As we better understand the challenges facing adolescents, research trying to address these problems is also becoming more prevalent. Children and adolescents are subject to abuse by adults, bullying by peers, have sexual and reproductive rights and needs, prone to accidents, suffer mental health issues, substance abuse problems, and so on. In many parts of the world, a minor teenager (let’s define that as being under the age of 18 for the purposes of this post) may be a child de jure but is an adult de facto.
As I think through the various issues that come up in my own research with adolescents, I come to realize that there are some grey areas with respect to issues surrounding consent. It does not help that different jurisdictions, or even different institutional review boards (IRBs) in the same jurisdiction can have differing views on some of these issues. That, in turn, leads to a lack of guidance for researchers, who want to do the right thing. It also causes inconsistencies and potential for violations in contexts where the requirements and oversight might be laxer.
Below of are some of the issues we are currently researching and grappling with in our own work. Feedback, guidance, and your own experiences with ethical dilemmas and how you tried to resolve them are all welcome – please use the opportunity to comment below to join the conversation…
Waiver of consent for electronic health records (EHR)
This is not necessarily an issue that is hugely different for minors vs. adults. As big data and administrative EHR data become more common – not just in high-income countries but elsewhere in the world – there is a tension between researchers trying to use EHR and requiring consent from individuals whose records are being sought. The proposed use is generally retrospective, although increasingly there are examples of prospective evaluations that seek to use administrative data from the outset. There is no doubt that if researchers are seeking to obtain administrative records, which do not have to be electronic, without consent from each individual (say, clients at a hospital or students at a school, applicants at a government facility, etc.), they should be seeking de-identified data. The anonymization can be done by someone at the institution holding the administrative records, who already has access to such data (such as a medical professional bound by doctor-patient confidentiality at a hospital), who can simply run code to remove personally identifiable information (PII), while creating a unique identifier for each individual. This is what was done in the (supplementary materials ). If there are other public administrative datasets containing PII, these can also be merged by the same people, allowing researchers to have de-identified data on outcomes from different sources linked to the EHR.
However, this type of a process to do potentially important work on public health or policy is not routine. You can get denied the permission to obtain such a waiver to access administrative records without consent. Conversely, in some settings, you may not only obtain to have access to such data but even be given PII, which allow you to contact people for follow-up purposes. As there is no one body or an accepted set of guidelines for such circumstances, the amount of leeway researchers get can vary widely. I enjoyed reading  in the U.S.) and encouraging better data handling practices., published in in 2016, which argues that when the only expected risk to the individual is a breach of privacy and the likelihood of that breach is minimal, obtaining waivers to access EHR without individual consent should be more or less routine (and, preferably, governed by one body in charge of such affairs). This would not only enable important research into public health/policy, but also save time, effort, and money – allowing IRBs to spend their time on more important and thorny ethical issues. The task of defining the threshold of “minimal risk” may not be most straightforward – there are definitions provided in the paper linked above – but it makes sense to (a) compare breach of privacy risks with paradigmatic risks, such as that involved in driving; and (b) remember that the marginal risk due to researcher involvement is very small: these data are already there vulnerable to breach in the first place. The involvement of a set of researchers may not only not increase that risk, but, in some cases, may decrease it by requiring better servers (secure and HIPAA-compliant
Needless to say, when the risks are more than minimal, individuals should seek client consent to access administrative records. As Mann et al. (2016) note, consent is not overriding even in such cases: greater than minimal risk simply means that there is a substantially higher burden on the researcher to obtain waivers by showing that (a) consent is not practicable and (b) the public health needs are great. There may be such cases, such as public health emergencies associated with outbreaks or epidemics, but they are far from your average case…
Community- or household-based research involving children
Until recently, economists mainly conducted household surveys, which may have involved measuring infants and young children, talking to minor adolescents, etc. In most such cases, even if the study is mainly interested in the outcomes of the minors rather than the parents/guardians, it is practicable to seek assent from the minor and consent from the parents. As the allow the researcher to not divulge the exact purpose of the study to the parent providing consent, it is generally not a burden to seek parental consent – with the scripts of the assent and consent forms properly reviewed and approved by at least one IRB. As enumeration teams are typically in communities visiting households at their residence, seeking consent does not impose a significant burden to the researcher or to the study.
