Role of Traditional Chinese Medicine in Treating Severe or Critical COVID-19: A Systematic Review of Randomized Controlled Trials and Observational Studies

Background: The coronavirus disease 2019 (COVID-19) continues to spread globally. Due to the higher risk of mortality, the treatment of severe or critical patients is a top priority. Traditional Chinese medicine (TCM) treatment has played an extremely important role in the fight against COVID-19 in China; a timely evidence summary on TCM in managing COVID-19 is crucial to update the knowledge of healthcare for better clinical management of COVID-19. This study aimed to assess the effects and safety of TCM treatments for severe/critical COVID-19 patients by systematically collecting and synthesizing evidence from randomized controlled trials (RCTs) and observational studies (e.g., cohort). Methods: We searched nine databases up to 19th March 2022 and the reference lists of relevant publications. Pairs of reviewers independently screened studies, extracted data of interest, and assessed risk of bias. We performed qualitative systematic analysis with visual presentation of results and compared the direction and distribution of effect estimates for each patient’s important outcome. We performed sensitivity analyses to observe the robustness of results by restricting analysis to studies with low risk of bias. Results: The search yielded 217,761 records, and 21 studies (6 RCTs and 15 observational studies) proved eligible. A total of 21 studies enrolled 12,981 severe/critical COVID-19 patients with a mean age of 57.21 years and a mean proportion of men of 47.91%. Compared with usual supportive treatments, the effect estimates of TCM treatments were consistent in direction, illustrating that TCM treatments could reduce the risk of mortality, rate of conversion to critical cases, and mechanical ventilation, and showed significant advantages in shortening the length of hospital stay, time to viral clearance, and symptom resolution. The results were similar when we restricted analyses to low-risk-bias studies. No serious adverse events were reported with TCM treatments, and no significant differences were observed between groups. Conclusion: Encouraging evidence suggests that TCM presents substantial advantages in treating severe/critical COVID-19 patients. TCM has a safety profile that is comparable to that of conventional treatment alone. TCMs have played an important role in China’s prevention and treatment of COVID-19, which sets an example of using traditional medicine in preventing and treating COVID-19 worldwide.


INTRODUCTION
Coronavirus disease 2019  is an acute infectious disease caused by the severe acute respiratory syndrome coronavirus virus 2 (SARS-CoV-2), which has become the most widespread global pandemic that human beings have encountered in the past 100 years (WHO, 2020) and continues to threaten human life and health. As of March 8, 2022, more than 433 million patients have been affirmed worldwide; of those, more than 5.9 million have died (WHO, 2022a).
According to the World Health Organization (WHO) severity definitions, COVID-19 patients are categorized into non-severe, severe, and critical (Agarwal et al., 2020). In the clinical practice, severe or critical patients have the top priorities in the treatment of COVID-19 because of a higher risk of mortality. Despite global efforts to identify effective therapeutic strategies for severe or critical COVID-19, only systemic corticosteroids, interleukin 6 (IL-6) receptor blockers (tocilizumab or sarilumab), or baricitinib combined with corticosteroids were strongly recommended by the WHO after evaluating more than 200 drugs involved in more than 400 randomized controlled trials (RCTs).
Integrated traditional Chinese medicine (TCM) and western medicine has played an extremely important role in the fight against COVID-19 in China during the pandemic (Ge et al., 2021). According to the WHO Report on the evaluation of TCM in the treatment of COVID-19, TCM has been demonstrated to be beneficial for mild-to-moderate COVID-19 patients; however, they did not make a conclusion for severe/critical patients because only one RCT was included for evaluation (WHO, 2022b). At present, encouraging evidence has been documented that TCM presented substantial benefits for severe or critical patients such as reducing the risk of mortality (Chen G. et al., 2020;Wang et al., 2021a;Sun et al., 2021;Zhang L. et al., 2021;Zhou et al., 2021) and rate of conversion to critical cases (Chan et al., 2020;Xu Z. et al., 2020), shortening time of nucleic acid conversion (Wang et al., 2021b;, and improving the local and systemic inflammatory response Fan et al., 2020;Runfeng et al., 2020). Given there is still no globally uniform therapeutic strategy for severe or critical COVID-19 patients, a timely evidence summary is crucial to update the knowledge of healthcare for better clinical treatment of COVID-19.
To provide trustworthy evidence to elaborate on the role of TCM against COVID-19, we conducted a systematic review with a rigorous methodology to comprehensively summarize the efficacy and safety of TCM in the treatment of severe/critical COVID-19 patients by systematically collecting available evidence from RCTs and observational studies.

