Well functioning academic health science systems develop a culture that fosters innovation in pursuit of high value healthcare, goals shared by ‘Learning Health Systems’. In this article Dr. Bill Smoyer and Dr. Susan Moffatt-Bruce describe practical experience of implementing an LHS in a novel and effective manner - Professor Sir John Tooke


William E. Smoyer, M.D. and Susan Moffatt-Bruce, M.D., Ph.D. M.B.A.

The convergence of three major trends in medicine: prioritization of translational and health services research, the need to control healthcare expenditure, and the adoption of electronic health records (EHR), has created unprecedented interest and opportunities to develop systems that improve quality of care while reducing costs.  Operationalizing what the U.S. National Academy of Medicine has termed the “learning health system” (LHS) is sorely needed, yet presents significant challenges for healthcare providers.  Doing so at the local hospital or health system level requires systematic changes that have not yet been realized in clinical practice. 

Numerous efforts have been made to date to establish LHSs, but these have been oriented largely around the technical and operational integration of EHRs among multiple health care systems (i.e. a top-down approach) [1, 2].  While this approach has led to notable progress, many cultural and operational barriers have stimulated consideration of an alternative approach based on the development of local LHSs that begin by integrating research, clinical care, and quality improvement (QI) in individual health care systems (i.e. a bottom-up approach), followed by expansion of successful program features across and among institutions [3].  Such local LHSs also create an opportunity to bring together key stakeholders in individual health systems to better align their historically varying, and oftentimes disparate, interests around clinical transformation (see Table and [4]).



Varying Interests of Key Stakeholders




Primary Interests


(Chief Executive Officer)


Quality Care



(Chief Medical Officer)

Medical Staff Administration

Peer Review


(Chief Operating Officer)


Improved Value


(Chief Quality Officer)

Quality Improvement

Patient Safety


(Chief Information Officer)


Data Governance


(Chief Financial Officer)

Lower Care Costs

Increased Patient Volume


(Chief Research Information Officer)

Discrete Data Points

Data Interfaces

Data Accessibility


(Chief Medical Information Officer)

Usability of Medical Record

Quality of Care


Best Care

Opportunity to Improve Care


Ease of Documentation

Clear Care Guidelines

Hospital Staff

Standardized Care

Ease of Documentation

Policy Makers

Cost Containment

Improved Value

Health IT Vendors


Data Accessibility

Risk Management

HIPAA Privacy

Data Integrity

Patients and Families

Quality of Care

Patient Satisfaction


While key health system stakeholders have varying and sometimes conflicting interests, local learning health systems create an opportunity to better align these interests, thus enhancing the likelihood of achieving the goals of fully integrating evidence-based medicine with evidence-generating medicine to systematically drive improvements in clinical care.

We contend that today’s ideal LHS should achieve the following: (a) Drive both clinical quality improvement at reduced healthcare costs, thereby enabling the value proposition; (b) Provide ample research and data platforms to leverage our “living laboratories” within health systems to inform care innovations and generate evidence-based improvements in standards; (c) Engage all healthcare providers throughout the continuum of care (hospitals, clinics, pharmacies and home care agencies) and hold key stakeholders (CEO, CQO, CMO etc.) accountable for strong performance; and (d) Ensure health equity to our most vulnerable and underserved patients through innovative and patient-centric healthcare environments.

Despite strong interest, major challenges or “headwinds” must be overcome to achieve the goal of creating LHSs. Recent data from our pilot program (Learn From Every Patient ® (LFEP) Program) in a large academic health system have shown that full integration of clinical care, quality improvement, and research can be successfully implemented in large, complex health systems.  Learn From Every Patient ® (LFEP) represents a novel approach that leverages the known externalities to create an innovative program that integrates “evidence-based care” with “evidence-generating care” in the context of robust care coordination, thereby enhancing value (see Figure [5]):

  • Integration ensures all patients receive the current evidence-based standard of care, while simultaneously generating robust data to enable the systematic evidence-based improvements needed in future care.
  • LFEP process utilizes real-time clinical + research data collection as discrete data elements to enable ongoing data analytics to provide a closed loop process for systematic improvements in care and reductions in costs.  
  • Implementation of the Learn From Every Patient (LFEP) pilot program confirmed it is possible to develop a “local learning health system” at a large medical center that can systematically drive simultaneous clinical quality improvement and reduced health care costs.
  • Results from the pilot study at NCH demonstrated:
    1. Costs ($225K) ~16% of first-year hospital charge reduction ($1.36 M; $10,151/patient)
    2. $6 saved for every $1 spent for patient care
    3. Multiple improvement “learning projects” initiated
    4. “Learning projects” have directly improved care and reduced costs
    5. The LFEP model is ideally suited for Accountable Care Organizations (ACOs)
  • Those health systems that are able to embrace the necessary changes may avail themselves of many significant opportunities, including:
    1. Reductions in health care utilization and expenditures
    2. Systematic improvements in clinical care outcomes
    3. Expected market advantage for robust delivery of evidence-based care
    4. Unprecedented ability to collect and integrate genomic and phenotypic data
    5. Career advancement of academic faculty (publications)
    6. Incorporation of Patient-Reported Outcomes (PROs) (desirable for FDA and/or EMA Registration)



Figure.  The Learn from Every Patient ® (LFEP) Model for the Integration of Evidence-Based Medicine with Evidence-Generating Medicine.  Starting with the introduction of a standardized evidence + expert-opinion-based clinical protocol (far left), each patient encounter includes collection of both clinical and research data, which are transferred to a data mart, where they are integrated with other patient-related data sets (proteomics, genomics, metabolomics, etc.).  These data are then analyzed (with IRB approval) to answer specific clinical or translational research questions.  The analyzed data are then published in peer-reviewed journals (i.e. dissemination of new knowledge), and the new knowledge is used to drive incremental improvements in cost effectiveness and standards of care.  This “quality improvement” is then applied systematically to the care of all patients in the form of specific evidence-based improvements in the original standardized clinical protocol.  Continued revolutions of this “improvement cycle” then drive continuous systematic improvements in future evidence-based clinical care (adapted from [5]).



  1. Embi, P.J., S.E. Kaufman, and P.R. Payne, Biomedical informatics and outcomes research: enabling knowledge-driven health care. Circulation, 2009. 120(23): p. 2393-9. PMID: 19996023.
  2. Payne, P.R., P.J. Embi, and J. Niland, Foundational biomedical informatics research in the clinical and translational science era: a call to action. J Am Med Inform Assoc, 2010. 17(6): p. 615-6. PMID: 20962120.
  3. Moffatt-Bruce, S., et al., Engaging the health care team through Operations Councils: strategies to improve population health from within. Adv Health Care Manag, 2014. 16: p. 51-67. PMID: 25626199.
  4. Smoyer, W.E., P.J. Embi, and S. Moffatt-Bruce, Creating Local Learning Health Systems: Think Globally, Act Locally. JAMA, 2016. 316(23): p. 2481-2482. PMID: 27997662.
  5. Lowes, L.P., et al., 'Learn From Every Patient': implementation and early results of a learning health system. Dev Med Child Neurol, 2016 PMID: 27545839.


About the authors

Dr. William Smoyer is a member of the Section of Nephrology at Nationwide Children's Hospital, Vice President and Director for the Center for Clinical and Translational Research at The Research Institute at Nationwide Children's Hospital and a Professor of Pediatrics at The Ohio State University College of Medicine. 

Dr. Moffatt-Bruce is Executive Director of University Hospital and a tenured professor in Surgery, Molecular Virology Immunology and Medical Genetics and Biomedical Informatics at The Ohio State University.