Governance of a Learning Health Care System for Oncology: Patient Recommendations

PURPOSE:

The learning health care system (LHS) was designed to enable real-time learning and research by harnessing data generated during patients’ clinical encounters. This novel approach begets ethical questions regarding the oversight of users and uses of patient data. Understanding patients’ perspectives is vitally important.

MATERIALS AND METHODS:

We conducted democratic deliberation sessions focused on CancerLinQ, a real-world LHS. Experts presented educational content, and then small group discussions were held to elicit viewpoints. The deliberations centered around whether policies should permit or deny certain users and uses of secondary data. De-identified transcripts of the discussions were examined by using thematic analysis.

RESULTS:

Analysis identified two thematic clusters: expectations and concerns, which seemed to inform LHS governance recommendations. Participants expected to benefit from the LHS through the advancement of medical knowledge, which they hoped would improve treatments and the quality of their care. They were concerned that profit-driven users might manipulate the data in ways that could burden or exploit patients, hinder medical decisions, or compromise patient-provider communication. It was recommended that restricted access, user fees, and penalties should be imposed to prevent users, especially for-profit entities, from misusing data. Another suggestion was that patients should be notified of potential ethical issues and included on diverse, unbiased governing boards.

CONCLUSION:

If patients are to trust and support LHS endeavors, their concerns about for-profit users must be addressed. The ethical implementation of such systems should consist of patient representation on governing boards, transparency, and strict oversight of for-profit users.

INTRODUCTION

In 2006, the Institute of Medicine (IOM) convened patients, providers, researchers, and policy makers to discuss a vision for the future of health care. Their objective was “development of a learning health care system that is designed to generate and apply the best evidence for the collaborative health care choices of each patient and provider; to drive the process of discovery as a natural outgrowth of patient care; and to ensure innovation, quality, safety, and value in health care.”¹ This initiative was born out of the promise of emerging technologies, such as electronic medical record (EMR) databases that could harness data generated during routine clinical encounters to enable real-time learning and collaborative research.²

In response, ethicists proposed a new framework in which providers and patients alike have a moral duty to actively contribute to learning systems.³ Obligations to patients include respecting rights and dignity, providing the best care, and avoiding unnecessary risks and burdens.³ Unsurprisingly, ethicists and institutional review boards (IRBs) have struggled with the dilemma posed by learning health systems (LHSs): how does one oversee such systems in a way that will foster research progress without harming the physician-patient relationship?^4,5

In an era of public distrust of big tech companies^6,7 and concern over the possible exploitation of patients,⁸ attention to legal and ethical uncertainties surrounding the governance of patient data-sharing has grown. Questions have arisen regarding how users and uses of patient data should be governed and by whom, with consideration of patients’ opinions being vital.^6-8

Surveys, interviews, and focus groups have been used in previous studies to investigate attitudes concerning LHS consent processes and patient trust regarding different users and uses of EMR data.^9-11 However, traditional data collection methods are inadequate in accessing patients’ informed and well-considered opinions about issues central to policy decision making, given the complex scientific, legal, ethical, and historical considerations that are unfamiliar to the average layperson.¹²

Such policy questions are appropriate topics for public deliberation—“a method of public consultation in which members of the public come together to engage in informed dialogue about difficult or complex social issues” in ways that can ground and legitimize policy decisions.¹³ Hence, this study used democratic deliberation¹² to address the challenging policy issues surrounding LHS governance and to elicit patients’ informed and considered opinions.

MATERIALS AND METHODS

The University of Michigan IRB determined this research to be exempt. Full methods are reported elsewhere.¹⁴ Between June 2017 and May 2018, the research team hosted four democratic deliberation sessions focused on ethical issues associated with CancerLinQ, a real-world LHS established by ASCO.^15,16 A total of 217 patients with cancer participated in the sessions.

At each deliberation, experts presented educational content grounded in the concept of the LHS as articulated by the IOM and applied to oncology.^1,17 The presentations were followed by participant discussions that were recorded and transcribed. The first round of discussions focused on disclosure and consent; these were examined separately and reported elsewhere.¹⁸ We present here our analysis of the second round of discussions, which focused on the governance of potential users and uses of data.

Experts presented LHS aims, ethical and logistical issues, and policy choices and then asked participants to consider whether policies should permit or deny certain users and uses of secondary data. Participants were instructed to choose policies that they thought would be best for other patients and for society in general. Votes were cast at the start and end of the discussion, and participants were prompted to explain the factors influencing their votes.

