Ethical dilemmas in conducting qualitative, public health research on social media: using a study on Facebook as a case.

AIM: Platforms on social media are increasingly used for public health research. While social media provides an exceptional opportunity to explore communication about public health topics, this practice is not without ethical dilemmas. Our aim was to identify and unfold some of these dilemmas and to suggest possible solutions and ways forward for future research. METHODS: Using our own research within a closed forum for people experiencing suicidal thoughts as a case, we explored certain dilemmas and possible answers relating to whether what is to be researched falls under a public or private social media domain; we investigated avenues for obtaining access to participants in an evolving online environment; how to secure informed consent from participants; and ways of ensuring anonymity. RESULTS: We provide recommendations and reflections that we hope will offer inspiration for researchers embarking on similar social media public health research within and beyond suicide research. CONCLUSIONS: The ethical framework commonly referred to in health research, based on confidentiality, anonymity, informed consent and doing no harm must be adjusted to be relevant for a social media context where technologies and regulations are constantly being altered.

The impact of influences in a medical screening programme invitation: a randomized controlled trial.

BACKGROUND: Invitations to screening programmes may include influences that are intending to increase the participation rates. This study had two objectives: (i) to assess if different categories of influences had a significant effect on the intention to participate in a screening programme for a fictitious disease and (ii) whether participants were aware of the influences, and if the intention to participate was associated to this awareness. METHODS: A seven-armed randomized controlled trial. Six hundred passers-by were randomly allocated to receive one of seven pamphlets inviting to a fictitious screening programme (neutral, relative risk reductions, misrepresentation of harms, pre-booked appointment, recommendation of participation, fear appeals, all combined). Participants were surveyed to assess (i) intention to participate (ITP) in the screening programme and (ii) awareness of an exerted influence. Chi-squared test was used to calculate the effect of the influences on ITP and the association of ITP with indicating awareness of an exerted influence and correctly locating an influence. RESULTS: Five hundred and eighty-nine participants were included for analysis. ITP was significantly increased (P < 0.05) in three pamphlets (misrepresentation of harms, fear appeals, all combined) [adjusted odds ratio (OR) 4.84, 95% confidence interval (CI): 2.54-9.23; OR 2.45, 95% CI: 1.31-4.59; OR 9.02, 95% CI: 4.44-18.34]. A percentage of 60.0-78.3 participants did not indicate awareness. Awareness was associated with a decreased ITP for those who could locate the influence (OR 0.39, 95% CI: 0.21-0.72) and those who failed to locate the influence (OR 0.47, 95% CI: 0.30-0.74). CONCLUSION: The application of influences should be carefully considered for interventions where an informed choice is desired.

The right to a second opinion on Artificial Intelligence diagnosis-Remedying the inadequacy of a risk-based regulation.

In this paper, we argue that patients who are subjects of Artificial Intelligence (AI)-supported diagnosis and treatment planning should have a right to a second opinion, but also that this right should not necessarily be construed as a right to a physician opinion. The right to a second opinion could potentially be satisfied by another independent AI system. Our considerations on the right to second opinion are embedded in the wider debate on different approaches to the regulation of AI, and we conclude the article by providing a number of reasons for preferring a rights-based approach over a risk-based approach.

Categories of systematic influences applied to increase cancer screening participation: a literature review and analysis.

BACKGROUND: Health authorities can influence citizens in subtle ways that render them more likely to participate in cancer screening programmes, and thereby possibly increase the beneficial effects. If the influences become too severe, the citizens' ability to make a personal choice may be lost on the way. The purpose of this analysis was to identify and categorize the influences while questioning whether they still permit the citizens to make their own choices regarding participation. METHODS: A two-stringed approach was used to obtain empirical examples of systematic influences that aim to raise participation rates in cancer screening programmes: First, a systematic literature search was conducted on three databases. Second, relevant experts were contacted via internationally based e-mail lists and asked for examples of systematic influences in cancer screening. The present analysis was based on direct, conventional content analysis to address different categories of systematic influences. RESULTS: The literature search yielded 19 included articles and the expert inquiry yielded 11 empirical examples of which content analysis of the empirical examples generated six major categories of systematic influence: (i) misleading presentation of statistics, (ii) misrepresentation of harms vs. benefits, (iii) opt-out systems, (iv) recommendation of participation, (v) fear appeals and (vi) influencing the general practitioners and other healthcare professionals. CONCLUSION: The six categories of identified influences work through psychological biases and personal costs and are still in widely use. The use of these types of influence remains ethically questionable in cancer screening programmes since they might compromise informed decision making.