One area in which economists can do better during field experiments is taking reporting requirements and linkages to related services and treatments more seriously. For example, if you are doing research of abuse and exploitation of children, you will likely be legally obligated to report suspected abuse to authorities. The practicalities of such a duty can be less than straightforward, the issues surrounding it for the subject and the research project may involve difficult tradeoffs, and the costs of setting up processes to do so can be high. For example, I worked with the to evaluate one of their interventions for early adolescents in a post-conflict setting and I was impressed with how seriously they took the issues of reporting requirements and how thoroughly they worked to link subjects to treatment at local facilities, by providing training to providers in each study community. We can all learn from such partners to become more responsible researchers…
Facility-based research involving children
What if you are doing research at, say, a health facility (or a number of health facilities) and study recruitment involves subjects seeking a certain type of treatment at that facility. Let’s say, for the sake of argument, your interventions are completely routine and risk-free: you are not trialing a new drug, or treatment, or a medical product, but might be simplifying some administrative processes at the hospital to make the client’s experience smoother, or you may be offering some discounted services to them, etc. We can think of two cases: (a) you want to use only their de-identified administrative data, or (b) you want to conduct follow-up interviews with them at some point in the future. In the former case, the lessons from EHR discussed above may apply. In the latter case, you need the subject’s consent to contact them later. This is fine, just seek their consent (at the outset or at the end of their visit – as appropriate) and you can receive administrative data with PII from the facility for consenting clients and de-identified data for those who chose to withhold consent.
What if, however, the subject is a minor? Do you need parental consent? This is where things get a lot more complicated. The first thing to note is that the legal definition of a minor might be contingent on the particular treatments or procedures involved in the proposed research. Suppose, for example, that the research concerns teen pregnancy, birth control, or family planning. In such cases, many jurisdictions allow minors to legally consent to treatment, meaning that they are considered adults for the purposes of consenting to treatments such as STIs, contraceptives (except permanent methods such as sterilization), substance abuse, mental illness, as well as surgical procedures and medical treatments for herself and her child relating to the delivery of the child, etc. See, for example, some of the relevant laws in the state of Virginia, U.S. and . There might also be local laws related to who is considered to be an emancipated minor, such as married minors.
You would think that if someone can consent to such treatment by themselves, without parental consent, they would also be allowed to consent (rather than assent) to participate in your research study, especially if that study simply involved some short follow-up interviews that would be private and confidential. However, it is not that simple: while some jurisdictions might allow minors in these circumstances to provide their own informed consent, don’t be surprised if your IRB turns your request down. My understanding, from talking to a few collaborators in the U.S. and elsewhere, is that there is heterogeneity in how strict IRBs can be in allowing minors to provide informed consent. The passage below was obtained from the of the website of the Office for Human Research Protections (OHRP) at the :
In my mind, there are three issues that lean in the direction of allowing the minor to provide informed consent herself in such circumstances:
- Seeking parental consent may not be practicable: It is possible that many of the clients seeking a treatment that is sensitive/private in nature may be at the facility without the parent or the guardian from whom the IRB insists you seek consent. Contacting such people may be expensive, even impossible at times. Unlike the case of community- or household-based research, where subjects are being visited in their places of residence, this may constitute a large burden, which needs to be weighed against the risks of obtaining informed consent from the minor only.
- Seeking parental consent may have a chilling effect on the minor’s view of the health facility in particular and the health system in general: The procedure to obtain parental consent would involve seeking assent from the client being enrolled in the study at the facility for her participation, then asking her - after explaining that since she is legally a minor and, hence, unable to provide informed consent for herself - if it is OK to contact her parents or guardians for their permission. Even allowing for the fact that the researchers are allowed to be less than fully truthful to the parent about the study purposes and using the best script/language possible to enable the minor to understand that their parents will never know that they were visiting this facility, seeking treatment for a sensitive issue, the possibility of a negative effect on the subject’s concerns about her privacy and the confidentiality of her visit cannot be ignored. Try to put yourself in the shoes of such a teenager, seeking treatment for, say, an STI: would you be comfortable with your provider asking if it is OK to call your parents – even if they did it in the most caring, soothing, and responsible manner possible? Maybe or maybe not…
- Consent bias: Related to the point above, it is a reasonable hypothesis to put forth that the characteristics of those who provide vs. withhold consent will be different. In the example above, suppose that two otherwise similar clients presented for STI testing: one tested positive and received treatment while the other tested negative. If it is more likely for the former to withhold consent (worrying about what their parents might find out) than the latter, we might have biased impact estimates. Mann et al. (2016) cite two meta-reviews showing that consent bias is real and that consenting and non-consenting groups can be significantly different than each other – indicating that studies with consenting individuals only may not be the best guide for the population among whom we’re trying to learn about intervention effects.
As this post is already getting too long, I leave you with the on ethical considerations in planning and reviewing research studies on sexual and reproductive health in adolescents, earlier this year.
 Health Insurance Portability and Accountability Act.