Data Analysis

Two analysts (R.D.J. and C.K.) established a coding system after reviewing and discussing the transcript. Using Dedoose (version 8.0.35), they then coded each transcript independently and met to resolve inconsistencies and update the codebook. The analysts and the principal investigator (R.J.) organized the finalized codes into themes using a thematic analysis approach.¹⁹

RESULTS

Participants were 67% female and 61% non-Hispanic White; mean age was 60 years. Many were college graduates (51%). The most common cancer type reported was breast cancer (42%). A nontrivial minority indicated having fair to poor health (23%), and almost a third (30%) were told by a doctor that their cancer was metastatic or incurable.

Participants voiced opinions on potential users and uses of secondary data in LHSs. These were organized into two thematic clusters: their expectations and concerns, which seemed to inform their LHS governance recommendations (Fig 1; Table 1).

FIG 1.

Thematic map.

TABLE 1.

Major Themes and Exemplary Quotes

Expectations and Concerns

Participants’ expectations and concerns regarding potential users and uses of LHS data can be summarized in four points: (1) LHSs have value for research and quality improvement. (2) LHS data should facilitate physicians’ recommendations, and insurance companies should have limited involvement in medical decision making. (3) Research motivated by profit, which many associate with pharmaceutical and insurance companies, raises suspicions that patients’ best interests will not be prioritized. (4) Marketing is usually associated with profit and, thus, is often perceived as spreading unwarranted or misleading information.

Overall, most participants recognized that LHSs could be important for research and quality improvement. Their discussions highlighted the potential to advance medical knowledge and support evidence-based care. It was commonly expressed that LHS data should be freely available to physicians and researchers to ensure that patients receive appropriate and up-to-date treatments. The involvement of insurance companies was generally perceived negatively and was poorly understood. Many agreed that LHS data should be used to inform clinical decisions driven by doctors rather than insurers who might deny clinically effective treatments to reduce costs, although certain countervailing viewpoints acknowledged the role that insurance companies can play in controlling inappropriate costs by approving established treatments. As participants discussed how and why insurers are involved, some became more accepting of the ways that insurers can be helpful. In addition, the use of data for marketing was worrisome to numerous participants who voiced concerns that for-profit pharmaceutical companies, and even hospitals in some cases, might encourage the spread of misleading information that could affect patient-provider communication and medical decision making. However, a few thought that it would be helpful to generate more awareness by the general public about available health care options if the advertised information could be verified.

LHS Governance Recommendations

Table 2 shows participants’ final policy votes after group discussion. Most expected to benefit from the LHS through the advancement of medical knowledge, which they hoped would improve treatments and quality of care available to themselves and others. Therefore, most were accepting of hospitals, universities, and pharmaceutical companies using LHS data to develop treatments. Participants were primarily concerned with the possibility that profit-driven users of LHS data might manipulate the data in ways that could burden or exploit patients, hinder medical decisions, or compromise patient-provider communication. Consequently, they were more often against LHS data requests associated with marketing, patents, or insurance company recommendations and coverage.

TABLE 2.

Final Votes After Group Discussion

Participants’ discussions generated five recommendations for oversight of LHS users and data uses: (1) A balanced and unbiased (unmotivated by profit) governing board should represent a variety of stakeholders, including patients. (2) LHS user motivations and research outcomes should be made transparent to patients. (3) Certain uses should be subject to access restrictions, especially if users are profit-driven or unqualified to interpret health data. (4) Certain users should be required to pay for access, especially those motivated by profit. (5) Penalties for use violations should be outlined and enforced.

Participants desired assurance that patients’ best interests would be prioritized in all determinations made concerning the LHS. Several recommended that LHS governing boards include patient representatives along with experts, such as physicians, lawyers, and ethicists, who could ensure that patients are not burdened or exploited by the system. It was also specified that members of the governing board should remain unbiased and should not be beholden to outside influences or motived by profit. These boards would be expected to govern the users of the data and the types of uses that are permissible. A number of participants proposed that patients should be notified about important discoveries and matters that might affect their treatments or the quality of their health care, such as when potential LHS users have financial conflicts of interest that might lead them to push certain findings. Many wanted to prevent LHS data from being manipulated in ways that might adversely affect patient care. Their recommendations stipulated that for-profit entities, especially pharmaceutical and insurance companies, must make clear how they intend to use the data, and if necessary, access should be restricted, fees should be charged for data use, and sanctions should be imposed if the data are not used accordingly. Finally, some participants also pointed out the risk of LHS data being unintentionally misconstrued. They suggested that LHS data should not be made fully accessible to individuals who are not qualified to conduct analyses.

