UC Berkeley

Many households in developing countries rely on contaminated and untreated drinking water sources, contributing to gastrointestinal illness and other health risks. Even piped water quality is often unreliable because of poorly-maintained treatment or distribution systems. Household water treatment (HWT) systems aim to enable users to treat their water at the point of use, making it safe to drink. While some HWT options have been successful in improving health in developing countries, low adoption and sustained use outside pilot projects and epidemiological trials remains one of the current challenges with this approach. Furthermore, Quantitative Microbiological Risk Assessment models predict that the health benefits from water quality interventions drop significantly with even occasional consumption of contaminated water. Therefore, to be effective, HWT options need to achieve high user compliance rates and provide safe water reliably.

I begin my thesis with an interdisciplinary analysis of the field of water, health, and development, followed by a description of my research study site. Using an interdisciplinary research approach, grounded in the local context, I led the development of an ultraviolet (UV) water disinfection system for rural households. This included an iterative process of design and field tests to create a user-friendly system and laboratory research to improve the performance of the technology. I also collaborated with a non-profit organization based in Mexico in the design of an implementation program to support the adoption and consistent use of the UV system.

Then I present the design and application of a stepped-wedge cluster randomized trial in rural Mexico to evaluate compliance with the implementation program and field efficacy of the UV system. I developed a framework that disaggregates and measures the components of compliance from initial adoption of a safe water practice to exclusive consumption of safe water. I applied this framework to measure compliance across intervention and control groups and to test if additional program components that improve convenience to users can be a cost-effective approach to increase compliance. I present evidence that the implementation program significantly improved compliance with the habit of consuming safe water, when compared to the practice of purchasing water bottled in reusable 20 L containers in the control group. The additional program components proved to be a cost-effective strategy to increase compliance immediately post-intervention, but their impact degraded with time. By analyzing results across different compliance components, I find limitations of the current HWT approach. I present the rational for pilot testing strategies outside the current HWT paradigm, such as expanding a narrow focus on drinking water to making all domestic water safe to drink (as suggested by our observations of multiple water access points in the household) or switching from a product-based to a service delivery model.

As a second component of the randomized trial, I present a series of controlled comparisons to evaluate the field efficacy of the UV system using E. coli as a fecal contamination indicator in drinking water. I use an as-treated-analysis to isolate the impact of the system and contrast these results with an impact evaluation of the implementation program led by a research colleague. I also created a drinking water reliability framework to compare potential contamination impacts from different household water management practices and a logistic regression model to assess household risk factors for post-UV-treatment contamination. I show that treating water with the UV system and storing it in 20 L narrow-necked containers, allowed households to significantly improve their drinking water quality and gain access to a more reliable source of safe water. However, I also found evidence of post-treatment contamination. Through the logistic regression model, I identify that inexperienced system operators, poor household infrastructure, and pouring water in drinking glass are associated with increased risks of contamination. Considering the current unviability of monitoring water quality in real time, the reliability framework proved to be a useful tool to generate a more realistic representation of the variations in water quality that households are exposed to. The processed-based model was also useful in identifying areas that can be targeted by HWT programs to improve water quality outcomes.

In the final chapter I investigate the greenhouse gas (GHG) emissions associated with the use of HWT technologies in Mexico. I do that by carrying out a literature review of existing studies assessing energy use of water treatment technologies; using secondary data to perform a life cycle assessment (LCA) capturing the embedded CO2 equivalent emissions of individual HWT products; and developing model to calculate a metric of GHG emissions per volume of water used (kg CO2 eq/m^3) representative of the HWT sector in Mexico. Filtration, ozone, and UV disinfection technologies resulted in similar LCA emissions, while reverse osmosis had emissions five times higher than the average of the rest. I also find GHG emissions of HWT to be 30 times lower than water bottled in 20 L reusable containers. In a context in which mortgage institutions have created green credit mechanisms, this result is useful for expanding financing options for HWT products, which are often more cost-effective than bottled water, but require a higher capital investment.