Protocol and registration
This protocol was developed based on the recommendations from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) statements (Moher et al., 2015; Shamseer et al., 2015). The completed PRISMA-P list box is available as additional file 1. The protocol was registered on the Open Science Framework (OSF) with trial registration number: https://doi.org/10.17605/OSF.IO/MW3H8. Under the circumstance of any revision to this protocol post-dissemination, the reason and the date for the revision will be published in the Open Science Framework register to ensure tracking of any changes.
Inclusion criteria
- The following requirements should be met by studies to be considered in the review:
- Types of studies: population-based cross-sectional studies, cohort studies, and follow-up studies published from 1st May 2013 to 31st May 2020, which reported prevalence, mortality, and any type of stroke incidence in any gender aged 18 years and above in Ghana.
- Types of articles: only published articles in English will be eligible for inclusion. Conference abstracts that present enough information on the sample size, data collection process, and data analysis will be included.
- Population: healthy adults aged 18years and above living in Ghana.
- Outcome measures: the primary outcome is the prevalence or incidence, or mortality of strokes. Secondary outcomes include the proportion of different strokes and the presence of co-morbidities.
- Setting/context: the review will consider only studies done in Ghana.
Exclusion criteria
The following studies will be excluded:
- Population: studies involving persons younger than 18 years old in a defined population such as those having mental disorders.
- Types of articles: letter to the editor, review articles, short communications, case reports, book chapters, expert opinion commentaries, and thesis or dissertations.
- Type of outcome measure: studies whose primary outcome is a self-reported stroke or transient ischemic attack.
- Geographical area: studies conducted outside Ghana.
Literature search
Literature for the review will be sourced via a search of major electronic databases, specifically, MEDLINE (EBSCO), CINAHL (EBSCO), Web of Science, and PsycINFO (EBSCOhost) from 1st May 2013 to 31st May 2020. Additional literature will be searched from grey areas such as reference list of selected studies. The PICO approach (Moher et al., 2015; O'Connor, Anderson, Goodell, & Sargeant, 2014) will be used in the literature search based on these concepts:
- the population: (18 years old adults)
- type of measure: (mortality or incidence or prevalence)
- outcome: (increase in strokes)
- the geographic area: (Ghana)
Strategies for literature search will be based on MeSH terms (Medical Subject Headings) using an appropriate set of keywords to delimit the concepts ‘stroke’. These searches will be combined using the Boolean (AND / OR) logic operator to increase the yield of appropriate studies. The study team will review the proposed search terms with necessary adjustments made before running the search. The search strategy will be developed using MEDLINE (PubMed) and then adapted for other databases (Table 1). Two reviewers, JA and EW, will independently search the electronic databases for relevant studies to be considered for this systematic review. The relevant studies' bibliographies will also be screened for articles of interest. Authors of conference proceedings and abstracts that will be obtained from the database will be contacted where possible for full results of the abstracts. The search records of JA and EW will be compared, and entered into the Endnote X8 citation manager. Disagreements that may arise over the comparison of JA and EW's search results, as would be shown by a Kappa co-efficient of less than 0.6, will be settled in consultation with the third reviewer (KBM). The pool of records will be subjected to the removal of duplicates by JA before screening titles and abstracts.
Screening and selection of studies
This review's screening and selection process will include the titles and abstracts screening and the full-text screening. The titles and abstracts screening will be performed independently by JA and EW, with KBM acting as an arbiter when a consensus cannot be reached between JA and EW. Studies that pass this stage will further be subjected to full-text screening. The set inclusion and exclusion criteria would be employed in the screening stages. The citation list of studies that pass the full-text screening for final inclusion into this comprehensive review will be screened to identify any additional articles that meet the eligibility criteria. Full-text articles that are not available online will be requested from the correspondent authors. A PRISMA diagram indicating details of articles included and excluded at each screening stage of the study selection process will be provided (Figure 1).
Data extraction
The relevant data would be extracted independently by the two reviewers (JA and EW in a data extraction form designed in Microsoft Excel. Any disagreement in the extracted data will be addressed by consensus or dialogue with a third reviewer (KBM). The relevant data would be extracted using a Microsoft Excel sheet. The data to be extracted would include the name of the first author, year of publication, study period, study objectives, the geographic region in Ghana, study design, sample size, study period, significant findings, limitations, and conclusion.
Assessment of the quality of studies
The selected studies' quality will be assessed using a risk assessment tool (Mensah, Oosthuizen, & Bonsu, 2018) developed and validated for quantitative cross-sectional studies. The tool assesses the study validity based on responses to these questions:
- Was the sample likely to be representative of the study population?
- Was a response rate mentioned in the study?
- Was the instrument used reliable?
- Was the instrument used valid?
- Was it a primary data source?
- Was incidence or prevalence or mortality of stroke?
The two reviewers (JA and EW) will independently assess the selected studies' quality and describe them as having a low, moderate, or high risk of bias.
Data analysis and synthesis
The study characteristic will be presented in tables using descriptive statistics. The primary outcomes are incidence or prevalence or mortality of stroke. The correlation between these findings and the demographic, co-morbidities, geographical regions, and socio-economic factors in bivariate or multivariable analysis will be reported.
The prevalence or incidence or mortality data and other relevant factors such as co-morbidities or risk factors will be presented as a narrative synthesis. The prevalence or incidence or mortality will be presented as a proportion, and summary measures of co-morbidities or risk factor-stroke association presented as odds ratios. If statistically suitable, incidence or prevalence, or mortality estimates from studies with the same stroke type will be pooled together to give a single summary assessment through a random-effects meta-analysis (DerSimonian & Laird, 1986).
We will report pooled and crude estimates of prevalence or incidence or mortality of stroke in 5-year periods over 2013 - 2020 to present trends. Furthermore, we will assess the temporal trends in the incidence or prevalence, or mortality of stroke using cumulative meta-analysis (Kaze, Schutte, Erqou, Kengne, & Echouffo-Tcheugui, 2017; Lau, Schmid, & Chalmers, 1995).
We will examine heterogeneity by inspecting forest plots in addition to the chi-squared test on Cochran’s Q (alpha set at 0.1) statistic (Cochran, 1954). The I 2 test will also be used to explore the heterogeneity by setting the cut-points for low, moderate, and high degrees of heterogeneity at 25, 50, and 75%, respectively (Higgins, Thompson, Deeks, & Altman, 2003). A sub-group analysis will be carried out to evaluate the causes of any heterogeneity and assess the pooled incidence or prevalence or mortality of stroke by age group, sex, geographical region, methodology, and year of publication. We will undertake a sensitivity analysis to explore the outcome of excluding articles with a high-risk of bias on the overall incidence or prevalence or mortality and relationships. The reporting of bias will be established using funnel plot asymmetry if more than ten relevant articles are listed, and Egger’s test (Egger, Smith, Schneider, & Minder, 1997). The data analysis will be done using the IBM SPSS statistical package (IBM Corp. Version 20.0 Armonk, NY, USA) and RevMan 5.3 programs.
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) guidelines will be used to assess the selected articles' quality of evidence (Balshem et al., 2011). The evaluation of evidence by the GRADE criteria is based on the risk of bias, directness (generalizability), consistency (heterogeneity), precision (statistical significance of effect measures), and publication bias and rated as high, moderate, low or very low. The risk of bias will be adapted from the Hoy et al. tool developed for incidence or prevalence studies (Hoy et al., 2012).