We developed our methods following the instructions of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines for which the systematic review will be written. This systematic review protocol has also written in accordance with the PRISMA for systematic review protocols (PRISMA-P) statement (19, 20) (See additional file 1). This systematic review has been registered under the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42020154777).
Eligibility criteria
Quantitative studies that have used standardized psychological tools to assess cognitive and behavioural outcomes among children surviving malaria will be included. Studies will be selected according to the following eligibility (inclusion/exclusion) criteria (Table 1; See additional file 2):
Inclusion and exclusion criteria
We will include published randomized controlled trials (RCTs), longitudinal studies, case-control studies, cohort studies, cross-sectional studies, observational studies, controlled before and after or interrupted–time–series studies and case series including >5 individuals. The inclusion of longitudinal studies will allow us to evaluate the potential impact of malaria on cognition and behaviour over time. We will include only human studies of children below 18 years of any gender with malaria infection by laboratory diagnosis (i.e. microscopic diagnosis, molecular diagnosis, antigen detection, and serology). We will not exclude studies by community or hospital setting. However, it is likely that most studies will investigate patients undergoing treatment and therefore be in hospital settings. We will include studies with a comparator or a comparison group that report cognitive and/or behaviour score as determined by a standardized psychological assessment as a primary outcome.
We will only include studies published in English or with an English translation as we do not have access to scientific translation services. The publication will be considered if the study was conducted in a LMICs in Asia, Africa, and Central and South America, as determined by the 2020 World Bank Criteria (21). The World Bank classifies countries into four income groupings (low, lower-middle, upper-middle, and high) basing on their income/ economies. Income is measured using gross national income (GNI) per capita, in U.S. dollars converted from local currency for the previous calendar year (21, 22). For the current criteria and list of eligible countries, see Table 2; additional file 2.
We will exclude studies that 1.) consider malaria infection determined by clinical diagnosis only (patient symptoms and on physical examination); 2.) human studies with participants older than 18 years of age and do not assess cognitive and or behavioural function as an outcome; 3.) animal studies; 4.) studies without a control or comparison group; 5.) studies in high income countries or not within LMICs, as determined by World Bank Criteria. We will exclude non-English studies, reviews, opinion pieces and letters to the editor, commentaries, abstracts and case series including <5 individuals.
We will use the Participant/ population, Intervention, Comparator, Outcomes, Timing, Setting/ study design (PICOTS) typology to describe the important key question and clarify the context relevant to our review as below:
Participants
The target population for inclusion is children (≤18 years) with active or recent malaria (within 12 months) infection. Inclusion of longitudinal studies will allow us to evaluate the potential impact of malaria on cognition and behaviour over time (23).
Intervention/ Exposure/ target condition
We will include any active malaria infection or any recent malaria infection that has occurred within 12 months of the cognitive and behavioural assessments. When possible, we will classify the malaria according to the following definitions:
Asymptomatic malaria is a form of Plasmodium malaria infection that lacks typical clinical symptoms, but has sub-microscopic parasite densities detectable by microscopy, rapid diagnostic test (RDTs) or molecular methods (24). Asymptomatic malaria is commonly a sub-microscopic infection (does not necessarily produce gametocytes chiefly detectable by molecular methods than microscopy).
Uncomplicated malaria is a form of Plasmodium malaria infections that are accompanied by fever and/or other symptoms like nausea, vomiting, muscle aches, abdominal pains, chills and sweat that are indicative of malaria. These infections are, almost without exception, detectable by microscopy or rapid diagnostic test (24).
Severe malaria is almost exclusively caused by Plasmodium falciparum infection characterized by prostration, impaired consciousness, respiratory distress (acidotic breathing), multiple convulsions, circulatory collapse, pulmonary oedema (radiological), abnormal bleeding, jaundice, haemoglobinuria, severe anaemia with quick progression to life- threatening disease (25, 26). However, P. vivax (24) and P. knowlesi (27, 28) can also cause severe disease.
If the specific form of malaria infection is not clear from the publication, it will be labelled as “general malaria infection.” If the study cohort contains individuals with different forms of malaria infection, it will be labelled as “mixed malaria infection.” The purpose of the classification is to enable reporting on how the different forms of malaria affect cognitive and behavioural outcomes and any differences/ variations in outcomes observed.
Comparator
Studies with a comparator or control group used against the cognitive and behavioural assessment as one of the key outcomes.
Outcome
The primary outcome of this systematic review is an evaluation of cognitive and/or behaviour outcomes as determined by a standardized psychological assessment (questionnaire-based scales and or neurocognitive assessments) as seen in Table 3 (See additional file 2).
Timing
For this review, timing will reflect the length of time that separates the malarial illness and the completion of the outcome assessment. Additionally, this outcome may be assessed at a single time point or at multiple follow-ups, both of which will be considered.
Setting and study design
Only studies undertaken in low- and middle- income countries will be eligible for inclusion, due to the high burden and receptivity of malaria infection in these settings. Income classification of countries will be according to the World Bank criteria (21, 22). We will include study designs like Randomized Controlled Trials (RCTs), longitudinal studies, case-control studies, cohort studies, cross-sectional studies, observational studies, case series including >5 individuals.
Language
Studies published in English language and available in full text will be eligible for inclusion. Studies published in any language other than English will not be included. This criteria is due to the limited resources available in performing this review.
