IDEA4PS: The Development of a Research-Oriented Learning Healthcare System

Abstract

Leveraging opportunities to learn and then improve the delivery of care using experiences throughout the healthcare system is essential in efforts to transform healthcare delivery. We present the approach of one academic medical center in becoming a Research-Oriented Learning Healthcare System (ro-LHS). By reframing the role of research in improving outcomes, the organization was able to move beyond its focus on quality improvement to foster a culture in which feedback informs practice and research drives improvement. Our patient safety learning laboratory, the Institute for the Design of Environments Aligned for Patient Safety (IDEA4PS), funded by the Agency for Healthcare Research and Quality, has provided foundational infrastructure to connect stakeholders across the medical center and University and conduct rigorous research in the context of practice. With this new focus, research now informs our operations in a cycle of continuous improvement across our ro-LHS.

INTRODUCTION

A Learning Healthcare System (LHS) is described as an organizational approach to health services delivery where efforts to improve both efficiency and effectiveness are grounded in experience with the populations served.^1–3 In practice, a LHS seeks to optimize the structure and process of care delivery with the goal of improving the organization’s ability to get the right treatment to the right person at the right time. While these efforts have been greatly supported by increased access to data made available by Electronic Health Records (EHRs), developing a system in which feedback informs practice and research drives improvement remains challenging.

Considering the motivations that drive improvement efforts, Mittman ⁴ contrasted the approaches of Quality Improvement (QI) as problem-driven and Implementation Science as solution-driven. Mittman ⁴ claims that an implementation science approach emerges from a recognition that organizations experience translational roadblocks that result in a lag between what we know and what we do. In practice, there is a distinction between problem-driven and solution-driven organizations. Solution-oriented organizations frame their experiences so as to focus on identifying generalizable knowledge beyond the context in which it is generated. Problem-driven organizations, however, seek to improve outcomes by reducing the causes of variation, often focusing on immediate needs for improvement in an iterative, rapid-cycle approach.

In this context, we propose that a LHS can be either problem-driven, as a research-oriented LHS, or solution-driven, as a quality-oriented LHS. We define a “research-oriented LHS” as one that frames health service improvement efforts in terms of their ability to offer insights into the experiences of organizations that face similar challenges. This stands in contrast to quality-oriented LHS (qo-LHS) where the cultural focus is on health service improvement through addressing quality issues. The distinction between a qo-LHS and a ro-LHS thus offers two different perspectives to explain how learning is framed within the context of care delivery.

This paper describes the experience of a large, Midwestern academic medical center (AMC) in developing a “research-oriented LHS” (ro-LHS). While the mission of an academic medical center is education, research and clinical service^5,6, in many cases the research is siloed and as a result, the culture of a LHS can come to be dominated by these three mission components in differential ways. The adoption of a ro-LHS represents a strategic choice by the organization to expend the resources necessary to generate, integrate, and disseminate research throughout the institution and promulgate those findings for the greater good.

RESULTS

Making the Shift from a Quality Orientation to Research

The cultural shift towards a ro-LHS at this AMC has taken place over the past seven years, during which time Quality and Patient Safety has seen significant expansion in their role across the medical center. Functioning under the responsibility of a single leader, the Chief Quality Officer (CQO), individuals seeking to improve their ability to act as agents of change have embraced the Define, Measure, Analyze, Improve and Control (DMAIC) model as central to the QI process⁷, despite the absence of a formal deployment across the medical center. Nursing, physician and administrative teams in the clinical areas, called Operational Councils, take the lead on identifying issues, developing solutions and working within an accountable framework.⁸ However, while this grassroots approach contributed to multiple awards and high quality designations, the medical center’s approach was that of a qo-LHS.

In 2011, when the healthcare system adopted a new EHR, the CQO identified a window during which organizational culture could be reframed.⁹ Because the EHR implementation was facility-wide, it required an interdisciplinary approach to redefine practice patterns.¹⁰ The availability of new tools and data improvements, in concert with a general acceptance that practice patterns would be required to change, allowed for the organization to consider an implementation science approach, and the transition to a ro-LHS began.

Building a ro-LHS required that nurses, physicians, staff and researchers see themselves as equal partners in improvement efforts. While a number of investigators and clinicians had been conducting a variety of health services and clinical research projects across the institution, these efforts had been largely independent. However, with growing interest in implementation science, the CQO convened a stakeholder steering group composed of champions in each of these domains.

One of the first challenges of this group was to build a shared language, a common initial step in multidisciplinary collaborations. Additionally, members of the group had to develop a shared understanding of the constraints they faced with improvement efforts. For instance, research interventions often require associated changes to practice in ways that are new and untested. Under such conditions, it is necessary for all stakeholders to be invested in the process as equal partners, thus another step involved achieving stakeholder agreement that they would serve as ambassadors to the communities they represented with the understanding that interventions to be tested would be respectful of the needs and preferences of all involved. Moreover, stakeholders agreed to defer to the collective in exploring and expanding research and practice at the health system level. As a result, a process had to be developed to review what had been independent quality efforts, channeling decisions through the CQO who was responsible for ensuring that, when appropriate, the collective could weigh in on the potential for incorporating research designs where broader knowledge could be gleaned.

