2.5. Data collection and analysis
This protocol is extracted from research article:
Efficacy and safety of mechanical thrombectomy for cardioembolic stroke
Medicine (Baltimore), Jan 15, 2021; DOI: 10.1097/MD.0000000000024340

To determine the studies to be searched further, 2 review authors (ZZ and CW) will independently scan the titles and abstracts of all articles identified from electronic databases. Full-text articles will be scanned for all potentially relevant articles. If there is any disagreement on the selection of articles, they will be discussed with the 3rd author (WX). The selection process will be shown in a preferred reporting items for systematic review and meta-analysis (PRISMA) flow chart (Fig. (Fig.11).

Flow diagram of studies identified.

For all studies included, 2 review authors (ZZ and CW) will independently extract the relevant information using a standard data extraction table. Information will include publication of year, author, participants, intervention, control, duration of intervention, outcomes, and methodologic characteristics. Disagreements will be resolved by discussion by arbiter (WX).

To assess the risk of bias for all included studies, Cochrane Collaboration's bias risk tool will be used by 2 independent review authors (ZZ and CW) to assess the following areas: random sequence generation, allocation concealment, blindness to participants, people, and results, incomplete outcome data, optional reporting, and other biases. Any discrepancies in the deviation risk assessment will be resolved through discussion. Ultimately, the quality of the studies will be divided into 3 levels: low risk of bias, high risk of bias, and unclear risk of bias. Disagreements will be resolved by discussion by arbiter (WX).

All analyses will be conducted by using RevMan software (V5.4), we will select fixed effects model or random effects model to merge the outcome indicators in accordance with the results of heterogeneity test. The fixed effects model will be applied for data synthesis of low heterogeneity (I2 < 50%) while the random effects model will be conducted if the heterogeneity is significant (I2 ≥ 50%). It is considered that differences are statistically significant if the results of Z test show that P value is less than .05, and the 95% confidence intervals does not contain 0 (for continuous variables) or the 95% confidence intervals does not contain 1 (for dichotomous variables).

For insufficient or missing data, we will contact the authors by e-mail or phone as much as possible. All analysis will be performed based on intent-to-treat principle.

Heterogeneity will be identified by visual inspection of the forest and tested by standard Chi-squared statistic and a significance level of 0.1. Additionally, the I2 statistic will be used to examine heterogeneity for quantifying inconsistency in the included studies. Fixed or random effects models will be performed in meta-analysis. If I2 >0.5, random effects models will be used.[22]

Funnel plots will be used to assess the potential for small study bias if there are sufficient studies. Asymmetry of funnel plots will suggest possible small study effects and the results will be explained cautiously.[23,24]

Meta-analysis will be performed using RevMan 5.4 software. When there is no statistical heterogeneity among the results, a fixed-effects model will be used for meta-analysis. When there is statistical heterogeneity among the results, the heterogeneity source will be further analyzed and a random-effects model will be used for meta-analysis after excluding the effects of significant clinical heterogeneity. When there is significant clinical heterogeneity, we will use subgroup analysis or sensitivity analysis, or only descriptive analysis.

We will perform the following subgroup analyses if included data are highly heterogeneous: by age, sex, sample size, type of interventional stents, highly heterogeneous.

Sensitivity analysis is an important method primarily used to assess the robustness and reliability of the combined results of meta-analysis. It is a commonly used sensitivity analysis method to combine the effect size after eliminating each of the included studies, or after changing the inclusion or exclusion criteria or eliminating certain types of studies. For possible low-quality studies, sensitivity analysis is required. If the heterogeneity of the included literature is significant, in order to ensure the credibility of the research results, we will conduct sensitivity analysis by excluding each included study separately, so as to improve the research quality.

This systematic review and meta-analysis will not require ethical approval because there are no data used in our study that are linked to individual patient data. In addition, findings will be disseminated through conference presentations and peer-review publications.

In this study, the level of evidence on outcomes will be assessed using an approach based on the grading of recommendations assessment, development and evaluation (GRADE). The quality of the body of evidence will be assessed based on 5 factors, including study limitations, effect consistency, imprecision, indirectness, and publication bias. The assessments will be categorized as high, moderate, low, and very low quality.

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.