Nine electronic databases (Campbell Library, Cochrane Library, EMBASE, PubMed, Web of Science, CBM, CNKI, CQVIP, and WanFang Data) were searched for studies published until 23 March 2022. In addition, the WHO International Clinical Trials Registry Platform (ICTRP) Search Portal, ClinicalTrials.gov, gray literature, reference lists of articles, and relevant Chinese conference proceedings were also searched. The search strategy was developed with assistance from a medical information retrieval expert and included the use of free terms in combination with Medical terms (MeSH) (for details, see Additional File Pages 2–8).

RCTs and SRs that evaluated the efficacy and/or safety of TCM against COVID-19 were included in the study. The criteria of SRs were in line with the PRISMA-P protocol, which included articles that specifically stated the methods used to identify studies (i.e., search strategy), strategies for study selection (e.g., eligibility criteria and selection process), and detailed methods of synthesis (Shamseer et al., 2015). The inclusion criteria were as follows: 1) population: suspected, confirmed, or convalescent COVID-19 (treatment and rehabilitation) patient, high-risk population (prevention); 2) intervention: TCM or integrated Chinese and modern medicines 3) comparison: standard care (i.e., standard psychological or routine care without medication); or modern medicine; 4) outcome: no restrictions. The exclusion criteria were as follows: 1) duplicate reports; 2) studies with insufficient information (e.g., Abstracts, Letters, and Comments); 3) studies not published in peer-reviewed journals (e.g., studies only appearing on medRxiv or a similar preprint server).

The relevant literature was independently screened, extracted, and cross-checked by two researchers, and any disagreements were resolved via discussion or consultation with a third researcher. Missing data were retrieved by contacting the relevant authors. Duplicate articles were removed using EndNote X9 and Rayyan software, and the titles and abstracts of the remaining articles were screened. After removing irrelevant studies, the full texts were further analyzed for final inclusion. The following data were extracted: publication year, first author, country, study design, sample size/number of RCTs, population, setting, intervention, control, outcome, and associated p-values.

The risk of bias assessment tool (Risk of Bias, RoB) (Higgins et al., 2011) recommended in the Cochrane Handbook (version 5.1.0) was used to assess the risk of bias of the included RCTs. A total of seven items were included, and each item was classified as “yes” (low risk of bias), “no” (high risk of bias), or “unclear” (unclear risk of bias). A trial was categorized as “low risk” when all the seven items were classified as “yes” and “high risk” when one or more items were classified as “no.” Otherwise, a trial was considered an “unclear risk”. A Measurement Tool to Assess Systematic Review 2 (AMSTAR-2) (Shea et al., 2017) was used to assess the methodological quality of SRs. AMSTAR-2 consists of 16 items evaluated as “yes,” “partial yes,” or “no.” The quality assessment process was performed online, and the overall quality of the study (“very low quality,” “low quality,” “moderate quality,” and “high quality”) was automatically generated after the assessment was completed. Two researchers independently conducted the quality assessment, and any conflicts were resolved by discussion with a third researcher.

This study was based on the methodology of Global Evidence Mapping (Katz et al., 2003; Bragge et al., 2011) and Campbell evidence and gap map (White et al., 2020) with some modifications (Yang et al., 2018; Li et al., 2020; Li et al., 2021a; Li et al., 2021b). All authors, including experts on evidence-based medicine, EM, and TCM, thoroughly discussed and approved the framework of this study. The EPPI software, R software, and Microsoft Excel 2019 were used for data entry and management. According to the WHO Family International Classifications (WHO-FICs) (Schippers et al., 2010), Core Outcome Set for Clinical Trials on Coronavirus Disease 2019 (COS-COVID) (Jin et al., 2020), and the included studies, evidence mapping was conducted using an established coding system. A bubble chart was used to present the key features of the evidence, and a coordinate system of the “intervention-outcome” evidence frame was constructed. Each bubble in the frame represented a study, the colors represented different study populations, and the size of the bubble represented the study sample size/number of RCTs. The outcome measures of the studies were plotted on the horizontal axis and the intervention measures and corresponding p-values on the vertical axis. Descriptive analyses of interventions, outcomes, adverse events, and evidence gaps were conducted with the help of bubble charts.

As shown in Figure 1, 2210 studies were retrieved through the preliminary screening of databases and supplementary sources, of which 794 were duplicates. In addition, 1,254 studies were excluded after screening titles and abstracts. The full texts of the remaining 162 studies were screened, and 36 unrelated studies were excluded (Additional File Pages 9–10). A total of 126 studies, including 76 RCTs and 50 SRs, were finally included.

