All the studies we examined involve recruitment at birth or prenatally (typically at routine ultrasound appointment). Blood samples are obtained from mother, father (if available) and infant, along with other biological samples, such as breast milk, cord blood, saliva, urine and stool, with some variation among the studies in collection of these additional samples. All studies potentially involve genetic research studies within the cohort, though participants may have a choice to opt out of genetic studies and, for cost reasons, some cohort studies may not carry out genetic research will all participants. Some, such as CHILD, ELFE, GenR and NCS, involve visits to the home to collect samples of house dust or other environmental measures, or to administer questionnaires to participants. Cohort studies that focus on allergy and asthma, like CHILD and COPSAC, ask participants to undergo lung function and allergy tests. The six studies have received ethics approval and funding for a defined length of time, however, in most cases, interviewees reported that biological samples will be stored "indefinitely", or "forever". The COPSAC study has permission to hold samples until 2015, but may seek approval to retain them for a subsequent period of time. Similarly, the CHILD study plans to follow participants for six years but, with additional funding, may seek to extend the time for studying the birth cohort.
Findings in regard to each of the key ELS areas identified above are discussed here. In each section, we state how the studies we examined address the issue, then provide discussion that expands on debates and recommendations emerging from relevant literature. It is important to note that these cohort studies are subject to various levels of ethics review, which ensures that the studies' proposed handling of ELS issues are assessed in accordance with domestic ethics guidelines. Multi-site recruitment may require approval from ethics committees at each location where recruitment will occur. For example, the NCS requires approval of the National Institutes of Health institutional review board, as well as from review boards at each of the 105 study sites. The ELFE study required approval from several national bodies, including a data protection authority (Commission nationale de l'informatique et des libertés), a statistics body (Conseil national de l'information statistique), and a committee for the protection of persons (Committee du protection des personnes).
All studies require initial consent from the mother for her involvement in the research and permission for involvement of the child; paternal consent is also sought if the father is available and willing to participate in the study. The breadth of consent varies across the six studies we analysed. Potential participants in the BiB study are asked to give relatively broad consent to participate in the study as a whole, including collection and storage of biological samples for "future use" , but our interviewee anticipated that fresh consent would be obtained when additional information and samples are sought. Participants are advised that research partners outside the UK and Europe may have access to de-identified information and samples. Participants in the CHILD study are asked to consent to research that "will investigate some of the genetic, the immunological ..., the psychosocial and the environmental factors that may play a role in the development, the persistence and the severity of allergic disease" . While the consent is relatively broad, studies must relate to allergic disease and researchers plan to re-consent participants each year during the five year duration of the cohort. COPSAC researchers must obtain ethics review approval and participant consent for new studies not addressed in the initial consent. In France, studies involving biological sampling require details on how samples will be used and stored, as well as information about future studies and whether genetic analysis will be conducted. The GenR study distinguishes between active and passive participation. Consent for active participation (e.g. contact with researchers for physical measures and questionnaires) is sought at four intervals: prenatal; birth to 4 years; 4 to 16 years; and > 16 years. Consent for long-term use of data (passive participation) is also sought at each phase. The NCS has a multi-stage consent process. Women who are eligible to participate are asked for their consent to participate in the study as a whole and to permit their child to participate. If a woman agrees to participate, she subsequently receives information about collection of biological and environmental samples to allow a separate consent for those aspects of the study. A woman may consent to participate only in questionnaire instruments and can refuse to provide physical samples. Participants who agree to provide samples are given a choice to opt out of genetic analysis.
While participation of a child in a birth cohort study depends on parental (or guardian) authorisation, the ethical and legal scope of a parent's role in permitting a child to become a research subject remains controversial . As an NCS document points out: "A critical factor in mounting such a broad-based research effort hinges on how individuals will give permission for their children's participation....." . Some ethics guidelines restrict parents' ability to permit a child's participation to studies involving no more than minimal risk [10, 11]. There is, however, no consensus on the level of risk involved in birth cohort studies involving genetic research [4, 12, 13], though some studies self-define the risks involved as minimal . We discuss below our position that children should have opportunities as they mature to confirm or reject parents' decisions to permit research participation.
