A hard look at the strategy of fighting open defecation

School sanitation at Mysore
School sanitation at Mysore

With over 620 million defecating in the open in India, do we need a new approach to curb this practice? The force of habit is such that even households with toilets have around forty percent of adults defecating in the open. But, does curbing open defecation necessarily lead to significant improvements in child health outcomes like diarrhoea, anaemia, parasite infection and growth?

Sumeet Patil, from the Network for Engineering and Economics Research and Management (NEERMAN), Mumbai, and School of Public Health, University of California, Berkeley discusses these in an interview with India Water Portal. 

Your study suggests that reducing the practice of open defecation does not bring about any significant improvement in health outcomes in children aged below five years. Please share the results of this large-scale, randomised, controlled study.

This would not be the accurate inference. The study found that the intervention we studied in Madhya Pradesh (MP) (Nirmal Bharat Abhiyaan implemented with help of Water and Sanitation Program in 2 districts of MP) does not deliver any health benefits based on a randomized controlled trial (RCT) we conducted between 2009 and 2011. The intervention is a “program aiming to build toilets and reduce open defecation (OD)”.  In fact one of the reasons we think that the health benefits are not seen is because the reduction in OD levels was probably not enough. The paper, editor’s summary and other documentation is available here.

We are currently doing further analyses to test the assumption that OD levels is indeed associated with health benefits; I presented draft results at Delhi School of Economics to get feedback from peers. Indeed, we find strong “association” between OD and height of child (BTW, this is only one of the measures of child health). The limitations of this “draft” finding are: (a) association does not prove causation; and (b) weak biological plausibility of height impacts in short term exposure to reduced OD levels. However, the more convincing proof can be provided only in future RCTs if they are able to achieve greater reductions in OD levels than what we observed.

Your findings are a deviation from the existing ‘formal’ assumptions in sanitation. Are there any other recent studies from India or elsewhere that support your findings?

We need to make a distinction between sanitation as a concept and sanitation programs that aim to provide specific facilities/interventions.

Tom Clasen and colleagues published their RCT results soon after our MP study was published. The intervention (sanitation program) they studied was implemented much more intensively (you learn about the program in their paper). They found much higher coverage of toilets, and thus, larger reduction in OD than what we found in MP. However, they also didn’t find any health impacts. Further, they measured environmental exposure to fecal pathogens through flies, water and hand rinses. They don’t find any change in exposure levels through these pathways either.

In the authors’ words: 'increased latrine coverage is generally believed to be effective for reducing exposure to faecal pathogens and preventing disease; however, our results show that this outcome cannot be assumed'. Like us, Clasen and colleagues also suggested that reduction in OD levels was probably not sufficient enough, but there could other possible explanations that improvements in household sanitation alone was insufficient. Other pathways such as hand contamination, food contamination or other sources such as animal feces can be important considerations.

Together, these two RCTs give compelling evidence that large scale sanitation programs may not deliver the health effects they are supposed to. Other than the above 2 RCTs, there is no other “peer reviewed and published” evidence of effectiveness of sanitation interventions/programs. An earlier RCT in Odisha that my former colleagues at RTI and I conducted (led by Subhrendu Pattanayak, now at Duke University) between 2005-06 focused on behaviour change effects of a pilot sanitation program and thus only the effect of the intervention on toilet use was published here.

The previous research has basically culminated in the “formal assumption” you referred to above and argues that WASH is a very effective public health intervention. This previous research was used to justify investments in sanitation.  However, the critique of most of the existing research on sanitation was that the effectiveness of sanitation was not proven “experimentally” (cross sectional analyses, analyses done using secondary or summary data,  or improperly designed/analyzed trials have severe limitation to prove “causation”; I will save you the academic debates).  

Currently, I am aware of at least one more trial from Mali which may provide evidence of positive health effects contrary to what we find. I believe that their intervention was much more intensive and at pilot scale. Even my colleagues and I are analysing health measurements from an earlier RCT in Odisha referred above and we find evidence of health benefits. Even this intervention was rigorous and implemented at pilot scale by professionals (not a scaled up program). Basically, there are forthcoming studies that may provide experimental evidence that sanitation programs can potentially deliver health. Let me not cite the results or share these working papers currently being reviewed by external experts. Obviously, replicating success from a small highly controlled pilot to a scaled up program is a different challenge.

Will the outcomes on child health be different if intervention campaigns are able to achieve higher levels of toilet coverage or if we allow for a longer duration of program exposure? Are there studies from India or elsewhere that corroborate this?

The previous answers address a part of this question. We have sided with a prevalent view in the sector that mere toilet coverage is not enough but behaviors must change. It is possible that over time behaviours may change slowly once people have access to toilets, but this is mere a hypothesis and can be tested only in future. We also acknowledge that some of the health benefits such as reduction in parasite infections may take more time because worms can remain viable for years.  

Are campaigns that focus on communication elements for demand generation for toilets and ‘behavioral nudges’ to address the issue of toilet usage more effective?

