National Children’s Study Advisory Committee 7 th Meeting
Thematic Sub-Group Breakout Meeting: Injury
September 15, 2003
Holiday Inn Select
Chair: Barry S. Zuckerman, M.D., NCSAC Member
Professor and Chairman, Department of Pediatrics, Boston University School of Medicine
Dr. Zuckerman convened the meeting, and after members of the Sub-Group introduced themselves, Ruth A. Brenner, M.D., M.P.H., NICHD, NIH, DHHS, summarized of the Injury Working Group’s activities to date:
- The Injury Working Group initially developed seven hypotheses. The Working Group received feedback on these hypotheses from the Interagency Coordinating Committee (ICC) and the National Children’s Study Advisory Committee (NCSAC). As a result of this feedback, the Working Group developed three broader, overarching hypotheses. All 10 hypotheses were presented to the NCSAC at its June meeting.
- At this point, the Working Group would like some clear direction from the NCSAC on what modifications are required in its hypotheses.
Review of Meeting Purpose
Members of the Sub-Group then discussed the purpose of this meeting.
Alan R. Fleischman, M.D., New York Academy of Medicine, and David C. Bellinger, Ph.D., M.Sc., Harvard Medical School, noted that they had provided comments on the Working Group’s hypotheses. Dr. Brenner indicated, however, that these comments had not been received by the Working Group.
Jan L. Leahey, Executive Secretary of the NCSAC, NICHD, NIH, DHHS, noted that the only feedback provided to the Working Groups was in the form of minutes. Participants agreed that the lack of feedback to members of the Working Groups is a weakness in the process.
Ms. Leahey explained that, unlike the other Thematic Sub-Groups, the Injury Thematic Sub-Group discussion would include members of the Working Group because feedback from the June meeting indicated the need for additional information in this Working Group’s hypotheses. Having the Working Group members present during the Sub-Group’s meeting would enable them to receive feedback directly from Advisory Committee members.
The Sub-Group agreed to review some of the hypotheses of the Injury Working Group to provide feedback on how the hypotheses should be modified. The Injury Working Group could then apply the suggestions made during this discussion to all of their hypotheses.
The following section summarizes discussion related to Injury Working Group hypotheses.
Physical Environment and Injury Risk
Dr. Brenner explained that this hypothesis was developed to meet certain criteria established for the Study, such as having public health significance and providing justification for a longitudinal study. The purpose of this hypothesis is to identify the characteristics of the physical environment that have significant risk and protective factors and how they are modified.
Dr. Zuckerman asked for an example of a specific hypothesis for unintentional injuries. Dr. Brenner suggested, "In neighborhood comparisons, traffic-calming devices are effective in reducing injury rates." The Working Group should define other specific hypotheses of this type. Dr. Zuckerman noted that because so many variables exist, especially around the physical environment, it is not clear how long a list of specific hypotheses the Working Group should develop.
Roderick Joseph A. Little, Ph.D., University of Michigan, asked whether unintentional injuries will be studied retrospectively, by asking Study participants at fixed times if they had experienced an unintentional injury since the last time they were surveyed. This would only be effective if the intervals between visits were relatively short. Dr. Brenner noted that the Mild Traumatic Brain Injury Workshop group that met on September 11–12 had recommended an active reporting system. Participants discussed whether such a reporting system might be needed for the entire Study.
John R. Lutzker, Ph.D., CDC, DHHS, pointed out the need for multiple sources of data on violence because no single measurement has demonstrable reliability. The most common measure of outcomes in child maltreatment is based on reports from child protective services (CPS), but these reports are subject to a host of biases. Dr. Brenner agreed that few epidemiological data are available on long-term outcomes associated with unintentional injury. Even incidence is not clear.
Dr. Fleischman stressed the importance of defining "injury" for each type of injury. Dr. Zuckerman agreed, noting that head injury, for example, might be defined by a report from the mother, a mark on the head, an x-ray, or a hospital record.
