After We Tried to Correct Claims About 'Deadly' Water Filters in Flint, We Were Accused of Scientific Misconduct — And That Was Just the Beginning

Retraction Watch Staff

May 19, 2022

The Sept. 10, 2019 PBS article accompanying the Frontline documentary "Deadly Water" was topped by a provocative headline: "The EPA Says Flint's Water is Safe — Scientists Aren't So Sure." The PBS story relied on a study of adverse health outcomes for people given point-of-use (POU) water filters during the Flint Federal Emergency.

We were astonished. Several of us worked closely with residents to first expose the problems with lead and Legionella that defined the Flint Water Crisis. We were supportive of later humanitarian efforts to provide Flint residents with the free point of use (POU) lead filters, since they effectively remove lead from water used for drinking and cooking. These off-the-shelf water filters are routinely used in about a third of U.S. homes, so we were mystified as to how they could have wrought such devastation when deployed in Flint.

We turned to the scientific study that PBS had cited, published in the International Journal of Infectious Diseases (IJID) in early 2019. The study, by Gina Maki and colleagues at Henry Ford in Detroit, claimed that the human test subjects in Flint had a 70% rate of severe pneumonia, 10% rate of sepsis, and a 10% rate of folliculitis and half died from "bacterial infections" – implied to be Legionnaire's Disease – as a result of using the POU filters. Such results —if they were accurate – would indeed be shocking. The PBS Frontline article and the associated IJID study called into question the competence of the government agencies who supplied the filters.

We critically reviewed the article, which consisted of just 350 words and one table. There was little explanation of methods, but we did our best to dissect the logic supporting the conclusions.

The timeline was especially perplexing. Specifically, the 2014-2015 Flint Legionnaires Disease outbreak that killed at least a dozen people from pneumonia ended around September 2015. But the high lead in water was not widely recognized until early September 2015 and the POU filters were not widely distributed until later. How was it possible that the POU filters distributed to Flint residents after September 2015 could be responsible for a 2-year Legionella outbreak that had already ended?

The Frontline article also noted that PBS reporters were responsible for collecting the water samples from residents that were used in the Maki et al. study. This also struck us as highly unusual.

We decided to seek assurance that the POU filters were installed in homes of Flint residents before the infections described in the paper occurred. We also wondered how a human subjects study with PBS reporters collecting water samples, involving a total of 10 homes using filters in Flint and one control home using a filter in Detroit, could be approved by a Institutional Review Board (IRB). Such a review should have required assurance that a human subject study design was sound and supported the generalizable conclusions.

We corresponded with IJID editors, determined a mutually agreeable course of action, and decided to write a letter to the IRB at the Henry Ford Medical Center (Henry Ford IRB) to inquire about specifics of the study and the approval process.

o our surprise, the day after we sent our inquiry to the Henry Ford IRB, two of us (Edwards and Pruden) were accused of human subject misconduct – including a failure to obtain IRB approval for our own early work in Flint that exposed the problems with lead and legionella.

It turned out that Henry Ford's Marcus Zervos, the senior author of the IJID study, had somehow learned about our query to the Henry Ford IRB. He immediately wrote a letter alleging the scientific misconduct to the Department Chair of Civil and Environmental Engineering at Virginia Tech.

Additional allegations about our misconduct were then made to Virginia Tech IRB. Edwards and Pruden only learned about the misconduct allegations, because Zervos made accusations to other individuals at VT not involved with VT-IRB. VT-IRB maintained strict confidentiality about the concerns raised , as required by IRB protocol.

After a thorough investigation conducted over the next 5 weeks into the allegation made by Zervos against Edwards and Pruden, it was determined that they had obtained appropriate human subjects approval for the research. There was no evidence of scientific misconduct.

It turned out that was the easy part. Getting answers to simple questions about the IJID study has taken two and a half years – and counting.

Following our questions about the involvement of PBS Frontline reporters in collecting water samples for the IJID study, the PBS article was quietly corrected. The original text read:

In 2018, a team including Zervos tested water filters from 10 Flint residents' homes that they suspected were infected, using samples collected by FRONTLINE.

The text "using samples collected by FRONTLINE" was suddenly deleted and the only indication of this change now appears discreetly at the end of the article:

[Correction: This story has been updated to accurately describe the 2018 study on water filters.]


In response to our questions submitted to Henry Ford IRB, IJID editor Eskild Petersen wrote us that "the corresponding author of the conference abstract, Marcus Zervos, has promised me to discuss your questions directly with you."

That promise was never fulfilled. None of the technical questions we submitted in writing were ever answered. Henry Ford IRB only replied with a simple statement that all necessary human subjects approvals to the study were obtained. They did not provide us with any documentation.

After our questions were ignored for a month, we asked Petersen to intervene. Peterson repeated that "the authors of the abstract in IJID you refer to promised to be in contact with you," and added: "If that does not happen I have no possibility to enforce that. The COPE [Committee on Publication Ethics] guidelines do not empower an editor to obtain the information you ask for from an author."

Petersen did offer us an opportunity to publish a letter to the editor expressing our concerns. But we did not see the point in writing such a letter until we had answers about the timing of the filter installation that allegedly resulted in the deaths of five Flint residents.

e discovered at least two other communications were sent to the IJID authors posing questions similar to ours about the research. Hernan Gomez and colleagues wrote a letter to the editor of IJID noting "an astounding 50% mortality rate in their study sample." They feared there had been a "selection bias" in selecting homes for the study and directly questioned the conclusion "that the use of the filters caused the deaths."

Gomez and his co-authors also expressed concerns about the possibility the IJID article conclusion would reduce "the use of POU filters, thereby unnecessarily increasing the risks of lead exposure for the residents of Flint."

