WV Chemical Spill of MCHM - doing the math on drinking water safety

The chemical spill of unknown amounts of crude MCHM, of unknown toxicity, and for an unknown amount of time has left a bad taste in my mouth. The whole thing stinks of too many unknowns. Over 300,000 people across nine counties have had their household water affected.

Just over 400 people have gone to the hospital with illnesses reportedly associated with chemical exposure and the subsequent health problems from lack of clean washing water. . And, these represent a small portion of the overall number of people that believe they’ve been chemically poisoned: The WV Gazette reported that, “the West Virginia Poison Center had received 2,302 calls about the chemical leak by Saturday evening…. Of those, 1,862 were human-related, 98 were animal-related and the rest were requests for information only.”

The government tap water advisory is not health-protective enough

Here I’m going to talk about the knowns and unknowns of calculating the health advisory level that is supposed to let people know when the water is safe to drink. In short, I’ve used the same data and the same assumptions that the government used for its calculation of a 1 ppm drinking water advisory, and it is 40X less protective than the data suggests. The difference is that the government used the mid-dose (100 mg/kg-day) of the study as a no-effect level, and I have used the lowest dose (25 mg/kg-day). Below I analyze the study results to show why this is justified. In addition, since the study reported toxicity even at the lowest treatment dose, I added an additional uncertainty factor of 10X. The number I calculate, 0.025 ppm for the general public including sensitive populations, is much closer to the government’s advisory for pregnant women. The government has advised pregnant women to avoid water until the chemical is no longer detected;  this means 0.01 ppm (1/100th of a ppm, or 10 parts per billion, ppb), which is the laboratory detection limit for this chemical.

I recommend that this value, which is 40X more health protective, be used not only for pregnant women, but also for nursing mothers, infants (especially those drinking formula made with tap water), young children, the elderly, and people with chronic illnesses.

The industry-sponsored study reports toxic effects at the lowest dose tested

The rat 28-day oral feeding study used four doses – control (no chemical), 25, 100, and 400 mg/kg body weight per day fed to each rat. Just five male and five female rats were in each dose group. The study authors – Eastman Chemical - state that the mid-dose of 100 had no effect, and therefore should be used to calculate a “safe” level for people. But, that’s not what the data shows. In the study, male rats had body weight decreases – compared with controls – in all dose groups including the lowest dose of 25 mg/kg-day and the mid-dose of 100 mg/kg-day that the Company called a “no-effect” level. The effect even showed a dose-response trend, increasing with increasing dose, but the trend was not statistically significant. In female rats, the low- and mid-dose groups weighed “slightly more than controls” but no specific measurements were reported and we are to trust the study authors that it was not statistically significant.

It is hard to get statistical significance in studies like this where so few rats are tested at each dose, given the natural variation between individual animals. In statistical-ese we say that the study is underpowered – its ability to detect an effect is very poor because the sample size is small. So, when such a study actually finds an effect, like this body weight decrease, despite its lack of power, it suggests a treatment-related effect. The chemical company may “overlook” the effect at low doses, but government scientists charged with protecting the public’s safety shouldn’t be so dismissive of the toxic effects seen at the mid- and low dose!

The study also reported that absolute and relative (to body weight) kidney weights were higher than for chemical-dosed male rats, when compared to controls. And, this effect was statistically significant for the lowest dose. But, the company dismissed it anyhow, because it didn’t follow a dose-response trend in the higher doses.

So, body weight decreases in the low dose were dismissed because they weren’t statistically significant, even though they showed a dose trend, whereas kidney weight changes were dismissed in the low dose because they didn’t show a dose trend, even though they were statistically significant.

I should point out that in this study no specific numbers were provided – the study only says whether organ and body weights were increased, decreased, or normal compared with control (no chemical dose) animals.  This is NOT a standard way of reporting data, and should not have been considered acceptable by government reviewers since it prevents independent experts from doing their own analysis.

Do industry-sponsored toxicity studies often underestimate risk?

Why did Eastman interpret the data in a way that underestimates the risk from exposure to MCHM? I've reviewed dozens and dozens of industry-sponsored chemical studies just like this one and overwhelmingly I have seen the pattern that when government or independent scientists review the same company-supplied data, they come up with different conclusions than the original company study.

In fact, in a famous poisoning incident in the 1990s' - right about the time that Eastman was conducting this MCHM study - dozens of people in California were sickened by pesticide-poisoned watermelon over a July 4th weekend, including some who  were hospitalized. Many of those who  were sickened had been exposed at levels the company said were "safe", but government scientists determined were way too high. (I published the story here)

Why don’t we know more about the health risks from MCHM?

The chemical 4-methylcyclohexanemethanol, or MCHM, is the main component of a mixture that is used to clean coal. MCHM was “grandfathered” in under the existing toxics law, called the Toxic Substances Control Act (TSCA), with almost no toxicity data, and no data required. The law grandfathered some 62,000 chemicals when it came into force in 1976, and did not require that EPA test them for safety or ensure that they met a standard of safety. And the law made it extremely difficult for EPA to require companies to produce data for any of those 62,000 chemicals, which is why nearly 40 years later, so little information on health and environmental effects of most of the chemicals in commerce is available to the public.  Those central problems of TSCA were laid bare when MCHM contaminated West Virginia’s drinking water, people turned to the government for answers, and none were available.

