Sustained Outrage

Revisiting the reporting on the latest C8 reports

On Monday, a commentator named Jon Entine, working for a group called the Statistical Assessment Service, or STATS, published a piece called “Toxic Alert: There’s a Killer, C8, Lurking in Your Kitchen, Says the Associated Press — Oops, Maybe Not!”

Mr. Entine alleges that I (and Vicki Smith from the AP) botched coverage of the recent release of status reports from the C8 Science Panel, the three-person committee investigating potential health effects of C8, also known as PFOA. His piece concluded:

Ward appears to embrace a conspiracy theory. He’s resistant to the likelihood that the C8 scientists are actually doing their job, independently, letting the ideological chips fall where they may. He and the AP’s Smith seemed determined to “correct” the public record on the C8 panel’s behalf, presenting a narrative they believe the scientists should have written, not what they actually found. The result was two misleading pieces, including the AP’s C8 Godzilla, unleashed into cyberspace.

I’m writing this post to address and respond to Mr. Entine’s allegations. I encourage you to first click here and read what he wrote, and then come back and read through this post. You can read my initial blog post on this story here and the Gazette print edition here. The AP story is available here.

We’ll take the findings of the C8 Science Panel’s study of liver function markers first. Here’s what Mr. Entine wrote:

The summary report on C8’s impact on the liver functions was highly anticipated because it was designed to address a question posed earlier in the research: what’s the relationship between increased levels of C8 and three markers, bilirubin, ALT and GGT. Analyzing data from 47,000 adults, the panel found no direct link between increased levels of C8 and bilirubin or GGT. Ward and the AP ignored these unexpected but scientifically significant findings.

Rather, they both led with the only sensational finding, an association, described by the panel as “small,” between increased levels of C8 in blood and ALT. The scientists also stressed the study’s structural “limitation,” which it wrote “makes it impossible to know whether PFOA can cause changes in liver markers.”

But the relationship between GGT and bilirubin is not as clear as he makes it sound. The Science Panel did not report that it “found no direct link” between those markers and C8. What they said was that neither of those two markers “showed a clear relationship with C8 exposure.”

Keep in mind: At this point, we haven’t seen the data the C8 Science Panel relied on. All we have is their summary “status report” filed with the Wood Circuit Court and posted on their Website. As we reported in a follow-up story on Sunday

In an email interview, panel member Tony Fletcher of the London School of Hygiene and Tropical Medicine said the detailed data about the other enzymes and C8 would be released once a formal scientific paper is published.

But, Fletcher said, the pattern of comparisons between C8 exposures and the other enzymes was “not completely flat,” but that the “pattern is less convincing” than with ALT.

While Fletcher said his personal conclusion is there is “little or no evidence of an association, he also said the results “are insufficient to prove an association, but also insufficient to conclude there is no association,” Fletcher said.

In addition, even the status report the Science Panel released wasn’t as clear-cut in arguing no impacts from C8 as Mr. Entine would have it sound. While he describes the association between ALT and C8 exposure as “small,” Mr. Entine truncated the quote. Here’s the entire sentence:

In conclusion, a small but clear linear association between PFOA and PFOS serum concentrations and ALT, a marker of liver injury, was observed in this large population.

Mr. Entine didn’t mention this further elaboration by the Science Panel:

This association with ALT is unlikely to be due to chance, as it is highly statistically significant. It is not explained by the other detailed information included in the statistical models — age, physical activity, body mass index (BMI), average household income, educational level, race, alcohol consumption and cigarette smoking.

What’s more, ALT is the most significant marker here, the one that is the most accurate in predicting liver disease and also the least likely to be influenced by other factors, such as alcohol consumption.

And, Mr. Entine left out any mention of this additional finding:

Increases in another similar chemical called PFOS, which was not released from the DuPont plant, were associated with increased levels of the two of the markers: ALT and bilirubin.

Contrary to Entine’s bottom line — that the Science Panel’s findings are good news regarding any potential association between C8 and markers of possible liver disease — the panel actually reported:

These results show a positive association between PFOA and PFOS concentrations and ALT serum levels, a marker of liver cell damage. This observed association between PFOA and PFOS with ALT is consistent both in terms of direction and magnitude with previous findings of occupational studies. Notably, these results are also consistent with data coming from a general population survey where background concentrations of PFOA are much lower compared to those reported here.

