You are here

Article Text

Article menu

Download PDFPDF

Original research
Remote monitoring of medication adherence and patient and industry responsibilities in a learning health system
  1. Junhewk Kim1,
  2. Austin Connor Kassels2,3,
  3. Nathaniel Isaac Costin2,4,
  4. Harald Schmidt2
  1. 1 Dental Education Research Center, College of Dentistry, Yonsei University, Seoul, Seodaemun-gu, Republic of Korea
  2. 2 Medical Ethics and Health Policy, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
  3. 3 School of Medicine, Medical College of Wisconsin, Milwaukee, Wisconsin, USA
  4. 4 Hackensack Meridian School of Medicine, Seton Hall University, Nutley, New Jersey, USA
  1. Correspondence to Dr Harald Schmidt, Medical Ethics and Health Policy, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA 19104, USA; schmidth{at}upenn.edu

Abstract

A learning health system (LHS) seeks to establish a closer connection between clinical care and research and establishes new responsibilities for healthcare providers as well as patients. A new set of technological approaches in medication adherence monitoring can potentially yield valuable data within an LHS, and raises the question of the scope and limitations of patients’ responsibilities to use them. We argue here that, in principle, it is plausible to suggest that patients have a prima facie obligation to use novel adherence monitors. However, the strength of the obligations depends considerably on the extent to which data that adherence monitors generate are, in fact, used to further the goals of LHSs. The way in which data ownership is structured in the USA poses a considerable challenge here, while the European Union framework offers a more promising alternative.

  • ethics
  • behaviour modification
  • health promotion
  • information technology

https://doi.org/10.1136/medethics-2019-105667

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    A learning health system (LHS) is one ‘in which knowledge generation is so embedded into the core of the practice of medicine that it is a natural outgrowth and product of the healthcare delivery process and leads to continual improvement in care’.1 LHS sees to ‘continuously, routinely, and efficiently study and improve themselves’.2 Traditional research ethics or clinical ethics frameworks typically considered clinical care separately from the research context, and are generally insufficiently equipped to address the unique context of an LHS that directly integrates research and practice with new roles and responsibilities for all involved.3–5 In response, Faden et al proposed a new ethics framework which comprises seven obligations (table 1).3

    View this table:
    Table 1

    Seven obligations of the learning health system ethics framework by Faden et al

    While the first five obligations reflect traditional bioethical concepts—including respect for autonomy, beneficence, nonmaleficence and justice—the last two are more bespoke and respond to the respective stakeholders’ roles in an LHS, arguing that healthcare professionals, institutions, payers and purchasers of healthcare also have responsibilities to contribute to the knowledge base of the healthcare system. In addition, patients have ‘an obligation to contribute to, participate in, and otherwise facilitate learning’.3

    We unfold the underlying reasoning further below, and start here from the assumption that the basic contractual justification of patients’ obligations under the LHS has considerable plausibility. We apply and test the framework and the articulation of patients’ obligations by addressing the central question of whether it imposes responsibilities on patients to use novel approaches that enable remote monitoring of medication adherence, and, once they commit to using a device, to share data for genuine research purposes. We argue that prima facie patients can be obliged in this way. The strength of the obligations depends on how potential issues around intrusion, privacy, trust and reasons for non-adherence are managed, that have rightly been the focus of debate to date,6 7 and, more fundamentally, on how data from monitoring devices are used for the purposes of the LHS. We see the generation of useful data as a necessary, although not sufficient, condition. In approaching our overarching question, we seek to clarify, in particular, who typically handles and manages data generated by monitoring devices, and what the responsibilities of monitoring device manufacturers are.

    Patient responsibilities

    When it comes to taking medication, in general, there are no legal provisions that would require patients to be unconditionally adherent. An exception is the infectious disease context, especially in the treatment of tuberculosis (TB). Based on the harm principle and the epidemiological need to reduce the spread of disease to the wider population, many countries permit coercive legal sanctions in the case of insufficient treatment adherence.8 Here, we focus mainly on non-infectious chronic condition in the clinical (as opposed to the public health) context that the LHS is predominantly concerned with.

