Design
This is a protocol to conduct a systematic review. The review will strictly adhere to the current guidelines for the Preferred Reporting Items for Systematic reviews and Meta-Analyses protocol (PRISMA-P) for 2020 [10]. Additionally, the protocol has been registered with the prospective registers for systematic reviews with a Registration Number CRD42023480528. This protocol will be conducted in five steps.
Search for literature
Data will be searched using a two-step procedure. Firstly, published studies will be searched from research databases, including PubMed, ScienceDirect, Google Scholar, Web of Science, and African journals online. Secondly, searching for research articles from references of included studies, systematic reviews of a similar topic, and hand searches of unpublished data from institutional libraries and repositories. A primary search string has been developed for the PubMed database (attachment 2) and will be modified for search for the rest of the databases. A broad search will be built using key terms, including all terms and synonyms, Boolean operators, and database-specific thesauri. The key terms include malaria vaccine, readiness, and Africa. The search will not be limited by language and date.
Eligibility criteria
The inclusion and exclusion criteria has been defined based on the Population, Exposure, Comparator and Outcome (PECO) framework guide [11]. For inclusion, studies must meet the following:
-
P: at-risk children (under five years) in Sub-Saharan Africa.
-
E: cross-sectional studies assessing the implementation of a novel malaria vaccine.
-
C: studies with a comparison group, if available, comparing different levels of readiness or acceptance.
-
O: studies on community readiness and acceptance of the novel malaria vaccine.
If any of the included studies have any of the following characteristics, they will be excluded from the analysis:
-
Missing on the outcomes that is readiness or acceptance.
-
Commentaries or letters to editors.
-
Full-texts not accessible for data extraction.
Study selection
After performing the database search, articles will be exported to Rayyan software in the research information system (RIS) standardized tag format for screening. A four-stage selection process including screening titles and abstracts, retrieving full articles, screening full texts, and selecting full texts will be employed [12]. Initially, studies will be included or excluded based on titles and abstracts. All included studies will undergo full-text screening and selection, performed independently by three reviewers who will be blinded. Any disagreements will be resolved through dialogue or, if necessary, by the principal investigator. The entire process will be documented in a flow diagram [10].
Data extraction from included studies
A prewritten data extraction form designed in Microsoft Excel (2013) will be developed and pretested on a few studies before use. Two independent reviewers will extract the data separately. In case disagreements occur, they will be resolved by discussion and, if necessary, consultation with a third reviewer. For articles obtained but with missing data, the corresponding authors for those studies will be contacted to provide the needed data. The data to be collected will include the author, title, year of publication, country, sample size, sampling strategy, demographic factors, level of willingness, level of acceptance, and associated factors, among other relevant data.
Risk of bias assessment
The risk of bias for the included studies will be assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for cross-sectional studies [13]. However, no study will be excluded based on its quality. Studies will be categorized as having high or low quality based on a set of eight questions [14].
Data management and synthesis
Before data analysis, the data in the extraction sheet will be cleaned and organized in a form that can be read by the analytical software. The degree of observed heterogeneity and feasibility to conduct a meaningful quantitative synthesis will be assessed. If heterogeneity is beyond what can be adequately managed through statistical methods and subgroup analysis, a narrative synthesis will be conducted to provide a summary of the level of community readiness or acceptance for the novel malaria vaccine. In case the data are adequate, a meta-analysis will be performed to estimate the pooled prevalence of readiness or acceptance for all included studies. This will be presented on a forest plot with individual studies included. The effect of each study will be shown as a square box with a horizontal line as a 95% confidence interval, whereas the overall effect (pooled prevalence) will be shown as a diamond, with its length symbolizing the confidence interval. To choose whether to use a fixed-effects or random-effects model, the variability between included studies will first be assessed through the I2 statistic. Where high heterogeneity exists (more than 50%), the random-effects model for obtaining the common effect will be used [15]. The index of heterogeneity (I2) will be calculated with uncertainty intervals to indicate its level of precision. I2 is based on the Q statistic, obtained as the sum of squared deviations of each study’s estimate from the overall estimate. When at least two studies report on a given predictor of readiness or acceptance, odds ratio (OR) with a corresponding 95% confidence interval will be assessed. These will be indicated on the forest plot graph. Each study in the plot will be represented by a square whose size is proportional to effect size (OR) and a horizontal line indicating the 95% confidence interval. The diamond will represent the OR and its length will symbolize the corresponding 95% confidence interval. Where the odds ratio crosses the null value (1.0) indicated by the solid vertical line in the plot will indicate no statistical significance.
To preserve the validity of the review, publication bias will be assessed and, if present, controlled. As a rule of thumb, publication bias is assessed graphically through analysis of funnel plot asymmetry when at least 10 studies are included in the meta-analysis [16]. Asymmetry of the funnel plot will indicate potential publication bias. Additionally, a non-significant p value (p > 0.05) of Egger’s test will indicate the absence of publication bias. Additional subgroup analysis and sensitivity analysis will be performed. Subgroup analysis will be performed based on the expected sources of heterogeneity, such as regions, year or publication, and quality of included studies. If over 10 studies have been included in the analysis, leave-one-out sensitivity analysis will be performed to assess the effect of inclusion of each study on the overall estimate [17].
All the statistical analysis will be performed using STATA version 17 (StataCorp LLC, College Station, TX) with the metan package for pooling effect sizes and conducting overall meta-analyses.