Another major debate in long-term studies involving biobanking centers on the ethical and legal acceptability of broad or open consent [14–16], which refers to consent for unspecified future uses of information and samples, opposed to specific, informed consent for each use. Some ethics guidelines stipulate that broad consent should be used only exceptionally. For example, a 2008 UNESCO report on consent states: "It is not acceptable to ask a participant in a research project to give an overall prior consent (so-called 'blank consent') to the effect that they would agree to any study that can be carried out with the data/material they provided, unless the data/material be irretrievably unlinked to the participants" [, p. 24, emphasis in original]. Some study documents we examined note the challenges of informed consent in a long-term study, including the fact that all future uses of the information and samples cannot be predicted at the time of initial collection. For example, the BiB protocol states that "... samples should be processed in such a way as to enable the widest possible range of analytical tests to be conducted. This is in recognition of the fact that it is not possible to predict from the outset the scope of specific analyses" [, p. 24].
The ways in which the studies we examined are handling consent reveals how they are navigating the complexities of long-term studies where all future uses of information and samples cannot be predicted. Notably, none of the studies adopts a blanket approach to consent where participants are asked to give one-time consent to unlimited future use. Rather, as all studies involve ongoing contact with participants, opportunities are available to refresh the consent given at the time of original enrollment and obtain consent for specific activities (such as home visits, blood draws and physical measures like allergy testing) where particular risks can be discussed. Continuing contact with participants also provides opportunities for them to ask new questions and to exercise their right to withdraw. Recognizing that participants may have special concerns about genetic research, some cohort studies also have specific consent processes to enroll cohort members in studies involving DNA analysis. A requirement to obtain ethics review for new research outside the original scope is also common across these cohort studies.
Assent and Consent from the Child
The NCS has established a policy that assent will be sought from children when they are between ages seven and eight. Upon reaching the age of legal majority in their state of residence, participants will go through a full informed consent process. Because the CHILD study will follow children only to age five, the protocol does not address the issue of assent or consent, though in our interviewee's experience, assent is typically sought around age seven to eight in longitudinal studies with children. At the time of our interview, the BiB study had not established an assent policy, but state they will need to address assent and consent from children as they mature. Our interviewee stated that the law in the UK regarding minors and consent to research is unclear, though, in practice, adolescents are often asked for consent around age 16. Continuing permission from parent(s) would also likely be sought until the child reaches the legal age of majority. Likewise, the ELFE study has not yet decided this issue, but anticipates seeking assent around age ten to 12 and will seek further guidance from ethics committees about obtaining consent from minors as they mature. The COPSAC and GenR studies have established policies about obtaining consent from participants in adolescence. Consent will be sought from COPSAC participants between age 15 and 17; in GenR, consent will be sought at age 16.
A large body of literature articulates the ethical and legal imperatives for seeking assent and consent from children [18–20]. Birth cohort studies, in which children become research subjects at or before birth, raise special consent considerations. By the time young subjects reach an age where they are able to understand information about the cohort study and to express views about participating, biological samples and much information about them have already been collected and used for research. Also, while " [c]hildren are a vulnerable research population, in the sense that they lack the capacity for consenting to their participation ... children's vulnerability is temporary and does not arise from a disorder; most children will become healthy adult members of society."  Having an opportunity to affirm or reject a parental decision made on the child's behalf is critical to respect the developing maturity of the child and their interests in making autonomous choices.
Some ethics guidelines and legal rules require assent from minors [4, 11]. Children typically develop some ability to understand information about research and express preferences by age seven; by early adolescence, some young people have capacity comparable to an adult to make informed choices about research participation [1, 18]. Some jurisdictions recognise a "mature minor" principle that permits a minor to make legally autonomous decisions if they have sufficient capacity to understand the nature and consequences of their decision . Very long-term studies, such as those that follow a child for decades, will need to seek consent from participants as they attain the age of majority prescribed by local law, commonly in the late teenage years.
By the time children can make their own choices about research participation, extensive amounts of information about them will have been collected and used for research. This fact presents complex questions investigators will need to consider. As data- and biobanks maintained by birth cohort studies are valuable resources for other researchers, to what extent should investigators responsible for the study share information and samples about children with external researchers who do not have an institutional affiliation with the various universities and other organizations involved in the study? Some scholars have recently argued that researchers who collect and store children's DNA for general population biobanks, that is, not disease specific biobanks, should "not make these DNA samples (or individual genetic sequence data) accessible outside the biobank until donors are recontacted as adults and given their own informed consent" [, p. 819]. Others contend that such a policy would be too restrictive, delay advancement of research on children's health, and that privacy concerns could be allayed by robust standards for data protection and research ethics oversight [22, 23]. As children mature, an additional issue is whether they should they have a right to access information about themselves, such as results of questionnaires asking the parent to provide information about the child's health, development and behavior as an infant or toddler. Access to such information may have privacy implications for the parent. These types of questions have not been uniformly addressed in the studies we examined and are topics for further analysis.