    Patients’ obligations to contribute to research in the LHS are based on what John Rawls termed the common goods principle.9 The principle of common good argues that the members of a community have an obligation to contribute to arrangements that, ideally, benefit everyone in society. As Faden et al argue expressly, this obligation is not fulfilled by merely paying a fee to use healthcare services, such as insurance premiums. Instead, “a discharge of obligations of reciprocity occurs through an established practice of making an appropriate and proportional return—returning benefit with proportional benefit, with all alike sharing, as a matter of moral obligation, the burdens necessary to produce these benefits’, and patients have a prima facie duty to contribute to generalisable knowledge that can support and improve healthcare systems.10 (The reason that this obligation is qualified as prima facie is that obligations 1–4 can establish powerful limiting constraints.) Other authors, likewise, have argued that patients, as beneficiaries of care have an obligation ‘to provide something in return for the benefits… received, or expect[ed] to receive, from research’.11 The contractualist relationship has also been suggested to underlie patients’ obligations to adhere to prescribed treatment (in the absence of reasonable side effects).12 A different but related line of reasoning is that the principle of beneficence can support a duty for patients to participate in biomedical research, since expanding medical knowledge may contribute to new and improved practices.10 13

    A set of existing codified responsibilities for patients further supports the articulation under the LHS in the present context of medication adherence monitoring. For example, the British Medical Association’s Handbook of Ethics and Law notes adhering to recommended treatments as an example of patients’ responsibilities.14 In the USA, the National Health Council’s Principles of Patients’ Rights and Responsibilities stipulates that patients ‘should cooperate fully with providers in complying with mutually accepted treatment regimens’.15 The American College of Physicians’ Ethics Manual notes that patients can participate and share responsibility for their healthcare.16 While these statements are not legally binding and do not settle, by themselves, the questions we are concerned with here, we see their relevance as expressing broad consensus among different key stakeholder groups about the prima facie obligation of patients to adhere to agreed treatment plans. Accordingly, they can further support the case for exploring patients’ roles to use novel adherence monitoring approaches within an LHS.

    Novel approaches to medication adherence monitoring

    Adherence is defined as ‘the extent to which a person’s behaviour—taking medication, following a diet and/or executing lifestyle changes—corresponds with agreed on recommendations from a healthcare provider’.17 Historically, there has been an important development away from a paternalistic model in which patients are simply expected to do as they are told—which was reflected in the commonly used terms ‘compliance’ or ‘concordance’—towards a more person-centred model in a shared decision-making context, in which physician and patient jointly agree on treatment plans.18–20

    A systematic review of the economic impact of medication non-adherence suggests that non-adherence can have substantial impacts, which speaks to another facet of patients’ obligations to contribute to quality and value under the framework by Faden et al: in a review of 79 studies, the adjusted total cost of medication non-adherence across all disease groups ranged from US$949 to US$52 341.21 At the same time, the underlying evidence on the relationship between adherence and clinical outcomes has substantial gaps. For example, a current systematic review on adherence of cardiovascular disease patients notes that ‘few trials demonstrate improvements in both adherence and clinical outcomes’.22 This situation, again, underscores the potential value of gaining better insights into patients’ actual adherence under the LHS.

    A number of different approaches have emerged to ascertain to what extent patients’ response and non-response to drugs correlate with taking (or failing to take) them as prescribed.23 These include reminder systems through apps or text messages,24 25 various forms of electronic pill bottles26 that send signals to tethered platforms when pill boxes are opened, automated and in-person adherence checks by video conferencing, automated ingestion monitoring, alongside in-person direct observed therapy (DOT), which is the standard of care for monitoring treatment adherence in patients with active TB27; see table 2.28 29

    View this table:
    Table 2

    Salient medication adherence monitoring methods

    Ingestion monitoring relates to a drug that is combined with ingestible sensors which are embedded in the pill and send a signal to an external monitor when absorption begins. For example, the first Food and Drug Administration (FDA) approved ingestion monitoring system is activated by stomach fluid and sends a signal to a patch attached on the skin of the abdominal region.30 Other data, such as body temperature, heart rate, time of absorption, activity/sleep levels can also be collected via a smartphone app. Data can be stored to provide records and promote adherence. Data can furthermore be transmitted to a server of the device maker and be analysed further, as well as shared with others, such as family members, healthcare providers or payers. A related approach by another company that has also been recently approved by the FDA works similar, but does not entail a patch that would record other physiological data, and focuses centrally on the ingestion event.31 32

    Computer observed therapy seeks to automate and combine DOT and electronically DOT and combines automatic drug and face recognition with an artificial intelligence algorithm. An app takes an image of the pill and a video of the swallowing motion of the patient via a smartphone camera. If the algorithm confirms that the patient has swallowed the medication, the app stores and analyses the pattern of patient’s medication adherence. The app also sends the data to the main server, and again, data can be shared with third parties, too.29

    Approaches of this type can, in principle, support patients, promote clinical outcomes and help achieve a better understanding of the exact relationship between medication adherence and clinical outcomes as part of postmarketing surveillance.33 While much depends on the exact set-up of each approach, the basic flow of data is from the patient to an app, and from there to a data warehouse (storage), where some form of postprocessing may occur. Critically, the data typically remain with the device makers34 (figure 1).

