Epic EMR Adoption, Utilization, and Cost | Dale Sanders | Healthcare Blogs Skip to content Skip to navigation

Epic EMR Adoption, Utilization, and Cost

April 23, 2009
| Reprints
You need to dedicate the time and resources to constantly iterate, refine, and improve the utilization of an EMR over time

The point of this blog is: Purchasing and installing an EMR and hoping that you’ll realize a positive return-on-investment is not enough. You need to dedicate the time and resources to constantly iterate, refine and improve the utilization of that EMR over time, far beyond its installation and go-live. It’s a race without a finish line so you might as well train, budget and plan for that accordingly-- up front.

At Northwestern, we take great pride in being one of the first fully-institutionalized adopters of an EHR in healthcare. We’ve had an ambulatory EMR (Epic) for 12 years and an acute care EMR (Cerner) for 10. For the past 18-24 months, we’ve been studying not just the adoption of the EMR, but also studying the utilization of the EMR. Drawing upon an analogy to illustrate the point, if a carpenter buys a pneumatic nail gun, I call that “adoption” of a new tool. But if that carpenter is still using the nail gun to manually pound nails as if it were a hammer, has he achieved “utilization” of the tool? If you believe there is added value to an EMR over a paper chart—or even a word processor-- where does that added value reside and are we using it for those purposes? I’ve been slowly polling and collecting data in an attempt to understand EMR “utilization.” I doubt my data collection process would pass Gallup’s scrutiny, but I still believe that it paints an informative picture. Here’s what I have so far…

Qualitative Assessment of Epic EMR Utilization: “Do you personally use the Electronic Medical Record for the following purposes?”

  1. Remote, Internet-based access to the EMR for the benefit of physician convenience and faster patient treatment: 87%
  2. Data-driven reports and analytics which benefit patient care, such as disease management: 34%
  3. Data-driven clinical research: 73%
  4. Data-driven compliance, regulatory, or accreditation reporting: 19%
  5. Referral communications through a clinical In-Basket or automatic letter generation: 94%

If you are a physician or other clinician who uses an EMR and want to add your 2 cents, go ahead and fill out this survey: EMR Utilization.

Quantitative Assessment of Epic EMR Utilization

Based on data from the EMR itself, we run queries in an attempt to objectively measure how the EMR is being utilized in a few key areas of “added value” over a paper chart. To me, those key areas of added value of an EMR reside in the computable data that’s collected in medication orders and management; allergies management and awareness; problem list management and awareness; and family history of disease Below are the questions we asked in the queries of our data. The reporting period for the queries is December 2008 through April 23, 2009.

  1. Medications: What percentage of encounters included a medication order during the reporting period?: 68%
    1. Clearly, not all patient encounters will include a medication order, so we would expect this number to be less than 100%. What we are really trying to find here is the denominator of all Rx orders to better understand utilization. That’s a tougher number to find-- we’re looking at prescription pad utilization as an indicator.
  2. Allergies: What percentage of encounters included a review of patient allergies during the reporting period?: 68%
    1. Ideally, this should be 100%; there is a checkbox in the Epic EMR to indicate “Allergies Reviewed”, but note that the metric seems to be tightly correlated to medication orders, which is the most important issue, i.e., Are physicians reviewing allergies when they prescribe medications? It would seem so.
  3. Problem Lists: What percentage of patients have an entry in their EMR Problem List during the reporting period?: 64%
    1. Again, not all patients will have an entry in their problem list, but common sense tells me that it should be higher than 64%. Therefore, this metric hints that we are underutilizing our Problem List.
  4. Family History: What percentages of patients have an entry in the Family History section of their EMR during the reporting period?: 59%
    1. It’s possible that a patient has no knowledge of family history or nothing worth noting, but common sense tells me that this number should be higher than 59%. This metric suggests that we are underutilizing the Family History section of the EMR.

$$-- Total Cost of Ownership

Finally, to gain some idea of the cost-to-benefit ratio, we measured the total cost of ownership for Epic-- hardware, software, network connectivity, and labor-- Total Cost. The analysis for this TCO took about 3-4 months and the involvement of over 20 people. It was a very thorough analysis. The TCO for the Epic EMR on our campus, including patient registration and scheduling, is $856 per physician per month, or $10,272 per physician year. We have 640 physicians in our physician group, but over 3,000 Epic users on the campus including staff in the Hospital, private physicians, and researchers who benefit from Epic, but do not expressly support it, financially.

