In Defense of Epic. No, Really!

flying cadeucii

Today THCB is delighted to feature an excerpt from Robert Wachter’s much-talked about new book “The Digital Doctor: Hope, Hype and Harm at the Dawn of Medicine’s Computer Age (McGraw Hill, 2015). If you enjoy this piece, be sure have a look at the director’s cut interviews Wachter did for the book with Atul Gawande: “Computers Replacing Doctors“,  and John Halamka: “Black Turtlenecks, Data Fiends & Code.” — John Irvine

That Epic would find itself labeled a monopoly is in itself an extraordinary turn of events. In 2000, after 21 years in business, the company had only 400 employees and 73 clients, and did not appear on a list of the top 20 hospital  EHR vendors. Its big break came in 2003, when the 8 million–member Kaiser Permanente system selected Epic over two far better known competitors, IBM and Cerner. The cost to build Kaiser’s electronic health record: $4 billion.

Today, Epic has 8,100 employees, 315 clients, and yearly revenues of approximately $2 billion. The system is now deployed in 9 of the US News & World Report’s “Top 10” hospitals. In 2014, the company estimated that 173 million people (54 percent of the U.S. population) had at least some medical information in an Epic electronic record.

Epic Founder and CEO Judy Faulkner’s vision, built on several central tenets, has been vindicated many times over. The first principle was that the winning EHR vendor would be the one that solved the most problems for its customers.

While Apple’s App Store has made a modular environment seem feasible and even desirable, most healthcare decision makers want a single product that does everything they need right out of the box (physician notes, nursing notes, drug ordering and dispensing, billing, compliance, and population health) and does those things everywhere, from the newborn nursery to the urology clinic to the ICU.

Said David Bates, chief quality officer at Boston’s Brigham & Women’s Hospital and a national expert in health IT, “If you make a big matrix of all the various things that you want as an organization, Epic covers many more of the boxes than others.” When it was choosing which EHR system to purchase, Partners HealthCare (which includes Brigham, Massachusetts General, and several other Boston-area hospitals and clinics) created just such a matrix, and Epic’s system covered more than 90 percent of the boxes.  The nearest competitor covered just 55 percent. Choosing Epic, said Bates, “was not a close decision.”

Partners’ investment in its new EHR system will total about $1 billion. As Epic often makes clear, only about 15 percent of an organization’s electronic health record investment ends up in Verona. The rest is for training the organization’s own workers, for paying its own IT staff and for outside consultants to help with the implementation.

Second, while Epic’s software has been criticized for its lack of interoperability, most healthcare leaders don’t stay up at night worrying about that. Don’t get me wrong: they care deeply about moving information around; it’s a core rationale for EHRs in the first place. But their definition of “around” is not everywhere. Rather, they want a seamless flow of information around all the buildings they own (within the hospital, between the hospital and their clinics, that sort of thing). They also want interoperability between their system and an outside laboratory they use, their system and Aetna’s claims department, and their system and the local Walgreens.

And they do care—up to a point—about connecting to the patient. But this is where things get nuanced, because the information the hospital CIO wants to make available to the patient is finely calibrated: enough to make the patient happy (scheduling appointments, refilling medications, e-mailing her doctor), but not so much as to risk the franchise. If you come to UCSF for a kidney transplant, it’s good for business if you also get your  shots and primary care from us. If you go to see that famous Sloan Kettering oncologist for a second opinion, the cancer hospital would also like to administer your chemotherapy, even though there is absolutely nothing special about its version of the medication, and you could probably get it cheaper at a clinic down the road.

Just think about it: if a patient can go online and shop around for the cheapest CT scan or colonoscopy (“Hey, Groupon’s offering a $75 MRI at the mall today!”), this creates a problem for the nation’s healthcare Meccas, which count on the profits from these activities to pay the bills.

