Parametric and Non Parametric Estimates of Malaria Attributable Fractions among Children in South West Nigeria

摘要

Malaria in children remains a major public health problem particularly in the tropics. Fever is one of the major symptoms of malaria; however, not all fever cases is as a result of malaria. Therefore this study estimated the proportion of fever cases that can be attributed to malaria among children in south west Nigeria. Records of all children (≤ 12yrs) eligible for the study were used for the analysis. Malaria attributable fraction (MAF) was estimated using 3 approaches: Classical method, parametric (logistic regression model) and non parametric regression methods (local linear smoothing). The proportion of children with fever increased during the first 3 years of life and thereafter decreased but was not significant with age. The prevalence of parasitemia decreased with increasing age. The MAF obtained from the classical method was 0.27 (95% CI, 0.1784-0.3520). The parasite density specific attributable fraction estimates from logistic regression analysis were 0.0385 (95% CI, 0.0105- 0.0665) and 0.3215 (95% CI, 0.2738- 0.3691) for ≤999μl and ≥1000μl categories respectively. The nonparametric estimate of the MAF obtained was 0.24 (95% CI, 0.1785- 0.3119). The prevalence of malaria disease and fever during childhood remained high. The results suggest a monotone increasing function and shows that the logistic regression method is a better method in determining malaria attributable fractions.