Population Preferences for Performance and Explainability of Artificial Intelligence in Health Care: Choice-Based Conjoint Survey.

BACKGROUND: Certain types of artificial intelligence (AI), that is, deep learning models, can outperform health care professionals in particular domains. Such models hold considerable promise for improved diagnostics, treatment, and prevention, as well as more cost-efficient health care. They are, however, opaque in the sense that their exact reasoning cannot be fully explicated. Different stakeholders have emphasized the importance of the transparency/explainability of AI decision making. Transparency/explainability may come at the cost of performance. There is need for a public policy regulating the use of AI in health care that balances the societal interests in high performance as well as in transparency/explainability. A public policy should consider the wider public's interests in such features of AI. OBJECTIVE: This study elicited the public's preferences for the performance and explainability of AI decision making in health care and determined whether these preferences depend on respondent characteristics, including trust in health and technology and fears and hopes regarding AI. METHODS: We conducted a choice-based conjoint survey of public preferences for attributes of AI decision making in health care in a representative sample of the adult Danish population. Initial focus group interviews yielded 6 attributes playing a role in the respondents' views on the use of AI decision support in health care: (1) type of AI decision, (2) level of explanation, (3) performance/accuracy, (4) responsibility for the final decision, (5) possibility of discrimination, and (6) severity of the disease to which the AI is applied. In total, 100 unique choice sets were developed using fractional factorial design. In a 12-task survey, respondents were asked about their preference for AI system use in hospitals in relation to 3 different scenarios. RESULTS: Of the 1678 potential respondents, 1027 (61.2%) participated. The respondents consider the physician having the final responsibility for treatment decisions the most important attribute, with 46.8% of the total weight of attributes, followed by explainability of the decision (27.3%) and whether the system has been tested for discrimination (14.8%). Other factors, such as gender, age, level of education, whether respondents live rurally or in towns, respondents' trust in health and technology, and respondents' fears and hopes regarding AI, do not play a significant role in the majority of cases. CONCLUSIONS: The 3 factors that are most important to the public are, in descending order of importance, (1) that physicians are ultimately responsible for diagnostics and treatment planning, (2) that the AI decision support is explainable, and (3) that the AI system has been tested for discrimination. Public policy on AI system use in health care should give priority to such AI system use and ensure that patients are provided with information.

The 'Expiry Problem' of broad consent for biobank research - And why a meta consent model solves it.

In this response to Neil Manson's latest intervention in our debate about the best consent model for biobank research we show, contra Manson that the 'expiry problem' that affects broad consent models because of changes over time in methods, purposes, types of data used and governance structures is a real and significant problem. We further show that our preferred implementation of meta consent as a national consent platform solves this problem and is not subject to the cost and burden objections that Manson raises.

Control, trust and the sharing of health information: the limits of trust.

Clinical information about patients is increasingly being stored in electronic form and has therefore become more easily shareable. Data are collected as part of clinical care but have multiple other potential uses in relation to health system planning, audit and research. The use of clinical information for these secondary uses is controversial, and the ability to safeguard personal and sensitive data under current practices is contested.In this study, we investigate the attitudes of a representative sample of the Danish population towards transfer of clinical data from their general practice for secondary use. We specifically study: (1) patients' trust in different types of healthcare professionals, (2) their interest in being asked about secondary use of data and (3) their willingness to dispense from a requirement of informed consent based on their trust in healthcare professionals.We find that adult Danes are positive towards research that use patient data, and they generally trust general practitioners, hospitals and researchers to treat their data confidentially.Nevertheless, they feel that they have a right to control the use of their data, only 7.3% disagreeing, and that the data belong to them, only 14.0% disagreeing. Answers to further questions about the relation between trust, information and consent show that although trust modifies the wish for information and consent, there is still a strong view that the patient should control the use of data. We find no differences between those who have frequent contact with the healthcare system and those who do not.