    Figure 1
    Figure 1

    A data flow diagram of novel medication adherence monitoring systems.

    LHS, adherence monitoring and patient responsibilities

    Insofar as other structural barriers such as cost of medication do not stand in the way of taking medications as prescribed, the ethics framework for the LHS can, in principle, support the case that patients have a prima facie obligation to use novel adherence monitors to promote the LHS’ goals.35 Primarily, patients agree to take medications out of self-interest, to regain or maintain their health. And often, they have genuinely good reasons not to take medications, for example, where the balance of benefits over side effects is not a positive one.20 36 As noted, our basic understanding of the relationship between the frequency and timing of taking medications, and clinical outcomes remains incomplete. And it is common that trial populations in which drugs are tested are not reflective of the wider population in which the drugs are eventually used. Further evidence on how drugs function in a real-world setting is therefore of considerable interest, and novel ways of monitoring adherence (as well as reasons for non-adherence) hold promise in producing generalisable knowledge that can help promote health of other, or future patients.

    The wider literature also supports this conceptualisation,37 and it has been argued on both contractualist and utilitarian grounds that patients have an ethical obligation to share their health information for research purpose when the research (1) is not interventional; (2) is a public sector health data; (3) conducts for the public good and (4) satisfies minimum requirements.38 While this argument is made for the context of public research, it can plausibly be extended to the use of remote adherence monitoring under the LHS, since adherence data, by themselves, are of prima facie utility in contributing to improved care for individual patients and to knowledge about the relationship between adherence and clinical outcomes that may benefit future patients.39

    It does not, however, immediately follow that patients should therefore universally be either encouraged, incentivised, or perhaps even coerced into using particular forms of adherence monitors. First, their uses raise concerns around several deeper ethical issues, including those around privacy, confidentiality and trust, that have already been unpacked in helpful detail for the context of novel ways of monitoring adherence.6 7 Second, close alignment of patient and provider perspectives is necessary, as currently, all approaches are open to workarounds that enable feigning adherence, while actually not taking medications: such records would be counterproductive to the overall aim of improving adherence, and better understanding its relevance. Third, and most fundamentally, any talk about patients’ obligations under the LHS would need to be motivated substantially by the potential value that the data have for its core objective—but here we need to explore first the regulatory context in which these approaches are used.

    Not least due to salient differences for the question at hand, we focus on the US and the European frameworks. By way of an overarching generalisation that speaks to our main objective, a major difference between the two regions is how data ownership is conceived. While this is a highly technical legal discussion,40 broadly, the US framework permits companies gathering data from adherence monitors to relate to data in a quasi-ownership way, which enables the selling of data to others (much like one might own, and sell, a car or house). This approach gives a potentially stronger role to device makers in making adherence data available for research purposes in the LHS, and stands in considerable tension to the common good principle. Under the European framework, by contrast, data from adherence monitors do not belong to a party in a quasi-proprietary sense, and the law focuses instead on data subject and data controllers. This structure gives a potentially lesser role to device makers, aligns better with the common good principle and a more prominent role to patients’ interest, in using adherence data in an LHS.

    US regulations

    In the US, adherence data from monitoring devices can have dual status.41 On one hand, data can constitute medical records covered by the Health Insurance Portability and Accountability Act (HIPAA) and the Common Rule, insofar as data flow between the healthcare provider and the patient. On the other hand, adherence records can constitute non-protected personal data, which are not governed by any laws, insofar as data are collected by the company or the payer in deidentified form. While the concept of deidentified or anonymous health data is becoming increasingly questionable,42 it is currently nonetheless most attractive for device makers to focus on deidentified data, to avoid needing to establish arrangements that can ensure HIPAA compliance.43