Transitioning to a ro-LHS required stakeholder leaders to commit both financial and non-financial resources necessary to be successful. While QI-driven processes seek to solve administrative and logistical variation in practice, shifting to a research-driven culture requires that stakeholders pay the transaction costs associated with doing health services research in a world dominated by clinical imperatives. These costs were often experienced in terms of the dedicated time necessary for clinical provider, technicians, and health services and outcomes investigators, to identify if workable solutions could be developed that might offer generalizable knowledge. When the possibility was identified, the stakeholder groups invested time and energy appropriately framing interventions to focus on the contextual factors that might explain the heterogeneity of treatment effect, create data conduits to explore if the interventions matter, and eventually develop grant ideas together and respond to calls for submissions. Additionally, effective leadership and coordination were imperative to the success of the ro-LHS such that the practice of research was consistently supported and rewarded appropriately.

Integration of problem-driven thinking into practice at the AMC was seen as key to success not only operationally but also in the ability to formulate research questions that had not previously been asked. For instance, in 2010, the AMC adopted Crew Resource Management (CRM) across the institution with the goal of training over 8,000 faculty, staff and nurses in how to work most effectively in teams and how to escalate patient safety concerns appropriately. ^11,12 This effort served as the cornerstone for many collaborative research efforts and has remained the cornerstone of the AMC’s dedication to continuous improvement.

Starting Small: Precision Application of the ro-LHS

With a common model for communication in place, we collectively sought to identify opportunities to increasingly engage the entire healthcare delivery system. Given the significant issue of falls as a source of risk institution, the stakeholder group identified gaps in expertise in human factors, analytics and data infrastructure focused on research and began to build a framework that could be used to drive organizational change. Within that context, we designed and deployed the “Falls Wheel” – a visual display presented for every patient that identifies the potential risk for a patient to fall”.¹¹ Daily nursing assessment of a patient’s risk now involves nursing who mark the wheel to identify a patient’s risks using a four color coding system. Green, for example, indicated a low risk of fall and universal nursing precautions could be taken, whereas red indicates a very high risk of fall with injury necessitating additional strategies such as bed alarms and hourly rounding. After a pilot study showed the intervention was effective, the Falls Wheel was deployed across the AMC, and represented one of our first attempts to engage patients, nurses, physicians and researchers in an implementation science approach consistent with a ro-LHS.

Another project centered on the Agency for Healthcare Research and Quality’s (AHRQ’s) Patient Safety Indicators (PSIs). While these PSIs were initially meant to trigger awareness about unsafe situations in the hospital, they are also now used in ranking and rating systems. As a result of increased focus on PSIs, our corresponding effort to manage these indicators has been substantial. With research and analytics we have shown that increased accountability for PSIs requires a management strategy rather than an improvement strategy.¹² Through ongoing collaboration among data analytics experts, researchers and QI strategists, we have been able to improve our measurement processes, and have enabled operationally-focused data managers to consider research opportunities rather than only local QI initiatives.¹³

In a third example, we sought to use a research frame to overhaul the process whereby we monitor patients within the acute inpatient care setting. Using established guidelines for telemetry, we sought to measure compliance with policies around appropriate levels of care. Based on the robust measurement model the collaborative developed, we considered how alarms tones might be used to differentiate issues and provide sound cues as part of the alerts about the criticality of the events. With thoughtful intervention and support from human factors engineering, the organization was able to reduce telemetry days and increase ED throughput with no effect on mortality.¹³ Furthermore, with active engagement through focus groups and real-time intervention by the clinical teams, researchers were able to innovate around alarm tones and expand their work to other types of alarms.^11,14

Formalizing the Research-Oriented Learning Health System

In the Spring of 2015, the Agency for Healthcare Research and Quality (AHRQ) released a Request for Applications (RFA) entitled Patient Safety Learning Laboratories: Innovative Design and Development to Improve Healthcare Delivery Systems (RFA-HS-15–001). The RFA spoke to a recognized scarcity of programmatic activity that engaged in design and systems engineering efforts focused on more than singular patient safety concerns. Specifically, the RFA sought to support “laboratories” that could enable multiple develop-test-revise iterations of promising design features and subsystems of the sort that could normally be found in larger-scale engineering projects. AHRQ asked for two to four closely related projects that focused on well-known, costly, patient safety harms in a given clinical area, and for which new and innovative design approaches were needed.

Building on processes and programs already implemented, some of which have been described above, researchers and practitioners in our AMC together outlined a research approach designed to identify and explore how feedback of information could be used to shape the development of robust practices that could lead to improved patient safety across the care continuum. This perspective was grounded in both the literature and our experience, and we believed we could advance a compelling argument in support of using an “information engineering” approach in which clinical practice could be advanced by designing, testing and exploring information flows.