Crude MCHM, is described in more detail in its Material Safety Data Sheet (MSDS here). But, the MSDS sheet has a lot of blank spaces. I think Daniel Horowitz said it best, in this Washington Post article:

Horowitz of the Chemical Safety Board said that the safety data sheet for the chemical “has a great many fields which say ‘no data available.’ ” Under the section titled “most important symptoms and effects, both acute and delayed,” Eastman’s form says “no data available.” Under toxicological effects of inhalation, “no data available.” It was the same for whether it causes cancer, affects reproduction or affects specific organs. “There is very little available testing data on its toxicity,” Horowitz said.

Since the spill, the National Library of Medicine Hazardous Substances Data Bank has information on the health risks from exposure to MCHM, identifying the chemical as a strong skin irritant and a moderate eye irritant. That information was only made available after the spill, in response to the public demand for information. Eastman  Chemical Company has links to its toxicity studies on aquatic organisms, skin sensitization, and other relevant information and Q&A information. But, it’s all just the same few industry-sponsored studies. There is nothing on the potential for long-term health impacts or respiratory irritation.  It seems that no one, including Eastman Chemical that makes and sells the chemical, has done any actual studies of  potential long-term impacts such as potential reproductive, developmental, or cancer risks.

Why we need effective government regulation of toxic chemicals

An excellent report by WV Gazette reporter Ken Ward Jr explains why we need to do more than love our waterways – we need to actively protect them. He points out that, “Threats to West Virginia's rivers, streams and lakes aren't hard to find. Mountaintop-removal coal mining has buried hundreds of miles of streams. Waste from coal mining -- like the coal-cleaning chemical "crude MCHM" that leaked into the Elk River -- are pumped into huge slurry impoundments. The boom in natural gas drilling in the Marcellus Shale presents a host of new water-quality problems.”

My NRDC colleagues, Jon Devine and Daniel Rosenberg have written about the weaknesses and loopholes in the federal and state laws that should have prevented this predictable disaster.

The Chemical Safety Improvement Act – legislation introduced in Congress and touted as a  “fix” to the many problems of TSCA -- is strongly supported by the chemical industry.  But the bill as written is riddled with so many fundamental flaws, loopholes and landmines that it would actually be a step backward – preventing EPA from taking timely and effective action to test and ensure the safety of most chemicals, while taking away the authority of states to take action to protect their citizens, where the EPA has failed to act, or when the state wants to be more protective. (more detail here from NRDC’s Daniel Rosenberg) 

The biggest reason why there are so many loopholes in the regulation of hazardous chemicals is because the chemical manufacturers and polluters fought for and won those loopholes and why they continue to do so every day. When people start realizing that a polluted environment isn't just "icky", but it can sicken or even kill us – linking our health with our environment – then maybe we will get the regulatory oversight that is needed to protect our families and communities.

For those that like details, let’s do some math

With very little data to go on, the U.S. Centers for Disease Control and Prevention's (CDC) National Center for Environmental Health calculated the drinking water health advisory – a 1 ppm limit in drinking water - using a standard formula that EPA uses for a 10-day drinking water health advisory in circumstances such as these. The calculations are based on a child of 10 kg (about 22 lbs) that drinks 1 liter of water (about 4.2 cups) each day. The toxicity of MCHM was derived from a study sponsored by Eastman Chemical Company that makes MCHM, and was conducted in 1989. The Eastman study involved feeding the pure MCHM chemical (not the crude MCHM that actually spilled, which is described in the Material Safety Data Sheet  as a mixture of a few chemicals) to rats for 28 days to determine a dose that had no effect (called the No Observed Effect Level, or NOEL).  The NOEL dose level was then adjusted using uncertainty factors: a factor of 10 for extrapolating from a rat study to human risk, another factor of 10 to address differences among people in sensitivity to the toxic chemical, and a final factor of 10 to account for weaknesses in the toxicological database because there only one study that was relevant to estimating human risk (which is a darned big weakness in the database!).

The most relevant info I can find for this situation is the EPA Drinking Water Advisories. Among those, I think that the most relevant for this situation would be the 10-day Health Advisory. This document provides a template for calculation a 10-day health advisory (see pages 45/46). Here EPA uses standard calculations based on a 10 kg child that drinks 1 L/day, which is a standard assumption of child body weight and daily water consumption. This includes a 1000X UF (10 inter-species, 10 intra-species, 10 for using LOAEL instead of NOAEL). 

So, the CDC calculation looked like this:

(NOEL of 100 mg/kg/day x Body weight of 10 kg) / (Uncertainty factors of 1000 x 1 L of water/day)  = 1 mg/L or ppm

Whereas, the CDC calculation should have looked more like this:

Let’s assume that the lowest dose in the study, 25 mg/kg-day, can be used as a reliable measure of the lowest-observed-effect level (LOEL) because of the statistically significant increase in kidney weight and the measurably reduced body weight. This means we need to add an additional 10X uncertainty factor, following standard EPA protocol when the study only identifies a lowest-observed effect, and not a no-effect level (NOEL).  In this case, the calculation would look like this:

(LOEL of 25 mg/kg/day x Body weight of 10 kg) /  (Uncertainty factors of 10,000 x 1 L of water/day) =  0.025 mg/L or ppm, or 25 ppb

This value of 0.025 ppm is 40X more health-protective than the 1 ppm drinking water advisory now in effect.

About the Authors

Jennifer Sass

Senior Scientist, Health program

Join Us