In Conclusion





I gather from your statement "don't short-change the investment" that the financial payback for an EHR may take years and that is not yet proven..Also when you developed the costs did you include any additional infastructure costs such as the costs of the compliance function and other administrative type costs?..

Dear notmd,

In short, yes, I think it takes years to realize a positive ROI from an EHR, even though the intangible benefits emerge almost immediately such as remote access to a patient chart, lab results, or order entry.

IMO, the most substantive ROI comes from mining the data which an EHR collects i.e., analyzing the data from an EHR for the purposes of quality improvement, cost reductions, and clinical research. But, it takes years to collect adequate volumes of data to support this type of analysis. Also, because adequate volumes of data are necessary in order to draw meaningful conclusions from that data, I believe small physician offices are unlikely to gain from this aspect of an EHR's ROI, unless they contribute their data to a larger pool, such as community, regional or national data warehouses. On a related theme, CSC Healthcare just published a great white paper on the relationship between an EHR's deployment strategy and Core Measures quality reporting (Core Measures: All About the Data).

Although the CSC paper doesn't say it explicity, I will: We should build our EHR products and deployment strategies by asking ourselves, "What questions do I want to answer with the data?", and then working backwards from that. But, we also need to minimize using the physician as a data collector, and maximize their value as a data synthesizer and analyst. That topic is probably worth another separate blog.

I did not include the costs of compliance or administrative infrastructure because I believe those to be the same or higher for a paper-based medical chart when compared to an EHR. But, I encourage you to contact me or post another reply and expand on this thought if you think I'm missing something.


Dale, I think you really touched on something in this sentence: "But, we also need to minimize using the physician as a data collector, and maximize their value as a data synthesizer and analyst." Unfortunately, no one seems to be taking this into consideration, and I think it will hamstring all adoption efforts for years to come.

To Dale, et al.,
A nice summary of opinions and insights which
confirms by its diversity that it is NOT clear how healthcare
IT will be integrated and beneficial, but that it will be a
prolonged and expensive venture. I would observe that
the evolution of the typewriter to the "Selectric" and then
to dot-matrix and ink-jet and "laser" does NOT improve
the quality of literature - but may facilitate lots of writing.
The art and science of medicine needs to continue to
evolve and improve, for sure, but I do NOT envision EMR/
PHR as a defining influence in this regard, and I think that
there is a "risk" to having IT be defining the business of
medicine rather than the "art and science" utilizing the IT
tool for its delivery.

Two quick points. 640 physicians times more than $10,000 per physician per year yields costs in excess of $6,000,000.

Number two, EHR's and the printer analogy doesn't work. Now if the typewriters and printers came with embedded knowledge, spell/grammar checker, and "The Elements of Style" then we would all be better writers to some degree.

The beauty of modern EHR's is interaction checking, safety mechanisms, clinical guidelines, clinical decision support etc.
I have to agree to disagree that how healthcare IT will be integrated and beneficial is very clear. IT must be used universally and it must communicate across a universal standard. The fact that we are so far behind on this is pitiful. The fact that some companies are charging ridiculous amounts of money for systems that don't communicate across platforms is truly sad. The final influence likely will not be the market or "consumer" driven healthcare (LOL, until they pay for it out of their own pockets!), but it will be that almost every new medical graduate coming out realizes the utility of technology assisted documentation and care.