The right to withdraw from research is a fundamental right [24, 25] and the studies we examined were consistent in informing participants of their right to withdraw without explanation or adverse consequences. Recognizing the research value of data and samples collected over many years, the studies offer participants withdrawal options: (1) withdraw from further contact, but allow ongoing use of previously collected data and samples and continued record linkage (if the study links to other sources of data, such as health records); (2) the same as (1), but with no permission for continued record linkage; and (3) withdrawal and no permission for further use. In the latter situation, data and samples would be removed from the study, with the exception that any data/samples that have already been used and analysed cannot be withdrawn. Permission for continuing use of samples and data may be conditional on anonymisation to prevent re-identification of withdrawn participants. In addition to advising participants of the right to withdraw, the studies we examined also permit cohort members to choose not to participate in specific data or sample collection activities.
Protection of Samples and Data
All the studies we examined emphasise strong data protection measures to reduce risks of informational privacy breaches. Samples and personal information about participants are coded with an identification number and access to participant names and contact details is restricted to a small number of key personnel. ELFE has particularly stringent procedures to protect confidentiality of participant data. Separate files with unique identifiers are created for each recorded measure about an individual. For example, a mother's maternity survey will have a different identification code than the survey she completes at two months post-partum. A third party holds the key that permits linkage of the separate files, but does not have access to the file contents. Researchers who receive the file contents do not know the specific identity of the participant. As more information about individual participants is compiled over time, additional procedures may be implemented to guard against risks of re-identification that may arise from access to a longitudinal file that contains many indirectly identifying details. For example, if researchers need to know a participants' place of residence in France, they will receive that information, but not other information, such as age and profession, that could potentially lead to re-identification.
Some studies have established special committees to advise on data handling issues, including requests for access by third parties. For instance, the NCS has a Data Access and Confidentiality Committee http://www.nationalchildrensstudy.gov/about/organization/dacc/Pages/default.aspx to establish policies on access to NCS data and a Sample Oversight Group http://www.nationalchildrensstudy.gov/about/organization/oversightgroup/Pages/default.aspx to review studies proposing to use biological and/or environmental samples. Researchers who are not directly involved in the cohort studies we examined can generally apply for access to samples and data or to propose an add-on study. Such requests will be reviewed for scientific merit and the necessity of accessing cohort materials to answer the research questions. Other oversight committees have also been established for some studies to provide guidance on ethical and scientific issues. The ELFE study has a scientific council to advise on study design issues and an ethics committee comprised of experts external to ELFE to provide guidance on ethical issues that may arise.
Policies regarding data access, particularly by third parties external to the cohort study, should be established in advance of starting recruitment into the study. To give informed consent, potential participants require details about how information and samples will be handled, such as de-identification measures, access to identifying information, and access to datasets and samples by third parties, especially those outside the jurisdiction who may not be subject to domestic law governing personal information and biological materials.
A growing body of literature examines risks of re-identification of research participants, particularly through linkages of ostensibly de-identified databases and analyses of genomic data [15, 26, 27]. Reports that publicly available data from genome-wide association studies can lead to individual re-identification has prompted major funders such as the National Institutes of Health and the Wellcome Trust to tighten their policies on publication of genetic data . A recent legal analysis argues forcefully that the "surprising failure of anonymization" should "trigger a sea change" in laws governing collection, use and disclosure of information and legislators should impose restrictions such as limits on size and public release of databases . While there continues to be debate about the degree of effort and cost required to re-identify specific individuals from de-identified data and samples, those responsible for the repositories of information and samples compiled, linked and used in cohort studies must be alert to changing standards and best practices for data protection and safeguard appropriately the privacy interests of participants. This is particularly important where researchers seek permission from participants for broad future uses of data and samples. Such requests for future use are generally accompanied by assurances of confidentiality and robust data protection; those promises must be kept to maintain public trust.