Dupont’s Washington Works Plant south of  Parkersburg.

Regarding the Science Panel study examining relationships between C8 and pregnancy outcome Mr. Entine started with this:

Ward and the AP also trumpeted the association between C8 and preeclampsia in the study of 12,000 local pregnancies. Once again, the panel noted this link was “small”—an informative qualification explicitly noted by all the reporters except Ward.

Actually, both my print story and my blog post quoted directly the portion of the Science Panel status report that found the association with preeclampsia:

… is small but clearly present based on a variety of ways of examining exposure.

The Associated Press story, by the way, also reported accurately that the preeclampsia finding was “small but clearly present.”

Mr. Entine continued:

Counter to expectations and contradicting prior less robust studies, the panel discovered “no associations” to the most feared outcomes: miscarriage, stillbirth, preterm birth, term low birth rate and birth defects. The AP at least mentioned those findings. In contrast, Ward ignored them.

Actually, in my blog post, I wrote:

Unlike some previous studies, the Science Panel did not, however, find associations between C8 exposure and reduced birth weight.

And in my print story, I wrote:

That same report, though, also said the panel’s work had not found the same association between C8 exposure and low birth weight identified in studies by other scientists who looked at other populations of newborns.

What Mr. Entine doesn’t explain is that the Science Panel’s work is not really a refutation of what may be the most significant previous study looking for relationships between C8 exposure and pregnancy outcomes. I talked to David Savitz of Brown University, the Science Panel’s lead researcher on these issues, and this is what we reported in Sunday’s paper:

Savitz cautioned that the new Science Panel report does not contradict the findings of a 2007 Johns Hopkins study that linked C8 exposure to low birth weights. The Science Panel only examined whether babies were above or below a certain weight of concern — 5.5 pounds — while Johns Hopkins researchers looked much more precisely at small changes in actual weight of babies compared to C8 levels in umbilical cord blood.

As for other birth outcomes examined by the Science Panel — miscarriage, stillbirth, pre-term birth and birth defects — Entine doesn’t make clear that just about the only previous study on these issues was a prior one by the Science Panel that actually did find an association between C8 exposure and birth defects.

Savitz told me that the newer study was much more reliable, as it included 10 times as many pregnancies. But again, is the issue as clear-cut as Jon Entine makes it sound? I asked Dr. Savitz:

“This is very strong evidence,” Savitz said. “But one study is one study. To be absolutely certain requires replication. I don’t want to say the case is closed. It’s not.”

In his commentary, Mr. Entine is especially troubled by the AP headline, “Panel: High death rates tied to heavy C8 exposure,” as well as the similar Gazette headline, “C8 panel says it has found cancer death rise at DuPont.” Mr. Entine wrote:

The Gazette and the AP headlines were over the top, signaling to the reader the crude sentiment: C8 kills. But that’s not what the panel concluded in its three two page summaries.

But he doesn’t discuss this study, concerning worker mortality at the DuPont Washington Works Plant, in any detail — at least not in his published commentary. In a prior email message to me, Mr. Entine explained his concerns about coverage of the worker mortality study this way:

The AP wrote that the panel concluded that “high death rates” were “tied” to heavy C8 exposure. You can come to that conclusion only if you’re determined to twist the overall findings.

Your key quote: “Significantly increased rates of death among more highly exposed workers … from kidney cancer and non-malignant kidney disease.”

The report said: “the only one [causes of death] which was significantly increased compared to the US population was mesothelioma, which was not “tied” to C8 in any way. You left this out.

In explaining the significance of this study, the panel stated that the overall death rate from cancer was 26 percent lower than in the overall population. Did you mention that? No.

