In the late 1990s, scientists studying children’s health pondered crucial questions they couldn’t answer: Conditions as diverse as asthma and autism were increasing in prevalence, with no clear reason why. Many suspected that a child’s early environment—even exposures in the womb—were connected to medical problems that manifested years later. For example, is risk of asthma influenced by the stress a mother experiences during pregnancy? What role does air pollution play? What about diet? Those links proved difficult to study because by the time a child shows signs of asthma, it’s too late to take a blood sample during pregnancy, or analyze the air the newborn breathed.
To address those challenges, leading pediatric researchers in the U.S. envisioned an ambitious study. They wanted to track 100,000 American children from before birth until the age of 21 by collecting detailed data, biological specimens such as blood and urine, and environmental samples, including dust from childhood homes. In 2000, Congress authorized the National Institutes of Health (NIH) to go ahead with the project, which would be called the National Children’s Study.
In the 14 subsequent years, the government has spent $1.3 billion, working with hundreds of scientists at dozens of sites around the country. There’s little to show for the effort. While some 5,700 children have been enrolled in pilot studies since 2009, researchers stopped collecting data on Dec. 12, when the NIH concluded that the project could not be salvaged and moved to shut it down. Families that had signed up received letters saying their contributions would no longer be needed.
“The goals of the study were laudable and they remain laudable,” Francis Collins, NIH director, says. “Most of us believe it should now be possible to accomplish those goals at a substantially lower cost and higher efficiency.”
How did the plan to track a generation of children unravel? There’s no single answer, according to interviews with more than half a dozen researchers and officials involved in the project, along with a review of public documents. A few explanations stand out. Science and technology overtook the study’s designers, who weren’t able to incorporate developments such as social media or data collection using mobile devices into their plans, Collins says. Funding for the full study was delayed for years as researchers argued over its design. Early estimates low-balled cost, leading to finger-pointing that threw the study into turmoil. And a toxic rift between NIH administrators and scientists at the 40 universities and hospitals conducting the study undermined the project.
“I’ve watched it with great interest, with rising and falling hopes—sort of phases of hope and despair,” says David Savitz, vice president for research at Brown University, who advised on the early design of the project and led a pilot site at the University of North Carolina until 2005. Researchers spent years debating which questions to investigate, Savitz says, leading to ballooning ambitions that were never curtailed. “At some point, you have to come to grips with reality: This much time, this much money, make it happen,” he said. “It never happened until they realized that reality was incompatible with where they were headed. That was last week.”
The study originally aimed to trace the influence of environmental factors—anything from exposure to pesticides to socioeconomic status—among a random sample of children who statistically represented the U.S. population. By having a huge number of participants, scientists could get useful data on even relatively uncommon conditions. For example, cerebral palsy affects just a handful of every 1,000 babies born. In a sample of 100,000 births, researchers would expect to see a few hundred cases of the neurological disorder, potentially enough to yield meaningful insights into its causes.
Most of the $1.3 billion spent on the National Children’s Study went into research at pilot sites. “The moment you start doing these trials where you’re going door to door, that’s a team of people that need to be hired—and need to be hired in several parts of the country,” says Russ Altman, a professor of bioengineering, genetics, and medicine at Stanford University who co-chaired an NIH advisory panel that recommended the study be canceled. “That gets incredibly expensive, incredibly quickly.” While about 100 scientific papers based on that pilot research have been published, many of them focused on methods of study design and data collection, rather than on the underlying medical questions the research was meant to address.
Additional tensions strained the National Children’s Study from inception. Debates over how to recruit women early in pregnancy—or even before conception—carried on for years. The initial approach—to reach potential participants in select neighborhoods by knocking on doors—proved difficult and costly, compared to enrolling women through their doctors or at hospitals. Researchers also went back and forth as to whether the investigation should begin with clearly articulated hypotheses or serve as a broader “platform study” that collected data for future scientific inquiries. The risk with the latter approach, scientists warned, was that crucial measurements would be omitted because investigators didn’t plan which questions to answer ahead of time.
For the seven years after Congress ordered the study in 2000, the NIH spent $54 million on planning. Lawmakers didn’t approve the much larger sums needed to begin recruiting participants and collecting data until 2007, when Congress approved $69 million to start research. But fundamental questions as to how the research would be conducted remained unsettled.
At the same time, revelations that the project could cost significantly more than the $3 billion that had been estimated for its lifetime angered supporters in Congress. Original director Peter Scheidt was accused of misleading lawmakers about cost and was ousted in 2009. A new director, Steve Hirschfeld, was put in charge, and several researchers blame a series of decisions under Hirschfeld for the study’s collapse. He instructed study sites to build local software systems, rather than using a centralized system for the entire project. As a result, each university had to be vetted separately according to federal security standards, which kept some sites from putting researchers in the field for almost a year, says Dean Baker, a researcher at the University of California-Irvine who ran a pilot site. Some investigators’ frustrations with management escalated in 2011 at the study’s annual meeting, when NIH administrators suggested a plan to change the target enrollment from 100,000 children to 250,000. Scientists were shocked. The cost of randomly selecting a sample that large would be prohibitive—while building a non-randomized sample would mean that the results could not statistically reflect the U.S. population and would likely curtail how much data could be collected about each participant.
Five lead investigators wrote a 16-page letter to Hirschfeld and his boss at NIH in September 2011, warning that “the present direction and conduct of the study places it at high risk of scientific failure” and suggesting ways to correct course. The warnings were never acknowledged, says Nigel Paneth, an investigator at Michigan State who signed the letter. “I think there was a scientific dysfunction at the heart of the study that wound up being played out in administrative dysfunction as well,” he says.
The NIH declined to make Hirschfeld available for an interview, and he did not respond to an e-mail or phone message seeking comment. Collins says he isn’t interested in pointing fingers. There were already “substantial problems” when Hirschfeld was appointed to lead the study, he says. “Certainly, though, as the leader of the enterprise, he carried a substantial amount of the responsibility to try to turn this into something workable,” Collins adds. “It may be that this particular project was more complex and demanding in ways that he was not prepared to tackle.” The NIH director also says lawmakers may share some of the blame: “There was an effort initially put forward by the Congress which maybe from the start was likely to run into serious trouble,” he says.
A Congressionally mandated review of the study by the National Academy of Sciences, published in June 2014, expressed “deep concern about the overall leadership and management of the NCS.” But Baker, the only lead investigator from the study asked to sit on the review committee, says the group didn’t conclude that the enterprise was beyond saving. That was a decision the NIH made. “We know a study of this nature is feasible, and we even identified a pathway,” Baker says. “The NIH leadership decided to ignore those recommendations,” he continues. “Everybody agreed there were issues scientifically—in management and design—that needed to be addressed, and they decided not to.”