[Today’s guest post is by Dr. Griffin Myers, whose innovative primary care clinic for Medicare patients, Oak Street Health, opened its doors in September 2013.]
Welcome back; this is the second in a series of guest posts I’ve been invited to write for GeriTech. As mentioned in my previous post, one nice way to think about our model of primary care for Medicare beneficiaries at Oak Street Health is using the list of recommendations put forth in the Institute of Medicine report “Best Care at Lower Cost.”
This post is about the first recommendation, which falls into what the report groups into the category of “Foundational Elements.” Ironically we’ve found these to be the hardest objectives thus far. We’re still in a bit of flux as we get settled into our new setup, but I’ll give an introduction.
“Recommendation 1: The digital infrastructure. Improve the capacity to capture clinical, care delivery process, and financial data for better care, system improvement, and the generation of new knowledge.”
This immediately brings to mind a super-EHR: one product for charting, practice management, financials, claims, population health, and clinical decision support.
But, surprise: there is no off-the-shelf product that integrates all of this for primary care, never mind primary care for seniors. At least not that we’ve found after an extensive search. There are products for each of those functions, but not a single package. And asking providers to use multiple non-integrated products isn’t a real solution. Here’s our approach.
How we chose an EHR & practice management system
Our vision
We began out with an idealistic vision. We wanted to align our efforts with secular trends in technology: mobile/mHealth, cloud, device agnostic, software-as-a-service, big-data, social, etc.
Sadly, the technology isn’t quite there yet. We had to adjust, and we ended up favoring other factors in the name of practicality.
(We experimented with the Google Chrome OS and a fleet of Chromebooks/Chromeboxes, which no one supports. It’s a really easy to manage, very secure environment. It would have been outstanding. We then changed to Macs, but ultimately our product wasn’t well supported on Mac. So we’re now on our third set of computers: standard PCs. Thanks to the Apple Store in Lincoln Park for taking our return.)
Meaningful Use
Everyone tells you to start with meaningful use. “Meaningful use,” for those unfamiliar with the term, is a series of financial bonuses practices can receive from the government for using certain features in an EHR. We’ve found this to be rather odd, to say it simply. Some things make sense (patient portal), but others seem to table-stakes (using the EHR for medication lists). Our plan? Exceed the meaningful use requirements because we think many of these features offer benefit to our patients. Getting paid for something we were planning on doing anyway is nice as well.
Selection Process
I want to stay away from recommending specific products, as fit varies by practice and many other factors. There is no exclusively-for-older-adults product, so we started from scratch. Here’s the best (free) summary of EMR products that we’ve found. It was our starting point for what was a very structured process.
Before I get into our process, a quick warning: vendors don’t make it easy. Our experience is that vendors will promise everything, give demonstrations in highly controlled environments, and wait for a contract to show much more. At least that’s been our experience. And many EHR companies now make their money not on the software itself but on doing the claims submission for providers…and taking a slice (i.e., 3 to 12% of all claims). It’s not a bad idea for some practices, but it is an expensive one. (You’re essentially outsourcing your claims staff…) Be prepared for the upsell. Now for our method…
Step 1: We started with a long list of features: price, web-based v. not, Mac v. PC, whether or not it requires a server on site, clinical decision support, level of support, claims servicing v. doing our own claims, size IT staff required to support, population health features, customizability, depth of the data utility, etc.
Step 2: We scored each product along those dimensions to create a ranking of products in the consideration set. We cut the bottom of the list and now had a small set to really investigate.
Step 3: For that short list, we attended demonstrations and did some Q&A with each vendor. This got us to a final decision based upon five factors:
(1) Installed base: How many people/practices use this product? New products only get better by making mistakes, fixing bugs, and responding to user feedback. So a new product or one with few users will cause you problems. (We’ve learned this the hard way.)
(2) User reviews: How do current users feel about the product? It’s really hard to get honest feedback, but that’s important. We’ve found vendors will send you to prized, happy customers. Try to find a typical user and get his/her thoughts. We cold-called practices using products we were considering; folks were happy to talk, and those discussions were always valuable.
(3) Integration of charts and practice management: Can you create charts and run the practice/schedule/claims from the same product? There is so much information that needs to be shared here, that we simply weren’t willing to try multiple products for this.
(4) Ability to use/manage data: How is data treated, how can you use it, and how customizable is the data utility? Ultimately we’re comfortable with our setup because of this alone. It’s the trick to making our practice systematic and consistent. We want be systematic about indicated screening, preventive maintenance, etc. This is how.
(5) Customize for a geriatric population: No product comes ready for population-specific assessment. What do I mean? No EHR product comes with a way to capture Timed-Get-Up-And-Go scores, Katz ADL and Lawton IADL batteries, PHQ-9depression scales, or Mini-Cog assessments. So we wanted a product we could customize to do this for us. We create powerful data in our patient interactions, and we want to use that data to improve our practice for individual patients and for our population.
With that introduction, I’ll pick up next time with recommendation number 2, “the data utility.”
Griffin Myers, M.D., M.B.A. is a founder and the Chief Medical Officer at Oak Street Health in Chicago. He is currently in his final year of an emergency medicine residency in Boston. You can contact him at griffin (at) oakstreethealth (dot) com.
Disclosures
The author wishes to disclose a financial interest in the primary care model discussed above. Furthermore, he is a trainee in a postgraduate clinical training program, and neither the program nor the affiliating university endorses, owns, or has any formal or informal relationship with the primary care model.
Just from a curiosity standpoint, can you go into #4 a bit more?
For example, how do you use the data to indicate preventive maintenance? What actual reports/tools/etc. do you put together? How much effort goes into writing custom solutions for using the data?
Thanks for the comment.
I should tell you that our leadership team is highly empirically driven. We argue over the best coffee on the north side by the numbers, that that influences our perspective.
Your question is around how we’re using data now, and I have to answer a slightly different question. We’ve been operating now for 2 months, so the questions for us are:
(1) How do you plan on using data as time goes on?
(2) How have you setup the system to do this?
In a very real sense, we are investing resources in primary care and preventive maintenance in hopes of reducing acute care needs down the road. That means we have to (first) design interventions and (second) apply those interventions to the right individuals. Our hypothesis is that data helps most with that second task. Then only with large n numbers does it help with the first, namely evaluating and comparing interventions.
We are still in the process now of building the “custom solutions” to which you refer. We ended up choosing our vendor/product based upon the ability to do just that: to structure data. When I say that, I mean we don’t want text data, we want structured fields we can study. Our hypothesis there is that HPI, standard physical exam, and reviews of systems are not easy to study/mine and probably don’t contain wisdom in numbers. Conversely, validated tools such as the Katz ADL, Lawton IADL, Morisky medication adherence, PHQ-9, etc. in combination with updated medication lists, diagnosis codes, social support variables, and acute-care utilization data… We do believe those items contain secrets to better care. So we are setting up our systems, both in the clinical practice and the IT infrastructure, to create and capture those items we think have predictive power.
Neat and insightful, thanks for sharing. It's nice to hear from providers who care about structured data and what can be done with it. I expect to see only more of that as time goes on.
But more importantly I have to ask: what's the best coffee on the north side?