Seriously, in this day and age why isn't my complete medical history still not easy to find on a safe, secure global network? We are practicing nonsense medicine. Every single day I'm in the ER, in Urgent Care, in the hospital, or in my office I cringe that my patients don't have a complete, obtainable, transferable, and usable electronic health record. And there are affordable solutions in existence that can do it.

Just yesterday I saw an 86 year old woman with mulitple medical problems on multiple medications with tachycardia and hypertension. I couldn't view her medication profile, I couldn't see her old ECG, I couldn't review her last cardiology consult, etc. This is modern medicine? Give me a universal open source EHR with collaborative sharing of all the records, newest treatments, newest computerized tools for monitoring care, and which is vastly more affordable (and they are out there like commercialized versions of VistA). PLEASE! ASAP, too.

Simple calculation of $6,000,000/year. I have to believe the vast majority of hospitals cannot afford that or realize anywhere near that number in ROI.

Gwen's comment of "open source" is poorly termed. As open source EHRs gain more steam around their fast, collaborative, and proven implementations from the VA, to West Virginia's Department of Health and Human Resources , to Midland Memorial Hospital, and now Silver Hill Hospital open source properly referenced will begin to show faster, more affordable, and highly adopted IT solutions for many hospitals.

Open source defined by some is an approach to design, development, and distribution offering practical accessibility to a product's source (goods and knowledge). EPIC is not an open source vendor and should never be referred to as such. I do commend Dale on his "transparency" of the enormous cost and effort to make EPIC a success at his institution. I also wonder as others do if those calculations included the huge upfront cost that most hospitals cannot afford in their capital budgets.

Ultimately, the measure should be patient outcomes. Then perhaps clinical metrics as Dale described which ostensibly lead to better outcomes. Smart hospital executives should make rational budget decisions and wise enterprise wide deployment of healthcare IT that gets them started on the road to efficiency and improvement. Its startling to me that the VA has done it so well, but community hospitals have been slow to copy the VA's template of success. It is almost delusional in my eyes that some major institutions are spending enormous amounts of money for massive implementations. Keeping up with the Jones's?

With open source your get community involvement, proven usage and outcomes, much better pricing, faster implementation, continual software enhancements and a better ratio of cost to ROI.

Robert, I enjoyed your comments.  I'd like to understand your six million dollars per year number above.  I couldn't relate it to Dales numbers, such as per physician year.

With regard to your "Ultimate measure" comment, there's broad agreement on that, as well as challenges.  See Anthony's post on PolicyLand, for a discussion of RealityLand.

Jonathan Teich, MD, PhD, in a presentation titled
"ARRA and Other National Initiatives: Implications for Quality of Care" shared the following graphic:

Jonathan contextualized this as levels at which measurement could be done.  This was developed as part of a CMS task force some time back. 

The point is pragmatism.  "Usage count" is the last point of the chain before which things get complicated and messy.  I'm sharing this because I think it provides a valuable framework for lots of things, including 1) defining meaningful use, 2) breaking down where OSS is distinct from 'all other approaches', and perhaps, most importantly, 3) where and how we define costs and ROI.

Thanks Dale, for this extremely clear article. You were true to your 'point of this blog.'

I'd like to offer further validation of your points:

From my article, Homework First, I cited the supporting work of the Advisory Board Company:

"[Perceiving Go-Live as an end-state] ... failure to recognize and invest in the [post-go-live phase] is one of the four most common, reasons for [EHR project] failure.
The other three, by the way, are
failure to establish metrics of success
failure to design for optimal end-user workflows
, and
settling for partial adoption and co-existence of multiple platforms. Sound familiar?"

They surveyed dozens of large, EHR project in developing their best practices research.

The other observation relates to "... financial payback for an EHR may take years and that is not yet proven."

I think you covered this well. I'd just like to re-iterate that the accounting tools in common use do not capture the economic value created and attributable to the EMR. That wasn't their intent.