Expanding and formalizing our original collaborative process, we established the Institute for the Design of Environments Aligned for Patient Safety (IDEA4PS). Engaging 23 investigators from more than 8 departments from across the University campus (see Figure 1), IDEA4PS is also supported by a Patient Safety Advisory Committee and Patient Safety Champions, as well as an internal advisory team and external advisory committee with representatives from across the globe (see Figure 2).

Figure 1.

The Institute for the Design of Environments Aligned for Patient Safety (IDEA4PS) engages 23 investigators from more than 8 departments from across the Ohio State University campus

Figure 2.

The Institute for the Design of Environments Aligned for Patient Safety (IDEA4PS) consists of three projects and three cores, including the Administrative and Implementation Core, the Human Factors and System Engineering Core and the Informatics, Analytics and Data Core. Additionally IDEA4PS is supported by an Internal Advisory board, an External Advisory Board, and the Patient Safety Advisory and Link to Patient Safety Champions hospital committees. Patient safety pilot projects are supported by the three cores.

DISCUSSION

Lessons Learned

In our development of a ro-LHS, we made a number of discoveries that were, for our organization, unexpected. First, our experience has reinforced our belief that you cannot build research on bad process nor with insufficient resources and insufficient infrastructure. Solving problems in a clinical environment that is not under statistical process control can result in scenarios where the effect of an innovation is indecipherable from the noise in which it is embedded. Similarly, without resources and a dedicated understanding of the benefits of a ro-LHS, the tension between QI and research will endure. Our approach is not to silo research and quality, but to explore how the two perspectives can work synergistically, with a particular focus on how information can be used to shape the process over time. In this way, clinicians and researchers, inclusive of health services experts, analysts and information technology (IT) staff work together to improve operational excellence and performance management processes.

Second, research is successful when there is sufficient institutional commitment to change practice regardless of whether a project receives funding. The three projects that were chosen for the study in IDEA4PS were based on initiatives that we had already started - telemetry and alarms, surveillance of healthcare-associated infections, and CRM and secure messaging within an inpatient portal. These projects had been identified through a collaborative and iterative process that sought to identify where the organization might both improve practice and contribute to generalizable knowledge. The Telemetry and Alarms project was designed to focus on the signal-to-noise problem experienced by clinicians with and aiming to improve both safety and care quality through designing, testing and retesting tones and sounds as alarm replacements. The Surveillance project involved research focusing on hospital safety events, including healthcare-associated infections (HAIs), to develop processes for real-time recognition and location-aware problem solving, sometime referred to as digital hot spotting, using real-time detection of infections as an innovative way to save lives. Finally, the Secure Messaging project was focused at how information flows coming from a newly implemented inpatient portal would impact care team communication dynamics. This inpatient portal introduced a secure messaging feature through which patients and providers could communicate, but little was known about how this new portal would impact workflow or patient care.

Each of these cases suffered from a similar challenge. Each case was grounded in how the implementation of a new technology could present challenges to practice while at the same time offering an opportunity to manage the concerns that the new technology might bring. In considering potential projects, we sought the most compelling rationale for selecting a given project. Because federal research applications are evaluated on two factors – significance and broader impact – we adopted the rule that significance could be measured by the dollars the organization was willing to invest to solve the problem. In doing so, the collaborative adopted a heuristic that the promise of federal funding should not drive practice or even experimentation.

We also have found that we are most successful when we build research on established relationships. Research in the context of practice requires a significant amount of knowledge and trust. Building a team that is inclusive allows for the development of collaborative resources that would not be otherwise available. Increasing collaboration requires reaching out and identifying those with an interest in research who might want to partner. In some cases, we have sought national or international partners to bring in specific knowledge, and increasing this collaborative capacity can pay significant research dividends.

Selecting among competing projects required that we identify when there was the capacity to do things differently, while at the same time provide the opportunity to explore new and important concepts, constructs and perspectives. The domain of possible research that is important to the institution, the patients and the providers is incredibly wide, and when that research is of common interest, it is much easier to successfully test the assumptions associated with a proposed intervention. With supportive infrastructure, including data and administrative resources, additional research questions can be asked and answered even when no additional funds have been secured. In this respect, the IDEA4PS is more a laboratory than an experiment where social capital and trust are the reagents and the collaborative structure is the instrumentation. Having a solid ro-LHS with guidance and direction is incredibly beneficial in this regard.

CONCLUSIONS

We have presented the approach one academic medical center pursued to become a ro-LHS. Reframing the role of research in improving outcomes has offered our institution the ability to build the capabilities, willingness, and resources to establish a different kind of learning culture. The premise of the LHS is one that has been embraced and theoretically endorsed for years.^1–3 Translation of this approach may be difficult, but we suspect that the model we have presented can be applied in other organizations. We are hopeful that our story will inspire other academic medical centers to take their operations, QI and research infrastructure and similarly leverage the potential of a ro-LHS.