APPROACH TO REVIEW

In the analysis that follows, we relied heavily on the National Academies report Ethical Considerations for Research on Housing-Related Health Hazards Involving Children (National Research Council and Institute of Medicine, 2005). This study concluded that compliance with the Federal Regulations for the Protection of Human Research Participants (45 CFR Part 46), while necessary, may not be sufficient for complex studies. History teaches that more may be required, at least in the court of public opinion. The adverse publicity, litigation, and ultimate settlement of claims arising from the study of partial lead abatement strategies conducted in Baltimore by investigators at the Kennedy Krieger Institute are illustrative. An appropriate metric for the NCS may be whether the study is acceptable in its entirety to the research participants and to society. Even though each individual step arguably poses no more minimal risk when considered independently, it is important to recognize that this study in the aggregate will be quite burdensome and intrusive for the participants.

CRITERIA FOR GIVING INFORMATION TO PARTICIPANTS

A major issue is what information about their own health and development participants will receive. The research plan recognizes that casting the NCS as a strictly observational study is not desirable and that providing information is important for its own sake as well as useful as a means of increasing retention of participants (NCS Research Plan, Vol. 1, Sec. 12.7). The plan states that participants will be given routine information, such as growth parameters and hematocrits. On a purely mechanical level, the plan needs to clarify whether assays will be conducted pursuant to the requirements of the Clinical Laboratories Improvements Amendments of 2007, since conveying noncompliant laboratory results that may influence participants’ health care behavior may actually be illegal.

With regard to assessments of participants’ health and development, the plan is to provide to participants with findings that are “clinically relevant and actionable,” with the caveat that results of genetic tests will be offered to participants and not automatically conveyed (NCS Research Plan, Vol. 1, Sec. 12.7.1). While laudable in many ways, this plan for sharing information warrants further development. The concept of “clinically relevant and actionable” needs to be defined. The research plan and the IRB draft document provide insufficient detail about how decisions will be made about what to disclose, saying only that these decisions will be made centrally to ensure uniformity throughout the country (IRB draft, Sec. 9.4). One measure of the uncertainty that is still present for this important aspect of the study is the statement that “the use of NCS fetal ultrasounds for identification of fetal defects remains under discussion” (IRB draft, Sec. 3.3).

An additional problem is that the point of the NCS is to identify previously unknown factors that influence the health of children. The critical but unacknowledged question, then, is what parents and children need to be told about emerging research findings. The research plan needs to address this issue and indicate whether the NCS plans to rely on standards that have been previously been articulated by such bodies as the National Bioethics Advisory Commission (1999). The NCS documents show little awareness of the vigorous ongoing debates about the scope of the obligation and the desirability of returning research results. For some, the proposed plan will not provide sufficient information to participants. Parents are often eager to get research results, regardless of their clinical utility or the availability of effective intervention. Many investigators have also argued vigorously that many or all research results should be communicated. Some have urged that respect for persons means that participants are entitled to receive results (Shalowitz and Miller, 2005). Another commentator has suggested that research participants should be allowed to decide at the beginning what results they would like, which could then be provided by a secure website (Kohane et al., 2007). Others, however, worry that participants may be misled if they receive a lot of information with uncertain clinical relevance (Beskow et al., 2001), especially since research results often are not replicated (Ioannidis, 2005; Moonesinghe, Khoury, and Janssens, 2007). The balancing between keeping participants informed and maintaining confidentiality is important, and can be dealt with by asking participants whether they want periodic updates.

Recommendation 5-1: The NCS should define the criteria and the process for deciding what individual clinical and research information, such as environmental assessments, test results, and survey scales, will be given to children and their families.

One entity that might assume this responsibility is the Data Safety Monitoring Board, working with the Federal Ethics Committee. One substantive criterion that may be particularly desirable is requiring that internal or external validation studies be available before individual research results are disclosed to children and their families.

