The protocol has been registered. It has been registered in the PROSPERO(CRD42020176572) and will follow the PRISMA and Meta-analysis of observational studies in epidemiology to develop detailed search strategies and data analysis method[25, 26]. Our systematic research protocol will comply to the PRISMA-P checklist[27].
Criteria for studies
Study population and study design
Population: All patients diagnosed with COVID-19 with no limitation about age, race, sex differentiation.
Intervention: Any moxibustion interventions issued by the therapeutic regimen will be included, such as indirect moxibustion or direct moxibustion.
Outcomes: Only in those studies addressing definitions of the COVID-19-related will be included in this systematic review.
Study design: We will only include randomized controlled trials (RCTs) that associated with moxibustion. Moxibustion as a part of a complex (mixed but not add-on) intervention will not be included in our research. Observational studies, such as cohort, case–control, case-crossover and self-controlled cohort studies will be excluded. Placebo, usual or standard care, wait-list controls and other positive interventions will be taken as controls.
Definitions and outcome measures
Specifically, studies addressing time of disappearance clinical manifestation symptoms of COVID-19 are primary outcomes:
- Fever
- Fatigue
- Shortness of breath
- Dry cough disappearance rate
- Temperature recovery time
- Serum cytokine levels
Studies addressing test result of COVID-19 are secondary outcomes:
- Negative results rate for two consecutive times
- Addressing clinical accompanying symptoms disappear rate
- Radiographic improvement on chest computed tomographic imaging
- Average hospitalization time
- Occurrence rate of common type to severe form
- clinical cure rate and mortality
Search methods for identification of studies
The literature search was performed between May and July 2021. Two authors (JG and XJZ) conducted a systematic search of 5 English bibliographic databases including Medline, Ovid, Embase, the Cochrane Library, the Allied and Complementary Medicine Database (AMED), 3 Chinese bibliographic databases including Chinese National Knowledge Infrastructure (CNKI), Chinese Biomedical Literature Database (CBM), VIP Database and Wan fang Database. In addition, the unpublished data were quoted for ongoing trials such as Netherlands National Trial Register (NTR), the Chinese Clinical Trial Registry (Chi CTR), and ClinicalTrials.gov. With no restriction about language, manual search will be conducted for potential eligible articles as supplements. Included references are traced to expand the database. Firstly, the search strategy will be conducted in Pubmed and then adapt for use in the other databases by using the strategy described on the Additional file 1. The three search subjects ‘COVID-19’, ‘moxibustion’, ‘RCTs’ will make Mesh subject headings and free-text words.
Data collection and analysis
Study selection
Records and data from different sources will be managed using a data collection form with excel file. All study records of the title and abstract identified by the search strategies will be directly imported and assessed based on the eligibility criteria. Two independent methodological trained reviewers, (JG and XJZ) working in pairs, will screen against eligibility criteria first at the titles and abstracts level, then full text level. Any disagreement shall be resolved by consensus. A flow-chart of article inclusion and exclusion will be drawn in the systematic review to document the study selection, as PRISMA guidelines[25]. The process of the study selection will be illustrated using a flow-chart on the Additional file 2.
Data selection and management
Two independent methodological trained reviewers (JG and XJZ) will design and pilot data extraction form until they achieve convergence and agreement. Data extraction from study characteristics will include: first author and year, study design, country of publication, outcome, sample, intervention, type of measures, risk of bias assessment and findings. The selection of the extraction form will be cross-checked by the two authors. The further disagreements or conflicts between the reviewers will be arbitrated by consulting a third review author (CGX). We will also request missing data from the original study author by email to provide additional detail information in the principle study. If there is no positive response from the corresponding authors (up to three contact attempts in one mouth by email).
Data synthesis and analysis
Relevant characteristics of the data for statistical analysis will be put into an excel file. To calculate the risk ratio, dichotomous data will be used with 95% CIs model. We plan to analyze the continuous outcomes by using standard mean difference (SMD) with 95% CIs or a weighted mean difference (WMD). The same scale or assessment instrument will be performed in the WMD; different assessment tools will be carried out in SMD. We will use frequentist framework of a network meta-analysis to operate the net-meta package in R software, to integrate direct and indirect evidences of included RCTs. We will perform a network plot to illustrate the results for primary outcomes if subsets of studies are sufficiently homogeneous. The assessment of statistical heterogeneity will be evaluated based on a standard χ² test. The random-effects model will be utilized when fixed-effect model is adopted if there is no significant heterogeneity .
Other sensitivity analyses will be clinically and statistically conducted the potential heterogeneity and inconsistency in special subgroup, such as age, gender and disease duration, trial blinding, evidence quality and so forth. We will adopt the meta-regression to quantify the inter-subgroup difference and explore statistical significance.
Quality assessments
The quality assessment for the methodological of eligible RCTs will be carried out applying the instrument: The Cochrane Collaboration’s risk of bias tool and the modified Downs and Black instrument[28].The two tools focus risk bias on 6 aspects bias: selection bias, performance bias, detection bias, attrition bias, selective reporting bias and other bias[29] .According to the tool, we will judge the each domains as‘low risk of bias’,‘high risk of bias’or‘uncertain risk of bias’.
Two independent methodological trained reviewers will access the quality of RCTs via these tools. Any Discrepancies between the pairs of reviewers will be solved by discussion and consulting a third review author (CGX). A funnel plot and Egger’s test will be performed graphically for reporting bias assessment[30, 31].If the publication bias is considered to exist, another method such as the trim-and-fill method will be used to modify the bias.
Two reviewers/authors (GJ and LZY) will independently evaluate the quality of the included trials through assessing the risk of bias using the following tools, when appropriate.
Patient and public involvement
This study will use the existing database. Patients and the public will not participate in the design of this study. This article will evaluate the comparative efficacy and safety of moxibustion for the therapeutic method of COVID-19 patients. The insights provided in this study can be used in clinical practice to improve the prognosis, especially to provide an effective adjuvant treatment.