Hi everyone. My name is Sophie Boisson, I work at the London School of Hygiene and Tropical Medicine. And today I'm going to talk about health impact, of household water treatments, and safe storage intervention. You have seen from previous lectures, the three, let's say the three elements to achieving health impact. First of all, while of course water has to be contaminated by microbes and it has to be a predominant route of transmission for diarrheal diseases and pathogens. Second, we need to have a technology that is effective at treating those pathogens. And third, we need to make sure that people are using those technology consistently and correctly. So this slide is to remind you that transmission of a fecal-oral pathogens is complex. As you can see from this diagram household water treatment and safe storage addresses only one pathway of many. So, the key question is how much diarrhea can be prevented by improving water quality alone? Over the past 30 years, there's been a lot of studies measuring the impact of water quality intervention on diarrheal disease. And this figure show show the results from different systematic reviews that were published in the last ten years. And key findings from those reviews were that overall water quality interventions were associated with about a reduction in diarrheal disease of 30, 40%. They also found that there were some differences in terms of impact between water quality intervention at the source and those at the household level. The impact on diarrhea was generally found to be higher when intervention focused at the household level. They also found that there were some differences between different treatment methods. Now we're going to talk a little bit more about the studies that measure the impact on diarrheal disease looking specifically at the type of design the outcome being measured, and highlighting some of the key challenges in conducting health impact studies. So in terms of evaluation designs the gold standard for measuring health impact of intervention is the randomized control trial. So we have a group of participants, households or villages that are randomly allocated to two different groups. One group will receive the intervention and the other group will act as a control. The unit of randomization can vary. And the participants may or may not be blinded to the intervention. And we will cover that in later slides. We can have other type of study design such as observational studies which includes comparing diarrheal disease rates among all participants before and after the introduction of an intervention. We have case controlled studies which identify cases and non-cases of diarrhea and compare the exposure, adjusting for potential confounding. And we have ecological studies. We also have modelling studies and I think you've already covered that in the last session using examples of studying the impact of different levels of compliance on diarrheal disease. So, all the studies have been looking at the impact on diarrhea, but what is diarrhea? Diarrhea, is a function of stool consistency, and frequency. So WHO definition is the passage of three or more loose stools, within a 24-hour period. And as you can imagine, it's pretty hard sometimes for people to remember how many loose stools they've passed in the last day. And also stool consistency, okay, it may vary, may vary by diet and so that's why it's very hard to measure it reliably. So potential issues is associated with measuring diarrhea, so we have reporting bias. So reporting bias may happened for example in the randomized control trial when participants in the intervention group receive a brand new filter. They may be more likely to report that, to the reporter, the enumerator who comes to visit them, that they feel healthier, that they, that their health has improved. It can be because they want to please the enumerator and that's often called a courtesy bias. On the other hand, participants in the control group who have who have not received anything may be more likely to complain and say that their health has not improved because they haven't got any filter. The placebo effect, it's possible that participants who've received a filter or chlorine tablets for example, may consciously or subconsciously think that their health is improving. The Hawthorne effect which is interesting because a lot of studies have looked at most of those studies measure diarrhea over time, and what they often find is that diarrhea decreases in both intervention and control group over time. And one possible explanation for that is the Hawthorne effect, which is participants may adopt, safer water handling or hygiene practices if some researchers come to their house regularly. It's also possible that over time participants are tired of seeing the enumerators regularly and and they don't want to bother answering questions or they just want to answer very short, or give very short answers so that the interview doesn't last very long. So these are potential issues to be considered when, when measuring diarrhea. So here is to illustrate this reporting bias that I mentioned before. So the systematic review conducted by Clasen and colleagues in 2006 shows that report report the impact on diarrhea according to whether the intervention was blinded or not. So blinded means that the participants and assessors, double blinded means that participants and assessors do not know which treatment they have received. Whereas open means that everyone knows what treatment they've received. So what they found was that there was an important difference in effect size. While open trials reported diarrhea reduction in the range of 30, 40%, blinded trial reported no effect. And although there were certain issues with those blinded studies, other researchers have documented that the lack of blinding of subjective outcomes such as diarrhea tend to exaggerate the magnitude of the effect by 25%. Now researchers, a group of researchers, Hunter and colleagues, they used this estimate to adjust the 30, 40% estimate that was obtained from open trial, and found that after adjustment for this bias, due to lack of blinding, the effect on diarrhea dropped to 15%. So of the many studies that have been conducted looking at the impact of water quality intervention on diarrhea the effect estimates often vary. And so why do we get very