One way to prevent such shopping around is to be sure the hospital’s IT system doesn’t connect to outsiders that the hospital views as competition. In fact, one technique that hospitals use with their Epic system is to buy it and offer nearby clinics free or subsidized access to it, but only if they admit their patients to that hospital or are otherwise part of its network. Epic implicitly encourages this by not selling its system directly to small practices. A clinic that chooses to stay outside the major medical center’s orbit may find its staff standing at the fax machine when it needs to exchange information with that hospital.”

Third, Faulkner was patient. While other companies were busy merging with one another in order to grow, or futzing around with new product lines, Epic remained confident that, someday, the market for electronic health records would take off. Faulker’s employees focused on building a solid system that would solve more problems, and solve them better, than their competitors’ systems. It worked. Fueled by superior customer ratings and word of mouth, Epic’s client base grew slowly and steadily.

“We were making a healthy profit before HITECH,” said Epic’s president Carl Dvorak, showing me a curve of customer growth from 2000 to 2014. While the graph depicted an unmistakable uptick after 2010, it also illustrated steady year-over-year increases before the federal incentives kicked in.

Point taken. But it is equally true that the EHR business represented a sluggish corner of the healthcare economy before 2010, so nobody but insiders really noticed Epic, or cared very much about it. The signing of the HITECH Act instantly changed that, and Epic was ready—if not for the politics and the attention, then at least for the business. “The reason that Epic got so much of the market is that its product is simply better,” said David Bates. I agree. To frame Epic as somehow having orchestrated HITECH—for its dominance in the past few years to be portrayed, as it sometimes is, as a massive conspiracy launched from the Wisconsin farmlands—is just plain silly. Given its 30-year history of quiet competence, it would also represent a sleeper cell story of, well, epic proportions.

From The Digital Doctor by Robert Wachter, reprinted with permission from McGraw-Hill. Copyright 2015.

Robert Wachter is a professor of medicine at the University of San Francisco, California. 

Categories: Uncategorized

Tagged as: , , ,

20 replies »

  1. Can anyone point me in the direction whether EPIC can filter notes in a way that IP can pro-actively screen for surgical site infections. Specifically COLOs and HYSTs ? If so how?

    Anna Christenson RN, BSN

  2. Pts do not want their records, generally, and those who do comprise, the worried well. This scenario is being repeated internationally.

  3. Can Epic be defended without a detailed analysis of its impediments to care, the errors it causes, the delays it causes, the intellectual distraction it causes, the deaths from its unavailabilities, and others?

    Evidence please.

  4. One cannot forget two things. The one’s most involved in the physician’s medical records are physicians. The medical record is a log of what a physician does originally intended for recollection in the future. It is not a newspaper report of fact, rather a document of some fact and a lot of presumption where the context is best derived by the one that wrote the note. Another physician might write a similar note and differ on the facts and the presumptions. Context (the important feature of a clinical note) is difficult to convert into the 1’s and 0’s (yes or no) that prevail in computer language.

  5. One cannot forget two things. The one’s most involved in the physician’s medical records are physicians. The medical record is a log of what a physician does originally intended for recollection in the future. It is not a newspaper report of fact, rather a document of some fact and a lot of presumption where the context is best derived by the one that wrote the note. Another physician might write a similar note and differ on the facts and the presumptions. Context (the important feature of a clinical note) is difficult to convert into the 1’s and 0’s (yes or no) that prevail in computer language.

  6. Phil (and others on the chain):

    I support patients having full access to their own records (this seems like a no-brainer to me) but think that such access is being massively oversold as a panacea. It’s hard enough for a doctor — with 7-10 years of post collegiate education/practice — to make sense of the 10,000 facts that populate a sick patient’s chart. The idea that an average lay person will be see all of this data and not be overcome with anxiety or confusion is politically correct hogwash. My son had a bad headache the other night and made the mistake of looking up his symptoms on WebMD. Luckily, I was there to reassure him that he wasn’t having both diabetes and acute glaucoma, because reading the site convinced him he was having both.

    If everyone was an MIT grad student, then maybe… but everyone is not.