METHODS
This study was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist (Page et al., 2021).

Eligibility Criteria
We included studies that met the following criteria: 1) patients with confirmed severe or critical COVID-19 according to the national or international recognized diagnosis standard, aged 18 years or older; studies that included patients with both non-severe and severe or critical COVID-19 were eligible if more than 80% patients were severe and/or critical; and studies that included information of severe or critical patients that could be extracted from a subgroup-analysis also were eligible. 2) Comparison: on the basis of usual supportive treatment, treatment with TCM versus without TCM. We did not limit the form of Chinese medicine; granules, decoction, and injections were all considered to be included. 3) Outcomes: we decided the outcomes of interest according to a living network metaanalysis published in BMJ  and a core outcome sets of COVID-19 (Jin et al., 2020) mainly including: a. clinical efficacy (e.g., mortality, length of hospital stay, and rate of mechanical ventilation), b. clinical symptoms (e.g., fever, cough, expectoration, and tiredness), c. laboratory indicators (e.g., lymphocyte percentage, white blood cell (WBC) count, C-reactive protein (CRP), and tumor necrosis factor-α (TNF-a)), d. adverse events (e.g., nausea and vomit, diarrhea, and abnormal liver function). 4) Study types: RCTs and observational studies (e.g., cohort study and historical control study).
We excluded studies that mainly did not report information about the ethical approval, and the study design was protocol, case report, case report series, cross-sectional study, and controlled before-after study.

Literature Search
Systematic searches were performed under the guidance of an experienced librarian of nine databases from December 2019 to 19th March 2022, including PubMed, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science, China National Knowledge Infrastructure (CNKI), WanFang Database, Chinese Biomedical Literature Database (CBM), China Science and Technology Journal Database (VIP), and the L-OVE COVID-19 Repository. We employed an extremely sensitive search strategy, which only included search terms related to disease  and study design, without restrictions on publication language and interventions. (see Additional file 1: Supplementary Table S1 for detailed search strategies). We also tracked the references of relevant publications.

Study Selection
We used EndNote X8.0 to manage the initial searched records. After removing duplicate records, the remaining records were imported into online reference management software Rayyan (Ouzzani et al., 2016). In total, teams of 2 reviewers (LMT and XXL, LYF and ZHF, and PB and LHH), following training and calibration exercises, independently screened the title and abstracts of each record and downloaded the full text of potentially eligible studies for further reviewing to determine the final eligibility. Any conflict was resolved through discussion.

Data Extraction
For each eligible trial, teams of two reviewers (LMT and ZHF, SMY and LY, and LYF and TC), following training and calibration exercises, extracted data independently using a standardized, pilot-tested data extraction form. We used a data extraction form to collate data from the included studies. We extracted the following from the studies: general information (first author, year of publication, trial registration, journal, timeline of patient recruitment, and recruitment location), characteristics of patients (age, patient type, sample size, sex, and complications), details of interventions (specific treatment measures, dose, components of Chinese herbal medicine used, and treatment duration), type of study design, and outcomes of interest.

Risk of Bias Assessment
For each eligible RCT, teams of two reviewers (LMT and ZHF, SMY and LY, and LYF and TC), following training and calibration exercises, independently assessed the risk of bias using a modification of the Cochrane tool for assessing risk of bias in randomized clinical trials (RoB 2.0) (Sterne et al., 2019;MAGIC 2022a) from the following six domains: bias from the randomization process generated, bias due to deviations from the intended intervention, bias due to missing data, bias in measurement of the outcome, bias in selection of the reported results, and bias due to other sources (e.g., consistency between the registration information and the final report and completeness of the report).
The risk of bias of each eligible cohort study was assessed using a revised Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) (Sterne et al., 2016;MAGIC 2022b) from the following six domains: bias due to confounding, bias in selection of participants into the study, bias from the interventions, bias due to missing data, bias due to measurement of the outcome, and bias in selection of the reported results.
Based on the aforementioned criteria for identifying the risk of bias, eligible studies were categorized into four groups: 1) low risk of bias, 2) probably low risk of bias, 3) probably high risk of bias, and 4) high risk of bias. Reviewers resolved discrepancies by discussion and, when not possible, with adjudication by a third party (GL). Detailed guidance for assessment of risk of bias was presented in Additional file 1: Supplementary Table S2.

