Background: Recent changes to regulatory guidance in the US and Europe have complicated oversight of secondary research by rendering most uses of de-identified data exempt from human subjects oversight. To identify the implications of such guidelines for harms to participants and communities, this paper explores the secondary uses of one de-identified DNA sample collection with limited oversight: the Human Genome Diversity Project (HGDP)-Centre d'Etude du Polymorphisme Humain, Fondation Jean Dausset (CEPH) Human Genome Diversity Panel. Methods: Using a combination of keyword (...) and cited reference search, we identified English-language scientific articles published between 2002 and 2009 that reported analysis of HGDP Diversity Panel samples and/or data. We then reviewed each article to identify the specific research use to which the samples and/or data was applied. Secondary uses were categorized according to the type and kind of research supported by the collection. Results: A wide variety of secondary uses were identified from 148 peer-reviewed articles. While the vast majority of these uses were consistent with the original intent of the collection, a minority of published reports described research whose primary findings could be regarded as controversial, objectionable, or potentially stigmatizing in their interpretation. Conclusions: We conclude that potential risks to participants and communities cannot be wholly eliminated by anonymization of individual data and suggest that explicit review of proposed secondary uses, by a Data Access Committee or similar internal oversight body with suitable stakeholder representation, should be a required component of the trustworthy governance of any repository of data or specimens. (shrink)

This paper contends that a research ethics approach to the regulation of health data research is unhelpful in the era of population‐level research and big data because it results in a primary focus on consent (meta‐, broad, dynamic and/or specific consent). Two recent guidelines – the 2016 WMA Declaration of Taipei on ethical considerations regarding health databases and biobanks and the revised CIOMS International ethical guidelines for health‐related research involving humans – both focus on the growing reliance on health data (...) for research. But as research ethics documents, they remain (to varying degrees) focused on consent and individual control of data use. Many current and future uses of health data make individual consent impractical, if not impossible. Many of the risks of secondary data use apply to communities and stakeholders rather than individual data subjects. Shifting from a research ethics perspective to a public health lens brings a different set of issues into view: how are the benefits and burdens of data use distributed, how can data research empower communities, who has legitimate decision‐making capacity? I propose that a public health ethics framework – based on public benefit, proportionality, equity, trust and accountability – provides more appropriate tools for assessing the ethical uses of health data. The main advantage of a public health approach for data research is that it is more likely to foster debate about power, justice and equity and to highlight the complexity of deciding when data use is in the public interest. (shrink)

Section I examines historical philosophical understandings of expertise in order to situate the current institution of bioethics. Section II focuses on philosophical analyses of the concept of expertise, asking, among other things, how it should be understood, how it can be acquired, and what such expertise warrants. Finally, section III addresses topics in bioethics and how ethics expertise should or should not be brought to bear in these areas, including expertise in the court room, in the hospital room, in the (...) media, and in making policy. 2. A GUIDED HISTORICAL TOUR As Scott LaBarge points out, Plato’s dialogues can be viewed as an extended treatment of the concept of moral expertise, so it is fitting to begin the volume with an examination of “Socrates and Moral Expertise”. Given Socrates’ protestations that he knows nothing, LaBarge observes that it would be interesting to determine both what a Socratic theory of moral expertise might be and whether Socrates qualified as such an expert. Plato’s model of moral expertise is what LaBarge calls “demonstrable expertise”, which is concerned mainly with the ability to attain a goal and to explain how one did it. The problem with this account is that when one tries to solve the various problems in the model – for example, allowing that moral expertise is not an all-or-nothing skill – then one is immediately faced with the “credentials problem”. As LaBarge puts it, “... (shrink)

BackgroundEthics review is the process of assessing the ethics of research involving humans. The Ethics Review Committee (ERC) is the key oversight mechanism designated to ensure ethics review. Whether or not this governance mechanism is still fit for purpose in the data-driven research context remains a debated issue among research ethics experts.Main textIn this article, we seek to address this issue in a twofold manner. First, we review the strengths and weaknesses of ERCs in ensuring ethical oversight. Second, we map (...) these strengths and weaknesses onto specific challenges raised by big data research. We distinguish two categories of potential weakness. The first category concerns persistent weaknesses, i.e., those which are not specific to big data research, but may be exacerbated by it. The second category concerns novel weaknesses, i.e., those which are created by and inherent to big data projects. Within this second category, we further distinguish between purview weaknesses related to the ERC’s scope (e.g., how big data projects may evade ERC review) and functional weaknesses, related to the ERC’s way of operating. Based on this analysis, we propose reforms aimed at improving the oversight capacity of ERCs in the era of big data science.ConclusionsWe believe the oversight mechanism could benefit from these reforms because they will help to overcome data-intensive research challenges and consequently benefit research at large. (shrink)

