Toward a Learning Health Community: Challenges and Opportunities

After attending the annual American Medical Informatics Association (AMIA) Symposium, I enjoy reflecting on the trends and innovations that I have seen-and then share those reflections with anyone who will listen! You, dear readers, are the designated recipients of my AMIA 2014 musings. As I thought about the scientific reports I heard and informal conversations I took part in, four themes converged into a single take-away: We have the vision, technology, and scientific know-how to improve our country's health outcomes by designing and implementing a national Learning Health Community (LHC). This is an amazing opportunity, but one that requires both very hard work and will power. Let me tell you why.

A Learning Health Community

Perhaps the most ambitious proposal for using Big Data I have heard to date is "The Learning Health Community (LHC)" ( www.LearningHealth.org) , which currently is more vision than reality, but a vision that appears to "have legs" and has gathered considerable support. Initially, a 2012 Learning Health Care Summit developed core values to guide the design and operation of the proposed national-scale system and gained endorsements from major stakeholders. The core values included: person-focused, privacy, inclusiveness, transparency, accessibility, adaptability, governance, cooperative and participatory leadership, scientific integrity, and value (Friedman, Rubin, Brown, Buntin, Corn, Etheredge, et al., 2014). Concerned that the US health system is underachieving and lacks the capability to analyze and improve its own behavior, the proposed LHC would leverage Big Data to provide the information needed to identify opportunities for improvement. The proposal extends Senge's (1990) "Learning Organization" through a national initiative that maximizes the utility of "big" health data. Recognizing the many obstacles to achieving a truly "high-functioning LHS," an ambitious research agenda has been outlined by a group of informatics leaders who caution that "rising to the challenge of the LHS may require a novel emergent science of large-scale learning systems best seen as an evolution from the science of information systems, through a science of cyber-physical systems, and ultimately to a science of cyber-physical-social ecosystems" (Friedman et al., 2014, p. 4).

The proposed LHC presents an opportunity that nursing informatics professionals (and nursing as a whole) should not fail to recognize. And, indeed, the nursing informatics community has taken a very important first step. In 2014, the second Nursing Knowledge conference was held at the University of Minnesota and focused on Big Data & Science for Transforming Health Care. These conferences had two goals: 1) Ensure that nursing data are included in the massive data health databases currently being collected (and that will be collected if the LHC comes to fruition); and 2) ensure that the information gained will be used to guide practice, research, and education. An impressive action plan was formulated at the 2014 conference, and leaders were assigned to each action. Copies of the proceedings and related materials can be found at httpi://z.umn.edu/bigdata

For years a cadre of nursing informatics leaders has worked to develop terminologies that can be used to standardize the descriptions of what nurses do so that we can collect and utilize our collected evidence to improve our practice and patient outcomes. Recently, a group of nurse leaders collaborated on a Pressure Ulcer Prevention electronic measure that utilizes SNOMED CT, LOINC, and Rx Norm terminologies and may well serve as a model for future indicators (Cipriano, 2014). Given the impetus for the LHC, it is critical that nursing leaders collaborate and capitalize on their experience with terminologies and standardized measures to ensure that nursing data are included and utilized.

Also this year the National Institutes of Health and several other agencies asked CMS to begin capturing standardized data and measures from social and behavioral domains for which good evidence exists of their association with health outcomes. The IOM report recommends that the data in these domains be included in the criteria for certification of EHRs. Suggested domains include: alcohol use, race and ethnicity, residential address, tobacco use and exposure, census tract-median income, depression, education, financial resource strain, intimate partner violence, physical activity, social connections and social isolation, and stress. Data describing many of these domains are already being collected, but their measures are not standardized nor are they collected with the same frequency. However, these data would clearly enhance the ability of the LHC to understand the social and behavioral contributors to both individual and population health outcomes and should be welcomed by nursing. More details can be found at www.iom.edu/EHRDomains2 .

A Plethora of Sciences

In an informal, member-initiated session titled "Birds of a Feather," we shared the theoretical and methodological approaches we use. In the group of perhaps 50 scientists, ethnography, mathematics, ecological psychology, economics, sociology, and anthropology were just a few of the examples of how people were studying health information systems. This richness of ideas and methods can serve us well as we work to harness Big Data within the health care environment. For me, this session reinforced the opportunities (and necessities) for collaboration across disciplines-especially in such a multi-disciplinary domain as health informatics.

Building Actionable Information from Granular Data Easily

Several speakers addressed how we might improve our ability to access and use data in real-time to improve our workflow and/or our patient outcomes. Several new products may help meet these needs while they contribute to our ability to create a LHC at any level (from local to national). A technology getting positive vibes was SMART on FHIR, which uses open specifications to integrate other applications with health IT, EHRs, or portals. In simple terms, SMART on FHIR enables access to granular data, therefore enabling either population-based or individual patient queries by other applications. Because it is standards-based, but open-source, SMART on FHIR can work with any of the current vendor products. Detailed specifications and a videotaped interoperability demonstration can be found at http://smartplatforms.org/smart-on-fhir/ .