    The status of data moreover depends on the way the data are processed. Two basic models can be distinguished: one is a service-oriented architecture, in which the data stays raw and services (eg, the smartphone app) retrieve and process the data in real time. Another is data warehousing, where the data are aggregated and prepared for processing, although the data are not available in real time.40 Where data are gathered to process, not accessed in real time, the question of data ownership arises in its most prominent form.44 Current policy in the USA, which is highlighted in the Digital Health Innovation Action Plan,45 targets lower risk technology and, for makers of medication adherence monitors of the type considered here, does not bar further processing and selling of data.46

    EU regulations

    The European Union (EU) enacted the General Data Protection Regulations (GDPR) in 2018. The GDPR aims ‘more consistent protection of consumer and personal data across EU nations.’47 Article 5 establishes the purpose limitation principle, meaning that data must be collected for ‘specified, explicit and legitimate purposes’. The GDPR does not sanction the postprocessing of data if the intended uses have not been clearly stated at the outset. The only exceptions are if renewed consent is sought, if there is a clear basis in the law (for example, for forensic purposes), or if the processing is needed for compelling public interest in the public health sector. There could be a hypothetical situation when highly infective and fatal virus is spread and there is a need to access individuals’ adherence data to establish prevention measures. The GDPR also gives individuals whose data are collected control over data by establishing the right to portability, meaning that data can be moved from one service provider to another, and it establishes the in-principle right to have collected data be erased from any records.

    This purpose limitation principle would regulate the arbitrary use of the data from the data holders, especially commercial use from the company. But there is a concern that this will undermine the quality of the data and the insights they could provide. Some commentators criticised that the GDPR imposes overly heavy burdens, rendering the analysis environment ‘suboptimal and inefficient’.48 However, where data are ‘anonymous in such a way that the data subject is not or no longer identifiable’, or processed by pseudonymisation, which means ‘the processing of personal data in such a way that the data can no longer be attributed to a specific data subject without the use of additional information’,49 the GDPR permits data holders to use data without further restrictions.

    Data generated by adherence monitors of the type considered here are deemed health data under the GDPR.50 Device makers wishing to commercialise data would require patients’ express consent. The device maker (‘data controller’ under the GDPR) cannot sell the data under the EU regulations. However, the database could be owned by the party based on the sui generis rights and it would be permissible for a company to distribute certain types of organised data.46 At the same time, the reference to use of adherence data for scientific research that promotes the public good could cover studies that further the objectives of the LHS—but may also potentially be used as smoke screens to mine data for profit.51

    Comparing regulations: ethical implications

    The GDPR allows processing of personal data for public interest, that is, public health and scientific research in the healthcare area. We take the objectives of the LHS to be broader than the public good in the sense that it can extend to improving health outcomes of individuals through constant learning. If patient participation to learning is an obligation in a LHS, data generated by the adherence monitoring devices should be considered as a public good and it should not be used for private interests, alone.52 Therefore, an approach that would be more appropriate for public good research under the LHS, while limiting commercial interests, is set out in the data flow model below (figure 2). This model establishes a barrier between the device maker or the software developer and data gathered from the monitoring devices. The barrier could be physical (for example, locating data server outside of the device company), digital (for example, access control software), or both. Data access should be controlled and governed by the national authority responsible for protecting health data.53

    Figure 2
    Figure 2

    Proposed data flow diagram of novel medication adherence monitoring systems.

    Conclusion: towards just use of medication adherence data in an LHS

    The availability of novel medication adherence monitoring can, in principle, help to support a prima facie responsibility of patients to use them in an LHS, especially where there is lack of clarity about the exact relationship of adherence and therapeutic outcomes, where non-adherence due to side effects needs to be better understood, or where drugs are particularly costly, and non-adherence jeopardises effectiveness. However, patients’ responsibilities so conceived are not unconditional.

    First, and most fundamentally, they depend on the extent to which adherence data are, in fact, usable, and used, for public good research: prior to urging patients to use adherence monitors by appealing to their responsibilities to support an LHS, it is therefore imperative to clarify the responsibilities of device makers, and other parties involved in implementing monitoring and managing adherence data.

