A Scoping Review of Selected Studies on Predictor Variables Associated with the Malaria Status among Children under Five Years in Sub-Saharan Africa

Background/Purpose: In recent times, Sub-Saharan Africa (SSA) had been rated by the World Health Organization (WHO) as the most malaria-endemic region in the world. Evidence synthesis of the factors associated with malaria among children aged under-five in SSA is urgently needed. This would help to inform decisions that policymakers and executors in the region need to make for the effective distribution of scarce palliative resources to curb the spread of the illness. This scoping review is aimed at identifying studies that have used multivariate classical regression analysis to determine the predictors associated with malaria among children under five years old in SSA. Methods/Design: The search terms followed population, intervention, comparator, outcome, timing, setting (PICOTS), and were used in searching through the following databases: PubMed, MEDLINE, Web of Science, Cumulative Index of Nursing and Allied Health Literature (CINAHL), Scopus, and Measure DHS. The databases were searched for published articles from January 1990 to December 2020. Results: Among the 1154 studies identified, only thirteen (13) studies met the study’s inclusion criteria. Narrative syntheses were performed on the selected papers to synchronize the various predictors identified. Factors ranging from child-related (age, birth order and use of a bed net), parental/household-related (maternal age and education status, household wealth index) and community-related variables (community wealth status, free bed net distribution) were some of the identified significant predictors. Conclusions: It is timely to have a synthesis of predictors that influence the malaria status of children under-five in SSA. The outcome of the review will increase the knowledge of the epidemiology of morbidity that will form the basis for designing efficient and cost-effective distribution of palliatives and control of malaria in SSA.


Introduction
Malaria has remained one of the major global public health challenges of the last two decades, especially in low and medium-income countries, putting almost half of the world population at risk of infection [1]. In 2015, over 214 million estimated cases of malaria were reported, with over 450,000 deaths worldwide [2,3]. Surprisingly, in 2018, the number of malaria cases had risen to over 300 million worldwide [4], and in 2019 the estimated malaria cases from 87 endemic nations were approximately 230 million [3]. It is on record that the world experienced a steady decline in the estimated number of deaths from malaria cases from over 730,000 in 2000 to over 400,000 in 2019 [3,5]. In 2017, there were an estimated 430,000 deaths; Sub-Saharan Africa (SSA) contributed over 90% of the global malaria deaths, with over 260,000 being children under five years of age, control in Nigeria and SSA, much more critical is the scoping of these studies. This has not been done, especially concerning determinants of the prevalence of malaria among children under five years in SSA. Therefore, this study aims to bridge this knowledge gap.
The Purpose of the Scoping Review This scoping review aimed to find and evaluate the studies that describe the association between the socioeconomic, demographic, and contextual factors and the prevalence of malaria fever among children under five years of age in Sub-Saharan Africa.

Criteria for Inclusion and Exclusion of Studies
The scoping review followed the preferred reporting items for systematic reviews and meta-analyses extension for scoping reviews (PRISMA-ScR) checklists [21,22]. The review question was in line with the population, interventions, comparators, outcomes, timing and study design (PICOTS).
The population of the study was any child under five years in SSA countries, irrespective of their gender.
The exposures include socioeconomic, demographic, and contextual predictors. These interventions were either classified as child-related, parental/household-related or communityrelated. Where the study reported both adjusted and crude effect sizes, the adjusted was selected.
The comparator was between the children under five years in SSA that had malaria infection versus those that did not have malaria infection.
The outcome variable was the malaria status. This review considered studies on socioeconomic, demographic, and environmental determinants of malaria fever among children under five of both sexes that used standard testing procedures in identifying malaria fever status. The usual method of testing for the presence of Plasmodium parasites was by measuring the axillary temperature of 37.5 • C [23], and, carry out a microscopic examination of thick and thin blood smears that were positive with several asexual parasites per 200 white blood cells, while if white blood cells had a count of 8000 cells/µL [24]. In addition, studies, which identified malaria status through the rapid diagnostic test (RDT) and polymerase chain reaction (PCR) were included.
Studies and articles written in English and published between 1 January 1990 and 31 December 2020 were included.
The study design covers all observational studies (cross-sectional and cohort studies).