More effective than what? I believe you are asking about a subsidy-driven approach versus behaviour change approach. The intervention we evaluated in MP combined subsidies for toilet with behaviour change activities based on Community-Led Total Sanitation (CLTS) tools (I would not characterize the intervention as CLTS though). However, we do see larger effect on toilet construction and reduction in OD in BPL households who received higher subsidies than in non-BPL households. The Odisha RCT paper published in the WHO bulletin reported that behaviour change explained about 60+% of the effect and the rest (more than one third) is explained by subsidies. Additionally, my colleagues from UC Berkeley and I are conducting a multi-country analysis using data from different RCTs to investigate the reduction in OD levels through investment (money) and behaviour change pathways. It is too early to share the results but emerging consensus is that subsidy and behaviour change both are needed. 

Considering this, does the government need to revisit its current strategy under Swachh Bharat Mission of increasing the number of households that have toilets to reduce open defecation? Can you suggest specific refinements of Swachh Bharat Mission as regards creation of toilet infrastructure and emphasis on behaviour change?

This is a tough question! I believe Swachh Bharat Mission - Grameen (SBM-G) strategy already talks about both access to toilets and behaviour change. The strategy does favor post-construction incentives to construct toilets. It has dovetailed funds with National Rural Employment Guarantee Scheme (NREGS) – just like the intervention in MP – to provide funds to Below Poverty Level and eligible Above Poverty Level households for toilet construction.

However, I personally perceive that there is an “implicit” assumption that we know how to solve the problem of open defecation (let’s assume that use of toilets by all is the end point we are interested in and keep the debate on health benefits of sanitation aside for the moment). Based on available evidence in the sector and its quality, I doubt that we really know much about what will work and which is the more efficient approach. There is a lot of advocacy and opinion in the sector about what government should do under SBM. My concern is that most such design and implementation recommendations are not tested under different field conditions. There also should be newer innovative approaches for generating demand and changing behaviors.

Therefore, a key recommendation will be to first test the strategies, approaches, beliefs, opinions, tools at a pilot level at different sites across India and generate evidence that they will at least result in access to toilets and their regular use in short term. Otherwise, we risk learning of failures (or successes if we are lucky) after they have happened “at scale” and that is a very costly way of learning and improving program designs/implementation. However, we can learn from pilots much sooner and can provide objective evidence to the government about what works and where. A few suggestions are given below, but they also need to be tested in “real” field conditions.

First, economics incentives and their structuring is not explored enough as behaviour change tools.  Currently debates mostly center around pre-construction subsidy or post-construction incentive, but that is a very limited view. For example, there can be group discount strategies to create social pressures and group level demands. There can be conditional cash transfer approaches to bundle sanitation with other allied services. There is some ongoing-debate on direct cash transfer to beneficiaries versus direct cash transfer to vendors which can be tested out.

Second, in the true spirit of swachhata, there should be concentrated efforts in progressing towards integrated sanitation at community levels through household toilets, shared/community/public toilets, solid waste management, waste water management, overall cleanliness. While this may indeed be the “stated” objective under SBM and previous programs like TSC and NBA, the on-ground situation I observed (and I am working in India in this sector since 2004) was that household toilet construction trumped everything else. Water safety, food hygiene, hand washing and personal hygiene are also key components of swachhata and can deliver rich public health dividend. Even integrating different programs such as mid-day meal, deworming and sanitation have a great public health potential.  However, I don’t see such integrated approaches being tried out.  Obviously this is easier said than done, but that is why we should at least test different approaches on the field and generate lessons on what works and where.

Third, the monitoring and evaluation (M&E) system under SBM should be evolved as an implementation and learning tool than a mere monitoring tool. It may sound clichéd but M&E is still considered an audit tool; something to generate proof that toilets are indeed built or even used. There are serious efforts currently underway to monitor the use and sustainability too. But, the M&E system can be much more. It should be able to inform site specific adoption of the programs, advise on implementation, and find failures before they happen.  It can also integrate “learning” as I suggested above. However, even this “M&E as a implementation tool” concept must be tested in field.

Arghyam’s experience is that the so-called ‘improved’ sanitation systems are leading to contamination from leachates of on-site sanitation systems. This combined with improper well protection is leading to contaminated groundwater. Did you come across such experiences during your study?

We collected and tested water from drinking water sources and in home drinking water for e.coli (which indicates contamination with fecal matter but not necessarily pathogenic itself). We didn’t observe any difference between the intervention and control groups. Tom Clasen in his paper alludes to the exact possibility you site above but even he did not actually detect such problems in his data. However, I will agree that just because the name implies “improved”, one cannot guarantee that the toilet can indeed contain the fecal pathogens from mixing with the environment. Technically, the soil conditions, water table height, and construction quality should matter. In fact, the majority of water sources in MP were of “improved” type but had very high levels of fecal contamination. So, let’s think of the word “improved” water source or sanitation as something better than “guaranteed bad” types of facilities.

Post By: Amita Bhaduri