Background: We determined the prevalence, incidence, and risk factors for epilepsy in Nigeria.

Methods: We conducted a door-to-door survey to identify cases of epilepsy in three regions. We estimated age-standardized prevalence adjusted for non-response and sensitivity and the one-year retrospective incidence for active epilepsy. To assess potential risk factors, we conducted a case-control study by collecting sociodemographic and risk factor data. We estimated odds ratios (ORs) using logistic regression analysis and corresponding population attributable fractions (PAFs).

Results: We screened 42,427 persons (aged ≥ six years), of whom 254 had confirmed active epilepsy. The pooled prevalence of active epilepsy per 1,000 was 9.8 (95% CI: 8.6–11.1), 17.7 (14.2–20.6) in Gwandu, 4.8 (3.4–6.6) in Afikpo and 3.3 (2.0–5.1) in Ijebu-Jesa. The pooled incidence per 100,000 was 101.3 (95% CI: 57.9–167.6), 201.2 (105.0–358.9) in Gwandu, 27.6 (3.3–128.0) in Afikpo and 23.9 (3.2–157.0) in Ijebu-Jesa. Children's significant risk factors included febrile seizures, meningitis, poor perinatal care, open defecation, measles, and family history in first-degree relatives. In adults, head injury, poor perinatal care, febrile seizures, family history in second-degree relatives, and consanguinity were significant. Gwandu had more significant risk factors. The PAF for the important factors in children was 74.0% (71.0%–76.0%) and 79.0% (75.0%–81.0%) for adults.

Conclusion: This work suggests varied epidemiological numbers, which may be explained by differences in risk factors and population structure in the different regions. These variations should differentially determine and drive prevention and health care responses.

This is a cross-sectional door-to-door and case-control study. Data was collected through a population census. Data entered into Microsoft Excel and processed. Prevalence and incidence calculated using R epitools. The case-control data was analyzed using Stata 15. We used a Chi-squared test to compare categorical variables between cases and controls and between sites. The Wilcoxon-rank sum and the Kruskal-Wallis tests to compare continuous variables between cases and controls and sites. Multivariate logistic regression to calculate odds ratios. Multiple imputations by chained equation (MICE) to deal with missing data.

Attached are the datasets. Included along with the data sets are the corresponding ReadMe files.