    Second, scientific utility and accessibility are necessary, but not sufficient for motivating patient responsibilities. In the next step, patients’ prima facie responsibilities need to be balanced against possible concerns around intrusion, privacy and implicit or explicit coercion. This second step has rightly been the focus of much of the current debate, and helpful general considerations have been set out for some of the key forms of remote monitoring.6 34 Much in this second step will depend on which of the four different exemplary approaches noted in table 2 are being implemented, and how: for each approach differs in the mode of monitoring, comprises multiple subtypes, and in many ways the concrete implementation can differ and will plausibly affect patient acceptability. For example, both rewarding and penalising incentives tied to particular adherence levels can render a monitoring approach coercive, where it leads patients to take drugs they would not otherwise have taken due to side effects (absent the monitoring approach and incentives). Ethical issues can also arise from the exact wording of a reminder text message (combined with whatever monitoring approach). Wording can be perceived as either unconditionally supportive, or perhaps stigmatising, if the message shows on a phone in locked-screen mode, or is otherwise viewed in a public place that exposes the user to unwelcome scrutiny.

    Third, as the brief outline of the regulatory framework has shown, conceptually, both defining and describing who owns and/or controls data are far from straightforward. This poses a considerable challenge for the design of consent forms, and the consent process. For consent to be genuine, the description of what happens with data generated by monitoring devices needs to be appropriate to the level of understanding of the users concerned.

    For many device makers, gathering and selling data are a plausible, if not central, part of the overall business case, in addition to selling adherence monitoring platforms. The challenge is hence to create an environment in which gathered data can be made available for research in a way that does not represent an existential threat to the companies that generate them (see also figure 2). In policy terms, similar issues have arisen in the wider context of mobile health, where it has been recommended that ‘companies and society can generate economic value in ways that also create value for society’.54 One option would therefore be that independent third-party auditors develop practices and protocols for ‘the responsible use of personalised health technology using tangible and standardised metrics’. Unless, and until such arrangements have been put in place, it is premature to establish that patients have obligations to use novel forms of medication adherence monitoring on the grounds that doing so furthers the objectives of the LHS.

    Acknowledgments

    The authors thank Ken Goodman, Kayte Spector-Bagdady, and the JME reviewers for helpful comments that enabled them to improve earlier versions of the manuscript. Needless to say, all remaining errors are the authors’ alone.