Recent rapid advances in Artificial Intelligence and Machine Learning have raised many questions about the regulatory and governance mechanisms for autonomous machines. Many commentators, scholars, and policy-makers now call for ensuring that algorithms governing our lives are transparent, fair, and accountable. Here, I propose a conceptual framework for the regulation of AI and algorithmic systems. I argue that we need tools to program, debug and maintain an algorithmic social contract, a pact between various human stakeholders, mediated by machines. To achieve (...) this, we can adapt the concept of human-in-the-loop from the fields of modeling and simulation, and interactive machine learning. In particular, I propose an agenda I call society-in-the-loop, which combines the HITL control paradigm with mechanisms for negotiating the values of various stakeholders affected by AI systems, and monitoring compliance with the agreement. In short, ‘SITL = HITL + Social Contract.’. (shrink)

BackgroundSharing de-identified individual-level health research data is widely promoted and has many potential benefits. However there are also some potential harms, such as misuse of data and breach of participant confidentiality. One way to promote the benefits of sharing while ameliorating its potential harms is through the adoption of a managed access approach where data requests are channeled through a Data Access Committee (DAC), rather than making data openly available without restrictions. A DAC, whether a formal or informal group of (...) individuals, has the responsibility of reviewing and assessing data access requests. Many individual groups, consortiums, institutional and independent DACs have been established but there is currently no widely accepted framework for their organization and function.Main textWe propose that DACs, should have the role of both promotion of data sharing and protection of data subjects, their communities, data producers, their institutions and the scientific enterprise. We suggest that data access should be granted by DACs as long as the data reuse has potential social value and provided there is low risk of foreseeable harms. To promote data sharing and to motivate data producers, DACs should encourage secondary uses that are consistent with the interests of data producers and their own institutions. Given the suggested roles of DACs, there should be transparent, simple and clear application procedures for data access. The approach to review of applications should be proportionate to the potential risks involved. DACs should be established within institutional and legal frameworks with clear lines of accountability, terms of reference and membership. We suggest that DACs should not be modelled after research ethics committees (RECs) because their functions and goals of review are different from those of RECs. DAC reviews should be guided by the principles of public health ethics instead of research ethics.ConclusionsIn this paper we have suggested a framework under which DACs should operate, how they should be organised, and how to constitute them. (shrink)

Biomedical research funding bodies across Europe and North America increasingly encourage—and, in some cases, require—investigators to involve members of the public in funded research. Yet there remains a striking lack of clarity about what ‘good’ or ‘successful’ public involvement looks like. In an effort to provide guidance to investigators and research organisations, representatives of several key research funding bodies in the UK recently came together to develop the National Standards for Public Involvement in Research. The Standards have critical implications for (...) the future of biomedical research in the UK and in other countries as researchers and funders abroad look to the Standards as a model for their own policy development. We assess the Standards and find that despite offering useful suggestions for dealing with practical challenges associated with public involvement, the Standards fail to address fundamental questions about when, why and with whom public involvement should be undertaken in the first place. We show that presented without this justificatory context, many of the recommendations in the Standards are, at best, fragments that require substantial elaboration by those looking to apply the Standards in their own work and, at worst, subject to potentially harmful misapplication by well-meaning investigators. As funding bodies increasingly push for public involvement in research, the key lesson of our analysis is that future recommendations about how public involvement should be conducted cannot be coherently formulated without a clear sense of the underlying goals and rationales for public involvement. (shrink)