Dealing with Sensitive Information
Research involving direct contact with children and families, especially in home visits, presents the possibility of encountering situations of suspected child abuse/neglect or harms concerning other household members. Researchers may have legal obligations to report information of abuse or neglect to authorities. Four interviewees stated they were aware of specific laws in their jurisdiction that mandate reporting of abuse/neglect; two did not identify a legal requirement but noted that reporting to social service authorities or referral to a pediatrician would occur. Two studies reported that they do not have written policies on handling this type of sensitive information, while three studies stated that the legal obligation to report to authorities is mentioned in the consent form.
One interviewee distinguished between making direct observations in home or study centre visits about evidence of harm to a child and acquiring information over the course of the cohort (e.g. from questionnaires) that potentially identifies a child who is at higher risk of developing a disorder. While the former situation raises immediate obligations to notify child protection officials, the latter situation relates to long-term handling of results (addressed below).
Interviewees were asked if they have policies for handling situations where a research participant other than the child is experiencing harm, such as a new mother showing signs of post-partum depression or spousal abuse. Some questionnaires ask about a new mother's level of anxiety or depression, including, in some cases, questions as direct as whether the new mother "had thoughts about committing suicide or hurting myself" or "ending it all" . Interviewees stated that situations would be addressed on a case-by-case basis, with possible referral to a health care professional or report to authorities. Two interviewees stated that this issue was addressed in the consent documents.
As it can be anticipated that research with family units has the potential to reveal sensitive information, investigators should develop policies on handling suspected cases of neglect or abuse consistent with local law and should inform participants at the time of initial consent about legally mandated reporting duties. Policies should also address situations involving imminent risks of serious harm and explain to participants how such circumstances will be handled.
Handling of Results
All the studies we examined will publish non-identifiable results in scientific journals and will disseminate findings through media such as newspapers, television and radio. Some studies organise periodic information sessions for participating families or distribute newsletters to share updates about general research findings. Studies distinguish clinically relevant information from research results where clinical significance is not yet understood and none of the studies return the latter. Practices vary in regard to return of other results. In COPSAC, for instance, participants receive results of routine measures carried out at annual visits, such as physical measures (height and weight) and results of allergy tests. Similarly, in GenR, results from tests with known clinical relevance are provided to participants and their physicians; this provision of information is considered a recruitment incentive. In the CHILD study, results of procedures such as lung inflammation and breathing tests will be provided to parents. They are also advised they "will be informed of research findings in the event of the discovery of abnormal treatable test results."  The NCS policy is that individual results will only be returned if they are immediately available at the time of the participant's visit to a study centre. In effect, this limits disclosure to routine data such as blood pressure, height and weight. Our interviewee noted that research analyses of samples may only be conducted months or years after collection, which creates challenges in reporting of results. The BiB study will not return results to individuals.
A review of national and international ethics guidelines indicates broad agreement that results ought to be returned to research participants if they "meet the requirements of scientific validity, clinical significance, [and] benefit" , such as prevention or treatment measures. The extent to which participants should be able to request results of unknown significance is an open question. Some scholars contend that participants have a privacy-based right to choose not to know information about themselves, including genetic information , and various ethics guidelines state that researchers ought to respect individual choices not to receive results [25, 32]. A parent who gives initial consent for their child's participation in a cohort study will make decisions where researchers offer choices about receiving results, including results about the child. While many parents may wish to receive clinically relevant information about a child, some may not .
Cohort studies would benefit from comprehensive policies that address various categories of findings, including those with clinical significance, those of unknown significance, those with implications for biological relatives, and incidental findings . Potential participants should be informed during consent discussions of policies for handling results and, where researchers plan to offer to return results, participants should have an option to choose whether to receive information and whether disclosed results will be shared with their health care providers. The evolving maturity of child participants raises additional complexities in determining how to handle results. The choices made by researchers and parents at the outset and during early years of the study will inevitably have consequences for the children. For example, lead investigators who design the study make early decisions about what findings, if any, will be offered to parents. In turn, parental decisions will impact the interests of child participants by choosing to receive or not receive results and, if results are obtained, in making choices about the child's health care or other areas of the child's life in response to the results. As children mature and researchers provide opportunities for them to give assent/consent to continuing participation, discussion with the child should address handling of results and researchers ought to consider how to give the child options for receiving information about themselves, including information accrued in infancy and early childhood.