Evidence Synthesis
When there were two or more eligible studies with the same study design, intervention, and controls, a meta-analysis was performed using Review Manager software (RevMan, version 5.3, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) . If the aforementioned requirements cannot be met, a qualitative systematic analysis would be conducted based on the data of RCT and observational studies, respectively. For dichotomy, we calculated the risk ratio (RR) with corresponding 95% confidence interval (95%CI). For continuous outcomes, we calculated mean difference (MD) with 95%CI. We directly used effect measures reported in observational studies if they reported the results with adjustment of confounders. Findings from each individual study were graphically presented using the effect of direction plot, which showed the benefits and harms of the included studies for each outcome. The risk of bias for each outcome was shown in different color: green for low risk of bias and red for high risk of bias. The direction of the arrow represented the benefits (upwards) and harms (downwards) impact. If the arrow was closed and filled, the effect of intervention was statistically significant; the hollow arrow indicated that there was no significant difference. We drew clustered bar charts to represent the proportion or time for both treatment/exposure and control/non-exposure groups, and effect measures were calculated. The color of the bars shows the risk of bias. In addition, an effect direct plot and clustered bar charts also were used to represent type of patients (proportion of severe or critical patients), type of TCM treatments, study design, and sample size. We compared the direction and distribution of effect estimates of different TCM treatments across studies to show the benefit and harm of TCM compared with usual supportive treatment. We restricted analyses to studies with low risk of bias to observe the robustness of results. We considered that the results were robust if more than half of studies were low risk of bias for each outcome. If the number of studies was sufficient, we also performed subgroup analyses by severity of patients, type of TCM treatments, age, and comorbidities. A radar chart was used to show the rate of common adverse reactions in included studies, and the incidence rate of each adverse reaction corresponded to a coordinate axis respectively.

RESULTS
Our searches yielded 215,761 unique studies. Of the total, 248 were potentially eligible and for further full-text review. We identified 212 studies related to TCM intervention in the first step and further restricted to studies that focused on severe and/or critical patients, in which 21 studies (six RCTs and 15 observational studies) proved eligible ( Figure 1).

Risk of Bias in Included Studies
Additional file 1: Supplementary Table S5 presents the assessment of risk of bias of single RCTs. A summary of risk of bias of six RCTs is presented in Figure 2A. Only two RCTs Zhou et al., 2021) were judged at low or probably low risk of bias in all domains. All other RCTs had probably high or high risk of bias in at least one of the domains. The limitations mainly were from the concealment of random sequence number assignment and unblinding.
Additional file 1: Supplementary Table S6 presents the assessment of risk of bias of single observational studies. A summary of risk of bias of 15 observational studies is presented in Figure 2B. Seven observational studies (Chen G. et al., 2020;Wang et al., 2021a;Shu et al., 2021;Sun et al., 2021;Zhang L. et al., 2021) were judged at low or probably low risk of bias in all domains. Eight eight studies (Zhao et al., 2020b;Qin et al., 2020;Wang et al., 2021b;Chen L. et al., 2021;Huang et al., 2021;Zhang and Pan, 2021;Zhou et al., 2021;Xiong et al., 2022) had probably high or high risk of bias in at least one of the domains. The limitations mainly were un-adjustment of confounding factors. Figure 3 is an effect direction plot that presents the benefit and harm of included studies for each outcome. Although about half of the results were from studies with high risk of bias, it clearly showed that almost all studies had beneficial health impact on the outcomes of interest. Figure 4 and Figure 5 show the proportions or times for treatment/exposure and control/non-exposure group and the calculated effect measures for all outcomes. Figure 5 presents the dichotomous outcomes on the left and the continuous outcomes on the right. Additional file 1:

Effects of TCM Interventions
Supplementary Tables S5-S14 presents the detailed outcome data of intervention group vs. control group, and effect estimates with 95% confidence interval (CI).