As shown in Table 1 and Additional File Pages 11–43, 76 RCTs were included, of which 74 were conducted in China. Based on the objectives, the trials were divided into treatment (53, 69.74%), rehabilitation (20, 26.32%), and prevention (3, 3.95%), with 59 intervention categories. The primary intervention was Xuebijing injection (XBJ) (4, 5.26%), and the treatments were provided in a hospital setting (58, 76.32%). In all, 57 common efficacy outcomes and eight study population groups were reported in the relevant trials. The main outcome was the total effective rate (24, 31.58%), and the most prevalent population was that of non-severe COVID-19 patients (28, 36.84%). Adverse events were analyzed in 45 trials (59.21%). Further, 39 trials reported a lack of any serious adverse events or significant differences between the two groups.

As shown in Table 2 and Additional File Pages 44–67, 50 SRs were included in this study, of which 47 (94%) were conducted in China. All SRs were therapeutic with nine intervention categories, and TCM (not specified/NS) was the most prevalent treatment method (29, 58%). The primary setting of the interventions was hospitalization (22, 44%). A total of 27 common efficacy outcomes and five study population groups were reported. The main outcome indicator and study population were chest CT manifestations (36, 72%) and COVID-19 patients (not reported) (34, 68%), respectively. Moreover, 38 studies (76%) analyzed adverse events, and 30 reported a lack of any serious adverse events or significant differences between the two groups.

Among the 76 RCTs, 22 (28.95%) were categorized as high risk of bias, 50 (65.79%) as unclear risk of bias, and 4 (5.26%) as low risk of bias. As shown in Figure 2 and Additional File Pages 68–70, 53 (69.74%) trials described appropriate random sequence generation processes, and 56 (73.68%) performed appropriate allocation concealment methods. Only eight (10.53%) trials were blinded for participants and personnel, and 13 (17.11%) were completed with the outcome assessors blinded to grouping. More than 90% (70/76) of the trials had a low risk of bias in the incomplete outcome data, and two (2.63%) were selective in their data reporting. No other bias could be confirmed in any of the trials.

Among the 50 SRs, 30 (60%) were characterized as very low quality, 14 (28%) as low quality, two (4%) as moderate quality, and only four (8%) as high quality. In addition, items 11, 1, 9, and 16 were well reported; particularly item 11 (using appropriate statistical combination method), which was fully reported in 49 (98%) SRs, was well reported. In addition, items 2, 3, 7, and 10 had significantly lower reporting rates, particularly item 10 (reporting study funding source), which was only reported in six SRs (12%) (Figure 3, see Additional File Pages 71–73 for details).

As already mentioned, the studies were classified based on their objectives into treatment, rehabilitation, and prevention. To compare the characteristics of the different interventions and outcomes, we also mapped the studies based on sample size/number of RCTs, population, intervention, outcome, and corresponding p-values (see Additional File Pages 11–67 for data sources).

A total of 45 RCTs and 38 SRs analyzed the adverse events of TCM interventions against COVID-19. No serious adverse events or significant differences were observed between the two groups in 39 RCTs. Three RCTs reported a significantly higher incidence of diarrhea and nausea in the TCM-intervention groups (p < 0.05), and three showed that the control group had significantly higher rates of gastrointestinal bleeding, rash, insomnia, tremor, and pruritus (p < 0.05). In addition, 30 SRs did not report any serious adverse events or significant differences between the two groups. Three SRs showed that adverse reactions were significantly higher in the control group (p < 0.05), and five SRs reported unclear results for adverse events.

In this evidence-mapping study, RCTs and SRs of TCM-based COVID-19 management published before March 2022 were systematically searched. A total of 126 studies were included, of which 90% were from China. The studies focused on three objectives/health strategies, 64 intervention categories, 59 common efficacy outcomes, and eight population groups. Among the 76 RCTs, only four were categorized as having a low risk of bias. The primary health strategy was therapeutic, and non-severe COVID-19 patients comprised the most common population. In addition, XBJ was the most common intervention. Most studies evaluated the effect of a specific intervention on the total effective rate. Among the 50 SRs, only four were assessed as “high quality.” Furthermore, the objective of all the studies included in the SRs was therapeutic, and non-severe COVID-19 patients were the most common population. TCM (NS) was the most applied intervention, and most studies evaluated its effect on chest CT manifestations.