Neither you nor the AP noted that increased risk was found in 25 cases, 12 of them cancers, out of 5793 people studied. It found there was NO overall risk of these diseases for all workers, but for a tiny group of “highly exposed workers” there was a “trend.” Was it linked to PFOA? The report makes it clear there is no way to know if that’s the case (I guess the word “tied” is so fungible that the AP could claim it wasn’t failing its readers, but if so that’s a journalistically cheap defense). The report concluded that kidney problems “could possibly be” due to PFOA, as “the kidney is a site in the body where PFOA is found.” But that’s not clear. It’s a working hypothesis at this point based on tiny fraction of cases, nothing more. Is C8 tied to these cancers? Not if words mean anything. Moreover, the report shouted to the reader that focusing on mortality is not the best way to study the potential effects. These studies could be over or under stating the effect.

In other words, the scientists report was screaming (to a diligent journalist) for you and other reporters to PUT THIS IN CONTEXT. Instead we get alarmist headlines like “high death rates tied to C8 exposure” which essentially ignored the framing presented by these independent scientists.

I sent my story and the C8 Science Panel’s report on this issue to several science reporters, academics and public health specialists I respect and asked them for a cold, hard critique.

A general theme of the responses I received was that I could have done a better job of explaining the context here. So, at the suggestion of one science reporter, I spent a fair amount of time last week working on the follow-up story that appeared in Sunday’s Gazette. Here’s what it had to say about the worker mortality study:

In its status report, the Science Panel said it found no increase in kidney cancer deaths or deaths from non-cancer kidney disease when it compared DuPont workers to the general U.S. population. Overall mortality rates among DuPont workers were also low, when compared to the general population.

Science Panel members explained that is to be expected, because of something scientists call “the healthy worker effect.”

“Almost all worker populations, when compared to the general population, have a lower mortality rate for all causes,” said panel member Kyle Steenland of Emory University, the lead researcher on the study. “Working populations don’t tend to have sick people.

“If you just compare the workers to the general population, you often see depressed or lower rates of mortality among the workers,” Steenland said. “If you want to pinpoint the effects on the mortality in the workforce, it often helps to compare workers with different exposures.”

That’s just what the Science Panel did, and the results “significantly increased rates of death among the more highly exposed workers compared to the low-exposed workers” for kidney cancer and non-cancer kidney disease.

The Science Panel cautioned that these results, while “statistically significant,” were also based on small numbers — 12 kidney cancers and 13 chronic kidney disease cases.

But the panel also said, “the increased risk for the more highly exposed in relation to malignant kidney and non-malignant kidney diseases could possibly be due to PFOA; the kidney is a site in the body where PFOA is found.”

Most epidemiologists will tell you that understanding and accounting for the “healthy worker effect” is key to doing a good study of worker health — or reporting on such a study for a mass media audience. There’s no question I could have — and should have — done a better job of explaining this issue in my initial story on the latest Science Panel findings. Like one longtime science journalists told me, anybody who has been in this business for long has written something under deadline and space pressures that, on reflection, they would have done differently.

I focused on the worker mortality study in my story, at least in part because of previous controversies over DuPont’s efforts to downplay such findings, despite advice to the contrary from their own science advisory panel. Still, looking at my story now, I wish I had included a few more paragraphs to put the worker-to-worker and worker-to-general-public comparisons in better context.

At the same time, Mr. Entine’s commentary is puzzling. Remember what he wrote about the story by Andrea Lannom of West Virginia Media:

She dispassionately outlined the panel’s central findings. Demonstrating sophistication in writing science stories, she noted the instances in which the scientists found fewer health concerns than expected—so-called null findings.

But then look at what Ms. Lannom’s story said about the Science Panel’s worker health study:

In its second report, the panel studied DuPont workers’ death rate, looking at more than 5,000 workers employed between 1948 and 2002 and also looking into 92 causes of death.

The release stated the only significant cause of death was from mesothelioma, a cancer caused by asbestos exposure. Additionally, the report noted when compared to the U.S. population, there was not an excessively high amount of reported kidney disease among workers.

Since many of these diseases may not be fatal, the panel conducted another study based on disease occurrence. The findings from this study will be released in 2012.

No mention there of the “health worker effect.” The story discusses only the comparison between all DuPont workers and the general public. It doesn’t note the more important comparison between higher exposed and less-exposed workers.

