ROBUSTNESS OF POISSON-MIXTURE MODELS IN IDENTIFYING RISK FACTORS OF UNDER-FIVE MORTALITY IN NIGERIA

Abstract

Under-five Mortality (U5M) is a public health concern. There is a dearth of credible statistics on its risk factors attributable to the weak vital registration system in developing countries including Nigeria. Estimates of U5M have taken advantage of indirect methods but U5M risk factors have been identified using fixed statistical models with little considerations for the potentials of mixture models. Mixture models such as Poisson-Mixture models exhibit flexibility tendency which is an attribute of robustness lacking in fixed models. Therefore, this study was carried out to examine the robustness of Poisson-Mixture models in identifying reliable determinants of U5M. The data on 18,855 women used in this study was from the 2008 Nigeria Demographic and Health Survey (NDHS). A stratified, two-stage cluster sampling design was adopted for the study. Birth history of women aged 15-49 years, survival of under-five children and their socio-demographic characteristics were collected by personal interviews using a structured questionnaire. Six different Poisson-Mixture models namely: Poisson (PO), Zero-Inflated Poisson (ZIP), Poisson Hurdle (PH), Negative Binomial (NBI), Zero-Inflated Negative Binomial (ZINBI) and Negative Binomial Hurdle (NBJH) were fitted separately to the data. The Akaike Information Criteria (AIC) and diagnostic check for normality were used to select robust models within Mosley and Chen analytical framework. Infant and child mortality were also considered. All tests were conducted at p=0.05. The NDHS data on USM and the associated factors had different distributions at National and Zonal levels. The models and AIC values for U5M were: 38763.47 (PO), 38654.55 (ZIP), 44270. 77 (PH), 38526.26 (NBI), 38513. 71 (ZINBI) and 44269.30 (NBIH). The PO, ZIP, PH and NBIH met normality test criteria and the ZIP model was of best fit. Also, the Infant mortality models and their AIC values were: 15167.95 (PO), 15169.95 (ZIP), 17568.57 (PH), 15105.42 (NBI), 15107.42 (ZINBI) and 17566.18 (NBIH). Among the models, PO, PH, and NBIH which met the normality test criteria, only the PO model with the smallest AIC had the best fit. Similarly, the models and AIC values for Child Mortality were 22854.13 (PO), 22765.75 (ZIP), 26528.66 (PH), 22708.97 (NBI). 22692.48 (ZINBI) and 26530.64 (NBIH). The ZINBI model with the smallest AIC gave the best fit. These models identified breastfeeding, mother's weight, paternal education, toilet type, maternal education, place of delivery, birth-order and antenatal-visits as significant determinants of infant, Child and U5M at the national level, while breastfeeding, mothers weight, paternal education and toilet type had significant effects on U5M in North-East and North-West Zones. Maternal education had significant association with U5M in South-East and South-South Zones. Breastfeeding, toilet type, maternal education and woman's age were significantly associated with U5M in the North-Central zone. Also, breastfeeding, maternal education, birth-order and antenatal-visits were significantly associated with U5M in the South-West zone. For all models, maternal education was significantly associated with U5M. The Zero-Inflated Poisson model provided the best robust estimates of Under-five Mortality in Nigeria, while maternal education was identified as the most important determinant. The Poisson-mixture models are recommended for modelling Under-five Mortality in Nigeria.