Beautiful Nordic sunset in Copenhagen’s Nyhavn waterfront

Can we know if water is safe in real time?

UNICEF and the world’s water experts take on the challenge

Everyone who works in the water sector would love to have a quick diagnostic test for fecal bacteria, an indicator for the most common drinking water contaminants. The fantasy technology would be a dipstick, similar to the home pregnancy test, that provides instant results showing that the water is safe. But the road to better microbial water quality tests has been paved with high profile failures. Beginning in 2007, Gates Foundation funded the Aquatest project that spent over $10M and still has no product on the market. In 2011, the Stockholm Junior Water Prize was awarded to a high school student’s unverifiable rapid E. Coli test. In 2012 and 2013, WaterCanary took the TEDx circuit by storm with a nonexistent little yellow box that was purported to instantly detect cholera. The truth is, the reason these efforts have failed is that the science involved in what they are trying to do is very complicated.

Meanwhile, some donors and institutions have taken the opposite direction of establishing water research labs using traditional gold-standard detection methods. The bulk of the test equipment for water utilities in countries wealthy enough to afford well-regulated public water supplies is provided by well entrenched biotech companies that have invested in costly regulatory approvals for their products, resulting in a high entry bar for any potential innovations. This has resulted in many pristine but un-stocked and under-resourced labs in developing countries.

Given the challenges and the recent history of poorly conceived approaches, I was pleased to see that Innovation Group within UNICEF’s Supply Division, who typically provide emergency relief supplies, decided to convene an open meeting dedicated to technologies for rapid E. coli detection. The one-day meeting was held in Copenhagen in November 2016. In attendance were almost all of the serious firms in this industry, as well as academics and technical experts such as myself, to find out whether a faster, cheaper E. Coli test might be possible soon.

UNICEF Supply Division in Copenhagen, where pallets of emergency relief supplies are packed and sent all over the world

UNICEF’s ambitious targets

The UNICEF approach was to first develop a publicly available Target Product Profile (TPP) that explains the need for E. Coli testing, the number of tests they might conceivably buy in the future, and a set of minimum and ideal performance requirements. The TPP is ambitious, targeting a system that produces results in less than 3 hours as opposed to the current 24 hours at a cost of less than $1 per sample. Their needs were driven by three different use cases:

  1. Regulators — the traditional market for fecal indicator bacteria testing is water quality regulators or utilities. Findings from the Aquaya Institute show (paywall) that only 54% of the required testing in several African countries is actually performed. Clearly, in order to provide safe water there is a need for cheaper and simpler tests, suitable for less than ideal laboratory conditions.
  2. MICS surveys — this massive UNICEF household survey program is the main way that we know key health and demographic data in many countries lacking the capacity to accurately monitor these things. Recently, UNICEF and WHO have added a water quality module to these routine surveys, allowing, for the first time, to generate nationally representative snapshots of the of the quality of the water people actually drink. The water quality add on is popular since it helps address the SDG target for ‘safe and affordable drinking water.’ And this approach will be adopted in many new countries next year.
  3. Behavior change communication — UNICEF believes that sharing the results of a positive test with household members might motivate them to seek out better water sources, or invest the time and money in treating their water. For example, communities with safe water could receive special recognition. This could take the shape of a certification program, similar to the Open Defecation Free (ODF) status awarded to villages where everyone uses a toilet.

The industry response to these proposed requirements was not surprising to anyone who knows the water testing market. One large test maker said that they refer to these specs internally as the ‘holy grail’ of water testing. That said, most of the companies present expressed a high degree of openness toward working together with UNICEF to develop something better, as long as the resulting product could at least break even on the market. One challenge is that these requirements pose a number of tradeoffs to be made between performance and convenience. For example, if a rapid result is most important, it will likely come at the expense of sensitivity down to the WHO guideline of no detectable E. Coli in a 100 ml sample of water.

Microbiological water quality monitoring is a surprisingly difficult problem. Ranjiv Kush of Aquaya likens the WHO limit to being able to detect a single coffee bean in an olympic sized swimming pool. Right now, all of the methods in use at water treatment plants involve duplicating these coffee beans every 20 minutes (the time required for one E. Coli bacterium to divide into two) until exponential growth takes over and we have hundreds of millions of them, allowing us to visually detect their presence. Using specialized light sensing devices instead of the human eye could help, but the problem is that to see something small you either need to look at a very small portion of the water sample or somehow rapidly scan a large volume.