Collecting data other than self reports is especially challenging because of implications for consent and for complying with the Health Insurance Portability and Accountability Act. Participants agreed, however, that hospital/physical data will be needed for the Study, including injuries.
Dr. Fleischman noted that it will be more difficult to collect medical data on a probability-based sample than a center-based sample. Dr. Zuckerman predicted that such data will be easier to collect once electronic medical records are the norm.
Dr. Brenner explained that the Working Group developed a list of the neighborhood and household environmental factors associated with different injury types. Some of these environmental factors are risk factors, while others are protective factors.
In addition, neighborhood-level variables must be collected; this has implications for sampling. Dr. Fleischman added that on-site environmental surveys also will be required. If these factors do not tend to change, they may only need to be measured once.
Dr. Zuckerman raised the following questions:
- How will land use be defined, and who will be asked about it?
- How will traffic be defined and who will be asked about it?
- Where will presence of crossing guards be measured—on the child’s way to school or within a certain distance from the home?
- How will fire department response time be obtained?
- Where will the volume and speed of motor vehicle traffic be measured—on the child’s block?
- Do standard instruments exist?
- Have these kinds of measurements been done previously?
- How will the data ultimately be coded and analyzed?
These issues are appropriate to address in the Study, but the NCSAC needs to know how these data will be obtained. Dr. Zuckerman explained that the ideas are conceptually appropriate but cannot be implemented in their current form. In addition, the hypothesis needs direction (for example, "A poor physical environment as measured by X is a significant risk for injury type Y and this risk is modified by Z and younger children are more vulnerable").
Dr. Fleischman noted that the Social Environment Working Group might also be interested in these community-level measures.
Recommendations for the Physical Environment and Injury Risk Hypothesis:
- The hypothesis must be able to be proven or disproved.
- The hypothesis needs to make clear what is to be tested so that power calculations and other design issues can be addressed. For example, it should specify the variables involved, including environmental factors, family factors, and type of injury.
- Collecting epidemiological evidence about the incidence of unintentional injuries could be a goal of the Study.
- Ascertainment beyond the data collected through self-reports should also be addressed.
- The Ethics Working Group consider the need to collect clinical data (such as visits to the emergency room or doctor’s office).
- The Working Group should select and prioritize a subset of the variables listed to examine.
- The Injury Working Group should ask the Social Environment Working Group about what kinds of community measures that Working Group is considering.
Cumulative Effects of Repeated Mild Traumatic Brain Injury (TBI)
The Working Group initially decided to focus on mild traumatic brain injury (TBI) because its members believed that answers were available for severe TBI. The Working Group is now convinced that the Study will play a unique role in helping answer questions about severe TBI. The Working Group also decided to focus on cumulative effects because it had been instructed to select questions that could only be answered by a longitudinal study in a large population.
Dr. Zuckerman asked how TBI would be measured, since no valid or standard measures are available in this field. He wondered whether pilot studies or white papers would be required for each issue for which no standard measures are available. Dr. Brenner stated that, ideally, tests could be conducted immediately following a TBI, and measures could then be repeated at frequent intervals. It will be difficult for the Working Group to select measures without knowing how often measurements are to be conducted in the Study.
Experts in TBI will be consulted to identify appropriate neuropsychological measures for the Study. However, the baseline neurocognitive measures will be determined by the Development and Behavior Working Group.
Dr. Little noted that the measures of TBI only need to be collected on the subgroup that experiences TBI, as well as for a matched control group. However, Dr. Brenner noted that the antecedent measures must be collected on the entire sample, because it cannot be known a priori which Study participants will experience TBI.
Participants also discussed how to determine whether a child had an injury. It might not be sufficient to collect this information through retrospective maternal reports; instead, a surveillance system might be needed.