In Defence of informed consent for health record research - why arguments from 'easy rescue', 'no harm' and 'consent bias' fail.

BACKGROUND: Health data holds great potential for improved treatments. Big data research and machine learning models have been shown to hold great promise for improved diagnostics and treatment planning. The potential is tied, however, to the availability of personal health data. In recent years, it has been argued that data from health records should be available for health research, and that individuals have a duty to make the data available for such research. A central point of debate is whether such secondary use of health data requires informed consent. MAIN BODY: In response to recent writings this paper argues that a requirement of informed consent for health record research must be upheld. It does so by exploring different contrasting notions of the duty of easy rescue and arguing that none of them entail a perfect duty to participate in health record research. In part because the costs of participation cannot be limited to 1) the threat of privacy breaches, but includes 2) the risk of reduced trust and 3) suboptimal treatment, 4) stigmatization and 5) medicalisation, 6) further stratification of solidarity and 7) increased inequality in access to treatment and medicine. And finally, it defends the requirement of informed consent by arguing that the mere possibility of consent bias provides a rather weak reason for making research participation mandatory, and that there are strong, independent reasons for making. CONCLUSION: Arguments from the duty of easy rescue in combination with claims about little risk of harm and potential consent bias fail to establish not only a perfect duty to participate in health record research, but also that participation in such research should be mandatory. On the contrary, an analysis of these arguments indicates that the duty to participate in research is most adequately construed as an imperfect duty, and reveals a number of strong reasons for insisting that participation in health records research is based on informed consent.

The right to refuse diagnostics and treatment planning by artificial intelligence.

In an analysis of artificially intelligent systems for medical diagnostics and treatment planning we argue that patients should be able to exercise a right to withdraw from AI diagnostics and treatment planning for reasons related to (1) the physician's role in the patients' formation of and acting on personal preferences and values, (2) the bias and opacity problem of AI systems, and (3) rational concerns about the future societal effects of introducing AI systems in the health care sector.

The biobank consent debate: why 'meta-consent' is still the solution!

In a recent article in the Journal of Medical Ethics, Neil Manson sets out to show that the meta-consent model of informed consent is not the solution to perennial debate on the ethics of biobank participation. In this response, we shall argue that (i) Manson's considerations on the costs of a meta-consent model are incomplete and therefore misleading; (ii) his view that a model of broad consent passes a threshold of moral acceptability rests on an analogy that misconstrues how biobank research is actually conducted and (iii) a model of meta-consent is more in tune with the nature of biobank research and enables autonomous choice.

Should all medical research be published? The moral responsibility of medical journal editors.

This article reinvigorates a key question in publication ethics: Is there research that it is permissible to conduct but that ought not to be published? The article raises the question in relation to two recent medical studies. It is argued (1) that the publication of these studies may cause significant harm to individuals, (2) that editors of medical journals have a moral responsibility for such harm, (3) that denial of publication is inadequate as an instrument to fulfil this moral responsibility and (4) that internationally acknowledged publication ethics codes should incorporate this aspect of editors' moral responsibility.

Eliciting meta consent for future secondary research use of health data using a smartphone application - a proof of concept study in the Danish population.

BACKGROUND: The increased use of information technology in every day health care creates vast amounts of stored health data that can be used for research. The secondary research use of routinely collected data raises questions about appropriate consent mechanisms for such use. One option is meta consent where individuals state their own consent preferences in relation to future use of their data, e.g. whether they want the data to be accessible to researchers under conditions of specific consent, broad consent, blanket consent or not at all. This study investigates whether meta consent preferences can be successfully elicited by a smartphone application in the adult Danish population. METHODS: A smartphone app was developed for the elicitation of meta consent preferences. An invitation to use the app was distributed to a stratified, representative sample of the Danish adult population. The meta consent choices, the use of the app, user experience data, and demographic data were logged and analysed statistically using IBM SPSS version 20. RESULTS: Of 1000 potential respondents 221 used the app. One hundred eighty-eight of the respondents were female and 103 male. The age range was 19 to 79 years with an average of 51 years (SD 16). Most users indicate 1) that they find the choices they are asked to make easy to understand (>75% find it 'Easy' or 'Very easy'), 2) that the application is easy to use (>75% find it 'Easy' or 'Very easy'), and 3) that this kind of choice should be offered to people (89% find it 'Absolutely' or 'Somewhat' important). CONCLUSIONS: It is possible to collect meta consent preferences in the general, adult population using a smartphone app.

