Hi everyone, I’m Sangita and I’m going to tell you about some research that Dean and I have done with PhyrumKov and Susanna Smets from the WSP Cambodia office on the linkage between sanitation and stunting in Cambodia.
Cambodia is an interesting country to look at because first, it has very high rates of open defecation, even slighly higher than India’s. Second, we have two recent Demographic and Health Surveys (one from 2005 and the other from 2010), from which we have both defecation and child height data. Third, there was a significant increase in child height between 2005 and 2010. Average child height increased by .13 standard deviations, which to put it in perspective, is the entire India Africa height gap. Also, there was a large reduction in open defecation over this period.So these two facts are what motivate our analysis: did the reduction in the rates of open defecation have anything to do with the rapid increase in average child height.The data we are using comes from the DHS (Demographic and Health Survey), which is a nationally representative population-based survey. Since DHS data is based on household surveys, our identification is not from latrines themselves but from the usage of latrines. In India there have been major governmental campaigns, while in Cambodia it isn’t the government.
Explain z-score: The height-for-age z-scores reported in the DHS are calculated by standardizing the height of an individual child using a well-nourished and healthy reference population for the same age and sex. A z-score of -2 means that a child is 2 standard deviations below the mean height of children of the same age and sex in the reference population. Dean talked a lot about why height is important, and in summary it is a long-term reflection of health and is a good indicator of outcomes later in life. The primary independent variable used in the analysis is OD in locality, which we calculate as an average of sanitation practices for all households living in the primary sampling unit. The idea is that this variable better captures a child’s exposure to open defecation. The greater the number of households in the area openly defecating, the greater the likelihood that fecal contaminants will get on a child’s hands and feet and the more likely they are to get sick from them.
The medical, epidemiological, and economic, literature linking sanitation to stunting is significant. There are a number of studies that link fecal-oral contamination and diarrhea to child stunting, and more recently there is a growing literature linking environmental enteropathy to stunting. We will hear more about environmental enteropathy in another session tomorrow, but briefly it is..Environmental enteropathy – a disease of the gut caused by fecal pathogens that prevents nutrients from being absorbed by the body. Does not necessarily develop into full-scale diarrhea.
Of the economic literature, we have already hear about a study linking the sanitation motivation through the TSC to taller children in Maharashtra. Dean has done other work linking differences in child height between India and Africa to differences in sanitation. Cameron et al (2013) show that a Total Sanitation project in Indonesia increased the height of children living in households that did not have latrines at baseline.
Here we are looking at height of children at every age depending on whether they are exposed to open defecation. Two conclusions are visible in this figure. First, a larger fraction of households openly defecating in the PSU is associated with greater stunting on average at every age. Second, the effect of the sanitation environment compounds. When a child is born, the effect of exposure to open defecation is not prevalent, but as the children age the effect keeps on adding on. And by the time a child reaches the age of two, the damage of the disease environment has mostly taken its toll and is irreversible.The difference between rich and poor doesn’t matter for newborns but the effect of open defecation becomes more important as time goes by, and compounds.
Here, we are looking at the relationship between height and exposure to open defecation for children growing up in households that both do and do not openly defecate. The plot notably shows that higher rates of open defecation in the locality are associated with greater stunting even when the household itself does not openly defecate. That’s why we see two downward sloping lines. When the fraction of households openly defecating in the area approaches 100 percent, children in households that do not practice open defecation are almost as short as children in households that do on average.
Could there be other factors that we are not including that are causing us to see this relationship?
This figure is pictorializing the fixed effects. Each line represents a region in Cambodia. The line connects the average height and rate of open defecation in 2010 to the average height and rate of open defecation in 2005. A downward sloping line means that as open defecation decreased, height increased. There are 38 lines on the graph, and 24 are sloping downward. So, does that suggest that a reduction in open defecation is associated with an increase in height? To explore this, I want to conduct a thought experiment. If there really were no relationship between open defecation and stunting, each of the lines would be equally as likely to slope up as down. If this were the case, it would be quite normal to see something close to 50 percent of the lines sloping up and 50 percent sloping down. But here we are seeing 63 percent of the lines sloping downward. Is that a lot? Actually it is, if there really were no relationship, we would only see 24 lines sloping down 7 percent of the time. So this is pretty suggestive that there is a relationship between open defecation and stunting.
