ENVIRONMENTAL AND NUTRITION OUTCOMES - Food Security, Food Prices and Climate Variability - page 170

Food availability and nutrition outcomes: West Africa
Research focused on understanding the consequences of an environmental shock can use a
variety of data to estimate nutrition outcomes. Ongoing food aid programs, assistance for child-
hood feeding, health and other nutrition campaigns, and other activities all affect the nutritional
status of children under five years of age ( Johnson and Brown, 2013). These efforts to reduce
overall food insecurity are occurring during a period of increasing rainfall and resulting vegeta-
tionproductivityandsigniicanteconomicgrowth(USAID,2012;USDA,2012).Research
bringing NDVI and DHS information together can provide insights into the consequences of
climate variability on childhood mortality and nutrition. These outcomes are measured through
direct sampling of individual children in a population, and represent nutrition outcomes result-
ing from broader socio-economic conditions, despite efforts from a wide variety of programs
and government interventions. This is a very different scenario from the FAO undernutrition
statistics, which are measured indirectly through analysis of national level population and food
production statistics, both of which have errors and uncertainties.
NDVI can be used as a variable in a regression analysis to remove the impact of weather and
agricultural growing conditions on nutrition outcomes. The study links NDVI to the reference
child's date of birth to child nutritional status (stunting and wasting) from DHS variables and
child survival data from four countries: Benin, Burkina Faso, Guinea and Mali. The association
between environmental conditions (as proxied by vegetative growth) and child survival and
nutritional outcomes is assessed using multivariate methods ( Johnson and Brown, 2013). This
initial research and subsequent variations on these analyses using satellite-derived information on
the vigor of the growing season vegetation will allow us to make stronger arguments about how
climate change (warming trends and anomalous weather patterns), via its impact on growing
conditions, can be expected to affect child nutritional status and survival.
Nutritional status and survival data, as well as other individual-level characteristics were
drawnfromtheDemographicandHealthSurveys(DHS;fundedbytheUnitedStatesAgency
for International Development (USAID)) program for the following surveys: Benin 2001,
Burkina Faso 2003, Guinea 2005 and Mali 2001 and 2006. DHS surveys were collected from
nationallyrepresentativeprobabilitysamplesselectedusingstratiiedtwo-stageclusterdesigns;
more information about each survey can be found in its respective final report (Benin: Institut
NationaldelaStatistiqueetdel'AnalyseÉconomique(INSAE)etORCMacro2002;Burkina
Faso:InstitutNationaldelaStatistiqueetdelaDémographieetORCMacro2004;Guinea:
DirectionNationaledelaStatistique(DNS)(Guinée)etORCMacro2006;Mali:Cellulede
Planification et de Statistique du Ministère de la Santé et al. 2002 and 2007) ( Johnson et al .,
2013). West Africa is expected to be particularly hard hit by the effects of climate change due
to the region's dependence on rain fed agriculture as a primary livelihood, and poor regional
capacity to adapt to changing conditions. The four countries have diverse ecosystems,
economies and levels of rainfall.
As in the Malawi study, DHS GPS data were overlaid with the average AVHRR NDVI
data from July, August and September to extract the NDVI value at each DHS cluster cen-
troid. The methods used were a one-way ANOVA with a bivariate analysis of the relation-
ship between NDVI and the outcome variables of interest, with additional variables to control
for confounding factors ( Table 8.4 ). Similar to the Malawi study, the results are expressed in
odds ratios that indicate the associations between the independent and dependent variables
without attributing causality.