    References

    1. ↵
      2. Olsen LA ,
      3. Aisner D ,
      4. McGinnis JM
      , eds. The Learning Healthcare System: Workshop Summary. Washington: National Academies Press, 2007.
    2. ↵
      2. Friedman CP ,
      3. Allee NJ ,
      4. Delaney BC , et al
      . The science of learning health systems: foundations for a new Journal. Learn Health Syst 2017;1(1):e10020.doi:10.1002/lrh2.10020 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31245555
      OpenUrlPubMed
    3. ↵
      2. Faden RR ,
      3. Kass NE ,
      4. Goodman SN , et al
      . An ethics framework for a learning health care system: a departure from traditional research ethics and clinical ethics. Hastings Cent Rep 2013;Spec No(s1):S16–27.doi:10.1002/hast.134 pmid:http://www.ncbi.nlm.nih.gov/pubmed/23315888
    4. ↵
      2. Largent EA ,
      3. Joffe S ,
      4. Miller FG
      . Can research and care be ethically integrated? Hastings Cent Rep 2011;41(4):37–46.doi:10.1002/j.1552-146X.2011.tb00123.x pmid:http://www.ncbi.nlm.nih.gov/pubmed/21845922
      OpenUrlPubMed
    5. ↵
      2. Goodman KW
      . Ethics, information technology, and public health: new challenges for the clinician-patient relationship. J Law Med Ethics 2010;38(1):58–63.doi:10.1111/j.1748-720X.2010.00466.x pmid:http://www.ncbi.nlm.nih.gov/pubmed/20446984
      OpenUrlPubMed
    6. ↵
      2. Campbell JI ,
      3. Eyal N ,
      4. Musiimenta A , et al
      . Ethical questions in medical electronic adherence monitoring. J Gen Intern Med 2016;31(3):338–42.doi:10.1007/s11606-015-3502-4 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26358284
      OpenUrlPubMed
    7. ↵
      2. Cohen IG ,
      3. Mello MM
      . HIPAA and protecting health information in the 21st century. JAMA 2018;320(3):231–2.doi:10.1001/jama.2018.5630 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29800120
      OpenUrlPubMed
    8. ↵
      2. Pritchett EN ,
      3. Schlossberg D ,
      4. Lovett-Glenn G , et al
      . Legal intervention for non-adherent patients in the treatment of tuberculosis. Int J Tuberc Lung Dis 2009;13(3):323–7.pmid:http://www.ncbi.nlm.nih.gov/pubmed/19275791
      OpenUrlPubMed
    9. ↵
      2. Rawls J
      . A theory of justice. Cambridge: Belknap Press of Harvard University Press, 1971.
    10. ↵
      2. Schaefer GO ,
      3. Emanuel EJ ,
      4. Wertheimer A
      . The obligation to participate in biomedical research. JAMA 2009;302(1):67–72.doi:10.1001/jama.2009.931 pmid:http://www.ncbi.nlm.nih.gov/pubmed/19567441
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      2. Resnik DB ,
      3. Ness E
      . Participants' responsibilities in clinical research. J Med Ethics 2012;38(12):746–50.doi:10.1136/medethics-2011-100319 pmid:http://www.ncbi.nlm.nih.gov/pubmed/22822200
      OpenUrlAbstract/FREE Full Text
    12. ↵
      2. Resnik D
      . The patient's duty to adhere to prescribed treatment: an ethical analysis. J Med Philos 2005;30(2):167–88.doi:10.1080/03605310590926849 pmid:http://www.ncbi.nlm.nih.gov/pubmed/16025851
      OpenUrlPubMed
    13. ↵
      2. Mittelstadt B ,
      3. Benzler J ,
      4. Engelmann L , et al
      . Is there a duty to participate in digital epidemiology? Life Sci Soc Policy 2018;14(1).doi:10.1186/s40504-018-0074-1 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29744694
    14. ↵
      1. British Medical Association
      . Medical ethics today: the BMA's Handbook of ethics and law. London: BMJ Books & Wiley-Blackwell, 2012.
    15. ↵
      1. National Health Council
      . Principles of patients' rights and responsibilities [online], 2018. Available: http://www.nationalhealthcouncil.org/resources/nhc-publications/principles-patients-rights-and-responsibilities
    16. ↵
      2. Sulmasy LS ,
      3. Bledsoe TA , ACP Ethics, Professionalism and Human Rights Committe
      . American College of physicians ethics manual: seventh edition. Ann Intern Med 2019;170(S2):S1–32.
      OpenUrlCrossRef
    17. ↵
      1. World Health Organization
      . Adherence to long-term therapies: evidence for action [online]. Available: http://www.who.int/chp/knowledge/publications/adherence_report/en/ [Accessed 11 Aug 2018].
    18. ↵
      2. De las Cuevas C
      . Towards a clarification of terminology in medicine taking behavior: compliance, adherence and concordance are related although different terms with different uses. Curr Clin Pharmacol 2011;6(2):74–7.doi:10.2174/157488411796151110 pmid:http://www.ncbi.nlm.nih.gov/pubmed/21592067
      OpenUrlCrossRefPubMed
    19. ↵
      2. Snowden A
      . Medication management in older adults: a critique of concordance. Br J Nurs 2008;17(2):114–9.doi:10.12968/bjon.2008.17.2.28140 pmid:http://www.ncbi.nlm.nih.gov/pubmed/18414285
      OpenUrlPubMed
    20. ↵
      2. Krueger KP ,
      3. Berger BA ,
      4. Felkey B
