An advisory committee that met yesterday to consider the design of the struggling National Children's Study (NCS) came down firmly in favor of one option: The study should recruit children from a geographic sample that represents the entire U.S. population. But whether the National Institutes of Health (NIH) will go with that plan isn't yet clear.
Proposed by Congress 12 years ago, NCS aims to enroll 100,000 pregnant women and follow their babies' health from before birth to age 21. In February, NIH announced that because the original plan to recruit women from certain addresses in 105 counties was too expensive, it was revising the design to cut 15% from the NCS's $193 billion budget. Instead of recruiting women from a probabilistic sample of the U.S. population, NIH said it might instead find women through doctors participating in health care plans, a so-called convenience sample of providers.
This proposal drew concern from many researchers and questions from Congress. Those who have run pilot studies at 40 "vanguard" centers that have enrolled thousands of women were also alarmed to learn that some sites would soon be transferred to a new contractor.
At yesterday's meeting of the advisory committee and ad hoc advisers in Bethesda, Maryland, Alan Guttmacher, director of the National Institute of Child Health and Human Development (NICHD), which oversees the study, said that NIH was looking for "an exploration by thoughtful, informed folks" of various sampling strategies. NCS Director Steven Hirschfeld later clarified that the role of the 105 counties had not yet been decided: "We don't know yet what locations will be in the sampling frame or not."
During the daylong discussion, committee members and others warned about the perils of a convenience sample of providers. "I'm very leery of a convenience sample no matter how convenient it may be," said Edward Sondik, director of the Centers for Disease Control and Prevention's National Center for Health Statistics. He said that a sample that doesn't represent the country's geographic diversity could miss important environmental exposures that affect health. Relying on health care organizations could also leave out certain groups such as undocumented immigrants and the working poor, others said. (One dissenting view came from committee member Jeffrey Krischer of the University of South Florida, Tampa, who suggested that the results could be "weighted" to reflect any population.)