Now, to delve into the econometrics. Here I have the regression equation that we are using. LHS variable is height for age zscore of child I in locality l in province p in time t. The idea behind using local area electrification is to include something that approximates the welfare and infrastructure of the PSU. We suspect that children in more well-off areas are taller for many reasons other than just the sanitation environment and by including this variable, we are aiming to isolate those effects from the effect of sanitation.Mom’s height reflects genetic variationMom’s BMI is much more than that – overall household food availability, quality of in utero nutrition, etc.
An increase from 0 to 1 in the rate of open defecation in a child’s locality is linearly associated with a decrease in children’s height by between 0.435 to 0.771 standard deviations depending on the model used.Accounts for…There are several important things to note. The first is that the point estimate for open defecation in the child’s locality stays remarkably stable as controls are added. Controlling for general infrastructure improvements, as proxied by electrification, and for increases in economic well-being, do not diminish the effect; despite overall infrastructural and economic advancement, these variables alone do not explain the improvement in child height from 2005 to 2010. Characteristics of the mother also do not seem to diminish the role of sanitation. Although maternal anthropometry itself robustly predicts child height, it does not change the estimate of sanitation’s effect. Finally, household open defecation eventually becomes statistically insignificant, which highlights the importance of the negative externalities associated with open defecation..42 this is just bigger than the amount that counterfactually would be necessary to explain the India-Africa height gap.Break this up into two slides. Include column numbers
This explores this slightly more rigorously. Here we have the relationship between height and open defecation for both 2005 and 2010. Each line represents a different year. As you can see, the lines are pretty close together. So the relationship between open defecation and height is pretty similar in both years. That means that the change in open defecation in the locality between the years accounts for a lot, if not all, of the change in child height between the two years.
1 sangita - dse conf-slides_13_7_28
Growing Taller Among
Evidence from changes in sanitation and child height in
Cambodia in 2005 and 2010
Phrym Kov, Susanna Smets, Dean Spears, Sangita
rice and WSP
Open defecation is widespread in Cambodia, but
significantly reduced from 2005 to 2010
Child height increased by 0.13 standard deviations –
the entire India-Africa height gap
Question: How did the improvement in sanitation affect
the height of children age 0-5 over the same period?
(Chekley et al 2008)
(Humphrey 2009, Lin et al
2013, Kosek et al 2013)
Hammer and Spears (2013) – randomized sanitation
motivation caused an increase in child height in
Spears (2013) - differences in child height between
India and Africa explained by differences in sanitation
Cameron et al (2013) - Total Sanitation project in
Indonesia increased the height of children living in
households that did not have latrines at baseline
Two is the Magic Number
Data from Cambodia DHS 2010.
Good Toilets Make Good Neighbors
Data from Cambodia DHS 2010.
Fixed effects analysis: control for persistent regional-
level differences by looking at change over time within
Were regions where open defecation decreased by more
the regions where child height increased by more, on
Include control variables
Eilpt: Household socio-economic characteristics
household electrification; indicators for owning a radio, TV, and
refrigerator; 10 indicators for floor material (an important indicator of
disease exposure); 19 indicators for household size; indicators for level of
literacy of the child's mother; indicators for cooking fuel type; and province-
level measures of average consumption and population density.
Dilpt: Child demography
13 indicators for birth order and 11 indicators for birth month.
Hilpt: Health and healthcare
an indicator for the child having a health and vaccination card; an indicator
for having had an institutional delivery; and indicators for being breastfed
immediately and within the first day.
Age-in-month dummies for girls and boys
Latrines spillover (effect)
Improvement in sanitation access likely played a big
role in increasing average child height
The change in sanitation accounts for almost the entire
change in height