Another issue that has not been addressed is that of “look-back liability,” a term used to indicate that people who collect clinical data may be liable for damages if they fail to recontact the patient or subject if the data are later found to be clinically relevant (Hirschorn et al., 1999; Sharpe, 1999). The research plan needs to address what responsibility, if any, the study or the local investigator may have to convey the implications of previously collected information to the mother and child.

In addition to deciding what information to convey, more attention needs to be devoted to what research personnel will do when they detect other risks in the course of the study. The research plan does acknowledge that interviewers may detect evidence of child abuse or neglect that, according to statute, must be reported. (We note that many states have more expansive reporting requirements than the reporting intentions identified in the research plan. For example, some states would require that the interviewer report.) Other proposed responses to identified risks may be judged by IRBs to be inadequate, particularly in a study of this magnitude. Many IRBs, for example, require that researchers put into place a system for referral of respondents who are found to be depressed; no such system is considered here. Similarly, it probably is not enough to offer information about help for domestic violence only to respondents who ask for it, which is the current plan (NCS Research Plan, Vol. 1, Sec. 12.3). Suggested international norms require that resources be identified and provided to interviewees regardless of whether they request them (Department of Gender and Women’s Health, Family and Community Health, and World Health Organization, 2001). Responsibility for assisting participants at risk cannot be avoided by having them enter their answers directly into computers so that the interviewers do not know their responses—the information to determine risk will be in the central repository. The research plan also seems to lack a specification of how thoroughly the data repository will be screened to identify participants at risk.

A number of factors could be considered in deciding when intervention is appropriate:

• what interventions the participants expect;
• if intervention will not be forthcoming, whether this is truly understood by the participants and their communities; and
• the likely public response to nonintervention.

The plans for and consequences of detecting and responding to risk need to be addressed in the consent process and with the local communities. The study’s IRB application, for example, acknowledges that parents need to know that answers to certain questions may trigger reports to children’s services (IRB draft, Sec. 9.3).

PROTECTION AND RELEASE OF INFORMATION

The issue of privacy is enormously important for the NCS. At the data collection stage, clear protocols need to be developed for encryption of information on interviewers’ laptops or tablet computers and for what to do in the inevitable event that these instruments are lost or stolen (New York Times, 2006). The steps currently planned by the coordinating center to protect data during the data collection phase appear to conform to current best practices, but these steps need to be reviewed periodically as best practices evolve. Instructions to interviewers should be exceptionally clear about how to handle sensitive issues that arise regarding the reporting of pregnancy, infertility, and pregnancy termination.

Much more complicated are privacy issues involving the full-sample data set. As with other large-scale longitudinal data sets, the large number of variables about each NCS participant renders impossible the task of somehow “deidentifying” NCS information, especially given the proliferation of other public data sets that can be used for triangulation. The best one can do is to address whether information is more or less identifiable (Malin, 2005; Malin and Airoldi, 2007).

The inability to protect privacy completely, coupled with the enormous scientific and public health value of NCS analysis, poses serious trade-offs for sharing all-site data both with NCS investigators and outside analysts. The NCS proposes a three-tiered approach to data access:

1. NCS investigators would have access to all data after they are stripped of personal identifiers, using “secure systems with safeguards in place to limit access and protect the data.”
2. Non-NCS investigators would have to sign data use agreements before receiving access to data that have been more highly redacted.
3. Public-use files would be “maximally de-identified” and may be subject to further restrictions (IRB draft, Sec. 7).

The panel views this approach as fundamentally flawed. First, science and public health are best served by a competitive process in which a number of analysts initiate new analytic studies and attempt to replicate and extend existing and emerging research studies. Second, the enormous expenditure of public funding argues that the NCS data should be considered a public good and available, under strict data protection protocols, to all qualified analysts. Third, given the enormous value of linking environmental and policy information to respondent residential locations, redacted or deidentified data would be worthless for addressing many of the NCS’s key hypotheses. Fourth, redacted or deidentified data would hinder or render impossible many valuable replication studies. Fifth, the norm for investigator-driven, large-scale national data collections, many of which are supported at least in part by the National Institutes of Health,¹ is for data release, with appropriate privacy safeguards, as soon as the data are cleaned and documented.

