How We Obtained the Government’s Data on Agent Orange and Birth Defects

The Veterans Administration refused to release what it had learned about possible links between birth defects and exposure to Agent Orange.

ProPublica and The Virginian-Pilot found a novel way to obtain the information under procedures historically used for scientific research by academic scholars.

For nearly four decades, many of the veterans exposed to Agent Orange have shared information about their health problems with the government. Officials questioned them about their exposure to the chemical, their illnesses and whether their children had birth defects. Over time, the results of these exams became a registry of health data on more than 600,000 veterans.

Those records could help answer a question that has long tormented vets and their families: Were the chemicals used to defoliate Vietnam somehow afflicting a second generation of Americans?

Over the years, researchers at the VA made no effort to analyze the data they had at hand, even as vets grew increasingly vocal about the possibility that their children’s ill health was somehow linked to their military service.

Last year, reporters at ProPublica and The Virginian-Pilot launched a comprehensive investigation of this issue. We began by listening to the veterans themselves. We asked veterans and their families to share their experiences. Eventually, more than 6,000 veterans filled out a detailed questionnaire and told wrenching stories about their health and that of their families. Many wondered if their children’s birth defects were linked to their exposure to dioxin, a toxic chemical in Agent Orange that has been conclusively tied to health problems in vets and others.

After we learned about the VA’s trove of data, we filed a request under the Freedom of Information Act asking that it be released to us, without any identifying information about individual veterans. Our hope was to move from anecdotes to statistics. If the servicemen’s suspicions were correct, the numbers would show that the children of vets exposed to Agent Orange suffered birth defects at a higher rate than vets who were not exposed to the defoliant.

The VA turned down our request, arguing that release of information in the registry “would constitute a clearly unwarranted invasion of a living individual’s personal privacy” — even though we did not ask for any identifying information.

We appealed that ruling and lost. The VA said it could only provide us with summaries of the data — and that was not useful for the analysis we had proposed. At this point, we could have filed what would have been an expensive and time-consuming lawsuit in federal court.

But along the way, we hit on a novel idea. What if ProPublica and The Virginian-Pilot asked the government for the same data under rules that allow scholars to conduct academic research? It was an approach we had never used before. There was also an obstacle. All scientific studies involving human subjects have to be approved in advance by an Institutional Review Board, or IRB. The IRBs play an important role in university research, protecting the rights and privacy of research subjects.

Most universities have their own IRBs; news organizations do not. We asked academics for suggestions and one pointed us to a commercial company, Schulman IRB, which could be hired for this purpose. The reporters drafted a formal protocol that spelled out in detail exactly what they wanted to do with the data, something people requesting information from the government under the Freedom of Information Act are not required to do. Journalists rarely use IRBs because they don’t want to cede oversight of their work to an outside body. But in this case, we felt comfortable with the additional scrutiny.

As part of the IRB protocol, we agreed that we would never attempt to figure out the actual identities of the veterans described in the data. The IRB did not review our findings before they were published. But we did promise to keep the data in a secure server with access limited to the project team. (FOIA does not allow the government to impose such precautions.)

The IRB approved our approach, which we forwarded to the VA. There, it was personally approved by the undersecretary for health. The data finally arrived in September and was analyzed by Hannah Fresques, a ProPublica fellow with a master’s degree in quantitative methods as well as Olga Pierce, our deputy data editor, and Charles Ornstein, the lead reporter on this project.

What we found was well worth the effort. Our first-of-its-kind analysis revealed that the odds of having a child born with birth defects during or after the war was more than a third higher for veterans who said they handled, sprayed or were directly sprayed by Agent Orange than for veterans who said they weren’t exposed or weren’t sure. Our analysis controlled for such factors as a veteran’s age and health status.

We ran our findings by five leading experts, who offered us feedback and suggestions. We are publishing our findings today as part of a story, as well as separately in a lengthy methodology. We are upfront about the limitations of our review, but despite them, experts still told our reporters that their results should prompt the VA to look much harder at the multi-generational effect of Agent Orange.

“It’s like a sign that says ‘Dig Here’ and they’re not digging,” said Dr. David Ozonoff, a professor of environmental health at Boston University and co-editor-in-chief of the online journal Environmental Health. “It raises questions about whether they want to know the answer or are just hoping the problem will naturally go away as the veterans die off.”

Stephen Engelberg is ProPublica’s editor-in-chief and served as founding managing editor from 2008-2012.

Comments powered by Disqus