Statistical Analysis of Malaria Incidence Among Rural Households
Notice: This is a sample project for study and reference. Submitting it as your own work violates most universities' academic integrity policies.
Abstract
About This Research Topic
Malaria continues to exact its heaviest toll in rural sub-Saharan Africa, where preventive interventions, housing quality and health literacy intersect to shape household risk. In Nigeria, which accounts for 27% of global cases, rural children under five experience malaria prevalence more than twice that of urban peers. Understanding why some households experience repeated episodes while others remain relatively protected requires household-level statistical analysis that goes beyond facility aggregates.
This study examines malaria incidence among 320 rural households in Lere and Kachia Local Government Areas of Kaduna State between January 2022 and December 2023. It applies descriptive statistics, chi-square association tests, Pearson correlation, and count-data regression — Poisson and negative binomial — plus logistic regression for severe malaria. By integrating household survey data with primary health care records, the analysis quantifies incidence at 2.87 episodes per household per year and isolates modifiable predictors such as insecticide-treated net use, proximity to stagnant water, window screening and presence of children under five.
The work demonstrates how appropriate count-data methods address overdispersion common in epidemiological counts, providing a methodological template for similar endemic settings. For students learning to model disease counts, our guides to Poisson regression and public health data analysis explain when Poisson assumptions fail and why negative binomial often fits better.
Main Abstract
Malaria remains leading cause of morbidity in rural Nigeria, yet household-level statistical analyses remain scarce. This study conducted comprehensive analysis of malaria incidence among rural households in Lere and Kachia LGAs, Kaduna State, using 320 households surveyed January 2022 to December 2023. Cross-sectional design with structured 28-item household questionnaire triangulated with PHC records. Descriptive statistics, chi-square tests, Pearson correlation, Poisson regression, negative binomial regression and logistic regression were applied. Overall incidence was 2.87 episodes per household per year (95% CI: 2.61-3.13). Households using insecticide-treated nets had significantly lower mean incidence than non-users (1.94 vs 3.81 episodes, p < 0.001). Poisson regression identified proximity to stagnant water (IRR=1.84, p<0.001), absence of ITN use (IRR=1.71, p<0.001) and presence of children under five (IRR=1.53, p<0.001) as significant predictors. Negative binomial model fitted overdispersed count data better (AIC=1,842.3 vs 2,104.7 Poisson). Logistic regression identified same factors plus lack of window screens as predictors of severe malaria. Findings support intensified ITN distribution, stagnant water drainage campaigns and community-based surveillance in rural Kaduna.
Chapter One Preview
Background to the Study
Globally, the World Health Organization estimated 249 million malaria cases and 608,000 deaths in 2022, with sub-Saharan Africa bearing 94% of cases and 95% of deaths. Nigeria alone contributes 27% of global caseload and 31% of deaths, with children under five and pregnant women most vulnerable. Rural households face disproportionately higher burden due to limited access to ITNs and indoor residual spraying, proximity to breeding sites, poor housing and low health literacy. Nigeria Demographic and Health Survey 2021 reports 23.4% malaria prevalence among rural under-fives versus 9.8% urban.
Kaduna State in north-western Nigeria presents varied ecology — Kaduna Plateau, Zaria plains, low-lying Kachia and Lere — influencing vector abundance. Despite National Malaria Elimination Programme and donor investments, malaria accounted for 38.7% of outpatient visits in rural PHCs in 2022 per Kaduna State Ministry of Health. Facility-level aggregates mask household heterogeneity essential for targeted intervention.
Count data methods are fundamental in epidemiology for non-negative integer outcomes that are often overdispersed. Poisson regression assumes equidispersion (mean=variance), while negative binomial relaxes this via extra dispersion parameter. Many Nigerian studies still rely on descriptive and chi-square only, without accounting for confounding or overdispersion. This study integrates household survey with PHC records to support robust inferential analysis, contributing to WHO Global Technical Strategy 2016-2030 target of 90% reduction by 2030.
Nigeria Demographic and Health Survey 2018 – Malaria Indicators
Statement of the Problem
Despite decades of control programmes and substantial financial investment, malaria incidence in rural Kaduna remains unacceptably high, leading outpatient and admission statistics. Persistence despite available interventions suggests programmes are not adequately targeted at highest-risk households. Systematic household-level analysis of demographic, environmental and behavioural risk factors is urgently needed to provide evidence base for cost-effective targeting; without it, scarce resources risk low-impact activities while highest-risk populations are neglected. Methodologically, existing Nigerian studies often use aggregated facility data, descriptive only, fail to account for overdispersion or confounding. This study directly addresses both substantive and methodological limitations by applying Poisson, negative binomial and logistic regression at household level in Lere and Kachia.
Aim and Objectives
Aim: To conduct comprehensive statistical analysis of malaria incidence among rural households in Lere and Kachia LGAs, Kaduna State.
1. Describe sociodemographic and environmental characteristics of sampled rural households.
2. Determine malaria incidence rate per household per year and seasonal pattern.
3. Examine association between household-level risk factors (ITN usage, stagnant water proximity, house screening, under-five children, education) and malaria incidence.
4. Apply Poisson and negative binomial regression to identify significant predictors of malaria episode count per household.
5. Assess determinants of severe malaria episodes using binary logistic regression.
6. Make evidence-based recommendations for targeted malaria control interventions in rural Kaduna.
Research Questions
What are the sociodemographic and environmental profiles of rural households in Lere and Kachia LGAs?
What is the household-level malaria incidence rate and how does it vary by season?
Is there statistically significant association between ITN usage and malaria incidence?
What household-level variables significantly predict count of malaria episodes per year?
What factors significantly predict occurrence of severe malaria among rural household members?
