Value-based Interoperability: Less is more

flying cadeuciiInteroperability in health care is all the rage now. After publishing a ten year interoperability plan, which according to the Federal Trade Commission (FTC) is well positioned to protect us from wanton market competition and heretic innovations, the Office of the National Coordinator for Health Information Technology (ONC) published the obligatory J’accuse report on information blocking, chockfull of vague anecdotal innuendos and not much else. Nowadays, every health care conversation with every expert, every representative, every lobbyist and every stakeholder, is bound to turn to the lamentable lack of interoperability, which is single handedly responsible for killing people, escalating costs of care, physician burnout, poverty, inequality, disparities, and whatever else seems inadequate in our Babylonian health care system.

When you ask the people genuinely upset at this utter lack of interoperability, what exactly they feel is lacking, the answer is invariably that EHRs should be able to talk to each other, and there is no excuse in this 21st iCentury for such massive failure in communications. The whole thing needs to be rebooted, it seems. After pouring tens of billions of dollars into building the infrastructure for interoperability, we are discovering to our dismay that those pesky EHRs are basically antisocial and are totally incapable or unwilling to engage in interoperability. The suggested solutions range from beating the EHRs into submission to just throwing the whole lackluster lot out and starting fresh to the tune of hundreds of billions of dollars more. When it comes to sacred interoperability, money is not an object. It’s about saving lives.

Every EHR vendor flush with cash from the Meaningful Use bonanza is preparing to take its unusable product to the next level, machine interoperability is shaping up to be the belle of the ball. A simple minded person may be tempted to wonder why people who, for decades, manufactured and sold EHRs that don’t talk to each other, are all of a sudden possessed by interoperability fever. The answer is deceptively simple. After exhausting the artificially created market for EHRs, these powerful captains of industry figured out that extracting rents for machine interoperability is the next big thing.

The initial pocket change comes from selling machine interoperability to their current bewildered (or stupefied) clients, and to less fortunate EHR vendors. But the eventual windfall will not come from the health care delivery system or the hapless patients caught in its web. How much do you think access to a national and hopefully global network of just-in-time medical and personal data is worth to, say, a pharmaceutical company giant? How about life insurance, auto insurance, mortgage, agribusiness, cosmetics, homeland security, retail, transportation? Google built an empire by piecing together disjointed bits of personal data flowing through its electronic spider webs. What do you think can be built by combining everything Google knows with everything your doctor knows and everything you know about yourself?

Machine interoperability is not about patient care in the here and now. Interoperability is not about ensuring that all clinicians have the information they need to treat their patients, or that patients have all the information they need to properly care for themselves. Interoperability is about enriching a set of interoperability infrastructure and service providers and about electronic surveillance of both doctors and their patients. Machine interoperability is about control, power and boatloads of hard cash.

For example, if you are hospitalized, it makes sense that your primary care doctor should know that you are (not in the past tense), and when you are discharged, he or she should be appraised of what transpired during your hospital stay. In the old days, before the advent of hospitalists, this could be assumed. Today, thanks to more efficient division of labor, not so much. If the government was genuinely concerned about smooth transitions of care, it would mandate that upon discharge, hospitals must provide all pertinent information to the primary care doctor, and the patient, by any means necessary. If this meant that a piece of paper is stapled to the patient’s robe, and that the hospital employs an army of delivery drones for the purpose, so be it. Eventually, hospitals, which are big businesses, would come up with the most cost effective and efficient way to be compliant with the law.

That’s not how things currently work or how they are envisioned to work. Discharge summaries have a mandated format of structured data elements, complete with metadata, based on government approved standards that change with frightening regularity. Furthermore, to satisfy regulations, the summaries must be generated and transmitted electronically from one “certified” EHR to another, allowing for a host of intermediaries to access and collect said data or at the very least its metadata. Consulting with the PCP by phone for an hour doesn’t count. Sending the information from a non-certified software package doesn’t count. Printing and sending over information by special courier doesn’t even begin to count. Attempting to build a device that streams the information as it happens directly into the PCP medical record will get you excommunicated or burned at the stake.

If you refer a patient to cardiology service, and in a misguided senior moment decide to pick up the phone and talk to the cardiologist at length about this patient, it doesn’t count. If the cardiologist pens a concise and beautiful letter to you after she sees your patient, thanking you for the referral and summarizing her impressions and plan of care in proper English, it doesn’t count. The only thing that counts is a lengthy clinical summary containing all the sanctioned data elements sent from you to the cardiologist, copied in its entirety and returned from the cardiologist to you, hopefully with some indication about what happened during the consult. Having your EHRs talk to each other this way is considered interoperability. Whether you actually read the interoperated information is irrelevant. As long as the contents are captured by the network for other uses, it’s all good.

