Five hundred utilities in the United States currently provide drinking water with unsafe levels of arsenic, according to the Environmental Protection Agency. But determining the actual number of people who are getting too much arsenic in their water is much less clear, according to a study conducted in part by the Columbia Water Center and recently published in the American Water Works Association.
Arsenic, a trace element, has been regulated by the EPA since the passing of the Clean Water Act in 1972. It is, nevertheless, still the most persistent and widespread contaminant of drinking water in the United States. At high doses, arsenic is acutely toxic; for centuries it was known as “the king of poisons” for its use in murders and suicides. At lower doses, chronic exposure is associated with a host of ailments affecting almost every human organ, as well as promoting cancer of the bladder, lungs, skin, kidney, nasal passages, liver and prostate.
According to Katherine Alfredo, a scientist at the Columbia Water Center and a lead author of the study, while many utilities have water with arsenic levels above the EPA limit of 10 micrograms per liter (µg/L), determining the actual population-based health impacts of arsenic in the U.S. water supply depends heavily on how contamination data is handled and interpreted.
In 2001, the EPA revised its arsenic rules, replacing its old standard of 50 µg/L with a requirement that utilities keep levels below 10 µg/L. While most utilities have complied with the rules, a number of smaller utilities in areas with high levels of naturally occurring arsenic haven’t met the new standards, usually because of the cost of arsenic treatment.
“There are a bunch of different thresholds that are out there,” says Alfredo. “There’s the maximum contaminant level, which is the federal regulatory level. Then you have the maximum contaminant level goal—that’s purely health driven. So for arsenic, that’s zero. Arsenic—any amount—is toxic. So if you could treat to zero, you would want zero.”
But she says, treating to zero µg/L is extremely difficult if not impossible. “They could [set the limit] at zero, and then everyone would be in violation. Or you could set it at something that will protect human health, but will also not bankrupt all of our utilities.”
To better understand just how much arsenic was in water systems, and where it occurred, Alfredo and her collaborators retrospectively analyzed two different EPA datasets, from before 2001 and after.
The first set, the Arsenic Occurrence and Exposure Database, or AOED, consisted of a compilation of state compliance-monitoring databases, included only 25 states (whose monitoring the EPA thought suitable) with different reporting limits. Because it devised its new rules in 2001, the EPA was forced to rely on this limited dataset to set a threshold for the regulation of arsenic.
The second database, from between 1999 to 2005, is more comprehensive, including the occurrence of 69 drinking water contaminants from 45 states. By comparing the predictions the EPA made with this more robust dataset, as well as with actual reported violations, Alfredo’s team tried to see how well the agency was able predict the number of utilities that would be out of compliance. “What we were trying to look at was, how well did the EPA do” using a much less complete dataset?
According to Alfredo, even though the first dataset was limited by detection limits and coverage, it provided a fairly accurate prediction of how many utilities would be out of compliance with the revised arsenic limits. “After doing this and seeing what they were grappling with … looking forward, to see how many utilities are out of compliance… they actually handled little data very well.”
However, when it came to the question of how many people were actually impacted by lack of compliance, it was a different story. “When we switched to a population basis—what’s the population that’s going to be impacted—the numbers varied dramatically,” depending on how the data was analyzed.
In particular, results depended heavily on how the samples at a single utility were handled. In the end, each utility, regardless of the number of samples reported, was to be represented by a single arsenic concentration for the analysis. While the EPA used average values to represent arsenic concentrations at a single utility, Alfredo’s research looked at the average, 75th percentile and 95th percentile values in their analysis. Using these three different scenarios, Alfredo’s team came up with very consistent predictions on the number of utilities that were likely to be out of compliance, but wide variations in the actual number of people affected.
“Looked at on a population basis—and this is where it really matters—you’re making sure that you represent small utilities, and that you represent large utilities; that you have this spread that represents the actual population, where we can have these differences, and see who’s actually impacted. The fact is that most of the health-based analyses are done on cancers avoided in the population,” Alfredo explained. “So if we’re assessing the thresholds based on populations that are going to be improved due to cancer reductions, then we need to also assess how many of these people are actually impacted. So say there are 500 utilities,” out of compliance, “but they all serve five people—is that really a problem? I think that was really one of the main drawbacks to [the EPA’s] analysis.”
At the same time, says Alfredo, the agency seems to have underestimated how well utilities—especially small utilities—would be able to handle the new regulations in terms of cost. “Arsenic is the top non-compliance issue for the country. So a lot of those who are non-compliant are small utilities, using groundwater, not really any treatment, and are right around 10 µg/L. So you have to wonder: If the EPA had set a limit of maybe 20 µg/L, maybe it wouldn’t be such a national issue. And what would be the difference in health impacts of that? So that’s one of the lingering questions.”
“The other lingering question,” says Alfredo, is why, despite various funding mechanisms available, so many utilities are still struggling to come into compliance. “If we look at just the case of California, they have three different funding mechanisms, and they still have the highest number of violations,” she says. “They still have utilities that are either trapped in the process, or just can’t come into compliance with the funding that’s out there and available. So something is wrong with either how we’re funding these projects, how we’re asking utilities to deal with these issues, or, the levels that we thought were acceptable given the treatments that [are] out there, or the technology that’s available.”
So how can agencies like the EPA develop more realistic regulatory limits that still protect human health? In their new research, Alfredo and her colleagues are working on aggregating the health impacts of multiple contaminants from a given water source to determine how to regulate each together to provide the greatest health benefit for the cost, given population affected and current technology.
The project, Alfredo says, “was started to really help utilities. So if I’m a utility operator and I’ve got a little extra cash this year, and I want to put that cash towards one of these contaminants, or improving the health of the people who my water serves … I can compare and say okay, I have some nitrates in my water, I have some fluoride in my water, and I have some arsenic in my water: Where should I put my money? So you could look at it on a health index basis, and then move to cost.”
However, “after diving into all these regulatory determinations, we realized that … we can start to have a discussion within the government where you say oh, you’re putting all this effort into nitrosamines,” for example, says Alfredo, “but we’re not benefitting people as much as we would if we brought arsenic into compliance.”
“That kind of tradeoff is not necessarily being done when they talk about regulations. They talk about one regulation at a time.”