DISCUSSION

This study generated patients’ informed and considered opinions on the challenging bioethical policy issues surrounding the governance of LHS users and uses. Although previous surveys, interviews, and focus groups have explored patients’ level of trust regarding different users and uses of EMR data,^9-11 our use of democratic deliberation methods is particularly innovative in that the participants received education by experts and discussed their viewpoints with others before rendering their final decisions. Not only did these deliberations underscore patients’ expectations and concerns, but they also generated valuable recommendations regarding the policies and procedures that should constitute LHS governance.

In these discussions that focused on potential users and uses of LHS data, many participants voiced suspicions that for-profit entities, especially pharmaceutical and insurance companies, would be motivated by profit at the expense of patients’ best interests. Several participants vividly described their own personal experiences with having their coverage denied or their frustrations with the high costs of health care, which may have contributed to their fear of being exploited. Certain uses that were perceived to be associated with profit, such as marketing, obtaining patents, or determining coverage and reimbursement, were seen as concerning. However, our deliberations allowed the reasoning behind mixed or opposing viewpoints to be acknowledged and examined. Indeed, there were participants who reasoned that patents might be needed to incentivize research, that insurance companies might help keep costs under control or ensure that established treatments are reimbursed, and that marketing might be useful in spreading awareness about health care options. Thus, this deliberation suggests that patients can be engaged in sophisticated discussion of the complex and controversial ethical questions involved in the governance of LHS initiatives.

Reassuringly, the findings of this study confirm that patients’ expectations overwhelmingly concur with the vision that the IOM had initially outlined for the purpose of LHS initiatives.¹ The majority of participants voiced a strong desire for research and quality improvement that can advance medical knowledge and support up-to-date, evidence-based patient care. Many also emphasized the importance of aiding and upholding patient-provider communication and decision-making. These expectations demonstrate that patients understand the innovation of the LHS and have hope in the promise of its contribution to the future of health care. They simply want assurance that the advancement of medicine will remain the primary goal of such systems, that patients’ best interests will take precedence, and that potential abuses will be appropriately deterred. These convictions are in line with those reported in previous research studies that suggest widespread support among patients when it comes to data sharing for health research, but only under the desired conditions of reflecting patients’ values, minimizing risks, and ensuring transparency, control, trust, and accountability.²⁰ It has been previously asserted that such conditions should be implemented in a “value-based governance framework that incorporates the diverse patient and public values, needs, and interests.”²⁰

Indeed, a particularly compelling finding of these discussions regarding oversight of potential users and uses of LHS data was the desire for patients to have an ongoing voice in LHS governance decisions. Participants generally were receptive to the idea of including patients’ voices in governing processes in a meaningful way, as evidenced by their lively discussions and discerning comments culminating in recommendations for how LHS users and uses should be regulated. Participants proposed the inclusion of patients as stakeholders on formal governing boards to maintain balanced and impartial representation. They recommended restricted access, user fees, and penalties to hold users accountable and to prevent misuse of data. The importance of transparency and patient notification was stressed, especially in situations that might involve dubious user motivations or biased research findings. Hence, the results of the current deliberation incorporate patients’ informed and well-reasoned judgments into a detailed description of how a governance framework might be operationalized.

Of note, these governance recommendations reflect sentiments expressed earlier in the deliberation day when participants discussed their expectations for informed consent, as reported previously.¹⁸ Most wanted to be notified about LHS activities, with several citing the importance of protecting patients from various harms. Because broad or blanket consent was commonly perceived as limiting the amount of control that patients can exercise in uncertain situations, some expressed the desire to specify explicit restrictions over potential users and future uses of their shared data. Yet, more granular consent may not always be feasible in some complex, ever-evolving systems and overly burdensome consent processes may hinder important research advances. Ethicists have contemplated these and other challenges of informed consent arising from the intricacies of health research in the era of big data.^6,7,21 To compensate for such limitations, some have proposed a “systemic oversight approach” that can foster reflexivity and inclusiveness by engaging patient stakeholders.²¹ This might include the institutionalization of deliberative democracy approaches in the form of advisory boards that can solicit input from lay members of the community. Similarly, others have proposed that patients should be trained to be integral members of group-based governance committees. These committees would make informed decisions about “purpose-specific” data requests “contingent upon the user’s fulfillment of ongoing reporting obligations.”⁶ The results of the current deliberation underscore that patients grasp the importance of representation on formal governing boards that can provide this type of ongoing oversight. Their recommendations confirm that involvement at the level of system governance can be a means of exercising specific control that may not be achievable solely through consent obtained at the individual level.