Mortality
Ten studies (two RCTs Zhou et al., 2021) and eight observational studies (Chen G. et al., 2020;Wang et al., 2021a;Feng et al., 2021;Shu et al., 2021;Sun et al., 2021;Zhang L. et al., 2021;Zhang and Pan, 2021)) enrolling 12,252 severe/critical patients reported mortality ( Figure 4). Compared with usual supportive treatments, the effect estimates of all TCM treatments were consistent in direction, illustrating that TCMs had more benefits. Of them, seven studies showed statistical differences in reducing the risk of mortality. One study reported accumulate survival rate and showed that TCM have a potentially higher survival rate although no significant difference was observed. A national retrospective registry study enrolled 8,936 mixed patients (including severe and non-severe) to explore the association between QFPD  use and in-hospital mortality, and their subgroup analyses showed that there were no significant subgroup effects for sex and age (P interaction = 0.67, 0.86, respectively); however, a larger reduction in the risk of mortality was found in patients without prior medical history/comorbidities status (P interaction = 0.02). Another cohort study (Chen G. et al., 2020) enrolled 662 patients, and a multivariate logistic regression analysis showed that the risk of mortality increased by 5.0% with every 1 year of age increase.

Rate/Duration of Mechanical Ventilation
Three studies (two RCTs Zhou et al., 2021) and one observational study (Qin et al., 2020)) enrolling 215 severe/ critical patients reported the rate of mechanical ventilation (Figure 5 A). All studies showed consistent direction that TCM treatments could reduce the mechanical ventilation rate, and both RCTs showed significant differences between TCM treatments and usual supportive treatment. One RCT (Hu F. et al., 2021) including 40 severe COVID-19 patients showed that TCM could significantly shorten the duration of mechanical ventilation compared to usual supportive treatment (Figure 5 B).

Rate/Duration of Nucleic Acid Conversion
Two studies (one RCT (Wen et al., 2020) and one observational study (Wang et al., 2021b)) involving 111 severe patients reported this outcome (Figure 5 A). Evidence from RCT (Wen et al., 2020) did not show significant improvement in increasing the rate of nucleic acid conversion at 7 days; however, an observational study (Wang et al., 2021b) showed that TCM treatment could significantly increase the rate of nucleic acid conversion compared to usual supportive treatment. Four observational studies Huang et al., 2021;Shu et al., 2021;Xiong et al., 2022) enrolling 233 severe/ critical patients reported the time to the nucleic acid conversion (Figure 5 B). Compared with usual supportive treatments, the effect estimates of all TCM treatments were consistent in direction, illustrating that TCMs had more benefits. Of them, three studies Shu et al., 2021;Xiong et al., 2022) showed statistical differences in the shortening time of nucleic acid conversion, which ranged from 0.88 to 5.67 days.

Rate of Conversion to Moderate/Critical Cases
Four studies (three RCTs (Wen et al., 2020;Luo et al., 2021;Zhou et al., 2021) and one observational study (Xiong et al., 2022)) enrolling 339 severe/critical patients reported the rate of conversion to critical cases (Figure 5 A). Three studies (two RCTs (Wen et al., 2020;Luo et al., 2021) and one observational study (Zhang and Pan, 2021)) enrolling 141 severe/critical patients reported the rate of conversion to moderate cases (Figure 5 A). All studies showed consistent direction that TCM treatments could reduce the rate of conversion of severe to critical cases and increase the rate of conversion of severe to moderate cases. Two RCTs (Wen et al., 2020;Luo et al., 2021) showed statistical differences in reducing rate of conversion of severe to critical cases.

Rate/Time of Symptom Resolution
Two observational studies Zhang and Pan, 2021) enrolling 74 severe/critical patients reported the rate of symptom resolution, and three studies (one RCT  and two observational studies (Wang et al., 2021a;) enrolling 268 severe patients reported the time of symptom resolution (Figure 5 B). All studies showed consistent direction that TCM treatments could increase the rate of symptom resolution and shorten the time of symptom resolution. Two studies (Wang et al., 2021a;Luo et al., 2021) showed statistical differences in shortening the time of symptom resolution.