Introduction A recent publication by Ludvigsson et al [1] attempted to explain and justify the nature of health registries in Nordic countries. These registries contain de-identified medical information from each of the individuals who interact with the nationally-run healthcare system and are used for research and quality improvement purposes. According to current laws in these countries, individual informed consent is generally not required for large-scale, registry-based studies that are deemed ethical by an ethics committee. In Nordic countries, these regional ethics (...) committees essentially perform the role of IRBs for human subjects research including clinical research but have less formal regulation to guide their practice [2]. As this holds similarities to laws in other countries such as the United States, this is an issue of global relevance. Ludvigsson et al. made several philosophical arguments to justify the option of informed consent, including the social contract theory of government, the ability of a representative ethics committee to serve as a proxy for individual interests, the principle of social justice, and the principle of utilitarianism. Here, we present an alternative point of view to these arguments made in support of the use of de-identified medical data in the form of health registries for research purposes without informed consent [3]. Though the argument Ludvigsson et al. puts forth specifically focuses on national health registries in Nordic countries, regulations in the US also permit the use of de-identified health information in research without informed consent. As such, the ethicality of using de-identified medical data for research studies without informed consent is relevant both in the US and abroad. Argument and Response The first argument posited by Ludvigsson et al involves the social contract theory of government. This theory states that individuals who are part of a given society have a set of moral obligations according to a contract, or agreement, between the individuals of that society which allows the society as an entity to exist. By choosing to remain part of a society, individuals are therefore agreeing to these obligations. In the case of Ludvigsson et al’s argument, individuals have an assumed agreement to contribute their personal data for research in exchange for receiving free or reduced-cost healthcare from the state. Consequently, participation is compulsory and informed consent is not necessary. This parallels other government requirements that aim to support society in exchange for the benefits of living in the society, such as drafting individuals to serve in the armed forces and requiring individuals to pay taxes. One practical caveat to this argument involves individuals who obtain private health insurance in the US. In this case, an argument could be made that because these individuals’ healthcare is not paid for by government-funded insurance such as Medicare or Medicaid, these individuals are not subject to the same agreement to contribute their data as those whose healthcare is provided by the state. A similar argument could be made for relatives of individuals who reside in a different country. In this case, certain medical data about an individual who resides in a given country may be used to infer medical information about the individual’s family members or relatives. Examples of this may include hereditary medical conditions or individual DNA sequence data. If the individual’s information is then used in a research study, this could also lead to indirect inclusion of relatives’ medical data in that research study, yielding potential violations of the relatives because they are not subject to the same social contract theory. Furthermore, their information is being used in research studies without their informed consent. In each of these cases, however, Ludvigsson et al’s argument falls short if we consider individual privacy as a fundamental human right. In this case, a government or other agency’s use of personal data without permission represents a violation of this right, whereas requiring individuals to serve in the armed forces or pay taxes does not. A counterargument may posit that because the medical data stored in these registries is “de-identified” and thus cannot be used to uniquely identify an individual, storage and use of this data without consent does not violate an individual’s right to privacy. While this argument appears sound, there are two important technical and philosophical considerations: whether an individual’s medical data can be truly de-identified, and whether new technologies in the future might enable re-identification of data that was believed to be de-identified. For instance, given that DNA sequence data is unique for each individual, creating a registry of this data on its own could conceivably be linked back to specific individuals in the future, even if this cannot be done at present. Furthermore, because there is greater similarity between genome sequence data of closely related family members, family trees could conceivably be created using the genome sequence data of individuals in a population-level registry, leading to potential identification of the genome sequence data of specific families that are outliers in some way. A similar technique could be applied to identify minority groups with known patterns in their genome sequence data. Combining these data analyses with other information such as census reports or demographic surveys could enable researchers to further identify specific groups and individuals from “de-identified” data, leading to privacy violations. From a philosophical standpoint, if we cannot truly de-identify medical data while retaining sufficient accuracy to conduct scientifically valid research studies, any use of data for research studies could potentially be used to personally identify individuals and thus would require informed consent in order to protect privacy. As a result, Ludvigsson et al’s argument that individuals have an assumed agreement to contribute their personal data for research in exchange for free or reduced-cost government healthcare falls short if we value privacy rights and acknowledge that individual medical data cannot be truly de-identified. A second argument posed by Ludvigsson et al. stipulates that research study approval from an ethics committee may replace individual informed consent regarding the use of medical data in a national registry because ethics committees are believed to represent the general public. While this argument is plausible in cases where there is unanimous public agreement regarding use of