“Just leverage our technology to gather data, generate knowledge, and use it to support optimal clinical practice!”
It sounds so simple. With the advent of ARRA and HITECH, realizing a true learning health system shouldn’t be a particularly lofty or complex goal - especially in light of massive health information technology uptake in many clinical areas. Clinicians want technology and tools to bring them data they can use from electronic health records (EHRs) and they want those EHRs to support efficient practice.
However, EHR design is largely driven by regulatory guardrails and nonstandard clinical workflows. Maddingly (for this implementer) we devote only 0.1 percent of total health care spending to the evaluation of implemented health policies and only about 18 percent of clinical recommendations in primary care are evidence-based.
My role as an executive-level informaticist is to implement enterprise EHRs on a budget, usually in an incredibly short timeframe. The technical implementation of health IT is actually pretty easy with a competent IT staff. The much harder lift is achieving operational buy-in and ownership of systems still largely seen as an IT endeavor - designed towards ever-changing regulatory requirements, which we know aren’t often evaluated for their effectiveness, and accommodate wild (trust me, they’re wild) variations in practice, which we know aren’t generally evidence-based. Just for fun, toss into the equation the new world of acquisitions, affiliations, and the system incorporation of facilities like critical access hospitals, long-term care facilities, dialysis. The landscape of care delivery mandates rapid decision making about EHR design and data strategies in environments that often lack the governance and vision to optimize care delivery, agree on standards for system build, or reinvent the legacy world of data governance and use. My job is fun! No, seriously, it’s fun.
Making Something Beautiful Out of a Complicated Equation
Today, though, I’m wholly focusing on data. Specifically, I’m focusing on a few tidbits related to making something beautiful out of the data quagmire many of us find ourselves standing in. Most off-the-shelf EHRs come with a dizzying array of data and analytics tools out of the box. At any given go-live event, hundreds of new data tools may be introduced into a clinical ecosystem. Further, clinician and finance requests for data are numerous and driven by a variety of factors – trying to increase margins, assisting with work-life balance (by decreasing and streamlining documentation), optimizing care of the patient (predictive analytics, reports) and data exploration (through self-service reporting tools).
I’ll be honest - few things make me more wary than the notification of a new report/dashboard/super-cool data visualization tool that is “ready” for our end users. The fact is, the visual representation of EHR data is most reliable and valid when the system used to collect data in the first place exists in an optimal state. That means things like:
- EHR/technology training was adequate and is ongoing…and includes education on the basics of data.
- The design of the EHR makes it easy to put data in the right place (i.e., you don’t have 14 places to enter the same data elements, with only half of them tied to analytics)
- You don’t make the documentation system so burdensome that clinical staff resort to entering critical data elements in free-text notes, because they can’t figure out where to put it.
- There is a clinical validation process in place, so the technically valid reports undergo a second (and equally as important) round of validation by an individual or team with expert-level knowledge of the system as well as data structures.
Critical Steps Between the Technical and the Clinical: Report Validation and Translation
That last bullet point is critical. When your report is validated, technically, it is time to send it to the operational leaders. Do the frontline leaders think the numbers look right? Did the numbers on the new report rise or fall when compared to the previous method of data-gathering and representation? If you were previously abstracting data and now your numbers have changed, you can probably put your money on the fact that you have documentation inefficiencies to track down - manual abstraction covers a multitude of data-entry and system design sins. With about 95 percent certainty I can say that I can walk into any hospital with an enterprise EHR in place and find a dashboard, predictive analytics model, or report in production that isn’t actually valid. The good news is, half the time, our end users don’t even know they exist…which is another blog post.
The real trick here is that many clinicians don’t have a deep level of knowledge of analytics or data elements/data structures. They also tend to not have a deep level of knowledge about the overarching design of their EHR or an understanding of how it all works behind the scenes. There has to be an intermediary between the the technical and clinical team and sometimes, your standard informaticist isn’t the answer (they aren’t always data-savvy, either).
There is no single answer to the piece of this puzzle, but taking a new analytics tool and making sure the data elements are truly representative of the care being delivered requires a mastery of the design of the data-entry system (i.e., the EHR) as well as the basic concepts of data/analytics, to accurately diagnose gaps between the two. Careful investigation may unearth the need for revision of the analytics tool or revision of the EHR and that requires a solid governance structure and the existence of a framework for validation. This can only be realized through the development of a well-rounded interdisciplinary informatics team or a combination of interdisciplinary governance teams comprised of discipline-specific subject matter experts.
The health IT and clinical workforce of a hospital system in 2019 can’t look much like it did in 2008 if we want to be successful at harnessing our health IT universe of data into elements that actually offer meaningful input to our clinicians, patients, and payers.
Rebecca is speaking at AcademyHealth’s Health Data Leadership Institute Sept. 25-26. We are sold out for the 2019 event. Add your name to the waitlist, if a space becomes available, a member of AcademyHealth’s staff will contact you and instruct you on how to register. If we are unable to accommodate your registration, we will notify you early for the opening of the 2020 Health Data Leadership Institute registration.