Use this URL to cite or link to this record in EThOS:
Title: Bayesian space-time mapping of childhood malnutrition in Somalia
Author: Kinyoki, Damaris
ISNI:       0000 0004 6423 1334
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Background: Malnutrition is a leading cause of childhood deaths in low- and middle-income countries and has permanent consequences for cognitive, physical and metabolic development. Globally, it is estimated that 26% and 8% children under-five years of age are stunted and wasted respectively. Approximately 90% of the world’s malnourished children live in sub-Saharan Africa and Asia. Food insecurity, which is a major driver of malnutrition, has been shown to be linked to inter-annual variability in rainfall in most of the part of sub-Saharan Africa. In general, a seasonal rainfall higher than 500 mm in sub-Saharan Africa is required to sustain healthy agriculture, with frequent droughts and periods of flooding highlighting the tenuous nature of agropastoral livelihoods in many parts of Africa. Despite the high burden of malnutrition there is limited formal investigation of its spatial epidemiology globally, especially in the most affected countries. Most of the published research has focused on the demographic, socio-economic and individual factors associated with childhood malnutrition. Little is known, however, about its geographical and contextual determinants and how policies can be formulated using the subnational distribution of these factors. Aim: The main aim of my work was to describe the space-time distribution of wasting and stunting in Somalia from 2007-2010 and determine their ecological comorbidity with Acute Respiratory Infection (ARI) and diarrhoea among children aged 6-59 months in Somalia. Methods: Data from household nutritional surveys in Somalia from 2007 to 2010 form a total of 1,066 settlements covering 73,778 children were used for the analysis throughout this thesis. Advanced Bayesian geostatistical models using stochastic partial differential equation (SPDE) in integrated nested Laplace approximations (INLA) were used for the space-time analysis. This modelling technique permits for simultaneous modelling of related issues such as risk assessment, spatial dependence, predictions and quantification of uncertainty. In the first set of analysis, the marginal effects of predictors were computed to determine their inherent spatial variability across the country (Study I). Using a novel approach, the seasonal and inter-annual variation of wasting was computed by first carrying out year-season prediction in the four main seasons in the country from 2007-2010. To then compute the effect size of each season, the survey year was used to define the temporal effect while the seasons were used separately to define the seasonality effects of wasting for the survey year (Study II). In both approaches, time-varying covariates were incorporated in the models to inform the temporal trends. The prevalence and spatial distribution of stunting between 2007 and 2010 was estimated and the role of environmental covariates in forecasting the risk of stunting was explored (Study III). Finally, a joint modelling was undertaken of wasting, stunting and underweight; and stunting, acute respiratory infection (ARI) and diarrhoea (Study IV); to concurrently determine their correlation and shared components (Study V). Results: In the period 2007-2010, the prevalence of childhood malnutrition remained very high throughout Somalia with all administrative regions reporting above acceptable levels of wasting, as defined by the WHO as above 5% prevalence. The average prevalence of wasting, stunting and mid-upper arm circumference (MUAC) < 125 mm in Somalia from 2007 to 2010 was 21%, 31% and 36%, respectively, values which meet the thresholds classified as ‘critical’ by the WHO. In addition, there was evidence of significant geographical and secular variations in the burden of malnutrition in Somalia, with South having higher levels as compared to the North in the country and clear seasonal variation was observed with a relative rise during the dry seasons and reduction during the rainy seasons. Environmental factors like rainfall and vegetation were major drivers of these variations. This study also demonstrated that wasting, stunting and underweight in children 6-59 months in Somalia shared common risk factors with evidence of correlation in space. Finally, the study showed clearly that the spatial shared component between ARI, diarrhoea and stunting was higher in the southern part of the country. Conclusion: Understanding the seasonal and annual fluctuations of different forms of malnutrition and their drivers in different regions can be used to target interventions in communities at high risk during emergency humanitarian interventions. Integrated programming and interventions focused on the common risk factors of the three indicators and specifically in regions where the co-distribution is highly prevalent may be a more effective way of reducing the burden of malnutrition in Somalia.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: RA Public aspects of medicine