Search Strategy
The search strategy was first carried out in PubMed with the following terms as displayed in Table 1 and combined with appropriate Boolean connectors.

Sources of Information
The online databases for literature search using the search terms of PICOTS produced the following results as displayed in Figure 1. The results identified a total of 1157 publica-

Sources of Information
The online databases for literature search using the search terms of PICOTS produced the following results as displayed in Figure 1. The results identified a total of 1157 publications. PubMed (867), Cumulative Index to Nursing and Allied Health Literature (CI-NAHL) (14), Scopus (0) MEDLINE (WOS) (122), Measure DHS (154). The databases were searched for published works from January 1990 to December 2020. The searches were done on 13 and 14 December 2020.

Study Selection
The selection processes followed the PRISMA-ScR recommendation [25]: identify all potential papers from databases and other sources; screen to remove duplicate publications; exclude those that did not meet up with inclusion criteria and inclusion for the review.

Data Selection Process
The reviewer PEO extracted relevant data from the selected studies into a Microsoft Excel spreadsheet (Microsoft Corporation, Washington, DC, USA); the design for this extraction was verified by the supervisory team.

Description of Study Records
This review aimed to synthesize evidence from published articles describing the determinants (socioeconomic, demographic and contextual) of the malaria status of children under five years in SSA between January 1990 and December 2020. The flowchart diagram in Figure 1 shows the selection of studies included for review. A total of 1157 records were identified from all the databases consulted. Ninety-three (93) were retained for full-text examination, and only thirteen (13) unique publications (with 18 country-specific studies) met the inclusion criteria and were examined for this review. Table 2 describes the characteristics of the 13 extracted papers considered for this review. The information includes the author's name and date of publication, the title of the paper, study location, survey type, target sample, the prevalence of the outcome, sample size, statistical methods, and software used for computation. The prevalence of malaria in children under five, as reported in the papers reviewed, ranged from 18% in Tanzania to 39% in Uganda, as reported in Njau et al. (2013) [26]. The mean sample size per study was 30,775 participants.   Figure 2 displays charts A-G and describes the various study characteristics of the included papers. Chart A describes that the publication years for the included studies were between 2013 and 2020. In 2016 there were no publications extracted. The 2018 period had the highest number of publications (23%), while 2013 and 2019 had one publication each (representing 7.7%), and the remaining years had 15% each. In terms of the study setting (chart B), the number of studies from multi-countries (33%) was more than every other country-specific study. Nigeria, Malawi, Tanzania and Uganda had two studies each (11%), and the remaining countries had one study each.

Study characteristics
With more than 70% of the number of country-specific studies, the malaria indicator survey (MIS) was the most used survey among the included studies (chart C), followed by multi surveys (16%) and demographic and health surveys (11%). Likewise, chart (D) indicates that multivariate logistic regression methods (46%) were the most used statistical method. Furthermore, with 38%, STATA was the most preferred statistical software used in the selected publications. This was followed by SAS, SPSS and combined software applications. The least popular was JMP (7.6%). Funding sources (chart F) show that multiple sources of funding were the highest at (38%), while no funding was reported in 30% of the studies, and two of the studies did not disclose any funding source. Three diagnostic methods were reported to detect malaria infection status in children under five years. Among the studies included were rapid diagnostic test (RDT), microscopic test (of thin and thick blood smear), and polymerase chain reaction (PCR). Chart (G) shows that RDT and the combination of RDT and MT were the most popular, and one study reported using PCR. , and the remaining years had 15% each. In terms of the study setting (chart B), the number of studies from multi-countries (33%) was more than every other country-specific study. Nigeria, Malawi, Tanzania and Uganda had two studies each (11%), and the remaining countries had one study each.  With more than 70% of the number of country-specific studies, the malaria indicator survey (MIS) was the most used survey among the included studies (chart C), followed by multi surveys (16%) and demographic and health surveys (11%). Likewise, chart (D) indicates that multivariate logistic regression methods (46%) were the most used statistical method. Furthermore, with 38%, STATA was the most preferred statistical software used in the selected publications. This was followed by SAS, SPSS and combined software applications. The least popular was JMP (7.6%). Funding sources (chart F) show that multiple sources of funding were the highest at (38%), while no funding was reported in 30% of the studies, and two of the studies did not disclose any funding source. Three diagnostic methods were reported to detect malaria infection status in children under five years. Among the studies included were rapid diagnostic test (RDT), microscopic test (of thin and thick blood smear), and polymerase chain reaction (PCR). Chart (G) shows that RDT and the combination of RDT and MT were the most popular, and one study reported using PCR.