different reduction in diarrheal disease from those different studies? Well first of all, they may be conducted in different settings where the transmission dynamics may be different. So the transmission routes or the importance of different transmission routes may be different. For example, in a setting where sanitation and hygiene condition are poor, are very poor and where water quality is not a major concern, then of course lack of sanitation and lack of hygiene are more likely to be predominant routes of transmission for diarrheagenic pathogens. Also the etiology for diarrhea may be very different depending on the regions as the GEMS-studies have shown. Second, the treatment methods may not be, as you've seen before treatment methods are not effective against all class of pathogens. So you've seen the example of chlorine, which is less effective against protozoan cysts, so in a certain area where the study is conducted Cryptosporidium is a major etiology for diarrhea, then a chlorine intervention maybe less effective at reducing diarrhea. Then third: compliance. Of course if the technology or if the treatment methods is not being used correctly and consistently, then we don't expect to see any, any improvement or any impact on diarrhea. And in fact,Arnold and colleagues and Hunter, later Hunter and colleagues in 2009, they have illustrated that the impact of diarrhea tends to diminish over time. And one potential explanation for it is that compliance when, the longer the studies are conducted the less compliant participant become and the less likely they are to use intervention, which may explain why diarrhea decreased over time. And then there may be also some difference between efficacy versus effectiveness trials where efficacy studies may lead to a higher reduction in diarrhea because conditions are more controlled and more, let's say, perfect compared to effectiveness trial which measures impact on real life situations where the intervention may not be so delivered perfectly. So, so we've seen that measuring health impact is not easy. So measuring diarrhea in particular is very challenging in the field. It's very easy to do because you could just collect this information through surveys. However, there's lots of challenges. And so there's several ways through which we could improve the way we measure the health impact of the intervention. First, we talked about blinding. Okay. There's been lots of studies now trying to blind intervention, so blinding chlorination tablets, blinding filters and there are a number of challenges associated to that. So one of the challenge for example with filters, it will be really hard to create the placebo filter that that is not effective, or doesn't to some degree reduce contamination of the water. And also it may be really hard to blind it effectively so that participants are not able to find out which filter they received. And also there's some, it's potentially unethical for example if participants who were boiling suddenly receive an intervention, they may reduce this practice and use the recommended tablet, which may be a placebo or inactive tablet. And by switching intervention among the placebo group, a participant may be at increased risk of diarrhea. Another option, if blinding has some challenges, is to use alternative health outcomes, so not only rely on diarrhea, but use a more objective health outcome. So one of them is nutritional indicators. So for example, weight for age. which can be used as a proxy marker for diarrhea. And in fact there has been a study conducted by Schmidt and colleagues in 2010 showing that the more they looked at the 14 days windows and they found that the more days you had diarrhea during that period the lower was your weight. So this is still being explored and see whether this indicator could be a useful proxy for diarrhea. But obviously because weight is influenced by another, by a lot of other factors such as diet, et cetera, it may be not sensitive enough to to reflect changes in diarrhea. And then we have height-for-age, which is an indicator of stunting or impaired growth among children. And there's been a few studies that have looked at the impact of solar disinfection on stunting but there's very few studies, and I think the evidence so far remains weak. It's, at the moment, there's a lot of interest in looking at environmental enteropathy, which is the clinical condition characterized by a damage of the gut resulting in leakage of nutrients and malabsorption. And there's an increased interest in measuring impact of water quality or other environmental health intervention on this on this condition. A third possibility is to look at certain specific pathogens either in stools or look at serological markers for exposure to those pathogens. But again these methods also have some drawbacks. For example you need to use laboratory methods and you need to identify which pathogen you're going to test for. And sometimes being infected with a pathogen doesn't necessarily translate into being sick with diarrhea. But these are the outcomes that should be considered and there's definitely an increasing interest in looking at those in future studies. Thanks, Sophie, for a clear explanation of measuring health impacts, illustrated with nice examples. If I can briefly summarize, there have been a lot of health impact studies which have shown large reductions in diarrheal disease, of around 30% to 40%, but, that these studies may be overestimates due to the methodologies used. Especially, there's this issue of blinding of the interventions. These studies do show that it is definitely possible to measure health impacts, but also that it's not simple, and that a really high quality evaluation is complex, also expensive, and definitely something that should be designed by public health specialists, such as epidemiologists. Health impact studies could still be done for large programs. But even for large programs, and certainly for smaller ones, it's probably better to focus first on monitoring actual use and effective use before attempting to measuring health outcomes. Here you see some of the papers that Doctor Boisson drew upon for this module, the systematic reviews and the other papers discussing the evidence base, and finally, a few of the individual studies that provide examples of monitoring health impacts of HWTS.