The benefits of these types of tools are several: First, organizations don't need to replace their current health IT because it is the base structure for the "add-ons" that are built on top of it. Also, this is where the core data elements are housed (and where nursing data elements also must be housed). Second, the "add-ons" function much like the applications (apps) we are used to downloading onto our Smart phones or other devices and provide the diversity and flexibility that a centralized system designed to meet the needs of many entities cannot. Third, for either quality management or research, this kind of tool makes it easier to aggregate clinical data from a variety of sources within the enterprise's systems. Such capability is critical to developing a LHC.

Continuous, Clinically-led Design Improves Workflow and Health Outcomes

Jon Patrick described a novel, clinician-centered design approach that allows clinical teams to create and test an application iteratively by selecting options from a predefined array of design tools-without engaging a programmer. Having this flexibility makes it easy for the teams to design or modify work tools in real-time that improve their workflow, making for trouble-free implementation, easy maintenance and quick updates. The clinically-led design process has several core principles, among which are: a) Keep the design objects available to users minimal in number and easy to use; b) Assume that there will always be a need for continuous, dynamic design; c) Clinical staff should control their own work processes and drive the design process; and d) The design must support evidence-based processes.

Demonstrating this approach, an Emergency Department team designed their own workflow management system (including triage, clinical documentation, nurse tasks, disposition, and care summary). An evaluation study showed that the system saved clinicians time, as compared with their current enterprise system (Bishop, Patrick, & Besiso, 2014). View a videotaped demonstration. Consistent with the goals of a LHC, this design approach appears to enhance clinicians' understanding of their own work processes and make it more likely that the resulting software design will be both useful and usable.

Conclusion and Implications

Taken together, the information I gleaned from the 2014 AMIA Symposium suggests that Big Data and the technology used to access the data may indeed enable us to create an effective LHC. Given the research agenda identified by our colleagues as necessary for the project, it appears that informatics researchers will be very busy. Informatics specialists too will be in high demand to facilitate all aspects of the development process (analysis, design, implementation, and evaluation), as well as to facilitate effective, accurate communication between the health-care professionals and the technical staff. Once the LHC is implemented, health professionals will need the help of informatics and work process professionals to help them examine their current workflow to identify places where technology can help. The people doing the work can identify many of the issues; but because they may have been doing the job the same way for a long time, some problems may be difficult for them to see.

Ultimately, whether the nursing profession and the patients we care for achieve the potential gains from the LHC I've described depends in large part on whether standardized nursing data are included in the core health data that will inform the LHC. As Cipriano (2014) reminds us "What we measure, we can improve" (p. 3). To that end, I view the work on integrating nursing terminologies encouraged through the collaborative efforts of the University of Minnesota's "Big Data" conferences as both promising and important. We have an opportunity to improve patient care outcomes through the LHC that we cannot afford to squander. It will take both collaboration within and beyond nursing to build the envisioned LHC-but within the informatics community at large we have the expertise needed to achieve the vision. Do we also have the will?

References

Bishop, R. O., Patrick, J., & Besiso, A. (In press). Efficiency achievements from a user-developed real-time modifiable clinical information system. Annals of Emergency Medicine, Corrected proof retrieved from http://www.sciencedirect.com/science/article/pii/S0196064414004910 Doi:10.1016/j.annemergmed.2014.05.032

Cipriano, P. F. (2014). What we measure, we can improve. The American Nurse, 46(6), 3.

Delaney, C. W. (2014). Nursing knowledge: Big data and science for transforming health care.Nursing Knowledge Conference Proceedings, June 5-6, 2014, University of Minnesota School of Nursing. Retrieved from http://www.nursing.umn.edu/about/calendar-of-events/2015-events/big-data-2015/index.htm

Friedman, C., Rubin, J., Brown, J., Buntin, M., Corn, M., Etheredge, L., …Van Houweling, D. (2014). Toward a science of learning systems: A research agenda for the high-functioning Learning Health System. Journal of the American Medical Informatics Association. doi:10.1136/amiajni-2014-002977 Retrieved from http://jamia.oxfordjournals.org/content/early/2014/11/07/amiajnl-2014-002977.long

Institute of Medicine. (2014). Capturing social and behavioral domains and measures in electronic records: Phase 2. Report Brief, November, 1-4. Retrieved from http://www.iom.edu/Reports.aspx

Senge, P. (1990). The fifth discipline: The art and practice of the learning organization. New York, NY: Currency Press.