Data Synthesis Method
The narrative/tabulation syntheses of the outcomes/findings from multiple reports for this review following the narrative guidelines given in Popay et al. [38] were used. The objective of the review was to collate pieces of evidence on the association between the socioeconomic, demographic, and contextual indicators on malaria fever among children under five in SSA from 1990 to 2020. The results were appraised using the narration of the

Data Synthesis Method
The narrative/tabulation syntheses of the outcomes/findings from multiple reports for this review following the narrative guidelines given in Popay et al. [38] were used. The objective of the review was to collate pieces of evidence on the association between the socioeconomic, demographic, and contextual indicators on malaria fever among children under five in SSA from 1990 to 2020. The results were appraised using the narration of the descriptive statistics and odds of the likelihood of the risk factors and the outcome variable (malaria status).

Predictors associated with Malaria Status
The socioeconomic, demographic, and contextual determinants of malaria status among children under five in SSA were grouped into child-related variables, mother or caregiver-related variables, household-related variables, and environmental or Area-related variables and interaction terms. These variables include age of the child, weight, anemia status, birth order status; maternal age and education status, parent's knowledge, attitude and practices of some basic facts about malaria fever; the type of material used to construct the building, distance from a health facility and cluster altitude as factors identified that are associated with malaria status among children under five years in SSA. A factor was considered statistically significant concerning what each paper considered as the p-value cut-off (0.01, or 0.05, or 0.001). In a situation where the factor was classified into different categories or dummies, the factor was labeled as statistically significant if at least one of the categories or dummies compared to the reference category was statistically significant. Table 3 shows the evidence found on child-related variables. This study revealed the role that the age of the child plays in the tendency for the child to be infected with malaria parasites. Eleven (11) of the country-specific studies investigated a child's age being under five years as a predictor of their potential malaria status. Nine of the studies found that the child's age in at least one of the age groups was significantly associated with the prevalence of malaria among under five years in SSA. In most of the studies, it was found that as the child's age increases, the odds of contracting malaria fever also increase [24,33,35,37,39]. However, Semakula et al. [32] in their multi-country study found no statistical significance in Tanzania (OR: 1 [24] found no statistically significant effect. Also, three studies that explored the effect of birth order found statistically significant effects (OR: 1.03 CI: 1.01-1.06, p = 0.011) [27], when the child was second-order compared to the 1st child born: (OR: 1.43 CI: 1.04-1.96, p = 0.03) [28] and [marginal effect: 0.045, p < 0.01) [31].     In addition, maternal knowledge of malaria fever was found to be a statistically significant predictor of under-five malaria cases in SSA. Children whose mothers showed having knowledge of malaria fever were less likely to be infested with malaria parasitemia (β: −0.013, p < 0.01) [31] and (OR: 0.78 CI: 0.62-0.99, p = 0.037) [24]. Table 5 describes the distribution of significant effects of household-related variables on the likelihood of developing malaria infections among children under five years in SSA. The most widely assessed household-related predictors are household socioeconomic status (designated as household wealth), place of residence (whether urban or rural area), Household size, improved water source and improved toilet facilities. All eleven countryspecific studies that investigated household wealth as a predictor found at least one of the categories being a statistically significant predictor of malaria status. The higher the household wealth quintile, the less likely it that the child in the household would contract malaria fever. The thirteen country-specific studies that found a statistically significant effect of the place of residence all reported that it was more harmful to a child under five years in rural SSA than in urban areas in contracting malaria fever. Though Wanzira et al., 2017 [24] and Zgambo et al. 2017 [37] found no statistically significant effect of place of residence, yet they reported a more protective effect for urban children than rural children (OR: 1.74 CI: 0.92-3.29, p = 0.089) [24], (OR: 2.3 CI: 0.9-6.0, p = 0.075) [37].