Information sources
We will systematically search MEDLINE (via PubMed), Cumulative Index to Nursing and Allied Health – CINAHL (via EBSCO), PsycINFO (via EBSCO), Embase and The Cochrane Central Register of Controlled Trials (CENTRAL). We will also search the reference lists of included studies to identify other studies. We consulted a medical librarian at Indiana University to develop our search strategy. Our final search terms include text words; malaria (asymptomatic, uncomplicated and severe), cognitive, behaviour, child development; MeSH terms for cognition, behaviour, child, malaria with the child filter (birth-18 years) (see Supplement 1). We will also hand-search the bibliographies of relevant studies, studies that have cited those included in our review and review articles, as well as search Google Scholar and relevant websites i.e. WHO, Malaria consortium and OpenGrey (www.opengrey.eu/ - for grey literature) for additional potentially eligible articles for inclusion. Our search will not be restricted by earlier publication date and length of follow-up in order to compare outcomes across time. We will include studies published until the time of the search is performed. Note: Our preliminary scoping of the literature suggested no relevant citations would be retrieved prior to 1920.
Study screening, selection and data extraction
Studies will be included only if they meet the full inclusion criteria above and not met any exclusion criteria. The abstracts and full-text articles retrieved using the search strategy will be imported into Endnote software and duplicates will be removed. Using a two tier approach, two of the authors will independently screen studies for inclusion at two levels; first at title/abstract level and then at full-text based on the eligibility criteria. We will then screen full text articles of studies that will have met eligibility at title and abstract screening. The two reviewers will be blinded to each other’s screening results during this screening process. For any discrepancies between the two reviewers at both title/abstract level and then at full-text level, another author will serve as a third rater to create a consensus. The two reviewers will each independently extract data from one half of all the full text articles of the included studies and populate the tables using an Excel screening and data extraction spreadsheet developed a priori to capture study details (e.g., authors, year, country, setting, length of follow-up); sample characteristics (e.g. age, gender, number of participants included), study design (e.g. Randomized Controlled Trials (RCTs), case control, cohort, cross-sectional), descriptions of how malaria, cognition and behaviour are assessed (tests and cut-off values used), and study outcomes (Table 4; See additional file 2). We will then cross-check the other half of the data tables to guarantee accuracy. The Excel spreadsheet and Endnote library will be used to manage records and data throughout the review.
Assessment of risk of bias and grading strength of evidence
To reduce the risk of bias, studies will be assessed independently by two reviewers according to inclusion and the exclusion criteria and in case of discrepancies it will be resolved by consensus by a third reviewer. Technical experts (content experts) will be consulted when needed. We will assess for key aspects like allocation concealment, Random sequence generation and attrition for randomized trials; choice of controls for case–control studies; similarity of baseline characteristics, sample size, control group for observational study designs; sampling strategy, response rates for cross–sectional studies; attrition for cohort studies. We will conduct sensitivity analyses by Risk of Bias, using a Fixed Effects Model in order to explore the robustness of the results of our primary outcomes. To assess reporting bias, if the number of included studies are more than 6, we will use a funnel plot and Kendall’s test to explore publication bias. To evaluate the methodological quality and strength of evidence on the topic, we will use the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach (29, 30). Two raters not involved in the screening process will independently assess the strength of evidence in five areas: study design, quality, consistency, directness and precision. The raters will add or subtract points for each of these categories in line with GRADE guidelines. We will rate the overall quality of evidence for each outcome as “High”, “Moderate”, “Low”, or “Very Low”. By adding or subtracting points, GRADE helps us assess whether further evidence from newly published studies would change the conclusions of the review. The raters will resolve discrepancies through consensus and involvement of a third rater. We will develop a summary of findings tables for each outcome and identify areas for further research.
Data synthesis and statistical analysis
We will narratively synthesize the extracted data from all included studies to provide a narrative account of the data extracted from the included studies. A PRISMA flow chart will be developed to show each level of the review process. First we will use descriptive statistics and where possible Forest plots to summarize included studies and structured synthesis of data. Group-specific sample sizes, means, and SDs will be abstracted from included articles. Based on the fact that results of a single study can be either published in different or multiple publications, caution will be taken to ensure that the unit of analysis is the study rather than the different articles in order to avoid over or under counting studies.
It is imperative to note that the use of a wide variety of often, not well validated assessment tools is linked to the lack of culturally sensitive and appropriate tools to identify and assess cognitive and behavioural outcomes in children in LMICs (31). This may create complexity in compatibility and comparison between studies in our systematic review. However, for 2 or more subsets of studies that are sufficiently homogeneous in terms of sample characteristics, same assessment measures of cognition and behaviour, and methods (e.g., design, setting, length of follow-up), a meta-analysis will be considered (32). The meta-analysis will be conducted using the Cochrane Collaboration Review Manager Software package (RevMan Version 5.3). We will consider heterogeneity (variation across studies) and consider a random-effects meta-analysis that assumes that the underlying effects follow a normal distribution. We will synthesise and transform effect size measures appropriately; for binary outcomes, we will use odds ratios and risk ratios and for continuous outcomes, we will use standard mean differences (Cohen’s d, Hedges’ g). We will explore methodological and statistical heterogeneity between studies with higher percentages signifying higher variation across studies using Cochran’s Q and quantify this using the I-squared statistics. In the absence of statistically significant heterogeneity, we will pool quantitative data using the random effects model (33) and explore the robustness of the results of our primary outcome by conducting sensitivity analyses by Risk of Bias, using a Fixed Effects Model