Returning to the Roots of Clinical Decision Support

Long story short, I was re-reading this paper about the origins of the HELP system at Intermountain Healthcare. I was the Director of Medical Informatics at LDS Hospital from 2000-2004, many years after this paper was published. I was lucky enough to be tutored and mentored by all of the authors. While reading the paper with time to reflect back, I realized that the healthcare industry turned clinical decision support inside out when we almost subconsciously moved from very targeted and specific clinical decision support applications that were clearly benefiting patient care and cost of care, towards EHRs that were simply general data collection tools. We've been trying to squeeze the decision support blood out of the EHR turnip, ever since, to no avail.

When Pryor, Clayton, Gardner, and Warner (and later Classen and Pestotnick), were developing computer applications to support patient care at Intermountain Healthcare's LDS Hospital, they built very specific, target applications. They didn’t use a backend, longitudinal EHR as the basis for their applications. They collected the data that they needed about a patient in a particular clinical state, and they ran computerized decision support against that data. Our smart phones are good examples of specialized applications. We call upon specific applications when the need arises to perform a specific task. In medical informatics, it was these specialized, targeted clinical decision support applications that clearly made a positive difference for both patients and clinicians.

Somewhere along the evolution of computerizing healthcare, we turned all of this success from specific decision support into a general tool we call an EHR (or EMR) that has so far shown almost no value to clinical decision support. Going forward, we will return to the roots of this success. The data collection templates for patients will be tailored specifically for the patient type, and the subsequent computerized decision support will also be tailored specifically for the patient's state. The “EHR” of the future will look more like a smartphone with dozens of applications that support very specific patient types. The data we collect will not be the general data that is currently collected in EHRs, but rather very specific data for a given patient type. We will knit this specific data together, on the backend, to form a longitudinal record.

We (Health Catalyst) are in a good position to make this turn to the future, given that our applications and analytics are very specific to patient types.