Discharge Rate and Length of Hospital Stay
The rate of discharge was reported in four observational studies (Wang et al., 2021a;Wang et al., 2021b;Feng et al., 2021; on 447 severe/critical patients (Figure 5 A). Compared with usual supportive treatments, the direction of effect estimates of TCM treatments were consistent in reducing rate of discharge, illustrating that TCM had more advantages. Four studies (one RCT  and three observational studies (Zhao et al., 2020b;Qin et al., 2020;Feng et al., 2021)) with four TCM enrolled 261 severe/critical patients reported the length FIGURE 3 | Effect direction plot summarizing direction of health impacts from each single study.
Frontiers in Pharmacology | www.frontiersin.org July 2022 | Volume 13 | Article 926189 6 of hospital/ICU stay (Figure 5 B). Compared with usual supportive treatment, RDN significantly shortened length of hospital stay; XBJ and RDN also showed advantages in shortening the length of ICU stay.

Biochemical Indicators
We divided laboratory indicators into biochemical, inflammatory biomarkers, coagulation, and hematologic. The specific indicators included are detailed in Supplementary Table S13. Four studies (one RCT (Hu F. et al., 2021) and four observational studies (Qin et al., 2020;Wang et al., 2021b;Chen L. et al., 2021;Sun et al., 2021)) enrolling 489 severe/critical patients reported biochemical indicators. The results showed that after XBJ  treatment, alanine transaminase (ALT) and aspartate aminotransferase (AST) were significantly higher than those before treatment and better than usual supportive treatment, with statistical difference. TCMD (Sun et al., 2021) was statistically better than the non-exposed group in restoring serum creatinine (Cr) and creatine kinase (CK) to normal values.

Other Efficacy Outcomes
Four studies (one RCT (Hu F. et al., 2021) and three observational studies ) enrolling 330 severe/critical patients reported PaO 2 /FiO 2 (Supplementary Table S14). CS (Hu F. et al., 2021) and XBJ  could improve PaO 2 / FiO 2 compared to usual supportive treatment group. In an observational study (Xiong et al., 2022) of 111 severe patients, the time for relieving blood oxygen saturation in HSBD was better than that that in usual supportive treatment, with a statistically significant difference (Figure 5 B).
A study enrolling 55 severe patients reported coagulation index: D-dimer, and the results showed that there was no statistical difference between TCM and usual supportive treatment groups (Supplementary Table S13).
One RCT (Wang et al., 2021b) enrolling 128 severe patients showed that qigong exercise and an acupressure rehabilitation program during the treatment period could significantly shorten length of hospital stay and improve lung function and symptoms.

Safety of TCM Interventions
Eight studies (four RCTs (Hu F. et al., 2021;Liu, 2021;Luo et al., 2021;Zhou et al., 2021) and four observational studies (Wang et al., 2021a;Wang et al., 2021b;Zhang L. et al., 2021)) including 9,482 patients reported 76 types of adverse events ( Figure 6, see Additional file 1: Supplementary Table S15 for more details). Regarding the gastrointestinal bleeding and prolonged coagulation time, there exist statistical differences between the HSBD group (Liu, 2021) and the usual supportive treatment group. Compared with the usual supportive treatment, the XBJ group  showed statistical difference in reducing the incidence of acute respiratory distress syndrome and septic shock. The results of the SH group  showed significant difference in reducing the incidence of hypoalbuminemia, increased blood glucose, thrombocytopenia, increased total bilirubin, increased white cell count, increased blood urea nitrogen, increased neutrophil, increased aspartate aminotransferase, diarrhea, increased serum potassium, respiratory failure or acute respiratory distress syndrome, cardiopulmonary failure, cardiac arrest, thrombocytopenia, and increased D-dimer and multiple organ dysfunction syndrome when compared with the usual supportive treatment group. Commonly, patients in the TCM group had a lower incidence rate among the aforementioned adverse reactions. Adverse reactions reported in two or more studies included rash, constipation, diarrhea, septic shock, increased aspartate aminotransferase, thrombocytopenia, hypoalbuminemia, increased white cell count, increased neutrophil, and increased D-dimer. Except for rash and constipation, the overall incidence of adverse reactions in the TCM group was lower than that in the usual supportive treatment group.

Main Findings
During the COVID-19 pandemic, China completely utilized the unique advantages of TCM in treating pandemics and combined it with western medicine to make great contributions to the control of the pandemic. This review provided a comprehensive overview of the evidence for TCM treatments in severe/critical   We failed to perform subgroup analyses because of insufficient number of studies for each outcome. However, a cohort study showed that more benefit in reducing the risk of mortality was found in patients without prior medical history/comorbidities status. Another cohort study showed that the risk of mortality increased by 5.0% with every one year of age increase. In addition, although some benefits were from studies with high risk of bias, the potential advantages of TCM treatments have still been welldocumented.