Household-Related Variables
It is worthy of note that access to mass media, number of rooms in the household and type of wall material were found not to be statistically significant predictors of malaria fever among children under-five in SSA [26,35]. The variations in household ownership of livestock were a statistically significant predictor of malaria status in children under five years in SSA. Semakula et al., 2015 [32] reported consistent findings in their four country-specific studies that a child from a household that owns cattle has a lower odd of contracting malaria parasitemia than a child from a household without livestock (Tanzania

Environmental/Area-related Variables
In consideration of environmental-related predictors, three variables (regional variations, malaria endemicity, and community free bed net distribution) were attractive for investigation among the included studies. Table 6 reports that Njau et al., 2013 [26] found that the predicted marginal effects (ME) of malaria-endemic areas for malaria fever in Angola, Tanzania and Uganda were significantly ME: 0.01 (p < 0.10), ME: 0.095 (p < 0.05) and ME: 0.288 (p < 0.01) points, respectively. Additionally, the same authors [26] reported an insignificant increase in the predicted marginal effects of 25.1% points for free bed net in the community among malaria positive children in Angola, but a significant reduction of 1.5% (p < 0.1) and 8.2% (p < 0.05) in Tanzania and Uganda, respectively.  Also, significant regional variations were reported across the six geopolitical zones of Nigeria. Morakinyo et al., 2018 [30], which found reduced odds of malaria infections among  Morakinyo et al., 2018 [30] report on Nigeria study, Ugwu et al., 2018 [35] found insignificant effects on regional variations in South East, South-South, South West, and North Central when compared with North West, but found significant odd of malaria-positive cases among 6-59 months in North East (OR: 0.3059, p = 0.015) when compared with north West.

Interactions-related Variables
Interaction-related predictors were reported by two papers in four country-specific studies (Table 7). Njau et al., 2013 [26] reported a significant decrease of 4.6% (p < 0.05) points in the predicted marginal effects among malaria-positive children in Angola with respect to interaction terms of free bed net and wealth status, but found an insignificant reduction of 0.9% and 6.4% in Tanzania and Uganda, respectively.

Discussion
This study aimed to conduct a scoping review of the predictors that affect the malaria status of children under five years in SSA. The review found thirteen studies that identified factors associated with malaria fever among children under five in SSA. Though the search strategies covered the period from 1990 to 2020, the distribution of the publication years shows that papers conducted on predictors affecting the occurrence of malaria among children under-five in SSA were carried out in the last decade. All the publications (meeting the inclusion criteria) were from 2013 to 2020, and no publications in 2016 or 2012 and earlier were found. This may relate to the fact that the data set from nationally representative individual and household surveys in SSA, such as from malaria indicator surveys (MIS) designed a stand-alone survey [40] for use in areas where DHS and MCIS have not been used [41] were not often available until around 2012 and beyond [42]. Most of the MIS datasets collected from 2005 to 2012 remained unavailable within the period [42]. Furthermore, this review shows that the countries of the study were more concentrated in Southern and Central Africa, with just two recorded in West Africa. The reasons for this disparity are not clear. However, from a UNICEF report, the reduction in changes in the percentage of under-five mortality resulting from malaria between 2000 and 2017 was more drastic among the countries in Southern and Central Africa [6]. In addition, the regional disparity in the number of studies may be related to the fact that it is much easier for researchers to secure funding for their studies in Southern and Central Africa than in West African countries. There were more studies from the malaria indicator surveys (MIS) data set than from the Demographic and health surveys (DHS) data set. These differences are likely related to the fact that the timing of MIS is usually in the season where malaria infections are high [43,44], and it includes the use of biomarkers on the field and laboratory [44]. These reasons notwithstanding, technical assistance for both DHSs and MISs was provided for by DHS. In recent times, some country's surveys combined both surveys into one. For instance, Nigeria conducted Nigeria's malaria indicator survey (NMIS) in 2010 and 2015, conducted Nigeria demographic and health survey (NDHS) in 2008 and 2013 separately, but NDHS 2018 was a combination of both NDHS and NMIS [45].
From this review, the most vulnerable, in children under five in terms of age, are those between 2 to 5 years [24,30,33,35,37]. This finding agrees with report from another study [46]. The reasons may not be unconnected with the fact that most families, when they have new-born, intra-family attention and use of resources are shifted to the new-born. As revealed by this review, another factor of importance is the significance of the increase in the maternal/caregiver educational status has on protecting the child from having malaria parasitemia [24,27,28,31,33,35,37]. This is also in line with the findings in Mehretie et al. [47]. One of the pathways in which this can affect the malaria status of under-5 is through adequate knowledge of malaria symptoms, prompt response to seek healthcare attention [24,31].

