3.1. Protocol registration
The protocol for the systematic review and meta-analysis(SR-MA) documented in PROSPERO 2020 CRD42020207068, which could available from: http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42020207068. The protocol is draw up following the principles that guide the SR-MA protocol (PRISMA-P) statement in the Cochrane Handbook[22]. If there are any changes on the protocol, we will describe them in detail in the final draft.
3.2. Inclusion criteria of literature
3.2.1. Participants of studies
We will include PFP patients who have been diagnosed by clinicians, on the basis of the diagnostic criteria of PFP established by the American Academy of Neurology (AAN) or the Neurology Society of Chinese Medical Association. There will be no restriction on country, occupation, race, age, and gender.
3.2.2. Interventions of studies
Depending on the purpose of our review, RCTs with moxibustion in the treatment group will be included. The treatment group with moxibustion in spite of the shape, size and type of moxa, length of time. The intervention group can treat with moxibustion or associate other interventions, while comparison group will only include other treatments.
3.2.3. Outcome measures of studies
3.2.3.1 Primary outcomes. Effective rate and the H-B score are regarded as the primary outcomes.
3.2.3.2. Secondary outcomes
(1) Quality of life
(2) Facial symptoms: FDI
(3) Side effects and adverse events, such as dizziness, nausea, scald, cough, vomiting and so on.
3.2.4. Design types of studies
All usable RCTs of moxibustion in the treatment of PFP will be included. The language of RCTs will be limited to Chinese and English.
3.3. Exclusion for study selection
- Patients with another disease which would influence the outcome indicators.
- Duplicate publications.
- Specialist experience or theoretical research.
- Result data is missing or unable to get complete data.
3.4. Literature retrieval and screening
3.4.1. Retrieval strategy
To obtain relevant RCTs comprehensively, we will search CNKI, China Biology Medicine, VIP, WF, PubMed, MEDLINE, Embase, Cochrane Library, and Web of Science systematically from their inception to March 18, 2021. The MeSH and non-MeSH terms will be adjusted according to different database, which including peripheral facial paralysis, facial paralysis, facial nerve disease, Bell palsy, facial nerve paralysis, Ramsay-Hunt syndrome,moxibustion, thunder fire moxibustion, Du moxibustion, suspended moxibustion, needle warming moxibustion, taiyi miraculous moxa roll, randomized, RCT. In addition, grey literature were searched for supplement. The gender, race, age, or country of the participants will not be restricted. The concrete search strategy for PubMed shows in Table 1. The search strategy for Pubmed.
3.4.2. Study selection
The search results will be imported into NoteExpress 3.2.0, excluding duplicate literature by it. Before literature selection, all reviewers will discuss and decide the selection criteria together. Firstly, two reviewers will exclude articles which did not meet the inclusion criteria by reading the title and abstract individually. Then they will read the rest of the studies in full text to make the ultimate decision. If there are disagreements, we will discuss or negotiate with the three reviewer to reach a consensus. The process of document screening is shown in Figure 1.
3.5. Data extraction and analysis
3.5.1. Data management and extraction
Two reviewers will independently design an extraction form to collect general information from including RCTs, including author, country, year of publication, diagnostic criteria, sample size, intervention(s), control(s), treatment cycle(s), outcome measure(s), adverse events and other informations. In cases of any disagreement, we will consult and discuss with the three reviewer. When the RCTs' requisite data are incomplete or missing, we will contact the first or corresponding author via e-mail for obtaining the data.
3.5.2. Assessing the risk of bias in included studies
Two reviewers will independently evaluate the methodological quality of included RCTs which base on the Cochrane Risk of Bias Assessment tool of Cochrane Reviewer’s Handbook 5.0.24[23], as following: randomly generated sequence number, allocation sequence concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting and other sources of bias. The trial will be classified into each aspect “ high, low or unclear risk ”. The rating results will be cross-checked by two reviewers, and all disagreements will consult with the three reviewer.
3.5.3. Synthesis of results
RevMan 5.3 software will be prepare to conduct meta-analysis. If the result of overall effect test is P < 0.5, it means that difference is statistically significant. On the contrary, the difference is not statistically significant between the experimental group and control group. As for the index of effect size, we will use the Weighted Mean Difference (WMD) and 95% confidence intervals(CI) to evaluate the continuous variables, while using rate ratio(RR)and 95% CI for dichotomous variables. The results will be presented in the form of forest maps.
3.5.4. Heterogeneity test
I2 will be applied to assess the heterogeneity for included RCTs. When P>0.1 and I2≤50, it is indicated that the merged studies with a lower heterogeneity, we will use the fixed-effects model. Oppositely, it means that the heterogeneity of the merged studies is higher, we will choose the fixed-effects model. When heterogeneity is apparent, we will perform subgroup analysis or sensitivity analysis to explore the source of heterogeneity and its impact on the research results.
3.5.5. Evaluation of reporting bias
When the number of eligible RCTs≥10, publication bias will be evaluated by funnel plot developed by Egger[24]. If the funnel plot appeared to be symmetric, showing that studies are lower with publication bias; If to be asymmetric, showing that studies are higher with publication bias.
3.5.5. Analysis of sensitivity If heterogeneity is significant in combination of different research data, we will conduct sensitivity analysis in line with sample size, methodological components, quality of RCTs, and study characteristics. Sensitivity analysis will be applied to assess whether the results are stable.
3.5.6. Analysis of subgroup If the data of eligible RCTs are reliable, we will conduct subgroup analysis based on age, course, frequency of treatment, treatment cycle, interventions for the control group and so on. For instance, if included RCTs were divided into 2 subgroups basis of different moxibustion type, and the 2 subgroups put on heterogeneity test which had been proved to be homogeneous. If the data of the 2 subgroups were apparent heterogeneity, it revealed that moxibustion type was the source of heterogeneity. Conversely, it showed that the type has little impact on the study results.
3.5.7. Grading the quality of evidence
Two reviewers will independently assess the quality of eligible RCTs evidence by the software of GRADE profifiler 3.6. The specific quality of the included RCTs will be evaluated in 5 dimensions (inconsistency, limitations, imprecision, indirectness, and publication bias), and ultimately the quality of RCTs outcome measure will be defined as 4 levels, expressed as high, moderate, low, or very low[25].