Finally, a commonly discussed concern was the potential for abuse that could arise from the undue influence of for-profit companies. This apparent distrust of the profit motive and the desire to deter exploitative practices by certain users that are perceived to be less trustworthy seemed to be a primary factor informing participants’ governance recommendations. Correspondingly, concerns have grown about companies benefiting financially through the use of patients’ data.^7,8 Ethicists have proposed the community benefit model,⁸ in which members of the community would play an active role in deciding how they would benefit from the sharing of patient data. This could be accomplished through the establishment of a development fund board “controlled by representatives of the relevant patient community.”⁸ Companies that generate profits through the use of patient data would be required to pay the board, and the board could then decide how to invest those funds in the community. As the current deliberation suggests, patients are aware of the potential to generate profits from the use of their data and are worried about potential harms. Including patients on formal governance bodies, such as institutionalized democratic deliberation sessions, community development fund boards, and other such committees, seems to be a key way to build patient trust and ensure respect for their wishes.

Some limitations to this study should be considered. It is possible that the educational component may have swayed participants’ views; however, efforts were made to ensure balanced content and deliberation by incorporating input from a diverse group of contributors when developing the materials and by meticulously selecting and training discussion facilitators. For example, the idea that certain users or uses of data might require different levels of vigilance was a general concept raised in the educational presentations, and the different user/use combinations for voting provided some degree of framing such that the specific concerns about for-profit use and use by industry and insurers may have been more prominent. However, patients in each cohort raised specific concerns on their own about for-profit use by industry and insurers, often citing examples from their personal experience. In addition, patients who were too infirm or who lacked transportation resources would have been unable to attend the full-day, in-person session. Those who did attend might have inherently been less concerned about privacy issues or more supportive of research in general. Moreover, patients who have been diagnosed with cancer may be more likely to see a clear link between research, medical advances, and their own quality of care and well-being and, thus, may support access to their data more so than those who have never been diagnosed. Therefore, the results of this study might not be generalizable to a broader patient population.

Still, a notable strength of this study is its use of deliberation to elicit the informed and well-reasoned opinions of patients on complex policy issues surrounding LHS governance. Participants engaged in vigorous discussions that, on the whole, exhibited characteristics of well-executed deliberation, such as balanced participation, recognition of others’ viewpoints, consideration of society’s best interests, and reflective thinking to support personal opinions.²² Thus, the wealth of judicious feedback from patients amassed during these high-quality deliberations should be instrumental in apprising LHS stakeholders of the important dimensions that should constitute system oversight and governance.

In conclusion, if patients are to trust and support LHS endeavors, their concerns about for-profit users and the potential for abuse must be addressed. Beyond robust approaches to individual informed consent, the involvement of trained, informed patients in decision making at the level of system governance is essential. Those considering enrolling patients in such systems should select systems that include patient representation on governing boards; provide transparency regarding users, uses, and outcomes (including the requirement that users publish their findings as a way to earn and maintain patient trust); and ensure surveillance and ongoing oversight of for-profit users. Policy makers and system administrators should heed the reasoning and insights highlighted here to allow patients’ voices to remain central to the development of LHS initiatives.

AUTHOR CONTRIBUTIONS

Conception and design: Kent A. Griffith, Rebecca Spence, Angela R. Bradbury, Raymond De Vries, Reshma Jagsi

Financial support: Reshma Jagsi

Administrative support: Reshma Jagsi

Provision of study materials or patients: Robin Zon, Richard L. Schilsky, Reshma Jagsi

Collection and assembly of data: Rochelle D. Jones, Chris Krenz, Rebecca Spence, Raymond De Vries, Sage Bolte, Navid Sadeghi, Reshma Jagsi

Data analysis and interpretation: Rochelle D. Jones, Chris Krenz, Kent A. Griffith, Angela R. Bradbury, Raymond De Vries, Sarah T. Hawley, Robin Zon, Richard L. Schilsky, Reshma Jagsi

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Governance of a Learning Health Care System for Oncology: Patient Recommendations

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).