Potential Mechanism
Studies on the mechanism of TCM have gradually emerged in recent years and strived to explain the mechanism clearly in scientific methods. TCM treatment of COVID-19 has a wide range of effects and multiple targets, which can regulate the internal environment, enhance immunity, control systemic inflammatory response, improve patient symptoms, prevent acute respiratory distress syndrome (ARDS), and reduce risk of mortality. Control the local and systemic inflammatory response and a potentially life-threatening inflammatory cytokine storm presumably could reduce the severity and mortality rate of COVID-19 (Dhama et al., 2020;Mehta et al., 2020). Severe/critical patients were prone to decrease LYM and increase CRP levels (Hillas et al., 2010;Chan et al., 2020;Xu Z. et al., 2020), which aggravated the condition and endangered patients' life. Evidence from our review showed that TCM could restore LYM and CRP to a general level and reduce the rate of critical cases. Zhao et al. (2021) investigated the mechanisms of QFPD against that of COVID-19 from the levels of molecule, pathway, and network; after comprehensive network and pathway analysis, the study indicated that four compounds (baicalin, glycyrrhizic acid, hesperidin, and hyperoside) and seven targets (AKT1, TNF-α, IL-6, prostaglandin-endoperoxide synthase (PTGS) 2, heme oxygenase 1 (HMOX1), IL-10, and TP53) were key molecules related to QFPD's effects, and 55 important targets, which regulated five functional modules corresponding to QFPD's effects in immune regulation, anti-infection, anti-inflammation, and multi-organ protection, respectively, were identified. This was important for promoting body temperature recovery and improving lung imaging, preventing further deterioration and reducing mortality in COVID-19 patients (Zhao et al., 2020a;Wu et al., 2020;Xu T. F. et al., 2020;Yan et al., 2020;Yang et al., 2020). XBJ had advantages in reducing inflammatory response and ARDS based on five active ingredients (hydroxysafflor yellow A, paeoniflorin oxide, Ligusticum striatum DC., lactone I, and paeoniflorin) (Sun et al., 2010;Ma et al., 2020). The molecular docking results of XBJ showed that the following effective ingredients and target binding activities were the best: ethyl ferulate-glyceraldehyde 3-phosphate dehydrogenase (GAPDH), protocatechuic acid-albumin (ALB) Rutin-TNF, apigenin-epidermal growth factor receptor (EGFR), ethy ferulate-mitogen-activated protein kinase (MAPK) 1, benzoylpaeoniflorin-caspase-3 (CASP-3), cryptotanshinone-signal transducer and activator of transcription (STAT)3, rosmarinic acid-MAPK8, cryptotanshinone-PTGS 2, and salvianolic acid B-transcription factor AP-1 (JUN). These active ingredients could synergistically produce anti-inflammatory and immunomodulatory effects (Zheng et al., 2020). The main active ingredients in CS included bufalin, resibufogenin, and cinobufagin, which possessed strong antiviral effects (Qi et al., 2018;Zhan et al., 2020). The result of molecular docking indicated that bufalin had the highest binding efficiency to 3CL protease, transcription factor AP-1 (ACE2), RNAdependent RNA polymerase (RdRp), and spike protein and potentially suggested that CS had a good therapeutic effect on COVID-19 . The top eight compounds in HSBD were quercetin, baicalein, kaempferol, beta-sitosterol, stigmasterol, isorhamnetin, naringenin, and formononetin. These compounds had a strong affinity for ACE2 protein and SARS-CoV-2 3CL protein and could modulate signaling pathways to be effective in the treatment of COVID-19 (Tao et al., 2020). Major chemical components of XYP were andrographolide sulfate A, andrographolide sulfate B, and andrographolide sulfate C (Zhan et al., 2012), and a pharmacological study demonstrated that XYP had antiviral, antipyretic, and anti-inflammatory effects . The network pharmacology results of RDN showed that active components such as quercetin and trans-caffeic acid could regulate cellular inflammatory factors through pathways (Jia et al., 2021). Top 10 chemical components of YDZF were rutin, sennoside A, hyperoside, 4-hydroxycinnamic acid, sinapic acid, rhein, wogonin, atractylenolide III, emodin, and aloe emodin, and results of a network pharmacological study indicated that it could exert anti-inflammatory and immune regulation through TNF, PI3K-Akt, hypoxiainducible factor (HIF)-1, and the toll-like receptor signaling pathway (Ma, 2021). The main chemical components of SM included beta-sitosterol, stigmasterol, and ginsenoside rh2, which could dock with core targets, such as caspase, estrogen receptor 1, catalase, and nuclear factor-κB inhibitor α, and reduce the level of inflammatory cytokines . Key targets of HJSS, such as IL-6, TNF, and catalase relieved the acute lung injury by deregulating the sphingolipid signaling pathway by managing HIF-1 and NOD-like receptors .
TCM could shorten the duration of SARS-CoV-2 RNA persistence time, inhibit SARS-CoV-2 replication in Vero E6 cells, reduce the mRNA level (Lyu et al., 2021) and negative rate of nucleic acid. The components of TCM played a role in relieving the symptoms of COVID-19. For example, Semen Armeniacae Amarum (xinren) in QFPD could relieve cough and asthma, and Gypsum Fibrosum (Shengshigao) and Glycyrrhizae Radix (Gancao) contained in HSBD and QFPD could reduce fever, relieve cough, and regulate the immune system.