The published response from the authors was evasive. They wrote that "only patients with suspected infections from waterborne pathogens were invited to participate," which seemingly confirms the selection bias. But they did not provide assurance that the allegedly deadly filters were installed before the infections occurred.

Instead, the authors cryptically explained that they did not "draw conclusions about deaths; rather, this information only characterized the status of participants." Well, in terms of the "status" of the participants, we were never confused about the fact they were dead. But the relevant scientific question was whether their deaths resulted from using the filters as stated in the IJID article.

Gomez and colleagues also expressed alarm "that the ethical standards for proper scientific inquiry have not been met by the [IJID article] authors." To which Maki and colleagues claimed that it would have been "irresponsible to NOT share microbial results" about the allegedly dangerous filters with Frontline and others.

Another researcher engaged in Flint, pediatrician Mona Hanna-Attisha, also e-mailed Zervos about the paper. "This paper is scary," she wrote. "5/10 people died. Is there more to it?…..How was this paper published -a sample size of 10 with such limited (and scary) info? I worry about the unintended consequences of this paper."

Zervos responded that "the 10 homes in Flint were from people that consulted with me about their illness" so it is not "representative of Flint." Zervos said he had "hopes of doing an evaluation of this that is representative of Flint" and that could provide a "more detailed look at reasons for pneumonia deaths."

By that point, the authors of the IJID paper had at least three opportunities to clarify the timing of the filter installation relative to the deadly infections, or to disavow their scientific conclusion that the filters caused the deaths of Flint residents. They did not do so. The journal editor claimed there was nothing further that could be done.

So we brought the case to the Committee on Publication Ethics (COPE).

OPE eventually determined that it was indeed "the editor's duty to follow up on concerns about published content…[including the>]"time at which the infections took place relative to the water filter usage, cause of death of the patients…" The IJID editor was then directed by COPE to resolve our concerns or face sanctions, which could include being kicked out of COPE.

The IJID editor eventually responded to COPE, saying they repeatedly "advised Dr. Edwards to talk to Dr. Zervos and his institution." He did not mention that we previously submitted written questions about the timing of the deaths to Henry Ford IRB and received no response. IJID further stated they adequately responded by publishing the letter from Gomez and colleagues, and that beyond that "we refuse to be used in a dispute that we do not understand."

You read that right: The IJID editors claimed to not understand, that if the POU filters were installed after the infections occurred, then the published IJID conclusion was invalid.

At that point, the IJID response was reviewed by three COPE trustee board members "who determined the case did not merit a formal sanction." They did recommend "feedback to the journal on the importance of alerting readers about concerns relating to the validity of published conclusions." The case was closed. We asked to see the feedback sent from COPE to the IJID, but we were told that this was private.

hile it seems that we may never know if the infections (and related deaths) occurred before or after the filters were installed in the homes of the test subjects, this case is not over. The PBS Frontline story and the Flint filter fears are a central issue in upcoming criminal trials in which State of Michigan employees are charged with willful neglect and other felonies.

The troubles all started when many of the IJID article authors formed a large research team called the Flint Area Community Health and Environmental Partnership (FACHEP) that received millions of dollars in research funding from the State of Michigan in 2016.

On the basis of dubious sampling, members of FACHEP started rumors that the POU filters were causing a shigella outbreak in Flint. The researchers later found out that their rumors were without basis, and the U.S. Centers for Disease Control and Prevention determined that the Flint shigella outbreak was likely spreading by conventional person-to-person contact, and not from using water filters. But many Flint residents still believe the FACHEP rumors that, "we have shigella because we washed our hands" and that the POU filters grew dangerous bacteria.

In late 2016, the team then tried to get a peer reviewed paper published warning about alleged dangers of the POU filters in Flint. One of the authors of this blog, Dr. Susan Masten, was a senior member of FACHEP and co-author of that paper, and she later contacted her administration with ethical concerns about possible contractual violations and co-authorship issues. This 2016 paper was rejected following peer review.

The fear-mongering that the shigella was coming from potable water, failures to promptly address human subjects issues, and other concerns caused a lot of friction between the FACHEP team and the State of Michigan, who was funding their $3.4 million dollar grant. The filter friction eventually led to the allegations of "obstruction of justice" by Michigan Department of Health and Human Services' Eden Wells and Nicholas Lyon who were overseeing the FACHEP grant. An "extortion" felony charge against Rich Baird, one of Governor Rick Snyder's advisors, was added for "threat<ing> to cause harm to the reputation and/or employment of [the FACHEP] leader."

After the felony charges were filed, we began to publicly document numerous scientific concerns about FACHEP's Flint filter research up to the publication of the PBS Frontline article in late 2019 through a series of blogs. As part of the standard COPE investigation agreement, we had to agree to stop blogging on the issue while it was under review.

The felony obstruction of justice charges were later dropped against Wells and Lyon, but the "extortion" charge for "threatening the reputation" of one of the IJID article authors still stands. The punishment for this offense can range up to 20 years imprisonment and a fine of $20,000.

All of this raises the possibility that there might be more learned about the PBS Frontline and IJID filter death story before the Flint criminal trials finally come to a close. To be clear, we want justice for the wrongdoing that brought on the water crisis in Flint as much as anyone else. However, confusion about the science may only result in more injustice, and perpetuate the sort of unethical behavior that created the Flint Water Crisis in the first place.

Marc Edwards and Amy Pruden are University Distinguished Professors at Virginia Tech. Sid Roy is a Research Scientist at Virginia Tech. Kasey Faust is an Assistant Professor at the University of Texas at Austin. Susan Masten is a Professor at Michigan State University.