In reviewing the criteria for Study hypotheses, NCSAC members agreed that the TBI hypothesis addresses issues that are significant to public health and human development and are appropriate to a large prospective study. To strengthen scientific merit, a method of ascertainment must be identified. Measures of severity will be needed based on recall or medical records, but because children are likely to receive different types of imaging studies, determining severity may be difficult.
Dr. Bellinger noted that the NCSAC had previously expressed concern about disparities in access to health care and that means of ascertainment might differ by population subgroups. Dr. Brenner replied that the Mild TBI workshop is likely to recommend a pilot study to determine how well active ascertainment works. This issue is likely to pertain to other issues, in addition to injuries, because finding out about hospitalizations will be difficult if the intervals between Study contacts are long.
Participants discussed whether the hypothesis should be restricted to younger children because it might be impossible to collect the data on older children if the contact intervals are lengthy. Dr. Brenner explained that TBI is a major issue for adolescents, particularly those who participate in sports.
Recommendations for the Hypothesis on Cumulative Effects of Repeated Mild Traumatic Brain Injury (TBI):
- The hypothesis should be expanded to address both mild and severe TBI, and both repeat and single incidents of TBI.
- Rather than basing measurements on the intervals selected for the Study by the NCSAC, the Working Group should inform the NCSAC of the optimal measurement intervals based on the literature.
- Members of the Injury Working Group are more familiar with the injury literature than the Development and Behavior Working Group. Therefore, the Injury Working Group should provide guidance to the Development and Behavior Working Group.
- Add a fourth focus on parenting function. For example, a sub-hypothesis could stipulate that outcomes from TBI are worse among children whose families fit a particular profile.
Predictors of Child Maltreatment
Linda Anne Valle, Ph.D., CDC, DHHS, explained that child maltreatment is measured through parental reporting, child reporting (once the child is old enough), and data from CPS. Obtaining these data requires parental permission to ask questions of the child, as well as a relationship with CPS. Dr. Lutzker explained that some states have centralized reporting systems while reporting systems in other locales are on the county level. Dr. Zuckerman noted that parental permission would be required to access CPS data.
Participants discussed whether CPS designation of child abuse is too broad. Approximately 60 to 70 percent of abuse is neglect, and some reports are unfounded. The Study might be able to examine abuse in other ways, such as by identifying fractures that indicate physical abuse. Dr. Fleischman noted that other studies have used questionnaires for this purpose.
Moreover, punitive parenting may have different meanings in the various cultural settings in which the Study will be conducted. The impact of culture should perhaps be addressed by this hypothesis. Dr. Zuckerman framed the questions as, "Are we measuring the behavior or the context in which the behavior occurs? Which matters to child outcome?" The scope of the Study should enable better addressing the impact of parental attitude and history on parenting.
Some participants wondered if the outcomes of abuse and neglect are the same, and what the impact is on children once they reach adolescence. The outcomes of interest include functional outcomes (such as finishing school) and mental health outcomes.
Functional outcomes are also relevant to head and other kinds of injury. Dr. Zuckerman suggested that these outcomes "represent trajectories in any part of childhood gone awry."
The final proposal for the Study is likely to include hypotheses about injury, so the question is how to set up a system for measuring injury and determining what will be measured that will make it possible to address interesting hypotheses. Dr. Little expressed concern that hypotheses that are not linked to a design strategy are not very compelling.
Peter Scheidt, M.D., M.P.H, National Children’s Study Director and ICC Member, NICHD, NIH, DHHS, noted that at the last NCSAC meeting, concern was expressed about the ability to measure abuse and neglect because those participants identified as abused or neglected will not continue to participate in the Study. Including abuse and neglect in the Study could severely impair the relationship between Study personnel and potential subjects. Although Dr. Zuckerman indicated that, in his experience, parents will likely stay with the Study, Dr. Scheidt noted that this concern must be addressed directly.
Dr. Lutzker asked what would happen if, in the course of the Study, a danger to the child, such as lead, were identified. Specifically, if the danger were reported to the parent, and the parent failed to act on the warning, would Study personnel be required to report that parent for neglect? Dr. Scheidt replied that Study personnel would be required to report any newly identified risk of a known harm.
