Predictors of Complicated Pediatric Malaria Among Children Under Five in the Vihiga Highlands, Western Kenya

Abstract

Malaria remains a leading cause of morbidity and mortality among children under five years in sub-Saharan Africa. Complicated malaria poses a significant threat, necessitating early identification of predictors for timely intervention. This study aimed to identify clinical, hematological, and cytokine profile predictors of complicated malaria among children under five years in Vihiga Highlands, Western Kenya. A cross-sectional study was conducted on 309 children. The study participants were sampled purposively and grouped in the categories. Among the 309 participants analyzed clinical groups were categorized into uncomplicated (n=253) where actually (n=82) were healthy controls and (n= 71) uncomplicated malaria and complicated malaria (n=56). Demographic and clinical data were collected through interviews, medical records, and clinical examinations, while hematological and cytokine profiles were analyzed from blood samples using standard laboratory techniques and ELISA to assess disease severity. Statistical analysis included chi-square tests for categorical variables, independent t-tests for continuous variables, logistic regression modeling (LRM), and random forest modeling (RF) to determine significant predictors (P<0.05). Principal Component Analysis (PCA) was employed to rank predictors, and cross-validation was used to assess model overfitting. Of the 309 children analyzed, 81.9% had uncomplicated malaria, while 18.1% had complicated malaria. Clinical features such as fever (P<0.001), jaundice (P<0.001), generalized pallor (P<0.001), poor feeding (P=0.003), and cough (P<0.001) were significantly associated with complicated malaria. Hematological markers, including hemoglobin (Hb) levels (P<0.05), hematocrit (P<0.05), RBC count (P<0.05), MCV (P<0.05), and platelet count (P<0.05), were also strongly linked to malaria severity. Additionally, elevated cytokine levels of IL-6 (P<0.05), IL-10 (P<0.05), IFN-γ (P<0.05), and MIP-1β (P<0.05) were observed in complicated cases, indicating their role in immune response dysregulation. PCA ranking identified the most influential predictors being RANTES (rank score: 0.263), IL-8 (0.255), hemoglobin (Hb) (0.251), IL-6 (0.251), and IFN-γ (0.249). Logistic regression and random forest models achieved high predictive performance. A correlation heatmap further illustrated significant associations among predictors. The malaria severity risk score (MSRS) was developed as a clinical decision rule to classify pediatric malaria cases based on clinical, hematological, and cytokine predictors. The integration of clinical, hematological, and cytokine predictors into a clinical decision rule provides a practical approach to malaria severity stratification. The proposed MSRS enhances early detection and treatment prioritization. Healthcare providers should integrate hematological and cytokine biomarkers with clinical assessments to enhance early detection and classification of complicated malaria, while predictive models like the MSRS should be optimized for clinical use. Future research should focus on external validation and optimization of predictive modeling to improve accuracy and clinical applicability.