Implementing, Studying, and Reporting Health System Improvement in the Era of Electronic Health Records

Electronic Health Records (EHRs) offer great opportunity to improve safety, quality, and value, but many of these benefits are not yet being realized. Some of the possibilities and precepts are like those in the pre-EHR world, but many have changed – specifically, cohort identification, doing large numbers of practical randomized trials, CDS, opportunities for patient engagement, and leveraging such technologies as artificial intelligence are fundamentally different.

In a supplement to Annals of Internal Medicine, guest editors Andrew Auerbach, MD, MPH and David  W. Bates, MD, MSc, curated a selection of papers that raise issues and provide an overview of the opportunities available by EHRs.  The individual articles provide specific guidance to researchers focused in that specific area, and as a group frame a blueprint for digital health research using electronic health records.

  • Understanding the local, non-electronic environment for research, which can be as important at the intervention itself (Haynes et al).
  • EHRs bring new opportunities for research using observational data. EHRs enable gathering of broad and complicated data sets that provide the opportunity to develop ongoing cohorts, many of which were not previously possible, as well as to carry out effectiveness studies in the “real world” (Callahan et al).
  • Randomized trials are the gold-standard approach for determining whether one intervention works better than another, but they have traditionally been expensive, time-consuming, and far removed from care delivery.  However, EHRs can address these considerations – with them, trials can be done more quickly and at a fraction of the cost traditionally required (Pletcher et al.)
  • A framework for studies focusing on safety and quality improvement by using EHRs provides an approach to designing and reporting studies that emphasizes the individual sociotechnical issues in developing and reporting HER-based quality improvement and safety studies (Singh and Sittig).
  • Clinical decision support (CDS) is arguably the most important intervention type enabled by EHRs, though its effectiveness has been highly variable. This paper suggests that a substantial proportion of HER-based intervention will involve forms of CDS to influence care and discusses some of the ways CDS can have more impact (Kawamoto and McDonald).
  • The tension in informatics and EHRs between standardization (easier to spread widely) and customization (local uptake and performance may be better) is reviewed. The authors suggest that translation of interventions from one site to another often requires adapting and customizing at the translation step and a discussion of how papers might provide information about opportunities to achieve such translation effectively (Wright et al).
  • A pivotal issue with EHRs has been utilizing them effectively; if they are hard to use, require substantial extra time, and do not support improvement, they can both create provider burnout and fail to improve quality, safety, or value. This paper studies the workflow and workarounds in EHR-supported work, including the unavailability of the data to researchers (Zheng et al).
  • EHRs offer powerful tools for improving patient engagement, including personal health records and capture of patient-reported outcome measures among others. This paper discusses issues around understanding patient engagement in health using EHRs, and how studies that address this should be reported (Lyles et al).
  • EHRs offer tremendous insights into both the value of what is being done currently in health care, and how it could be improved.  This paper discusses how EHRs can be leveraged to improve value and covers key issues around research and value using EHRs (Rudin et al).
  • Medicine has lagged other industries in adopting artificial intelligence, but that is changing quickly and EHRs will undoubtedly enable this transition.  This paper discusses some of the challenges that artificial intelligence creates – such as models that are not static but evolve in real time – and how interventions using machine learning and artificial intelligence should be reported (Bates et al).

bvbvvbAndrew Auerbach, MD, MPH

Dr.  Auerbach is the Chair of the Clinical Content Oversight Committee for UCSF Health, the operational group responsible for developing and implementing electronic health record tools across the UCSF Health enterprise. Dr. Auerbach is a widely recognized leader in Hospital Medicine, having authored or co-authored the seminal research describing effects of hospital medicine systems on patient outcomes, costs, and care quality. He leads a 13-hospital research collaborative focused on new discoveries in healthcare delivery models in acute care settings.