March 6, 2020: Creating a Learning Health System Through Randomization (Leora Horwitz, MD, MHS)


Leora Horwitz, MD, MHS
Associate Professor, Population Health and Medicine
Director, Center for Healthcare Innovation and Delivery Science, NYU Langone Health
Director, Division of Healthcare Delivery Science, Department of Population Health, NYU Grossman School of Medicine


Creating a Learning Health System Through Randomization


Embedded research; Electronic health records; Randomization; Quality improvement (QI); Learning health systems

Key Points

  • Randomized QI projects are needed in order to know whether such system-level programs/interventions are effective.
  • The best candidates for QI randomization projects have a high volume of cases and short-term outcomes that are already being collected.
  • These randomized QI projects were developed in collaboration with the frontline care providers and staff to ensure seamless implementation with no additional burden. Also essential was buy-in from senior-level administration.

Discussion Themes

Statistical significance is not always the right threshold for decision making in a health system.

What amount of resources are needed to do these projects (eg, researcher/analysts, project managers)? Is the work self-sustaining?

Disappointing QI findings are important to know, so that ineffective activities can be discontinued.

Read more about this project in Creating a Learning Health System through Rapid-Cycle, Randomized Testing (Horwitz et al., New Engl J Med, Sep 2019).

#pctGR, @Collaboratory1