      . Medication adherence and persistence: a comprehensive review. Adv Ther 2005;22(4):313–56.doi:10.1007/BF02850081 pmid:http://www.ncbi.nlm.nih.gov/pubmed/16418141
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      2. Cutler RL ,
      3. Fernandez-Llimos F ,
      4. Frommer M , et al
      . Economic impact of medication non-adherence by disease groups: a systematic review. BMJ Open 2018;8(1):e016982.doi:10.1136/bmjopen-2017-016982 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29358417
      OpenUrlAbstract/FREE Full Text
    22. ↵
      2. Fuller RH ,
      3. Perel P ,
      4. Navarro-Ruan T , et al
      . Improving medication adherence in patients with cardiovascular disease: a systematic review. Heart 2018;104(15):1238–43.doi:10.1136/heartjnl-2017-312571 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29572248
      OpenUrlAbstract/FREE Full Text
    23. ↵
      2. Garfield S ,
      3. Clifford S ,
      4. Eliasson L , et al
      . Suitability of measures of self-reported medication adherence for routine clinical use: a systematic review. BMC Med Res Methodol 2011;11(1):149.doi:10.1186/1471-2288-11-149 pmid:http://www.ncbi.nlm.nih.gov/pubmed/22050830
      OpenUrlCrossRefPubMed
    24. ↵
      2. Oberoi S ,
      3. Gupta VK ,
      4. Chaudhary N , et al
      . 99 DOTS. Int J Contemp Med Res 2016;3:2760–2.
      OpenUrl
    25. ↵
      2. Donner J ,
      3. Mechael P
      . mHealth in practice: mobile technology for health promotion in the developing world. London: A&C Black, 2012.
    26. ↵
      2. Bangsberg DR ,
      3. Hecht FM ,
      4. Charlebois ED , et al
      . Comparing objective measures of adherence to HIV antiretroviral therapy: electronic medication monitors and unannounced pill counts. AIDS Behav 2001;5(3):275–81.doi:10.1023/A:1011396711486
      OpenUrlCrossRef
    27. ↵
      2. Karumbi J ,
      3. Garner P
      . Directly observed therapy for treating tuberculosis. Cochrane Database Syst Rev 2015;(5).doi:10.1002/14651858.CD003343.pub4 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26022367
    28. ↵
      2. Waltz E
      . Drugs go wireless. Nat Biotechnol 2016;34(1):15–18.doi:10.1038/nbt.3446 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26744969
      OpenUrlPubMed
    29. ↵
      2. Labovitz DL ,
      3. Shafner L ,
      4. Reyes Gil M , et al
      . Using artificial intelligence to reduce the risk of nonadherence in patients on anticoagulation therapy. Stroke 2017;48(5):1416–9.doi:10.1161/STROKEAHA.116.016281 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28386037
      OpenUrlAbstract/FREE Full Text
    30. ↵
      2. Dotolo D ,
      3. Petros R ,
      4. Berridge C
      . A hard pill to swallow: ethical problems of digital medication. Soc Work 2018;63(4):370–2.doi:10.1093/sw/swy038 pmid:http://www.ncbi.nlm.nih.gov/pubmed/30137573
      OpenUrlPubMed
    31. ↵
      2. Chai PR ,
      3. Carreiro S ,
      4. Innes BJ , et al
      . Digital pills to measure opioid ingestion patterns in emergency department patients with acute fracture pain: a pilot study. J Med Internet Res 2017;19(1):e19.doi:10.2196/jmir.7050 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28087496
      OpenUrlPubMed
    32. ↵
      2. Chai PR ,
      3. Castillo-Mancilla J ,
      4. Buffkin E , et al
      . Utilizing an ingestible biosensor to assess real-time medication adherence. J Med Toxicol 2015;11(4):439–44.doi:10.1007/s13181-015-0494-8 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26245878
      OpenUrlCrossRefPubMed
    33. ↵
      2. Vlahović-Palčevski V ,
      3. Mentzer D
      . Postmarketing surveillance. In: Seyberth H , Rane A , Schweb M , eds. Pediatric clinical pharmacology. Handbook of experimental pharmacology. Vol 205. Berlin: Springer, 2011: 339–51.
      OpenUrl
    34. ↵
      2. Klugman CM ,
      3. Dunn LB ,
      4. Schwartz J , et al
      . The ethics of smart pills and self-acting devices: autonomy, truth-telling, and trust at the dawn of digital medicine. Am J Bioeth 2018;18(9):38–47.doi:10.1080/15265161.2018.1498933 pmid:http://www.ncbi.nlm.nih.gov/pubmed/30235091
      OpenUrlPubMed
    35. ↵
      2. Goodman KW
      . Chapter 8, Biomedical research, from genomes to populations. In: Ethics, medicine, and information technology: intelligent machines and the transformation of health care. Cambridge: Cambridge University Press, 2015: 121–46.
    36. ↵
      2. Kvarnström K ,
      3. Airaksinen M ,
      4. Liira H
      . Barriers and facilitators to medication adherence: a qualitative study with general practitioners. BMJ Open 2018;8(1):e015332.doi:10.1136/bmjopen-2016-015332 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29362241
      OpenUrlAbstract/FREE Full Text
    37. ↵
      2. Vayena E ,
      3. Tasioulas J
