Study registration
This study was funded and registered on PROSPERO (CRD42020165562). It is reported according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocol statement [31–32].
Criteria for including studies
Types of studies
This study will only consider randomized controlled trials (RCTs) assessing the effectiveness and safety of EA for the treatment of UI in patients with SCI for inclusion. We will not limit their language and publication date to all included RCTs.
Types of interventions
The intervention of the trial group only used EA for the treatment of UI in patients with SCI.
The intervention of the control group could use any treatments, such as conventional therapy, medication, and any others. However, we will exclude EA or EA combined with other therapies as comparators.
Types of patients
Regardless of ethnicity, age, sex, educational background, any SCI patients who were diagnosed as UI will be included in this study.
Types ofoutcome measurements
The primary outcome is the change from baseline in the amount of urine leakage, as measured by the pad-weighing test or other tests.
The secondary outcomes are urination diary, bladder capacity, severity of UI, a 72-hour incontinence episode frequency, clinical symptom scores, the number of participants healed completely within study period and adverse events.
Data sources and search
We will comprehensively search the electronic databases of MEDLINE, EMBASE, Cochrane Library, PsycINFO, Web of Science, CBM, and China National Knowledge Infrastructure from origin of each database up to January 31, 2020. All literature resources will be searched regardless language and publication date. The detailed search strategy of MEDLINE is built (table 1). Similar search strategies of other electronic databases will be modified.
In addition, we will search other sources, such as Google scholar, dissertations, conference proceedings, and reference lists of eligible trials.
Data collection and analysis
Study selection
All retrieved literatures will be imported into Endnote X7 and all duplicates will be removed. Two investigators will independently screen the titles and abstracts of all searched literatures, and any unconnected studies will be excluded. Then, full-texts of the remaining studies will be obtained and read cautiously against all inclusion criteria. Any uncertainty between two investigators will be resolved through consulting a third investigator. The procedure of study selection will be exerted in a flow diagram.
Data extraction
Two investigators will independently extract all essential data from each included trial using a predefined data extraction form. Any divergences will be figured out with the help of a third investigator through discussion. We will extract the following information:
Study information: first author, publication year, country, et al.
Patient information: gender, age, race, diagnostic criteria, inclusion and exclusion criteria, et al.
Trial methods: details of randomization, concealment, blind, et al.
Specifics of intervention and controls: treatment duration, dosage, frequency, et al.
Outcome details: primary and secondary outcomes, adverse events, et al.
Missing data dealing with
If there is unclear or insufficient data, we will contact primary authors to request it. If we can not receive those data, we will only analyze available data. If necessary, we will discuss its potential affects on the study findings.
Study quality assessment
Two investigators will independently appraise study quality using Cochrane risk of bias tool through 7 fields. Each item is graded as high, unclear or low risk of bias. Any differences will be solved by a third investigator through discussion and a consensus will be reached after discussion.
Subgroup analysis
We will carry out subgroup analysis to identify the possible factors that may result in significant heterogeneity based on the different interventions, controls and outcome indicators.
Sensitivity analysis
We will perform sensitivity analysis to test the robustness and stability of study findings by excluding low quality trials.
Reporting bias
We will examine the reporting bias using funnel plot and Egger’s regression test if more than 10 included trials are included [33–34].
Data synthesis
RevMan 5.3 software will be used to undertake statistical analysis. To assess the extracted data, mean difference or standardized mean difference and 95% confidence intervals will be used for continuous data. For dichotomous data, we will use risk ratio and 95% confidence intervals. Statistical heterogeneity across eligible trials will be inspected using I² statistics. I² ≤50% means fair heterogeneity, and a fixed-effect model will be examined. If sufficient outcome data is extracted, we will conduct a meta-analysis. I²>50% suggests apparent heterogeneity, and a random-effect model will be used. We will carry out subgroup analysis to investigate the sources of heterogeneity. If the sources of heterogeneity can not be identified, synthetic analysis will not be performed and descriptive analysis will be adopted.
Quality of evidence
The quality of evidence for major outcomes will be appraised by two independent investigators using Grading of Recommendations Assessment, Development and Evaluation [35]. Any disagreements will be disentangled by another investigator through consultation.
Dissemination
We expect to publish this study on a peer-reviewed journal.