    The main advantage of real people having total access to their records is that it will help ensure that providers will have the same access. Managing one’s health and healthcare is simply not the same as managing your travel or finances. Perhaps one day the technology will be so good that healthcare will be safely democratized, but we’re going to see a lot of mischief, hype, and even some harm between now and then. God bless — and long live — Mr. Keating, but depicting him as a typical patient is like depicting me as Tiger Woods because we both play golf.

  7. Bob-

    Points all well taken. As long as we have a fragmented, fee-for-service health system, even with the ACA and Epic rolling out, care won’t be streamlined/coordinated.

    What are your thoughts about Steven Keating, the MIT PhD student who is arguing for “full patient data access to clinical records, in simple, standardized, and digital forms through an open API”? It doesn’t seem like we will get there with Epic or VISTA/CPRS or AthenaHealth…



    with best regards, Phil

  8. Thanks, Philip. I appreciate your nice words about the book.

    Re: the interoperability within Epic systems (ie, Epic-to-Epic connections), this kind of connectivity actually is a major selling point for Epic. At UCSF, we now are able to connect, without breaking much sweat, to most of the other places that have Epic. Since the company has nearly run the table in our region, that’s a huge number of places: all the Sutter hospitals, virtually all of the University of California hospitals, Stanford, Cedars Sinai, etc. It is one of Epic’s major selling points; at least it was for us. And it’s proven to be a real advantage. There are ways of connecting to non-Epic shops (“Epic Elsewhere,” as opposed to “Epic Everywhere”) but it is dicier.

    Sure, the VA is more seamless but that’s less an EMR issue than a broader organizational one: the VA is an actual health SYSTEM. There is only one identifier and it stays with the patient anywhere in the system — in essence, there is only one chart. Not gonna happen in our pluralistic mess of a non-system, outside the VA and Kaiser. As I wrote in the book, Epic users today actually have a higher degree of interoperability than users of any other system; ironic since Epic has dragged its heels on more general issues of interoperability (Epic to non-Epic).

    — Bob

  9. I enjoyed your book- it was excellent- especially the exposee on the TMP-SMX overdose. My issue is less with Epic’s lack of a user-centered design. Presumably the user interface will improve dramatically, just as Gmail and Facebook have improved immensely over the past decade. My principal concern is the lack of interoperability between hospital systems which use Epic. Since Americans don’t have a unique medical record number that travels with them, Epic doesn’t allow a doctor from one hospital to read records of the same patient in another hospital system which also uses Epic. This is to be contrasted to the VA’s VISTA/CPRS system. In VISTA/CPRS, you can read records from any VA in the United States. That’s incredibly useful if a patient got an EGD or MRI in VA-San Diego and now they are in VA-Atlanta. I don’t see that happening in our brave new Epic world.

  10. Thanks, Jeff. I appreciate your kind words about the book — they mean a lot, particularly coming from someone I admire so much.

    There is no easy answer, and no answer at all until someone produces a better product. I doubt there will be another contender that produces an enterprise-wide, does-everything product that can beat Epic (at least for huge health systems) anytime soon. The dethroning is more likely to come from being forced to become open (by legislation, if needed, or — as it is today — by the threat of legislation, combined with public shaming) and then for niche products to take away parts of the business, open API by open API. Little by little, this will chip away at the monolithic market dominance. But it will take a long time — like carving a glacier.

    Before being too thrilled by this prospect, it is important to reflect on Epic’s point of view: namely, that there are some real advantages that accrue from a fully integrated system built by a single vendor. Yes, I know it violates our app-store sensibility, but my Starbucks app doesn’t really need to integrate with my podcast app. A pharmacy inventory app better integrate seamlessly with a CPOE app and then a bedside bar-coding app, lest you have major tsuris. Yes, they are control freaks out in Verona, and yes, their decision to build and own the entire ecosystem is partly a business strategy. But their argument for integration holds some water, at least for some parts of the healthcare workflow — the particularly complicated risky parts.