If Mr. Entine thinks reporters have “cherry picked” information in a way that inflates concerns about toxic chemicals, then he attacks those reporters’ integrity. But if reporters ignore important findings that suggest chemicals could be dangerous, that’s scientifically sophisticated journalism.

Mr. Entine also wrote this:

[Ward] told me he was irked by the panel report because it did not find the serious health problems that he had expected. “Why would the Science Panel downplay associations with adverse health effects in its [prior] report to the community … ,” he wrote. “It’s difficult to put this particular study into the context of the many previous C8 studies that have noted associations with adverse health effects at levels that the general public is exposed to, compared to high-dose exposures for workers.”

Note that the first sentence there isn’t in quotations, perhaps because I never said that. And the part that is in quotations is truncated. Here’s what that email I wrote to Entine actually said:

In that regard, you might go back and compare some of the previous C8 Science Panel press releases and the wording they used in those to the journal articles that were later published on those same studies. What you will find, based on what I’ve seen, is that the journal articles oddly enough use wording that is much more suggestive of serious health effects than in the press releases that the Science Panel puts out. I don’t offer any comment as to why that is. You might ask the Science Panel that question.

Why would the Science Panel downplay associations with adverse health effects in its report to the community, but make more pointed statements about those associations in its peer-reviewed journal articles? And, how does the change in wording affect what the public is told about what this research is showing?

Reporters have generally covered the science panel’s release of its summaries, but not followed-up with more coverage when the journal articles themselves were published.

There’s an inherent problem there for journalists: Do we cover the un-peer-reviewed summaries issued by the science panel in order to — as Judge Beane has said should be done — provide the public in this community as much information as soon as it’s available? Or do you think we should ignore those summaries and not report anything until it’s peer-reviewed and published in a journal? I would be interested in your thoughts.

There actually is a significant issue here for reporters trying to inform the public about the Science Panel’s work. One key tool in covering specific scientific papers is to contact other scientists who weren’t involved in the papers — but working in the same field — and ask them to comment, to help put the data, methods and findings in better context. When the Science Panel releases its status reports, that’s almost impossible. I found that out again in trying to follow up on the complaints Mr. Entine raised about my coverage. I contacted a handful of other scientists and asked them to comment about the most recent three status reports … one consistent message I heard was: These status reports don’t contain the actual data tables, so it’s hard for me to say much.

The Science Panel, of course, doesn’t want to release the data tables with the status reports. If they did that, their chances of getting their work on this project published in more prestigious journals would drop significantly.

Journalists could wait for the peer-reviewed papers. But the people of the Mid-Ohio Valley paid for this research, and aren’t they entitled to hear about it as soon as possible? The Science Panel could help here, by having more frequent press conferences and even having a public meeting once in a while where they discuss their research and answer questions.

Those very real issues about the place where journalism and public health intersect weren’t mentioned in Mr. Entine’s commentary.

As I understand it, Mr. Entine generally believes some journalists have pre-constructed narratives — and we’ll try to cram whatever facts we can that fit those narratives into our stories. But reading his commentary on the C8 studies, I wonder if he’s not really the one who does that.

For example, Mr. Entine makes much of his concern that corporations will be “stampeded into outrageous settlements based on fear mongering by advocacy groups and tort lawyers.” But in trying to frame the DuPont C8 story that way, Mr. Entine gets the basic facts and chronology of the story wrong. He writes:

In 2001, some residents and workers in Parkersburg stirred by a campaign launched by the Environmental Working Group and partnering with a national team of tort lawyers filed a class action against DuPont, accusing it of damaging the environment and human health. The suit appeared to have the makings of a billion dollar case: “contaminated” waterways and anecdotal stories of cancer suffering workers and their children.

That’s just not what happened here. The C8 story in Parkersburg started with Wilbur Earl Tennant and other members of the Tennant family, who were concerned that pollution from a DuPont landfill were harming their livestock. The Tennants sued DuPont in June 1999.

Their lawyer was not a national team of tort lawyers, but a young guy named Rob Bilott, from a corporate defense firm in Cincinnati. As Callie Lyons explains in her book, Stain-resistant, nonstick, waterproof, and lethal: the hidden dangers of C8, Bilott’s mother grew up in Parkersburg and he had fond memories of visiting the area. His grandmother referred the Tennants to him.