Strengths and Limitations
There are several strengths identified in this scoping review. (i) This is the first scoping review that was carried out on predictors affecting the malaria status of under five years in SSA that used classical statistical regression methods on data from a secondary analysis of nationally representative surveys. (ii) The review was rigorous with intense supervision from the team that cuts across two institutions.
It is acknowledged that this review has some limitations. (i) All the studies considered in this review are the secondary analysis of nationally representative cross-sectional surveys. Causal effects are not established in the studies, and cross-sectional studies, which are carried out for a time point, cannot determine trends [48]. (ii) Very few studies reviewed in this project considered the contextual factors that may be associated with the malaria status of under-fives with appropriate statistical technique; this may have reduced the reliability of the results attained. Studies that can investigate the contextual factors related to malaria fever among children under five in SSA countries are urgently needed. (ii) Only thirteen studies were identified that met the inclusion criteria. The study may not have successfully identified all the papers as the only considered studies were those that were written in the English language [49]. (iii) In view of the scoping review study design applied [50], we also acknowledged that there was no publication bias and quality of study assessment done. (iv) Only studies that applied frequentist statistical methods were included. Therefore, the exclusion of studies that applied Bayesian statistical methods could have resulted in limitations in the findings. (v) The intervention terms were omitted in the search; this may have excluded some potential studies.

Future Work
There are a few areas not covered in the papers included in this review that require future investigations. Considering the limitations stated above, a scoping review that will take care of them should be the basis for future study. Malaria infection in children is comorbid and, as such, may have overlapping associative predictors. Studies that could explore this area are a potential study area for the future.

Conclusions
SSA is one of the high endemic malaria regions in the world, with a high mortality rate resulting from malaria morbidity. There is a more significant commitment on the part of government and partners to ensure that morbidity and mortality resulting from malaria fever in some countries is reduced to zero by the end of 2020 [5]. The target year is now passed, but it does not seem to have been achieved. The knowledge derived from a careful analysis of these many factors contributing to the rising burden could be used to fast track appropriate intervention mix as they become available [51]. For instance, children delivered in health facilities have a reduced risk of malaria infections when compared with those, who are delivered at home, so a child's place of delivery, child's anemia status, vaccination status, access to insecticide-treated bed net, maternal education status, number of births within five years, duration of breastfeeding, improved water source, availability of electricity, a constant residual spray of houses and environment, and community distance to healthcare facilities are areas that need government attention through policymaking and implementations to reduce malaria infection rates among children under five years. Generally, some population settings, especially the children under five, are more at risk than others, where over 70% of mortality from malaria occurs [52], and measures are needed to protect these vulnerable groups [53]. However, vector control (such as insecticide-treated mosquito nets, drug treatments and indoor residual spraying) is one of the main approaches most SSA governments have adopted to prevent and reduce the spread of malaria [52][53][54] caused through mosquito bites [55]. Inadequate knowledge of how the individual and contextual factors are associated with malaria contraction may jeopardize governments' ability to eliminate the malaria parasite [54].
Author Contributions: The conceptualization of this study was done by P.E.O. and K.K.; the formal literature searching, screening and drafting of the manuscript were carried out by P.E.O., while S.J.W., R.J. and K.K. participated in supervising, revising and editing of the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This study is an integral part of P.E.O.'s doctoral study at the School of Health and Related Research of the University of Sheffield, United Kingdom. The funding for the doctoral study was granted by TETFUND (Nigeria).

Institutional Review Board Statement:
This study is an integral part of a larger doctoral study. The ethical approval to carry out the doctoral research study was granted by the School of Health and Related Research (ScHARR) Ethics Committee of the University of Sheffield (Reference Number: 031534).
Informed Consent Statement: Not applicable.