Strengths and Limitations
Our study had several strengths. First, to the best of our knowledge, this study is an up-to-date systematic review based on all available evidence to evaluate the efficacy and safety of TCM for severe/critical patients. Second, we strictly followed the back-to-back principle for literature screening, data extraction, and bias risk assessment, which ensured the quality of the study in the methodological aspect. Third, considering the insufficient number of RCT evidence, we also included observational studies to supplement the body of evidence and improve the persuasiveness of our review. Finally, the implementation and reporting of the research followed internationally recognized standards to ensure quality and improve research readability. In addition, we focused on patient-important outcomes, which came from the integration of the "Therapeutics and COVID-19: living guideline" issued by the WHO and the "Core Outcome Set for Clinical Trials on Coronavirus Disease 2019." However, there were some limitations in this review. First, although we conducted a systematical search, continuously update, screened studies from more than 200,000 records, and included studies from both RCTs and observational studies, only 21 studies were included finally. Second, considering the limited number of included studies and different TCM treatments included, we only performed qualitative analysis without metaanalysis. However, in order to summarize the effect direction and enhance the reliability of the review, we drew clustered bar charts and effect direction plot to visual present the results. Third, although we performed sensitivity analysis by restricting studies to be low risk of bias and relatively robust results were found, our findings might be influenced by small sample size and high risk of bias studies. In addition, current pharmacological studies focusing on the mechanism of TCM failed to explore the mechanism from the perspective of chemical components and components' percentage, which hindered our further discussion of the potential mechanism of TCM in the treatment of COVID-19. Future research is necessary to address this issue.