But wait, there is more. If you practice, say, in St. Louis, Missouri and work for a huge health system or somehow managed to string together a machine interoperable network with the twenty or so specialists you use on a regular basis and the four hospitals where you have admitting privileges, that’s not good enough. Nothing is good enough unless any research lab in Hopewell, New Jersey or Bangalore, India can discover you on the (inter)national interoperability network and request data about a patient you may have treated five years ago, and nothing will be good enough unless any app store developer in Cupertino, California can discover your patient and subsequently obtain her medical data once she downloads a free diet app from iTunes.

Are you “just” a patient eager to be “engaged” in your own care? Picking a doctor who will spend two hours with you listening carefully and explaining things you don’t understand, and who will give you his cellphone number in case you have more questions, doesn’t count. Getting a team of physicians together on a conference call to brainstorm about your mom’s options, doesn’t count. Building a long term relationship with your pediatrician and having her come see your sick kid at home because your car is in the shop and your toddler can’t keep any food down, and now the baby won’t stop crying, doesn’t even register on the interoperability radar. Nothing counts unless you log into a website or an app, accept the cookies, the tracking beacons, the small print, and then click on some buttons to verify that you are a “Never smoker”, or to peruse machine generated visit notes that even your doctors don’t read anymore.

Perhaps machine interoperability on a national scale is a wonderful thing, but so is having arugula in every fridge. There is absolutely no evidence that either one will improve health and/or reduce the price of care. Every dollar spent on national machine interoperability is a dollar that was previously used, or could be used, to provide medical care. Where did we find the moral fortitude to demand that people experience adverse outcomes at least three times before letting them have a slightly more expensive pill, while spending billions of dollars to incentivize the purchase of unproven and often failing technologies? If we are supposed to be parsimonious in our use of health care resources, if we are supposed to choose wisely in all other areas, where is the comparative effectiveness research showing that expensive machine interoperability on a grandiose global scale provides more value than cheaper and simpler localized or human mediated communications?

  • Add one doctor visit for every Medicare beneficiary for the next 8 years
  • Give primary care a 20% raise for the next 4 years
  • Double the number of residencies for the next 3 years
  • Educate 60,000 new primary care doctors from scratch
  • Buy an iPhone glucose monitor for every diabetic patient and an iPhone BP monitor for every hypertensive patient (no, I’m not a “technophobe”)
  • Put a brand new playground, a gym teacher and a home economics teacher in every elementary school in the U.S.
  • End homelessness in America

These are some of the things we could do with the billions of dollars spent on machine interoperability. Which has more value for our collective health? How did health care become a fully owned subsidiary of the computer industry? Who authorized this unholy acquisition and how much were those brokers paid? Have we forfeited our right to choose, or even know, how endless fortunes are steadily interoperating out of our treasury and into the hands of global technology firms? Publishing fuzzy ten year plans on obscure websites, so the Technorati can tweak them, doesn’t count. Publishing thousands of pages of regulations in the federal register, so interest groups can preview the fruits of their labor, doesn’t count either. Raiding public coffers to please friends and family and to curry political favors is hardly a disruptive innovation, so let’s just call it what it is.

<em>Margalit Gur-Arie is an author, entrepreneur and a partner at BizMed</em>

Categories: Uncategorized

36 replies »

  1. Wow, what a fantastic piece! And what a great discussion, save for the HIT guy who seems to think that because it’s hard that proves we should do it.
    I personally think that some combination of name plus a thumbprint scanner would do the trick relatively cheaply. They cost about 100 bucks on Amazon. My meaningful use stage 1 check could have bought about 180 of them, and would have been money better spent.

  2. “the problem isn’t the goal. The goal is correct.”

    It may be correct, but it also isn’t really that important in the big scheme of things.

    We’d have been better off taking every penny spent on EMRs and putting it into childhood immunizations and school PE programs.

  3. “Yes there is a lot of clear evidence that interoperability will solve a lot of problems”

    Evidence, please.

  4. Oh no, Merle. I wasn’t describing a fault with MedKaz, which I think is wonderfully simple and effective idea.
    I was describing a fault with the “unconscious and naked at the ER” use case, which is invariably being used to justify billions and billions of dollars spent on massive interoperability, that will never address the specific use case, unless we are all tagged and cataloged like a bunch of dogs..

  5. Hi Margalit, Sometimes, we simply don’t recognize that the solution to a problem is staring us in the face!

    Truth is there will always be extreme but low-probability risks associated with any system so you can’t and IMO shouldn’t delay adopting a system that meets your needs most of the time. Not doing so could spell disaster.