Other methods that require our coffee bean to come into contact with a sensitive surface (DNA microarrays, antibodies like those used in home pregnancy tests) are difficult because the chance that a single bacteria will come into contact with that surface is incredibly small. In case you are wondering why we can detect many chemical contaminants in water almost instantaneously, consider that even at the lowest possible detection limit of almost any chemical of interest, perhaps 1 part per trillion (ppt), there are still about 3,300,000,000,000 (3.3 trillion) molecules of of that chemical in 100 ml of water. You can imagine that with that many molecules bouncing around in the sample, they are constantly colliding with sensitive surfaces such as electrodes or with other chemicals that we add to help detect them.

The current state of water quality research

The meeting benefited from contributions by academic researchers who deeply understand these challenges and are familiar with possible technologies that might help overcome them in the future. They also helped to caution against overconfidence in the value of a single indicator such as E. Coli, as well as the meaning of a single test result when the microbial world is incredibly complex and variable.

Prof. Joe Brown’s infamous graph of daily E. coli measurements taken from a single source over the course of one year.

Joe Brown, from Georgia Tech, outlined some of the known issues with using E. Coli as an indicator of microbiological water quality but concluded in the end that for all its known flaws it is probably one of the best indicators we can easily measure. Kelly Baker, of the University of Iowa, made a strong argument for better data sharing, pointing out that many of the problems UNICEF wants to solve by creating a more rapid test, such as providing timely feedback to households about the test results, could be addressed through mobile communication technologies.

Joe also brought up a problem that I find to be a major flaw in the well-intentioned desire to provide test results back to individual households or communities: microbiological test results vary greatly from day to day or even sample to sample. Referring to his famous zig-zagging graph of the results from an E. Coli test conducted every day for a year on the sample water source, he invoked Heraclitus’ adage that ‘you cannot step twice into the same river.’ In other words, we have to remember that we are taking a small sample from a complex system that changes constantly due to both local and external factors. These factors include household water contamination due to poor hygiene, seasonal changes in water supply and quality, rainfall events, new construction or land use changes, infiltration from poorly constructed latrines and sewers, intermittent piped supply due to rationing, and too many others to list here.

Anyone working in international development absolutely has an obligation to share the data we collect with the people we collect it from; yet we cannot honestly tell a household or water user committee that their water supply is safe based on the results of a single test on a single day. In addition, even though we can all think of anecdotes where a household tells us that knowing that their water quality was poor influenced them to take action about it, most controlled studies have not found much evidence that providing water quality results causes a long term change in behaviors, like choosing to treat water. At best, this is a subject requiring further research, particularly concerning the benefits versus the cost of testing.

Where does this leave us?

Remembering that the ideal water quality field test might be something like the dipstick used for home pregnancy tests, the consultation confirmed that we are years, if not decades, away from something like this. Meanwhile, it is entirely possible that the rapid pace of development of molecular biology tools that directly measure DNA and RNA will surpass any R&D effort we undertake. Most of the concrete ideas on near term progress seem to center around either accelerating the existing process of incubating bacteria, detecting their presence earlier, or a combination of both methods. I would question whether a project to develop this kind of rapid technology would really be worth the effort. Is it really that transformational to get results in 3–6 hours rather than 24 hours? That would still be a long way from our 1 minute dipstick test.

Rick Johnston’s slide on the possible technological pathways to incubation-free E. coli detection

In light of all of the complexities and disappointments in the water testing world, here at mWater we are hoping that donors like UNICEF will absorb some of the risk that developing new technology for an unproven market entails. In the meantime, we have decided to focus on making better use of technology that is available right now to make water quality testing easier and cheaper. By working extensively with front line health workers, we realized that procedures which seem simple to microbiologists are not so easy for untrained people. We tried to simplify every step possible, arriving at our current kit that includes all of the supplies needed for two simple tests that measure E. Coli at the safe drinking and safe recreational water limits. This also simplifies interpretation of the results: safe for drinking, safe for bathing, or unsafe.

The biggest innovation we need right now is not in the speed or accuracy, but in the cost of testing. Given that even utilities and regulators are only performing a fraction of the tests required, based on the population served, any reduction in the unit price could translate directly into better monitoring.

UNICEF is moving in the right direction by convening an open consultation with real experts from within the UN system, academia, and industry. I look forward to seeing where this path leads and hope that there will be some near-term incremental progress, perhaps on cheaper reagents or simplified filtration methods. As we move toward widespread implementation of the SDG target for safely managed water, it is an amazing achievement that we are no longer talking about whether to test water for safety. Instead, we are talking about how to do it faster and better.