      . The dynamics of big data and human rights: the case of scientific research. Philos Trans A Math Phys Eng Sci 2016;374(2083):20160129.doi:10.1098/rsta.2016.0129 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28336802
      OpenUrlCrossRefPubMed
    38. ↵
      2. Ballantyne A ,
      3. Schaefer GO
      . Consent and the ethical duty to participate in health data research. J Med Ethics 2018;44(6):392–6.doi:10.1136/medethics-2017-104550 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29358219
      OpenUrlAbstract/FREE Full Text
    39. ↵
      2. Price W ,
      3. Nicholson II
      . Big data, patents, and the future of medicine. Cardozo L Rev 2015;37:1401.
    40. ↵
      2. Raghupathi W ,
      3. Raghupathi V
      . Big data analytics in healthcare: promise and potential. Health Inf Sci Syst 2014;2(1).doi:10.1186/2047-2501-2-3 pmid:http://www.ncbi.nlm.nih.gov/pubmed/25825667
    41. ↵
      1. Vitality Institute
      . Ethical, legal, and social implication of personalized health technology [online], 2015. Available: http://thevitalityinstitute.org/site/wp-content/uploads/2015/07/Ethical-Legal-and-Social-Implications-of-Personalized-Health-Technology.pdf [Accessed 11 Aug 2018].
    42. ↵
      2. Rocher L ,
      3. Hendrickx JM ,
      4. de Montjoye Y-A
      . Estimating the success of re-identifications in incomplete datasets using generative models. Nat Commun 2019;10(1):3069.doi:10.1038/s41467-019-10933-3 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31337762
      OpenUrlCrossRefPubMed
    43. ↵
      2. Paul G ,
      3. Irvine J
      . Privacy implications of wearable health devices. New York: ACM. Proceedings of the 7th International Conference on Security of Information and Networks; September 9-11, Glasgow, Scotland, 2014.
    44. ↵
      2. Rosenbaum S
      . Data governance and stewardship: designing data stewardship entities and advancing data access. Health Serv Res 2010;45(5 Pt 2):1442–55.doi:10.1111/j.1475-6773.2010.01140.x pmid:http://www.ncbi.nlm.nih.gov/pubmed/21054365
      OpenUrlCrossRefPubMedWeb of Science
    45. ↵
      1. U.S. Food and Drug Administration
      . Digital Health Innovation Action Plan [online], 2018. Available: https://www.fda.gov/downloads/medicaldevices/digitalhealth/ucm568735.pdf [Accessed 11 Aug 2018].
    46. ↵
      2. Scassa T
      . Data ownership. CIGI papers No. 187; Ottawa faculty of law working paper No. 2018-26. Available: http://dx.doi.org/10.2139/ssrn.3251542
    47. ↵
      2. Kolluri S
      . Mobile medical applications: the regulatory framework in the U.S. and the EU. MPO [online]. Available: https://www.mpo-mag.com/issues/2018-06-01/view_columns/mobile-medical-applications-the-regulatory-framework-in-the-us-and-the-eu
    48. ↵
      2. Zarsky TZ
      . Incompatible: the GDPR in the age of big data. Seton Hall L Rev 2016;47:995.
    49. ↵
      1. EU
      . Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General data protection regulation); 2016.
    50. ↵
      2. Izmailova ES ,
      3. Wagner JA ,
      4. Perakslis ED
      . Wearable devices in clinical trials: hype and hypothesis. Clin Pharmacol Ther 2018;104(1):42–52.doi:10.1002/cpt.966 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29205294
      OpenUrlPubMed
    51. ↵
      2. Pormeister K
      . Genetic data and the research exemption: is the GDPR going too far? Int Data Priv Law 2017;7(2):137–46.doi:10.1093/idpl/ipx006
      OpenUrl
    52. ↵
      2. Ballantyne A
      . How should we think about clinical data ownership? J Med Ethics 2020;46(5):289–94.doi:10.1136/medethics-2018-105340 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31911499
      OpenUrlAbstract/FREE Full Text
    53. ↵
      2. Salerno J ,
      3. Knoppers BM ,
      4. Lee LM , et al
      . Ethics, big data and computing in epidemiology and public health. Ann Epidemiol 2017;27(5):297–301.doi:10.1016/j.annepidem.2017.05.002 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28595734
      OpenUrlPubMed
    54. ↵
      1. Vitality Institute
      . Guidelines for Personalized Health Technology: Final Report [online], 2016. Available: http://thevitalityinstitute.org/site/wp-content/uploads/2016/03/Guidelines-for-Personal-Health-Technology_March2016_FINAL.pdf [Accessed 11 Aug 2018].

    Footnotes

    • Contributors HS proposed the overall angle. JK wrote the first draft of the manuscript and led all revisions of subsequent versions. ACK, NIC and HS drafted substantive subsections and critically reviewed iterations of the manuscript. All authors discussed the main line of argument and approved the final version.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Patient consent for publication Not required.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data availability statement There are no data in this work.

    Other content recommended for you