Significance of the Study
First household-level analysis for Lere and Kachia fills local evidence gap. Methodological advancement over descriptive-only via count-data regression handling overdispersion. Findings assist Kaduna State Ministry of Health and National Malaria Elimination Programme in targeting ITN distribution, indoor residual spraying and environmental sanitation to highest-risk rural households. Provides replicable analytical template across other endemic states in sub-Saharan Africa. Contributes to evidence base for WHO Global Technical Strategy 2016-2030 90% reduction target and Nigeria's progress toward universal health coverage. Students can adapt similar framework using our tutorials on epidemiological statistics and logistic regression modelling for public health.
Scope of the Study
Geographically restricted to rural communities in Lere and Kachia LGAs, Kaduna State. Data collection January 2022 to December 2023. Unit of analysis is household defined as persons sharing dwelling and eating from same pot. Focus on malaria incidence (number of laboratory-confirmed or clinically diagnosed episodes per household per year), severe malaria occurrence, and demographic, environmental and behavioural predictors. Does not include entomological vector density or climate variables due to data constraints.
Operational Definition of Terms
Malaria Incidence: Number of new malaria episodes experienced by household members over one year, expressed as count per household.
Rural Household: Family unit residing in community classified as rural by National Population Commission, characterized by agrarian livelihoods and limited urban infrastructure.
Insecticide-Treated Net (ITN): Bed net impregnated with insecticide used as barrier against mosquito bites during sleep, primary individual-level prevention tool.
Stagnant Water Proximity: Presence of stagnant water bodies (ponds, ditches, puddles, containers) within 100 metres of dwelling, potential Anopheles breeding sites.
Severe Malaria: Episode requiring hospitalization or resulting in complications including severe anaemia, cerebral malaria, respiratory distress as recorded by facility.
Incidence Rate Ratio (IRR): Multiplicative change in expected episode count associated with one-unit change in predictor, derived from Poisson or negative binomial regression.
Overdispersion: Condition where observed variance substantially exceeds mean, violating Poisson equidispersion and necessitating negative binomial alternative.
Household Head: Person acknowledged as primary decision-maker whose sociodemographic characteristics serve as household-level proxies.
CDC – About Malaria and Prevention
Short Conclusion
Analysis of 320 rural households confirms high burden at 2.87 episodes per household per year, with clear modifiable risk factors. ITN usage halved incidence (1.94 vs 3.81), while proximity to stagnant water increased risk by 84% (IRR 1.84), absence of ITN by 71% and presence of under-fives by 53%. Negative binomial outperformed Poisson (AIC 1,842.3 vs 2,104.7) due to overdispersion, underscoring importance of appropriate count-data methods. Logistic regression extended same predictors to severe malaria plus lack of window screens. Recommendations centre on intensified ITN distribution with hang-up campaigns, community-led stagnant water drainage and larviciding, window screening subsidies, and community-based surveillance linking PHC records to household registers. Future research should incorporate longitudinal design, entomological data and climate variables. For practical field tools, see our guides on household survey design and community health data collection.
Frequently Asked Questions
Q: What was the malaria incidence rate in rural Kaduna households?
A: Overall 2.87 episodes per household per year (95% CI 2.61-3.13) among 320 households surveyed January 2022-December 2023 in Lere and Kachia LGAs.
Q: How effective were insecticide-treated nets (ITNs)?
A: Households using ITNs had mean 1.94 episodes vs 3.81 in non-users, p<0.001. Poisson regression IRR 1.71 for absence of ITN, indicating 71% higher expected count without ITN after adjusting for other factors.
Q: Why use Poisson and negative binomial regression?
A: Malaria episodes are counts (non-negative integers) often overdispersed (variance > mean). Poisson assumes mean=variance. Negative binomial adds dispersion parameter handling overdispersion, providing better fit (lower AIC 1,842.3 vs 2,104.7) and valid standard errors.
Q: What were strongest risk factors?
A: Proximity to stagnant water within 100m IRR 1.84, absence of ITN use IRR 1.71, presence of children under five IRR 1.53, all p<0.001. Absence of window screens also predicted severe malaria in logistic model.
Q: What is severe malaria and how was it analysed?
A: Severe malaria defined as episode requiring hospitalization or complications like severe anaemia, cerebral malaria. Analysed via binary logistic regression to identify predictors of occurrence among household members.
Q: Why focus on household level not facility level?
A: Facility aggregates mask heterogeneity in risk within communities. Household-level captures sociodemographic, environmental and behavioural factors (housing, ITN use, under-five presence) essential for targeting interventions to highest-risk households.
Q: What are limitations of the study?
A: Self-reported episodes may undercount home-treated cases, some clinical diagnoses without lab confirmation, cross-sectional design limits causal inference, self-reported ITN use may have desirability bias, and two LGAs limit statewide generalizability. Triangulation with PHC records mitigated some issues.
Q: How does this contribute to WHO malaria targets?
A: Provides local evidence for targeted ITN distribution, environmental sanitation and surveillance aligning with WHO Global Technical Strategy 2016-2030 target of 90% reduction by 2030 and Nigeria's National Malaria Elimination Programme.
Q: What interventions are recommended?
A: Intensified ITN distribution with hang-up and behavior change campaigns, community-led stagnant water drainage and sanitation, window screening subsidies, and community-based malaria surveillance linking household registers to PHC data for early outbreak detection.
Q: Can this method be replicated elsewhere?
A: Yes. Cross-sectional household survey with 28-item questionnaire, descriptive, chi-square, correlation, Poisson, negative binomial and logistic regression forms replicable template for other endemic LGAs in Nigeria and sub-Saharan Africa, adaptable to include climate and entomological variables.
Purchase to unlock the full material.