interpretation of minimal risk and, therefore, shares in its uncertainty or confusion.
That the “minimal risk” standard invites variable interpretation has long been clear. A 1981 article reported on a survey of pediatric researchers and department chairs that found variations in how they applied the risk categories in proposed federal regulations to several procedures used in pediatric studies (Janofsky and Starfield, 1981). For example, when asked to assess tympanocentesis (puncturing of the ear drum) in children up to 1 year of age, 46 percent of respondents classified the procedure as involving a minor increment over minimal risk, 40 percent thought that it involved more than a minor increase, but 14 percent thought that it posed minimal risk or less. The level of agreement on a risk category typically was less than 70 percent. An accompanying editorial called for better standards of risk assessment in children’s research and suggested (without results) that a task force of the American Academy of Pediatrics develop consensus opinions about the risks of different procedures (Lascari, 1981).
Some 20 years after the study by Janofsky and Starfield was published, a survey of chairs of IRBs again found variation in the assessments of risk that a variety of research procedures presented to children (Wendler, 2003; Shah et al., 2004). For example, when asked to assess the risk presented by a confidential survey of sexual activity, 29 percent of respondents classified it as presenting more than a minor increase over minimal risk, whereas 44 percent thought that it presented minimal risk. Fifteen percent of respondents classified a blood draw once a week for 6 weeks as minimal risk, but 32 percent classified it as more than a minor increase over minimal risk. Eighty-one percent judged a single blood draw as minimal risk, but 17 percent labeled it as a minor increase over minimal risk. For research involving lumbar puncture without conscious sedation, 32 percent of respondents classified the research as involving a minor increase over minimal risk when it involved ill children and 6 percent classified it as minimal risk; for healthy children, the corresponding figures were 16 and 2 percent. The respondents might have shown greater agreement if they had been reviewing real protocols, and it is not clear whether respondents were giving their own views or the views that they thought others held. Also, the study did not determine the nature and extent of the experience of the IRB chairs with protocols involving children. Nonetheless, the results still point to the considerable subjectivity of risk assessments.
Chapter 8 cites analyses of actual IRB determinations that show similar variability in risk categorization. Most involve determinations reached by multiple IRBs reviewing the same protocol for multicenter trials. This variability in risk categorization raises ethical concerns about whether children are being appropriately protected from research risks. It also presents practical problems for those involved in designing and implementing important