Similarly, whether Study personnel learn about abuse from the parent or another source, they have an obligation to report it. Dr. Fleischman has found that families in such situations do remain, as long as they have built a good relationship with Study personnel. Moreover, few children are currently removed from the kinship environment.
Dr. Little offered that if certain aspects of the Study might drive subjects away, these aspects could be optional. Dr. Fleischman pointed out, however, that Study personnel might still learn of the risk in question.
Recommendations for the Predictors of Child Maltreatment:
- The Working Group should justify explicitly the need for CPS data.
- Outcomes must be specified for each hypothesis.
- The Working Group should address the pros and cons of retrospective reporting versus a surveillance approach to help those designing the Study.
Dr. Fleischman raised the following questions:
- How will aggression be defined and operationalized for younger children and adolescents?
- How will children with childhood aggression be identified?
- Are standardized measures available?
- Will these questions be asked of everyone?
Measures will include parental reports of a child’s aggression (using scales of some type), magnetic resonance imaging, and genotypes. Ascertaining maltreatment and aggression will require direct observation of schoolyard, parent-child, and home interactions. Dr. Fleischman emphasized the need to make the case for direct observation because it is labor intensive and expensive. Perhaps it will only be required for a subset of the sample.
Measurements also will include teacher reports; the Study will need to interact with schools. Dr. Scheidt explained that the Study’s measures database includes proposed measures of the school environment, such as behavioral and physical exposures.
Recommendations for the Physical Aggression Hypothesis:
- Assessments should be conducted once in early childhood, once in latency, and once in early adolescence.
- Because it is unclear which instrument will be most sensitive to later aggression, the hypothesis should include several instruments.
- The Study will need to obtain academic records.
Functional Outcome of Injury: Influence of Antecedent Factors
Dr. Brenner explained that this hypothesis was added to the TBI hypothesis by the Mild TBI workshop. This hypothesis is broader, however, because it includes all kinds of injury and not just TBI.
Recommendation for the Hypothesis on Functional Outcome of Injury: Influence of Antecedent Factors: The Working Group should specify the kind of functional outcome and the kind of injury being studied. The same approach should be used for TBI.
Psychosocial Environment and Injury Risk
Dr. Zuckerman raised the following questions:
- What parts of the psychosocial environment are being referenced?
- What factors increase risk?
Dr. Bellinger noted that it is possible to examine risk factors while controlling for protective factors and vice versa.
As currently written, the hypothesis does not address the interaction between factors. Rather it only suggests that some factors increase and others decrease risk. Moreover, Dr. Fleischman noted that the hypothesis includes some interesting concepts (for example, supervision) because caretaker supervision has an impact on the likelihood of accidents.
Recommendations for the Psychosocial Environment and Injury Risk Hypothesis:
- The hypothesis should address controlling for protective factors while examining risk factors.
- The hypothesis should specify direction of the independent predictor in the regression model.
- This hypothesis should be coordinated with the Development and Behavior Working Group, which has several hypotheses about parental actions and practices.
- The hypothesis should clearly delineate how to measure parental and caregiver supervision.
Outcomes of Child Maltreatment
Dr. Zuckerman raised the following questions concerning sub-hypothesis A:
- Does "multiple" refer to two or more, three or more, or something else?
- How is "severity" defined?
- How is "maltreatment" defined?
- How are "moderate," "episodic," and "chronic" defined?
- How are "emotional/psychological," "cognitive," and "physical" defined?
- How and when will this be measured?
Dr. Little expressed concern about the intuitive nature of this hypothesis. Dr. Valle replied that past work on this issue has been of poor quality. Dr. Brenner added that what makes some children resilient to these outcomes is unknown. Dr. Bellinger pointed out that analyzing the question, as currently written, will not provide information about resilience.