registry medical data for clinical research, it falls short when there is disagreement. In such cases, an ethics committee’s blanket statement concerning medical data in a registry that includes data from individuals who do not agree with the ethics committee’s position would violate their autonomy. Therefore, preserving individual autonomy would require us to ask each individual whether he/she approves of the use of his/her data for a particular research study, which is essentially informed consent. As such, Ludvigsson et al’s argument that ethics committee approval may replace individual informed consent for the use of medical registry data in the context of clinical research only holds weight if public opinion surrounding a particular study is unanimous. A third argument posed by Ludvigsson et al. involves the principle of social justice. Here, social justice connotes a fair distribution of wealth, opportunity, and social privileges among the members of a society. The argument is as follows: Informed consent may be especially challenging to obtain from high-risk populations. Informed consent protocol for a given research study would therefore exclude these populations, resulting in selection bias and publication of research findings that may not be relevant to these populations. The unequal publishing of research findings relevant to these populations could lead to further marginalization of these populations, further perpetuating social inequalities. This represents a violation of the principle of social justice. A potential counterargument involves violation of individual autonomy in research studies performed without informed consent. This may represent a violation of social justice if informed consent is obtained from some individuals but not others if we consider the opportunity to exercise one’s individual autonomy through the informed consent process to be a social privilege. However, this counterargument does not apply if informed consent is not obtained. A fourth argument posed by Ludvigsson et al is a utilitarian argument. In this case, utilitarianism refers to the principle of maximizing utility, or producing the greatest benefit to the greatest number of individuals. Specifically, requiring informed consent would drastically reduce the number of study participants and thus statistical power of national population-level studies. Furthermore, the cost required to obtain consent from millions of individuals as part of a national health data registry would be exorbitant. Because this research has the potential to benefit many of the individuals whose data are being used, requiring informed consent would conflict with the principle of utilitarianism because it would prevent such research from being conducted. One counterargument here involves the nature of research. Given the inherent uncertainty of research studies, it is possible that the results of large-scale research studies using national registries would not yield results that benefit many of the individuals whose data was used Additionally, while conducting large-scale research studies may maximize benefits to society, it may not maximize benefits to the individuals whose data was used to conduct these studies. If there is a greater risk of harm than benefit to some individuals and we choose to conduct the research without allowing these individuals to choose whether their data is used, we set a dangerous precedent that justifies harming a few individuals in order to benefit society. This also represents a potential violation of the research ethics principle of favorable risk-benefit ratio, which states that the potential benefits of a research study should be proportional to or outweigh the potential risks, if we assume that this principle should apply to each individual participant as well as to society as a whole. One possible response to this objection is that if each research study must be approved by an ethics committee, it would be this committee’s responsibility to ensure that the research would not result in harm to a few individuals in order to benefit the many. Thus, Ludvigsson et al’s argument is sound if we value the principle of utilitarianism over individual autonomy and assume that large-scale research studies are likely to confer the maximum benefits to society. Indeed, from a utilitarian point of view the potential harms to the people whose data are used are quite low, particularly when protections are in place, while the potential benefits to society are quite large. This tends to tip the scale in favor of utilitarian-based violations of privacy rights [4]. Discussion and Concluding Remarks Ludvigsson et al’s arguments regarding social justice and utilitarianism appear to be sound. However, Ludvigsson et al’s arguments regarding the social contract theory of government and ethics committees serving as a substitute for individual informed consent appear to be flawed. In order to determine whether use of de-identified medical data for research purposes is ethical in the absence of informed consent, we must consider whether the principles of social justice and utilitarianism supplant individual autonomy. Several additional considerations are warranted in the context of de-identified individual medical data for research studies. These include potential benefits to data subjects as a result of de-identification of data, additional tools that can be used to protect de-identified data from the risk of individual privacy violations in addition to or instead of informed consent, and protection of both de-identified and personally identifiable research data from unauthorized access and use. Specific additional tools include technical tools to assess the level of de-identification of data and statistical risk of re-identification, additional legal regulations on the use of individual medical data to prevent potential consequences of privacy violations such as employment discrimination or insurance discrimination, more granular access controls for researchers and research groups working with potentially re-identifiable data sets, and increased education and awareness of best practices for research personnel working with potentially re-identifiable data. Each of these considerations will now be discussed. The first additional consideration includes potential benefits to research subjects as a result of de-identification of data. Even if there is a risk that the data cannot be completely de-identified, using de-identified data as opposed to personally identifiable data in research studies, whether these studies involve informed consent or not, reduces the risk of individual privacy violations if