Implications
The clinical understanding and treatment of diseases in TCM was not only differential diagnosis of disease but also differentiation of symptoms and signs. However, TCM treatment focuses on the difference of "syndrome," and could further understand the disease through syndrome differentiation. Evidence-based evidence had shown that a variety of TCM possessed advantages in the treatment of COVID-19, but each TCM had its own characteristics. According to the existing guidelines for the COVID-19 (CAIM, 2020;Wang and Huang, 2020;NHC, 2022), patients with dyspnea require mechanical ventilation. For patients accompanied by unconsciousness, irritability, cold sweat, and cold limbs, it was inner blocking causing collapse. Based on this syndrome type, XYP and RDN should be used in case of fever, cough and uncomfortable expectoration. If there were palpitations, shortness of breath, chills, perspiration due to deficiency of both qi and yin, and pulse deficiency, given SM. HSBD should be used if there exist the following symptoms: fever, cough, yellow and sticky sputum, or blood in sputum, fatigue, dry mouth, bitter and sticky, nausea, loss of appetite, and poor stool. XYP, XBJ, and RND were effective for syndrome of flaring heat in qifen and yingfen, that is, when the following symptoms occur: upset and thirsty, shortness of breath, delirium, dizziness, blurring of vision, or rash, or hematemesis, bleeding, or limb convulsions. In addition to TCM recommended in the published guidelines, the characteristics of patients included in the studies have shown that GZ 3 also had certain advantages for patients with inner blocking causing collapse . CHJD was suitable for patients with heat toxicity in the lungs and intestines. FZJF could be used in patients diagnosed with deficiency of vital energy (Wang et al., 2021a). The main symptoms of patients treated with YDZF and MHLJ were fever and dry cough (Zhao et al., 2020b;Chen G. et al., 2020). Similarly, more than 95% patients using HJSS had coughs .
At a time when the COVID-19 pandemic prevails around the world, China has posted excellent performance in pandemic prevention and COVID-19 treatment, maintained a low infection rate and a high cure rate of COVID-19 (NATCM 2022a; NATCM 2022b) and entered the "dynamic zero-case" stage of precise prevention and control. Among various prevention and control measures, TCM, as a treasure of traditional Chinese culture, once again safeguarded the health of the public and made significant contribution to the battle against the pandemic. In this context, the unique advantages of TCM should be utilized and should be ready to take the center stage globally and ultimately benefit the people all around the world.
According to "WHO Expert Meeting on Evaluation of Traditional Chinese Medicine in the Treatment of COVID-19 (28 February-2 March 2022)" (WHO, 2022a), there was insufficient evidence (one RCT) for the treatment of severe/ critical patients with TCM, but the safety of TCM was comparable to that of conventional western medicine. We included six RCTs and 15 observational studies in this review, TCM presented substantial advantages (e.g., reduction in the risk of mortality and rate of conversion to critical cases, improvement of inflammatory) in treating severe/critical COVID-19 patients. At present, TCM has been applied in 196 countries and regions around the world, and 86 countries have signed cooperation agreements on TCM with China. A total of 17 TCM centers have been established abroad, a few TCM standards have been formulated, and several TCM cooperation bases have been established (Wang and Wang, 2016;Wang, 2018). Among the "Three Medicines and Three Prescriptions" recommended in the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia issued by the National Health Commission (China), HSBD have received emergency use approval in the United Arab Emirates and Cambodia; XFBD has been approved for sale by the Natural and Non-prescription Health Products Directorate (Canada); In Pakistan, JHQG has completed a drug registration-oriented clinical trial, proving its clear efficacy in the treatment of COVID-19. From 2012 to 2020, LHQW has been approved for sale in Canada, Mozambique, Singapore, Thailand, Ecuador, Laos, Brazil, and other countries. In the aforementioned context, strengthening international exchanges and carrying out close international cooperation is conducive to conducting highquality and large-scale research on TCM and promoting TCM to the world. However, the application of TCM overseas must also consider the wishes of the local people comprehensively and be carried out in accordance with laws and regulations.
In the clinical practice perspective, the use of TCM should be as early as possible. In the existing clinical studies, TCM has shown that the earlier the interventional treatment was used, the better the prognosis of the patient would be (Liang et al., 2020;Yong et al., 2020;Zhang J. et al., 2021). In this regard, clinicians should fully consider the characteristics of the patient's disease, and use TCM in a reasonable and timely manner to maximize the therapeutic effect.

CONCLUSION
Compared to usual supportive treatment in severe/critical patients, TCM could significantly reduce rate of conversion of critical cases and mechanical ventilation, shorten time of nucleic acid conversion and length of hospital stay, improve PaO 2 /FiO 2 and laboratory indicators, accelerate symptom recovery, and ultimately reduce mortality. TCM has a safety profile that is comparable to that of usual supportive treatment alone. The roles of TCM against COVID-19 have been welldocumented, which sets an example of using traditional medicine in preventing and treating COVID-19 worldwide. However, the present conclusions might be influenced due to small sample size and high risk of bias in the randomization process generated and unadjusted confounders. More high-quality international multicenter researches and mechanism studies are still needed to further corroborate the effectiveness and safety of TCM in the treatment of COVID-19.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
LG, XS, JL, and LH designed the study. The search strategy has been developed by ML and YfL. ML, HZ, YfL, YoL, and MS screened the references resulted from the literature search and extracted the required data. ML and HZ carried out data synthesis and manuscript writing. LG, XS, and YZ were responsible for the revision and proofreading of the manuscript. All authors reviewed and critically revised the protocol and the manuscript.

FUNDING
This work was supported by the Fundamental Research Funds for the Central Public Welfare Research Institutes (No. ZZ15-WT-05). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.