2.4. Data collection and analysis
This protocol is extracted from research article:
Electroacupuncture in treatment of Parkinson disease
Medicine (Baltimore), Jan 22, 2021; DOI: 10.1097/MD.0000000000023010

All authors will be trained to reduce human factors in carrying out this review. During search period, EndNote X8.2 will be used for records management. According to the predetermined inclusion and exclusion criteria, 2 authors (Tianyao Qiang and Yuxin Zhang) will screened studies separately by title, keywords, abstracts, and full texts if needed. Each article excluded will be given a reason and recorded in summary. Any discrepancy will be resolved through discussion or consultation a third author (Cong Gai). The selection process will be detailed in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis flow chart shown in Figure Figure11.

Flow diagram of the study selection process.

Two authors (Tianyao Qiang and Yuan Chai) will separately extract the important data, according to a premade collection form. If discrepancies encountered, a consensus will be reached by consulting an expert reviewer (Cong Gai). If the data needed in the records were not given, we will contact the author for more. The following information will be extracted: study details (authors, country, publication time, journal name, title, contact information), participants (inclusion and exclusion criteria, PD diagnostic criteria, age, gender, race, disease duration, baseline data), study methods (registry platform, sample size, blinding method, randomization method, allocation concealment, incomplete report or selecting report), the interventions (type of acupuncture, needles, acupoints, electric frequency, treatment duration, treatment frequency, practitioner, dosage of L-dopa) and the outcomes (primary and secondary outcomes shown above).

The following 7 domains of Cochrane Handbook for Systematic Reviews will be applied to assess the risk of bias by 2 authors (Yuxin Zhang and Yuan Chai) independently: randomized sequence generation, allocation concealment, blinding of participants, personnel and outcome assessors, incomplete outcome data addressed, selective reporting, and other issue. On this account, the incorporated literature will be divided into 3 categories: low risk, high risk, and unclear. In the event of a divergence, discussion will be conducted. If no agreement can be reached a third party (Wandi Feng) is consulted for consensus.

Data analysis will be carried out using RevMan 5.3 software. Dichotomous data will be presented as relative risks, continuous variables measuring with the same scale will be presented as with mean differences or the standardized mean differences. Effect sizes will be indicated by 95% confidence intervals.

If multiple interventions are included, we will combine EA groups as the experimental group; if multiple groups using L-dopa plus EA, we will combined them into 1 group as control.

If the data information of the included trials are found to be incomplete, 2 authors (Tianyao Qiang and Wandi Feng) will contact the first authors and the corresponding authors by email. The unobtainable data will be excluded and only the complete data will be analyzed. In addition, the reason for the missing of the data and the impact for the analysis will be discussed.

Heterogeneity among studies is tested via chi-square test and I2 statistic tests. P < .05 and I2 > 50% indicated the existence of heterogeneity. Fixed-effect models are applied if there was no significant heterogeneity across studies (I2 < 50%); otherwise, random-effects models were applied.

If more than 10 articles are included, the funnel plot of Revman 5.3 software is used to analyze publication bias. The symmetrical distribution on both sides of funnel plot data indicates that there is no publication bias, while the asymmetry on both sides may lead to publication bias. If the number of articles is less than 10, the Egger test and Begg test of Stata 14.0 software are used for statistical analysis. P < .05 indicates the existence of publication bias, while P > .05 indicates the absence of publication bias.

Subgroup analysis is conducted based on intervention time, gender, age and quality score of included literature, and efficiency is taken as the basis.

Sensitivity analysis is performed when there are important positive results or critical results in primary analysis. After excluding abnormal results (too large or too small samples), meta-analysis was conducted to consider whether there was any change. If there is no significant difference between the results before and after sensitivity analysis, the results are proved to be stable.

According to the GRADE, the evidence quality is divided into 4 grades: high, medium, low, and extremely low.

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.