there is a data breach. While data may need to remain individually identifiable in order to perform certain analyses in some research studies, individuals should be made aware of the level of de-identification of their data as part of the informed consent process and the data should be kept as de-identified as possible to minimize participant risk. In particular, given that there are multiple ways to de-identify data and de-identification exists along a spectrum [5], explaining what steps, if any, will be taken to de-identify a participant’s data and the associated risks of re-identification as part of the informed consent process for a given research study will enable participants to make a truly informed decision regarding use of their data in a given research study. The second important consideration involves whether informed consent is the only tool that can be used to protect “de-identified” data from the risk of individual privacy violations. Clearly, requiring a participant to consent to use of his/her data in a research study protects against potential privacy violations because the participant has given permission for his/her data to be used in this manner. If the participant has not given informed consent, however, whether or not a given use of the participant’s data as part of a research study constitutes a privacy violation depends on whether the data may be used to re-identify the participant. As such, developing technical tools to better assess the re-identifiability of a given set of data, both alone and in conjunction with other information, may help assess the risk to individual privacy of using a given set of data as part of a research study. Legal regulations could serve as another potential tool to protect against privacy violations. For instance, laws preventing discrimination for employment or insurance coverage on the basis of individual health data such as genetic information could protect individuals from unauthorized use of their health data by employers or insurance companies in the event that their “de-identified” data is able to be re-identified. Current laws such as the Genetic Information Nondiscrimination Act of 2008 already exist to protect against genetic discrimination from employers and for health insurance coverage [6]. Another potential tool involves restricting access of specific research groups or research personnel to different data sets to reduce the likelihood of a single group or researcher obtaining complementary data sets that could be used to re-identify individual study participants. This may be analogous to conventional checks and balances used across large organizations and in government to prevent any single individual from wielding too much power. Finally, increased education of research personnel regarding the risks of working with de-identified data and a set of best practices for storing and analyzing this data may also reduce the risk of inadvertent data breaches or re-identification. Thus, while informed consent is a valuable tool to protect de-identified data from posing a risk to individual privacy, it is not the only tool that can be used for this purpose. The third consideration involves protection of research data, both “de-identified” and personally identifiable, from unauthorized access. Given that de-identified data can be re-identified and the likelihood of re-identification depends on the ethics of the entity analyzing the data and the other data the entity has access to, maintaining controlled access to de-identified data is crucial. Determining who is able to access this data requires both technical safeguards and approval from an ethics committee regarding which research studies and research groups are permitted to access and use a given set of data, along with how the data may be used as part of a given research study or by a particular research group. References [1] Jonas F Ludvigsson et al., “Ethical Aspects of Registry-Based Research in the Nordic Countries,” Clinical Epidemiology 7 : 491–508, https://doi.org/10.2147/CLEP.S90589. [2] R. Froud et al., “Research Ethics Oversight in Norway: Structure, Function, and Challenges,” BMC Health Services Research 19, https://doi.org/10.1186/s12913-018-3816-0. [3] “Privacy and Security Conditions Required for Research Using EHR and Other Electronic Health Data,” ASPE, November 23, 2015, https://aspe.hhs.gov/report/feasibility-using-electronic-health-data-research-small-populations/priv acy-and-security-conditions-required-research-using-ehr-and-other-electronic-health-data. [4] Luke Gelinas, Alan Wertheimer, and Franklin G. Miller, “When and Why Is Research without Consent Permissible?,” The Hastings Center Report 46, no. 2 : 35–43, https://doi.org/10.1002/hast.548. [5] Mark A. Rothstein, “Is Deidentification Sufficient to Protect Health Privacy in Research?,” The American Journal of Bioethics : AJOB 10, no. 9 : 3–11, https://doi.org/10.1080/15265161.2010.494215. [6] “The Genetic Information Nondiscrimination Act of 2008,” National Human Genome Research Institute, accessed December 16, 2018, https://www.genome.gov/27568492/the-genetic-information-nondiscrimination-act-of-2008/. (shrink)

Can we legitimately speak of ethicsexperts? Recent literature in philosophy and medical ethics addresses this important question but does not offer a satisfactory answer. Part of the problem is the absence of an examination of what it means to be an expert in general. I therefore begin by reviewing my analysis of expertise which appeared earlier in this journal. We speak of two kinds of experts: persons whose expertise is in virtue of what theyknow (epistemic expertise), or what theydo (performative (...) expertise). Applying this analysis to the domain of ethics, I argue that we may speak of ethical expertise in three epistemic senses: a) expertise indescriptive ethics, b) expertise inmetaethics, c) expertise innormative ethics, and in a performative sense: d) expertise inliving a good life. I conclude with a brief description of some social roles of ethics experts. (shrink)

Many are calling for concrete mechanisms of oversight for health research involving artificial intelligence (AI). In response, institutional review boards (IRBs) are being turned to as a familiar model of governance. Here, we examine the IRB model as a form of ethics oversight for health research that uses AI. We consider the model's origins, analyze the challenges IRBs are facing in the contexts of both industry and academia, and offer concrete recommendations for how these committees might be adapted in order (...) to provide an effective mechanism of oversight for health‐related AI research. (shrink)