ENVIRONMENTAL AND NUTRITION OUTCOMES - Food Security, Food Prices and Climate Variability - page 169

food production variability due to moisture conditions were controlled using the monthly
NDVI for the month/year during which the household was surveyed. The independent vari-
ables of interest included a categorical variable representing the percentage of forest cover
associated with each DHS sampling cluster (0-9 percent, 10-19 percent, 20-29 percent,
30-39 percent, 40-49 percent and 50-59 percent forest cover), and another categorical vari-
able reflecting decadal changes in forest cover over the period 2000 to 2010 (no change, net
forest loss and net forest gain). Forest cover data were derived from the MODIS vegetation
continuous fields at 250-meter resolution (DiMiceli et al ., 2011). The results are expressed in
odds ratios that indicate the associations between the independent and dependent variables
without attributing causality (Johnson et al ., 2013).
The study found that children living in DHS clusters characterized by a net loss of forest
cover over the past decade demonstrated a 19 percent decrease in the odds of having a diverse
diet and a 29 percent decrease in the odds of consuming vitamin-A rich foods, relative to the
reference category (no net change in forest cover) ( Figure 8.4 ). Conversely, children living in
communities with higher percentages of forest cover had increased odds of consuming
vitamin-A rich foods and decreased odds of experiencing diarrhea. Children living in clusters
with a net gain in forest cover from 2000 to 2010 demonstrated a 34 percent decrease in the
odds of experiencing diarrhea. No statistically significant associations were found between the
environmental variables and child stunting.
These results show that environmental dynamics and change are associated with different
nutrition outcomes, even if food availability is not a primary factor. The pathways between
environment and health are many. One study in Malawi will not conclusively demonstrate a
relationship between intact ecosystems and health outcomes, and thus the analysis will soon
be extended to Nepal, Uganda, Kenya and Mali. As the work expands to other areas, the
associations between food availability, food production and nutrition outcomes will be
strengthened.

2.5

Odds of consuming vitamin-A rich foods
Odds of experiencing diarrhea in past 2 weeks

30

25

20

15

10

0-9% 10-19% 20-29%

Percent of forest cover

30-39% 40-49% 50-59%

FIGURE 8.4 Results from the Malawi study showing odds of consuming vitamin-A rich foods and
of experiencing diarrhea in the past two weeks (source: derived from Johnson et al .,
2013).