Although EMR impact does show up on income statements, balance sheets and cash flow analysis, the impact can only be separated out with accounting tools like Activity-Based Management, that capture materials and labor. Or, another tool is Throughput and Flow analysis, with metrics like TDD, or throughput-dollar-days.  These quantify objectively the economic impact of sub-optimal processes.  They manifest themselves as bottlenecks, which have very real economic implications.  These are processes that are often improvable with EMR enablement.

Typical accounting approaches demonstrate that well-implemented EMRs increase organizational capacity. Capacity that is not utilized by more revenue generating events (admissions, clinic visits, etc) doesn't contribute to ROI. Yet, EMRs have contributed to the only extent they can.  If the capacity is not exploited or exploitable, then there is no objective, earned ROI.

To paraphrase a quote from John Halamka, "you don't need an ROI analysis to conclude that skydiving is better with a functioning parachute than without."

Simply on the basis of paper records versus EMR, there is economic value in EMRs. Legibility, concurrent access by multiple people, the ability to organize information for speed-to-focus to accomplish tasks underlying processes, and essentially instantaneous movement of information to and from departments (pharmacy, lab), all substantially reduce time, FTEs, and waste (poor quality/safety). All have real, large, positive impacts. To say that paybacks are 'unproven' is simply a distortion.

Rigorous, analytic study is critically important. I applaud your discipline. Unintended consequences are real and complicate the analysis. That analysis, though, should be used to refine and improve our use of care processes and the EMR. Not to discredit that the EMR doesn't automatically pay for itself so we shouldn't bother with them.

Thanks again, for your content and positive tone. Both are critically valuable to me.

Anthony and Dale,
     I agree with Anthony's observation and made a closely related point, in my post, "Live From HIMSS: The Waterfront of CDS issues embedded in ARRA"

Physician CDS-informed used of EMRs (item e- the risk of asking physicians to be data collectors, when that's not a good thing)For each of these, there are valid maturity and verification steps that physicians need to own or co-own:

a.  Is there justifiable disagreement?  What's the process and timing around agreement?
b.  More education and agreement necessary?
c.  Background and issues clear, but alternative options incomplete or unclear?
d.  Insertion of these artifacts into work problematic?  "Just-in-Time" often sounds better than it is.
e.  Are physicians being asked to do non-clinical documentation?  Requires thorough vetting.

     I'm not suggesting that this is cut and dried.  There are times when the physician is the best person to collect clinical data.  Especially when the capture and framing has implications to diagnosis and therapeutic decisions.  But, from a process perspective, there are clearly times when the physician is a lessor or least appropriate person to do data collection.  For example, several coordinated care delivery systems use pharmacists and nurses to collect data on drug compliance.  If a patient is having trouble taking a drug, often a pharmacist is far more efficient at sorting things out and getting the patient on the right track (in an ambulatory setting.)

     The point of the list in the box above, concordant with your point Anthony, is that the EMR should be used in designing and improving the process of care.  This is in contrast to blindly using an EMR to automate existing tasks.  For a wonderful treatment of this topic, see either Dr Jim Walker/Geisinger's article in this month's HealthAffairs, or the video of his grand rounds presentation (freely available on the internet) from Johns Hopkins last month.  Incidently, Geiseinger's EMR?  Epic.


Thanks so much for what obviously was a time-consuming post. As a layperson, I SO appreciate your direct, clear, and acronym-scant explanation. What truly strikes me, though, about this post and your others, is your willingness to be so professionally and personally transparent, as you allow the rest of us to learn from your experiences, findings, and outcomes, with no hint of ego or thoughts of keeping it close to the vest.

Your "open source" attitude is exactly what is needed if we've got any hopes of collectively maximizing the potential of EMR adoption, in this layperson's opinion. Thanks - I look forward to learning more through your insights!


I am a recruiter in the industry and one of the main issues is sourcing top talent.. Many of our clients have relied on us to really bring the project home! This is awesome information. I am going to share this with our clientbase. If you need any advise on sourcing talent please feel free to shoot your questions over. jamar.suber@thehcigroup.com