Despite these arguments, providing greater access to NCS investigators may seem attractive as recompense for their “sweat equity” in collecting the data. While recognizing the enormous value of the data collection efforts of NCS investigators, the panel views the compensation argument as flawed. Because data collection sites were selected with probability sampling methods by the NCS to maximize diversity and generalizability (NCS Research Plan, Vol. 1, Secs. 6.1-6.3.6), investigators who were not close to sample clusters had less opportunity to compete to become NCS study centers. Moreover, the study centers competed on their ability to collect the data, not on their ability to analyze it. Finally, the process of collecting data will give sites a significant advantage in carrying out timely analyses of them.

The conditions under which the all-site study data should be released to both insider and outsider analysts warrants very serious attention. The sensitivity of the data and the threat of identifying respondents argues for data release under highly restricted conditions. There are several successful models for releasing highly sensitive data to the research community for general analysis. Perhaps the most promising is the system of secure research data centers developed by the U.S. Census Bureau, which enable qualified researchers to analyze highly sensitive data compiled by the Census Bureau, the Internal Revenue Service, and the Social Security Administration.

All of these considerations argue for releasing data to all analysts, both inside and outside the study centers, but with appropriate privacy safeguards, as soon as they are cleaned and documented. The benefits of such a policy for the advancement of both science and public health greatly outweigh the costs.

Recommendation 5-2: NCS and non-NCS investigators should be given equal access to the full NCS data as soon as they are cleaned and documented. To protect respondent confidentiality, all analyses should be performed with the kind of strict safeguards employed by the Census Bureau research data centers.

The coordinating center plans to obtain a Certificate of Confidentiality, which is issued by the National Institutes of Health to “protect identifiable research information from forced disclosure” (Office of Extramural Research, 2007). Questions exist about the legal efficacy of these certificates, including whether such a certificate would protect against a query under the Data Quality Act or other recently enacted laws that permit greater access to data (Singer, 2005). The NCS may want to consider whether other mechanisms, such as seeking to come within the Confidential Information Protection and Statistical Efficiency Act of 2002 with its protections and penalties, would be more effective. It might also be useful to have some central guidance on HIPAA issues for every site.

IRB REVIEW AUTHORITY

The panel is concerned that the NCS is overly optimistic about the willingness of the study center IRBs to enter into collaborative agreements in which they would defer in whole or in part to other IRBs. Although IRBs are permitted under 45 CFR § 46.114 to defer their authority to another IRB, it seems unlikely that many IRBs would do this in a study with as high a profile as the NCS. Indeed, we predict that IRBs will differ substantially in their assessment of this study, which may seriously complicate its conduct.

INFORMED CONSENT

The research plan and IRB draft document clearly show an awareness of the complexities that surround consent in a study of this nature. The broader plan to evaluate various approaches to consent is excellent, as is the staged approach to obtaining consent, more or less on a need-to-know basis. The proposal to do a trial of a video-assisted consent process in which study personnel are present is sound. However, the way the process is described of allowing respondents to continue after they give an incorrect answer in the video consent process raises questions (IRB draft, Section 12.2). The viewer is left with the impression that respondents will be allowed to continue guessing until they get the right answer, at which time “the presentation will explain why the answer is correct….” More likely, what is meant is that a wrong answer will lead the study personnel to probe further to ensure adequate understanding and agreement, but this procedure should be made more explicit. In another domain, not all would concur with the ethical and legal reasons that are presented for allowing pregnant teens to consent for themselves and their children to participate in this research.

COMMUNITY ENGAGEMENT

The NCS proposes to engage the study community for the primary purpose of recruiting and retaining the sample. One approach to community involvement is called community-based participatory research (Israel et al., 2005; Minkler and Wallerstein, 2003). Under this approach, academic researchers and the community form a partnership and share control of all research processes and activities. Given that the study foci and design are already fixed, it is too late to employ this approach.

