in any experimental study that met our inclusion criteria. Our primary outcome measure (reduction in severity of disease) did not achieve convincing statistical significance in any of the primary studies. In relation to the secondary outcomes, the positive effects of LHQW, JHQG, and QFPD on various symptoms could be explained by bias in assessment of outcome (and in particular, the widespread use of the "symptom score" in TCM), and would need to be replicated before being viewed as definitive. Similarly, the positive impacts of different 3M3F remedies on radiological outcome (two studies), laboratory tests of biomarkers (one study) and length of stay (two studies) need to be replicated before being viewed as definitive.
With the exception of diarrhoea with JHQG, the 13 studies did not report any adverse events linked to 3M3F use. Adverse events have, however, been reported in the past when LHQW was used for influenza. 44,45 Previous studies have also reported some digestive system side effects from using JHQG to treat influenza, though not significantly more than the control group. 6 Duan and colleagues attributed the high incidence of diarrhoea in their treatment group to the high dose of JHQG they used to treat COVID-19, and also invoked classical TCM theories to suggest that diarrhoea may have a curative role in this condition.
Although we did not limit the publication language or geography, unsurprisingly all included studies were conducted in China, thus the findings may not be generalisable to other countries. During the editorial process of this manuscript, we noticed a phase three trial of LHQW in Singapore was registered, but the results were not posted yet. 46 There is also no placebo-controlled study, making it impossible to assess the effect of 3M3F when used alone. Most of the articles are of low quality and sample size, potentially limiting their use in informing practice. We also observe some concerning practices in these studies, for example, the number of trials registered in Clinical Trial Registry is small, and in one case we cannot even find the registered protocol using the protocol number given by the authors. Informed

| Strengths and limitations of this review
To our knowledge, this is the first systematic review and metaanalysis of a group of CHM specifically promoted for COVID-19.
Whilst some systematic reviews have examined the impact of integrating any CHM with conventional treatment, 48 49 we had provided more accurate and detailed information in quality appraisals of included studies and, comparing with, 50 independent analysis of outcomes of each intervention. We followed Cochrane interim guidance for rapid reviews during this pandemic, 19 and undertook independent statistical analysis of key findings from primary studies.
One limitation is the small number of primary studies identified.
The relative success of China in managing the initial and second waves of COVID-19 may have limited the ability to conduct trials after detailed protocols based on early clinical experience had been developed. It is also possible that the Chinese government had access to additional unpublished data before developing its official statement on 3M3F. At least 39 clinical trials for CHM interventions were registered in the Chinese Clinical Trials Registry by January 2021 before this review was initiated, though it is unclear how many of these relate to 3M3F. 47 If such data exist, we recommend that they are placed in the public domain, for example, by sharing and regularly updating data under their registries, to ensure clinicians, researchers and policy-makers are appropriately informed. Another limitation is that other traditional medicines used for treatment of COVID-19 were not included in our review. We prioritized 3M3F as it has been officially sanctioned and promoted by the Chinese government for use in China and other foreign countries.

| Suggestions for further research
Larger, multi-centre randomized placebo-controlled trials of CHM, and especially 3M3F, are urgently needed, with consistent inclusion criteria and objective outcome measures designed to contribute to meta-analyses. Better reporting of adverse events is needed to confirm the safety profile of 3M3F. It was beyond the scope of this review to explore the pharmaceutical properties and alleged antiviral mechanisms of the various ingredients; there is much scope for further studies in this area, perhaps with a view to developing new chemical entities for mainstream medicine. Many of these studies were performed before much as known about the disease, or which outcomes were most appropriate for inclusion. Only one study attempted to measure or report viral load of COVID-19 patients or whether this was reduced with the intervention; such variables should be included in further research. Additionally, as our examination focused primarily on the use of CHM in acute COVID-19 treatment, future research examination of CHM for longer-term symptomatic relief may be warranted given that many outcomes measured in the studies are also often reported as significant in post-acute COVID-19. 51

| CONCLUSIONS
The findings from this rapid systematic review neither support nor refute the official claim that CHM (specifically 3M3F) alters the severity of COVID-19 or provides alleviation of symptoms. While the limited studies appear to suggest that 3M3F, when used on top of usual care, may offer some relief for some symptoms experienced by mostly mild or moderate COVID-19 patients, the results do not support the high-level claims that 3M3F could prevent disease from progressing to a more severe type. Studies were few in number, small in size, and had significant methodological limitations (most notably, potential bias in assessment of outcomes), though the positive nature of some individual findings do suggest further examination may be warranted.
More rigorous multi-centre randomized placebo-controlled trials with decent sample sizes are required to properly ascertain the potential role of CHM in treatment of COVID-19.