    Re: the vendor lock question, the bottom line is I’m not sure what all the chess moves look like that end up with the dethroning of Epic. But I’m guessing people said the same thing in the early years of Microsoft, of Compaq, of Kodak, and of Enron. Companies stumble, pressures grow, new ideas emerge, and stuff happens. Healthcare is different, but not that different.

    What odds would everybody give that Epic will be leading the pack in 10 years? I’d say 60-40, maybe 70-30. But not 90-10.

  11. With Epic’s increasing dominance, what incentive is there for the company to address the usability issues you described so eloquently in your book? The switching costs are enormous for the customer and the barriers to entry are in the multi-billion dollar range for anyone foolish enough to try and dethrone them. Why doesn’t the field remain essentially frozen in Windows 95 space?

    (Your book was the best thing I’ve read about our field in a very long time!).

  12. The only thing worse than a monopoly is a government controlled monopoly. Epic is being targeted, but it is not the problem. Vendor lock may be the worst problem impairing innovation. Following in the footsteps of Windows, I guess.

  13. For those who accuse me of being “too easy” on Epic (something you might think if you read only this excerpt), let me point out that 1/5th of The Digital Doctor is devoted to a 39-fold overdose that occurred with an Epic system. I also describe the clunky interfaces, the relative lack of user-centered design, and some of the ethically problematic rules, such as the prohibition against sharing screenshots.

    But as the excerpt points out, Epic is sometimes blamed for decisions made by their users — health systems and hospitals — who also have an interest in holding their data close to their collective chests. Moreover, the reason Epic is getting all the heat now is that they are clearly winning the battle in the market because, at this moment, they are the best system that a big, complex hospital-based system can buy.

    I find this as troubling as others do, since monopolies are generally bad things… and in the specific case of healthcare, they are even worse — since Epic now controls a huge chunk of the data in the system and can set the rules about how such data are used. Moreover, our market doesn’t work like a normal one; the issue of “vendor lock” is real — once you’ve invested a few hundred million dollars in your Epic EHR, you’re not going to switch to a “better one” without a huge amount of angst and the rolling of many corporate heads.

    The pushback against Epic for being closed (by ONC, the media, their competitors, and patient advocates) is healthy, since the company’s DNA is to stay closed; this has been their winning strategy up till now. But they will open up (their data, their APIs, etc) if the pressure to do so overcomes their innate resistance. This appears to be happening, and I see it as a healthy development.

  14. Great analysis that turns conventional opinion on its head.

    One thing this does not overturn, however, is the view that payers in the market have failed to make a business case for hospitals about some information-related aspects of performance and that incentives might be used to accomplish this. For instance, payers have failed to convince hospitals that the costs of achieving HIE are outweighed by the benefits.

    In another post (https://thehealthcareblog.com/blog/2014/11/26/rip-meaningful-use-2009-2014/), you called for the sunsetting of meaningful use except for creating standardized and publicly available APIs. At the time, I thought the post made sense. But reading this one, I wonder if we should double down on MU, putting another large pool of incentives behind it to make hospitals care about using the APIs? The payments could then go, for instance, to the pass fail testing Fred Trotter described here: https://thehealthcareblog.com/blog/2015/04/03/learning-from-our-interoperability-failures/.

  15. Agreed. Right now there are a good number of patients who refuse to share relevant information with their docs…..and the number is going to grow as more become aware their intimate secrets are up in the cloud waiting to be hacked or used for surveillance and control by the social engineers and technocrats (most of whom are true believers that this will all be for the general welfare).

  16. This is a big deal to our culture: putting the most intimate details of everyone in effectively readable archives, that might be essentially permanent.

    I think the societal impact is going to follow a latent incubation period of a few years, while we assimilate and think about what has happened to us.

    There isn’t anyone without a medical secret. There isn’t anyone who cant be embarressed.

  17. Read the reports of Kaiser’s former employee, Justen Deal circa 2007. Wall Street Journal front page.