BACKGROUND

The National Academy of Medicine and others advocate the development of learning health systems (LHSs). Ideal LHSs integrate observational data, quasi-experimental designs, and pragmatic trials to track care, test interventions to improve care, implement successes, and share findings. This integration of learning across the system poses challenges for governance, because traditional governance models impose substantial burdens on activities defined as research, while failing to ensure participant protection in other learning activities. Although governance of learning is central to the missions of LHSs, we know little about how they carry it out or engage patients and families in the process.

OBJECTIVES

Our long-term objective was to describe governance structures and processes for learning activities within LHSs, with particular attention to the roles of patients and families. We pursued the following 4 aims: Describe how LHSs define, identify, and classify learning activities. Describe the governance mechanisms that LHSs use to oversee learning activities. Describe how LHSs engage patients and families in governance. Identify best practices for governance.

METHODS

This 3-year qualitative study combined semistructured interviews of LHS leaders, focus groups with community members, and iterative stakeholder input. Study sites represented a diverse set of LHSs selected from Association of American Medical Colleges Research on Care Community members and referred by subject matter experts. We interviewed 99 individuals from 16 LHSs, including chief executive, nursing, medical, quality, information, and experience officers; IRB directors; patient/family advisory council chair(s), members, or other patient and family leaders; and directors of research or their equivalents. We then conducted 8 patient and family focus groups with individuals who received care in 4 catchment areas representing 5 LHSs. Each focus group included a diverse group of 5 to 11 participants (N = 77). Interviews focused on governance structures (eg, Who is responsible for oversight?), processes (eg, How is governance integrated across learning activities?), dilemmas, and possible solutions. Throughout, we emphasized the roles of patients and families. Focus groups sought participants' views regarding patient and family participation in LHS governance. We recorded interviews and focus groups, then transcribed and imported them into ATLAS.ti. Qualitative analyses combined an inductive approach that built a coding structure using the constant comparative method with a deductive approach that began with a conceptual framework to guide the analysis. After initial coding, we conducted higher-order analysis, including axial coding, theme development, and generation of hypotheses.

RESULTS

Analyses focused on how LHSs define the line between research and quality improvement (QI) for governance purposes and on the roles of patients and families in governance. Our findings include the following: Interview respondents emphasized the fine line between research and QI. Characteristics that favor categorizing an activity as QI rather than research include that the activity aims to implement a proven intervention or to more effectively apply existing knowledge; does not intend to publish results; aims to affect care locally rather than more broadly; does not involve randomization; does not collect data from patients beyond those required for routine care; does not involve practices that deviate from routine care; does not involve external funding or collaborations; intends to implement findings in real time; does not impose significant additional risk on patients; does not alter clinician-patient relationships; and intervenes in health-system operations and processes rather than directly with patients. We observed some level of patient and family engagement in governance within many systems. We also observed several exemplars of high-level patient integration into the learning mission. Overall, however, systems were more likely to seek input from patients and families for specific activities or their component parts than to systematically partner with them in prioritizing, designing, overseeing, implementing, or disseminating learning. Although focus group participants endorsed patient and family involvement in governance, they particularly emphasized the need for transparency regarding the conduct and outcome of learning activities.

CONCLUSIONS

Despite the vision of bridging boundaries between QI and research, LHSs continue to distinguish between these activities for governance purposes. No single characteristic defines the boundary between these activities; rather, numerous features combine to favor characterizing a particular activity as one or the other. In addition, even in LHSs that have developed sophisticated engagement structures and processes, few have substantially integrated patients and families into governance of their learning missions.

LIMITATIONS

Limitations include the fact that (1) no standard definition or consensus list of LHSs exists; and (2) results are based on interviews of limited numbers of individuals at 16 LHSs, precluding comprehensive characterization of any individual LHS or of the landscape of LHSs as a whole.