Bilott settled the Tennant’s case two years later, but not before he filed a lengthy and detailed letter with various regulatory agencies, outlining the findings of his investigation on the Tennant’s behalf. DuPont tried to get a federal judge to conceal the letter, but the U.S. District Judge Joseph R. Goodwin declined. A few months later, Bilott and the Charleston law firm of Hill Peterson sued DuPont again, this time seeking class-action status to represent thousands of Mid-Ohio     Valley residents whose water was contaminated by C8.

National advocacy groups and larger class-action firms got involved in C8 and Teflon issues later, and when they did they were using information that was dug up through Rob Bilott’s earlier efforts on behalf of the Tennant family.

Mr. Entine describes the C8 class-action that resulted in the settlement creating the Science Panel has having “stalled” in 2005 until the court “engineering a pioneering settlement.”

I’m not sure where he came up with that description. I had the C8 trial on my calendar for mid-October 2004, when a settlement was announced on Sept. 9. Sure, the lawyers had taken part in court-ordered mediation, but such mediation is standard operating procedure in West Virginia before cases head to trial. And the idea to use settlement funds to study C8’s health effects came from the lawyers — not from the judge.

Mr. Entine also tries to write-off or downplay a $10.25 million (not $10.5 million as he wrote) fine paid by DuPont to settle a Toxic Substances Control Act lawsuit brought against the company over C8. He says DuPont sued EPA “for not reporting blood sampling — an administrative issue.”

What did this case really involve? As I wrote at the time (subscription required):

Specifically, EPA alleged that DuPont never told the government that it had water tests that showed C8 in residential supplies in concentrations greater than the company’s internal limit.

Also, EPA alleged that DuPont withheld for more than 20 years the results of a test that showed that at least one pregnant worker from the Parkersburg plant had transferred the chemical from her body to her fetus.

Also, Mr. Entine writes of the settlement between EPA and DuPont:

But in a blow to the complainants, the agency reiterated its belief that human health risks from C8 are minimal or unproven and declined to list it is a dangerous substance.

What? EPA said no such thing in the TSCA settlement.

Perhaps Mr. Entine has that confused with what EPA said later, in March 2006, when it announced plans for a voluntary phase-out of the use of C8 by DuPont and others in the industry:

The use of PFOA in the manufacturing process does not mean that people using these products would be exposed to PFOA. The agency does not believe that consumers need to stop using their cookware, clothing, or other stick-resistant, stain-resistant products.

Those EPA statements focused on whether consumer products made using C8 were dangerous, not on any potential effects of the chemical itself.

And in the five years since those statements, EPA has become more — not less — concerned as additional research on C8 and related chemicals is published.

In January 2009, for example, the Bush administration issued a provisional health advisory for C8 in drinking water.  EPA scientists have found that carpets and carpet cleaners may be major sources of human exposure to these chemicals. The federal ATSDR has warned residents in the Parkersburg area about drinking C8-contaminated water (remember that the city of Parkersburg’s water plant does not treat to remove the chemical).

And the most recent scientific paper I’ve seen from EPA scientists, has this to say about C8:

Gestational PFOA exposure induced delays in mammary gland development and/or lactational differentiation across three generations. Chronic, low-dose PFOA exposure in drinking water was also sufficient to alter mammary morphological development in mice, at concentrations approximating those found in contaminated human water supplies.

As Marla Cone of Environmental Health News explained in writing about this study:

… Scientists with three federal agencies who studied mice exposed in the womb to a chemical used to make Teflon found delayed breast development and impaired lactation. The effects were found in the mice at the concentrations detected in the water supply of an Ohio town near a DuPont Co. plant that uses the chemical, known as PFOA. Water supplies are not routinely monitored for it.

“If human exposures in distinct populations are approximating those provided in this study, concern over human breast health and lactational competency are justified,” said the authors, led by Suzanne Fenton, a mammary gland expert at the National Toxicology Program.

In November 2010, EPA added C8 and PFOS to a list of chemicals that would be screened for possible endocrine disruption effects. And the agency continues to move in the direction of regulating C8 and PFOS in drinking water supplies.