Dr. Zuckerman characterized sub-hypothesis B as a better hypothesis that subsumes sub-hypothesis A. But sub-hypothesis B should be more specific by, for example, defining how the cumulative effects of maltreatment will be modified. Caregiver age and education, socioeconomic status, and family composition should all be given direction and definition.
Dr. Little noted that the hypothesis might be more compelling if it addressed some kind of controversy, such as an area in which two different theories exist. Corporal punishment might be such an issue.
Recommendations for the Outcomes of Child Maltreatment Sub-Hypotheses:
- The Working Group should clarify whether outcomes a–d in sub-hypothesis A are likely to be the same. If not, the Working Group should specify the order of these four groups.
- Resilience should be incorporated into the hypothesis.
- The Study should control for maternal depression or violence in the home, which might have an impact.
- The Working Group should consider deleting sub-hypothesis A.
- The Working Group should utilize current literature to inform the direction of the hypothesis.
- The Study should ask parents about their physical and nonphysical disciplinary strategies.
- The Injury Working Group should solicit input from the Development and Behavior Working Group because that Working Group is focusing on some of these same characteristics of families.
Dr. Fleischman asked Dr. Scheidt whether the Injury Working Group can address the needed specificity in the hypotheses through bulleted points, without having to enhance its narrative. Dr. Scheidt replied that this is up to the NCSAC. If, however, the hypotheses could be strengthened without developing further rationale, this would be useful to the Study Program Office in compiling the set of measures that will comprise the protocol.
Dr. Brenner asked if the Working Group should rework all 10 of its current hypotheses. These are presently organized into a few categories, such as aggression and maltreatment. Only one of the Working Group’s hypotheses addresses a specific type of injury (head injuries) because these injuries are an enormous long-term problem and require a different set of measurements.
However, less justification exists for addressing other types of injuries through specific hypotheses. Setting up measures by cause of injury would not be helpful, although the data might be analyzed by cause of injury.
Dr. Zuckerman explained that the NCSAC Members attending this meeting accepted the aims of what the Working Group wanted to assess, but the NCSAC also was asking the Working Group to present these criteria as hypothesis terms, consistent with the format for any funded scientific study.
Drs. Zuckerman and Fleischman pointed out that some of the hypotheses in each category of injury would be in the final set for the Study.
Recommendations for Next Steps:
- The Working Group needs to further delineate hypotheses with definitions and measurements.
- The Working Group should retain all of its current hypotheses, prioritizing hypotheses in each category but not across categories.
- The Working Group should do some power calculations to determine, based on the frequency of some of these events, what size odds ratios could be detected. Hypotheses could then be sorted by feasibility.
- All of the hypotheses should address measurements, timing, and definitions.
- Dr. Brenner will provide Dr. Zuckerman with the Working Group’s revised hypotheses.
- Drs. Zuckerman, Bellinger, Fleischman, and Little will then review the hypotheses, and Dr. Zuckerman will pass the revised hypotheses to the NCSAC.
- Dr. Brenner will write a brief summary of this discussion and share it with her co-chairs before distributing it to the entire Working Group.
Thematic Sub-Group Members
Chair: Barry S. Zuckerman, M.D., Boston University Medical School
David C. Bellinger, Ph.D., M.Sc., Harvard Medical School
Alan R. Fleischman, M.D., New York Academy of Medicine
Roderick Joseph A. Little, Ph.D., University of Michigan
Peter C. Scheidt, M.D., M.P.H., NICHD, NIH, DHHS
Injury Working Group Members
Ruth A. Brenner, M.D., M.P.H., NICHD, NIH, DHHS (Co-Chair)
John Lutzker, Ph.D., CDC, DHHS (Co-Chair)
Gitanjali Saluja, Ph.D., NICHD, NIH, DHHS
Linda Anne Valle, Ph.D., CDC, DHHS (At-Large Member)
Observers and Other Participants
Jan L. Leahey, NCSAC Executive Secretary, NICHD, NIH, DHHS