    What do you think would have happened if everyone attending HIMSS15 were stricken with some unidentifiable disease and were rushed to a hospital for care? These people presumably — at least in their own minds — are the most informed about healthcare IT. How many of them do you think would have access to all their medical records from all their providers? My guess is one — me, because all my records are on my MedKaz! Put another way, if you or a loved one were critically ill, would you want your/her/his records available on a device you control, or would you prefer to be without them?

    As the CMIO of a major teaching hospital put it “It [MedKaz] gives the patient real power in accumulating their record from any location they receive care. It is a simple, inexpensive, and straightforward solution to a complex problem.”

    WRT to storing patient records in silos in the cloud or on local servers and then trying to access them on the fly when you need them, what happens when an earthquake, flood, tornado or an everyday power outage or system crash knocks out your providers’ computer systems? Thousands if not hundreds of thousands of patients and their providers are left without any patient information! Surprisingly, such events are happening with increasing frequency (think Katrina, Sandy, Irene, Joplin) and I, personally, think they pose a far greater risk with far more serious negative consequences than the situation you describe to fault the usefulness of MedKaz.

  6. Howdy Merle, hope you’re doing well too.
    Just wanted to point out that the hapless ER bound guy, could have been a victim of carjacking and his wallet was in the cup holder when he was dragged out of the car and tossed in the street where he hit his head on the curb and passed out…. so no wallet and no key chain…. at some point we will need to pause, I would think…..

  7. Sorry for arriving late to the party.

    Margalit, great to see you’re alive, well and feisty as ever. Brava!

    JP, there is a simple solution to all the problems you pose — and then some. It does so by aggregating a patient’s complete medical record from all his providers on a device he carries on a key chain or in a wallet. He gives it to any provider he sees anytime, anywhere. It is called MedKaz®. Our company, Health Record Corporation created it. It is the only available system that solves the Interoperability problem that finally is recognized as the missing link in healthcare IT.

    My own MedKaz demonstrates its scope and power. It contains records of more than 340 encounters, tests, etc. spanning 30 years, from more than 40 providers practicing in six cities in three states, using more than 14 different record systems, including paper and the leading EMR systems. Any doctor I see can electronically search for and read any record on it with only two or three clicks, even without Internet access—and I can, too, to add addenda to correct mistakes or add my comments. My doctors have been delighted to use it and it has saved Medicare a bundle of money.

    It easily solves the unconscious-patient problems you raise.

    If the patient is unconscious, EMT personnel or ER docs can access his basic medical information so they can stabilize him. They also can access the information they need to contact his doctors, emergency contacts, and insurance payers, and read his advance directives and emergency care orders, such as DNR or Organ Donor,

    Also, it contains a photo of the John Smith who owns the device so all an ER doc need do is compare the photo with the face in front of him. If they match, it’s safe to use the records on the MedKaz to treat John Smith, the ER patient. If they don’t match, you don’t use the records. Couldn’t be simpler! 🙂

    This short video (https://vimeo.com/126040751) explains what it is and how it works. You can learn a lot more about it and even buy one at http://medkaz.com.

  8. Superb. Unfortunately, I fear the Medicare “doc fix” will allow the EMR masters to keep up their momentum, as doctors’ pay will be very subject to measures as reported by EMR.

  9. I don’t think I am. Let me give you an example. There is obviously utility in having patients being able to communicate with doctors via email if they so desire. ONC should have mandated that all EMRs should have the capability to support that (assuming they wanted to get ahead of the market, because nobody had to “mandate” online banking).
    But that’s should have been all they did. EMR vendors would have responded by buying stuff off the shelf, or building their own stuff. There was no shortage of “secure” email products on the market, and they all work with any client.
    Instead, ONC, felt it necessary to create a brand new standard, enforce its “adoption”, create a web of certifications and roadblocks, etc. What on earth was the purpose of that? And now we’re throwing it all away, because there are newer shiny things….
    Furthermore, they then mandated that doctors and patients exchange a certain number of emails. Seriously?

  10. “The problems are multi-faceted. They are political, legal, technical and on and on it goes.”

    As I have written about at some, continuing length. Google “interoperababble” (I coined the term, those are my posts). They are in fact mostly “political” — in the sense of incumbency/market imperatives. I’m hardly the first one to make that observation.

    “It is not easy to create a data dictionary for every EHR system in existence.”

    If all we had to do was sign up for the checks, we’d all be millionaires.

    Moreover, you misunderstand/mischaracterize my assertion. It’d be relatively easy to create ONE “standard data dictionary” and then require everyone to write to it, deploying whatever schema and UX they wish for functionality and “look ‘n feel” features. That’s what the word “STANDARD” means, notwithstanding that ONC apparently doesn’t get that.

    Not that it would be a panacea. But, it would be foundational.

    I hope FHIR will comprise a practical solution. But, I rather doubt it. I await probative evidence, though.