Overall, most of the included studies focused on treating COVID-19 patients. A total of 53 treatment-related RCTs included 39 intervention categories, of which XBJ was the most common intervention, with four published RCTs. Only one RCT was published for each of the 31 other interventions. The high number of intervention categories may be related to the specific characteristics of TCM formulations (Tang et al., 2008; Fu et al., 2021). Compared with modern medicine, TCM emphasizes on the constitution of the patient more than the disease and considers individual differences. Therefore, TCM prescriptions are formulated based on the environment and physical and mental health of the patient. This considerably increases the difficulty of large-sample clinical research for a specific intervention. Therefore, further investigation into designing and conducting relevant studies according to the theories of TCM is warranted (ManKe et al., 2020). Only a few RCTs recruited critical COVID-19 patients. The sample size of severe COVID-19 patients was small, which could be attributed to the characteristics of the SARS-CoV-2 virus (An et al., 2021; Young et al., 2022). Given the efficacy of TCM against the non-severe COVID-19 population, it is reasonable to expect similar outcomes of TCM among severe/critical COVID-19 patient population.

Although most RCTs and SRs evaluated the efficacy of TCM interventions for a wide range of outcomes, the results were inconsistent for approximately 85% of the efficacy outcomes in SRs and 54% of the outcomes in RCTs. High-quality SRs should be conducted to overcome this limitation. Nevertheless, all therapeutic studies that analyzed “TCM syndrome scores” reported significant improvement with the interventions. Therefore, the abovementioned discrepancy could be because of different standards for evaluating the clinical efficacy between TCM and modern medicine (Qinggang, 2018; Yu et al., 2020; Xin et al., 2021). Further studies are needed to evaluate the efficacy of TCM solely depending on the Core Outcome Set and evaluation standards of modern medicine. In addition, most studies reported that TCM formulations were safe for COVID-19 patients (Jiang et al., 2021; Wang et al., 2022). Some common adverse events were diarrhea, nausea, and other minor reactions, which disappeared on their own and were not more frequent in the TCM intervention group compared with those in the control group (Ang et al., 2020; Liu et al., 2020). TCM has also shown promising results in rehabilitating COVID-19 patients (Huang et al., 2022). Most studies showed that TCM intervention could significantly improve the psychological function, respiratory function, lung function, and immune function of patients who had been diagnosed with COVID-19 or discharged from the hospital (Jingling et al., 2021; Li L. et al., 2021; Chen et al., 2022). In addition, TCM reduced the infection rate of COVID-19 and the incidence of cold-like symptoms in individuals at risk of contracting COVID-19 (Yan B. H. et al., 2020; Yan X. et al., 2020; Dan, 2021). However, only a few studies on COVID-19 prevention using TCM are available, which are not enough to draw a definite conclusion.

The main evidence gaps in this study involved the study quality, study populations, interventions, and outcomes. For instance, more than 28% of the relevant RCTs had a high risk of bias, and more than 80% of the SRs were of low or very low quality. In addition, only four RCTs were conducted in severe COVID-19 patients. No SRs or RCTs were observed including critical COVID-19 cases, and no SRs were found for rehabilitating and preventing COVID-19. The definition of specific interventions in the included studies was unclear, as were the respective efficacies of the different interventions because of a lack of continuous research. More than 50% of the SRs did not specify intervention programs, and 76 RCTs had 59 intervention categories. For studies on prevention and rehabilitation, only one RCT was published for each of the 23 different interventions. There is currently no consensus on the efficacy of some outcomes following TCM, which will require further high-quality RCTs or SRs. Finally, the differences in the evaluation standards for TCM and modern medicine and the Core Outcome Set for TCM, warrant further investigation.

We systematically searched for RCTs and SRs related to TCM interventions for COVID-19 and compared the specific interventions, outcomes, populations, and study quality. Evidence mapping can help identify the knowledge gaps that provide a reference for researchers. Meanwhile, this study provides an evidence matrix of TCM, which is a scientific and comprehensive reference basis for clinical policymakers in COVID-19 management. Nevertheless, this study has some limitations that need to be considered. First, given that RCTs and SRs are highly representative of a particular research topic, other types of studies, such as cohort studies, and case-control studies, were not included. Second, given the research purpose and evidence-mapping methodology, we focused on the analysis and presentation of evidence rather than quantitative statistics. Third, most of the included studies were conducted in China; thus, the generalizability of the outcomes is limited. Fourth, our results were based only on publications published before March 2022, and results need to be regularly updated as new studies emerge.