The Iron Law Comes for Stunting and Sanitation

One of the classic exceptions to Rossi’s Iron Law of Policy Evaluation- that the expected impact of a rigorously evaluated public policy is zero- is sanitation. Large pre-modern cities generally had higher mortality than smaller cities or rural areas in the same countries, a distinction that is generally and plausibly linked to worse sanitation as well as greater pollution. Once sanitation improved, mortality fell.  For example, one 2005 review of falling U.S. mortality in the late 19th century found that:

Mortality rates in the US fell more rapidly during the late 19th and early 20th
Centuries than any other period in American history. This decline coincided with an
epidemiological transition and the disappearance of a mortality “penalty” associated with living in urban areas. There is little empirical evidence and much unresolved debate about what caused these improvements, however. This paper investigates the causal influence of clean water technologies – filtration and chlorination – on mortality in major cities during the early 20th CenturyWe find that clean water was responsible for nearly half of the total mortality reduction in major cities, three-quarters of the infant mortality reduction, and nearly two-thirds of the child mortality reduction. 

In data geeks’ favorite example, the pioneering doctor-meets-geomapping-analyst John Snow used a map of water sources and incidences of cholera to link an 1854 London outbreak to wells contaminated by waste, leading to belated improvements in London’s water system and an eventual end to large-scale cholera epidemics in Europe. 

John-Snows-cholera-map-of-009.jpg

Infectious disease is the big exception to the law that no intervention makes that big a difference (which is why it is curious that vaccines have routinely come in for so much more criticism than far less effective elements of the medical system.) Sanitation is one of the most effective methods of controlling childhood exposure to infectious disease agents. Height in European countries began climbing at around the same time as sanitation was brought under control:

European-hight

So I don’t think that even very skeptical people found it hard to believe evidence that poor sanitation practices such as open-air defecation across India and other low-income countries was largely responsible for childhood stunting in these countries.

journal.pone.0073784.g001.png

Even as a tourist to India fifteen years ago, I saw many more visibly stunted kids in cities in Uttar Pradesh with poor sanitation than in cleaner parts of the country.

A recent large-scale set of multi-year randomized trials in Kenya and Bangladesh suggests, however, that, even if poor sanitation is involved in the processes that produce childhood stunting, stunting isn’t immediately amenable even to quite intense intervention. You can’t just expose households to sanitation interventions targeted to them and expect the next generation of kids from those households immediately to be taller, and even the combination of nutrition with sanitation produces small effects, not distinguishable from nutritional supplementation alone. As one quite clear write-up in an Indian newspaper notes,

“The studies, which targeted over 13,000 families, showed that water purification, sanitary latrines and hand-washing (WASH) interventions in select households were not enough to prevent stunting in those households…Three groups received the three individual WASH interventions, while a fourth received nutritional counselling and dietary supplements for children. The fifth group received all three WASH interventions, the sixth received WASH as well as nutrition, while a seventh served as a control. Once the pregnant women gave birth, stunting, diarrhoea and mortality rates were tracked among their children for two years…After two years, the Bangladeshi study found children in the WASH groups to be no taller than controls. Improved diet did not make a big difference either – it corrected only a sixth of the height deficit in the nutrition groups. The Kenyan study reported similar findings.”

In Bangladesh, it seems as though the researchers are possibly observing genuine benefits- a reduction in diarrhea incidence though not stunting for the groups exposed to water purification, sanitary latrines, and hand-washing, although this reduction was not present for the Kenyan children.

One author suggests that “We have forthcoming evidence that illustrates that although we did reduce exposure to faecal organisms somewhat, there was a large reservoir in the environment that children continue to be exposed to.” This would seem to be plausible but, depending on how broadly distributed that reservoir of disease organisms is and how difficult they are to eradicate, potentially disheartening, if merely, for example, playing in the dirt exposes kids to sufficient numbers of bacteria to negate the effects of improved sanitation. An optimistic alternative is that city-wide improvements in sanitation will have immediate effects not possible from household-level interventions. Another idea, not mentioned, is the idea that improved sanitation and nutrition could in fact have large effects on height, but only over multiple generations, through selection for greater height. This seems to have occurred and to be still occurring in Europe. 

IQ and height have a lot in common- similar heritability, for example, and a similar tendency to rise secularly during periods of economic development while being resistant to any individual intervention aimed at changing them.  Sanitation is among the very few One Weird Tricks that really work – a true exception to the Iron Law. I’m guessing that nobody is going to find that improved sanitation doesn’t help reduce cholera epidemics, no matter how far the Replication Crisis extends.  But even the best tricks can’t solve everything.

5 thoughts on “The Iron Law Comes for Stunting and Sanitation

  1. The effect size of that study looks reasonable to me, given the rate of past improvements. Whenever I look back at any broad improvement in health outcomes, it always surprises me that you can usually never see big discontinuities. Instead, what it looks like is any event we choose to remark on is only one in a long series of small improvements that build up to produce the result. Execution matters, especially patient, diligent, repetitive execution matters.

    Liked by 1 person

  2. I suspect that one cause of height increases in the West is C-sections, which have allowed women to give birth to abnormally large babies without dying (which would in the past have often resulted in dead babies.)

    Also, I don’t think you’re going to get very far just targeting households if kids are still playing in cities with open sewers running through the streets. It’s a nice start, but as said, still a lot of germs in the environment.

    Liked by 1 person

    1. I agree; I think a district or city-level intervention seems like the next thing to try, though it’s likely you’d need a long time horizon to get things going, since a city-level sanitation program will take a lot longer than giving individual moms sanitary toilets and treated water and nutrition supplements.
      Re C-sections, I ended up waiting for a bus with a young army medic in India once who volunteered out of the blue that he thought the prevalence of c-sections in the West was good for post-parturition sex.

      Like

  3. This week I assigned the famous example of Simmelweis’ work on reducing childbirth mortality from Hempel’s A Philosophy of Science to my students for discussion. This gives me some supplemental resources.

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s