Community needs assessments for each study center are being proposed to identify the most pressing local issues of concern. The current NCS documents state that results of these needs assessments will be considered for inclusion in the core or local substudies, yet the documents fail to describe just how such results will be incorporated. Nor is mention made of how diverse community needs will affect data collection efforts across sites and at each of the sites, particularly in light of the NCS’s goal of collecting standardized data across all of its sites. Collection of new data that are re sponsive to a community’s needs may raise additional concerns about data collection burden and the use of established and validated measures. Ethical concerns arise if promises for inclusion of measures in the data collection are being made to communities but study centers fail to follow through on such promises.

There are a number of approaches—community collaboration, community cooperation—that, while falling short of full community-based participatory research, should be considered and would yield numerous benefits to the study and to the community. Greater community input into the entire research process has many benefits, some of which are noted by the NCS. These include but are not limited to the promotion of trust between study centers and communities through shared knowledge; more effective and efficient implementation of research, including better recruitment of participants, as noted by the NCS; enhancement of both the quantity and quality of data collected; a sharing of control in the interpretation and presentation of research findings, which may benefit communities as they use the data to address some of their own pressing needs; development of culturally appropriate measurement instruments, which may be useful for those measures that still require validation for diverse populations; and, ultimately, a more accurate understanding of a community’s circumstances in comparison to other communities. Moreover, for some of the ethical concerns raised thus far in this chapter, such as the kinds of information that should be provided to participants about their own health or about potential services that may be beneficial to them, community input and perspectives should be sought and represented in decision-making bodies.

Communities should therefore be given opportunities for greater input into study-related issues and collaboration with study centers. The notion of community refers to both the geographic communities in which data are being collected and to members of organizations for whom specific child health issues, such as learning disabilities, asthma, or schizophrenia, are of interest.

We think that the NCS should make a greater effort to engage communities in its research planning. There appear to be no materials that have been developed, nor much discussion in the proposal, about how and to what degree the various communities will be engaged. Moreover, there is no mention of how communities will participate in the implementation and interpretation of the data or whether the data will be available to communities for their own use. Communities can help to identify relevant protective and risk factors that may have been overlooked or underemphasized in the current study design. Given that local needs assessments are being proposed and community-specific data might be collected in the sites, communities should be provided with opportunities for and even resources to support data analyses that are specific to local communities. To take this idea a little further, the NCS might provide opportunities for building community research and data utilization capacity.

Other ethical concerns revolve around previous community-based participatory efforts that have uncovered a high level of discomfort in many communities about the collection of such biomarker samples as urine, DNA, blood, hair, and placenta. Although there is little evidence of discriminatory use of genetic information, the public concerns remain. In the future, specific laws or other legal protections may help to allay these concerns.

A high degree of trust is needed to ensure that a community fully understands (1) the necessity of the collection of the samples, (2) the role that the biological samples will play in answering the research questions and whether the samples will be used for multiple purposes, (3) the procedures involved in collecting the biomarkers, and (4) the eventual interpretation of the information.

Recommendation 5-3: The NCS should engage communities in selected study implementation, data analysis, and data interpretation activities that go beyond recruitment. The NCS should consider requiring every study center to formulate a more detailed plan to engage and collaborate with local communities.

CONCLUSION

In sum, the NCS presents challenges in a broad array of ethical, social, and political domains as a result of its burdensomeness, intrusiveness, and raising of potentially controversial topics relating to children and families. A successful study will require ongoing and, at times, difficult conversations and collaborations with research participants and with the communities in which they live. The NCS should pay particular attention to defining processes for deciding which research results should be returned to participants, to revising the plans for investigators’ access to research data, and to developing more robust plans for engaging communities involved in the study.

Footnotes

1

Examples include the National Longitudinal Survey of Adolescent Health, the Health and Retirement Survey, the Fragile Families and Child Wellbeing study, and the Panel Study of Income Dynamics.