Informed consent and registry-based research - the case of the Danish circumcision registry.

BACKGROUND: Research into personal health data holds great potential not only for improved treatment but also for economic growth. In these years many countries are developing policies aimed at facilitating such research often under the banner of 'big data'. A central point of debate is whether the secondary use of health data requires informed consent if the data is anonymised. In 2013 the Danish Minister of Health established a new register collecting data about all ritual male childhood circumcisions in Denmark. The main purpose of the register was to enable future research into the consequences of ritual circumcision. DISCUSSION: This article is a study into the case of the Danish Circumcision Registry. We show that such a registry may lead to various forms of harm such as 1) overreaching social pressure, 2) stigmatization, 3) medicalization of a religious practice, 4) discrimination, and 5) polarised research, and that a person may therefore have a strong and legitimate interest in deciding whether or not such data should be collected and/or used in research. This casts doubt on the claim that the requirement of informed consent could and should be waived for all types of secondary research into registries. We finally sketch a new model of informed consent - Meta consent - aimed at striking a balance between the interests in promoting research and at the same time protecting the individual. Research participants may have a strong and legitimate interest in deciding whether or not their data should be collected and used for registry-based research whether or not their data is anonymised.

Meta Consent - A Flexible Solution to the Problem of Secondary Use of Health Data.

In this article we provide an in-depth description of a new model of informed consent called 'meta consent' and consider its practical implementation. We explore justifications for preferring meta consent over alternative models of consent as a solution to the problem of secondary use of health data for research. We finally argue that meta consent strikes an appropriate balance between enabling valuable research and protecting the individual.

Conflict of interest disclosure and the polarisation of scientific communities.

In this article, we introduce and define the notion of polarised scientific communities. We show how polarisation may generate genuine conflicts of interest that require specific efforts to be identified and reported, and suggest a simple heuristic for the identification and reporting of the polarisation of research. We use examples from the debates about breast cancer screening and the prescription of statins to people at low risk of heart disease.

The stigmatization dilemma in public health policy--the case of MRSA in Denmark.

BACKGROUND: Multi-resistant bacteria pose an increasing and significant public health risk. As awareness of the severity of the problem grows, it is likely that it will become the target for a range of public health interventions. Some of these can intentionally or unintentionally lead to stigmatization of groups of citizens. DISCUSSION: The article describes the phenomenon of stigmatization within the health care area by discussing the concept in relation to AIDS and psychiatric diagnosis. It unfolds the ethical aspects of using stigmatization as a public health instrument to affect unwanted behaviours e.g. smoking. Moreover it discusses stigmatization as an unintended albeit expected side effect of public health instruments potentially used to counter the challenge of multi-resistant bacteria with particular reference to the Danish case of the growing problems with Methicillin-resistant Staphylococcus aureus (MRSA) within pig production. We argue that using stigmatization as a direct means to achieve public health outcomes is almost always ethically illegitimate. Autonomy and dignity considerations count against it, and the cost-benefit analysis that might by some be taken to outweigh these considerations will be fundamentally uncertain. We further argue that interventions where stigmatization is a side-effect need to fulfil requirements of proportionality, and that they may fall prey to 'the stigmatization dilemma', i.e. the dilemma that arises when all policy options are potentially stigmatizing but stigmatize different groups. When this dilemma obtains the decision-maker should choose the intervention that does not lead to permanent stigmatization and that stigmatizes as few as possible, as briefly as possible, and as little as possible.

Doctors, Patients, and Nudging in the Clinical Context--Four Views on Nudging and Informed Consent.

In an analysis of recent work on nudging we distinguish three positions on the relationship between nudging founded in libertarian paternalism and the protection of personal autonomy through informed consent. We argue that all three positions fail to provide adequate protection of personal autonomy in the clinical context. Acknowledging that nudging may be beneficial, we suggest a fourth position according to which nudging and informed consent are valuable in different domains of interaction.