This post is a bit different from most of the policy points, institutional cases and reports of technical innovations that I’ve been reading on THCB in the past months. I want to pose the above Question to the readers of this blog, since many of you are uniquely positioned help answer it in your comments. And I have a hope that your responses to this Question will help nuance the technical and policy debate over EMR adoption.
First, let’s unpack the Question:
1. THE JOB. In the past several years, a number of public and private initiatives, most recently ABMS’s Improving Performance in Practice (a project with which I am affiliated) and NCQA’s Patient Centered Medical Home, have been making fitful progress toward a new post-reform model of primary care: patient-centered, accessible, care-coordinating, population-focused, prepared, proactive, and the rest. These collaboratives and demonstration projects have all stressed the importance of computerized ‘registry functions’ as the foundation for these progressive capabilities. One, the Health Disparities Collaboratives run by HRSA, went so far as to commission a registry program and provide it free to participating clinics.
A key hope and assumption of many of those leading this work has been that once practices have implemented EMRs, and once they have standardized their protocols and coding sufficiently, they will be able to use their EMRs to fully support the new, population-based model of care. Practices that have attempted to do this have faced daunting challenges. Many practices have ended up using both an EMR and a registry, with all of the duplication and rework that this implies.
2. The EMRs. Fifteen years ago, while working as a QI and informatics trainer, I had occasion to take a first-hand look at the database architecture of one of the popular outpatient EMR programs of that era. (It is still a very popular program, and will here remain nameless.) What I found became a cautionary example for my classes when I explained relational database principles: The database contained a table for patient data, a related table for medication data, and another for patient problems. But, probably for reasons of speed and simplicity, there was no ‘gerund’ table that would link medications to problems. While that data structure worked fine for displaying a list of patients, their problems and medications on a single screen, it was impossible to use the datatbase to answer such population-level queries as ‘which patients are currently prescribed aspirin for CVD prophylaxis?’ And no amount of massaging or exporting or data-warehousing could make it answer such a question – the required relational information was simply missing. Alas, my career path since those days has not put me back in the engine room of that or other EMR systems, so I am unable to say whether this was a fluke, or is now typical of EMR design.The upshot: If these kinds of lacunae still infest the architecture of today’s EMR products, then it would seem that widespread adoption of today’s EMRs will consign much vital clinical data to a kind of ‘black hole’ out of which it will be impossible build the queries and reports that are needed to support the kind of flexible, improvement-oriented, population-based primary care that we increasingly seem to need.
Finally, the question again, refined: “Will the data architecture of any of today’s outpatient EMR programs allow their data to be used – either directly or following transfer to an offline data warehouse, by means either of built-in reports or third-party applications – to support population-based care of the kind described in the PCMH?”
I look forward to your responses and discussion in the days ahead. Since this may be a sensitive subject for some, please mention your affiliations if they are relevant to your response.
Richard Scoville, PhD, is an independent consultant specializing in healthcare quality improvement and performance measurement. He is an Adjunct Professor in the Department of Health Policy and Administration at the University of North Carolina at Chapel Hill where he teaches courses in healthcare quality improvement and informatics. He serves as an improvement advisor to the Institute for Healthcare Improvement, NICHQ, the Cincinnati Children’s Hospital Medical Center, and the Federal Health Resources and Services Administration on a range of collaborative improvement and systems design projects.