    “After you have really studied these situations for three years,…”

    I started writing RDBMS code in a radiation lab Oak Ridge in the 1980’s (bench computational “apps,” and business mgmt systems). I’ve been immersed in the issues for a long time as well. So, spare me the patronizing.

    “if it was easy we would have done it already!”

    ONC missed the boat on doing so when it was relatively easy.

    Nice website, btw, despite the lack of ID’ing the Principals.

    You should spend some time over at Dr. Jerome Carter’s “EHR Science” website. Look up some of Fred Trotter’s work as well.

  11. Let’s see: Before MU, every EMR was able to connect to every willing pharmacy in the country and there were bi-directional standards for sending scripts, receiving renewals, checking formulary, checking med histories, all in real time. Before MU, every EMR was capable of sending lab orders and receiving results to any reference lab and any other willing lab, and we had standards for that too. Less common, but starting to emerge, was the ability to do the same for imaging. In addition, every EMR was capable of sending/receiving information about admissions, discharge, transfer, appointments, etc. and we had standards for that too.
    Also, and even before these information exchanges, every EMR was able to exchange claim and remittance information with every payer, and also to check eligibility in real time.

    So what is it exactly that HITECH added to the mix, other than collection of data that contributes nothing to patient care, and a bunch of taxpayer funded HIE organizations with no business model?

  12. The use case you described is uncommon (and that’s an understatement). Identifying a patient based on incomplete demographic data is not uncommon. However, you may want to check with surescripts, because they have a pretty good way of doing that….

    You can’t use SSN because not all people have on. You can’t use driver license because not all people have one. You can’t use insurance member ID because not everybody has one (and it changes). You can’t use a proposed Medical ID because, I guarantee that not everybody will have one either.
    You can however use multiple IDs, with or without a new Medical ID. It’s just a few more lines of code…. Once biometrics, and genetic data become prevalent, you won’t need any external IDs, and the something you know, something you have, thingies will become obsolete. So how much money should we spend on a temporary fix to increase matching accuracy, which is already pretty good, by a percent or two?

    My role is that of a rabble rouser….

  13. Of course the arugula is overkill, although….. people who have arugula in their fridge tend to be healthier, so here you go 🙂

    But what you said is exactly right. And I think that we are creating those inefficiencies because we don’t trust people to come up with their own solutions because they haven’t done so in the past. The thing to remember though is that technology has changed, how people use technology has changed, and I am willing to bet that if we just removed all the regulatory chains today, you would see huge changes in how technology is applied to health care in a very short time,

  14. It is not an uncommon occurence. My point was to explain to you one of the unsolvable problems faced by doctors who lack clinical information. The problem is one of patient identity. How do you determine which John Smith you have in an EHR system when you go to create a composite record of all his pateint data? You can’t even use a social security number for identification as not all people in America have a unique one. For exampler. all members of an Indian tribes share the same exact social security number.
    What is your role in healthcare?

  15. Yes there is a lot of clear evidence that interoperability will solve a lot of problems. It is a very long term goal though.

  16. The HItech act was not a failure by a long shot. A huge percentage of businesses have changed over to EHR systems from paper in a very short amount of time. Interoperability is a very long term goal. It is clearly not a waste of time.

  17. I am an interoperability IT architect and professional I have spent years studying these problems. This is what I do for a living. The problems are multi-faceted. They are political, legal, technical and on and on it goes. It is not easy to create a data dictionary for every EHR system in existence and to get the vendors to change their software. After you have really studied these situations for three years, then come and have a conversation. Think about it, if it was easy we would have done it already!

  18. The arugula in every fridge metaphor is overkill ..

    It trivializes (arugula is a silly, frivolous vegetable that silly people who shop at Whole Foods put on their silly salads) as we describe a goal that’s probably a good idea:

    Making closed systems open

    In information technology,the problem isn’t the goal. The goal is correct.

    It’s the culture and the institutions and the bloated processes that have propigated around the goal

    We’re creating complexity and inefficency when trying to create solutions

    That doesn’t work.

  19. Hold on. You can’t go after people when they build closed systems and then turn around and go after the movement to open them up ..

  20. Medical care is getting worse, not better, directly proportional the the $$$ spent on HIT.

    Agree. Spend it on medications, homeless, infant mortality, ghettos, nurses, nutrition.

    The government has treated the HIT vendors to $$$$. Instead of HITECH, they should have called it the “Great EHR Vendor Welfare Bill of 2008”.

    Would you not rather spend that money on the indigent and downtrodden, or hand it over the the millionaire EHR vendor CEOs?

  21. Now now, broccoli in every fridge, as long as it is ingested, will improve health!

    Hitech has been a collossal failure. MU does not improve outcomes. Interoperability will waste even more time of health care professionals.

    One EHR for all in the land is the only solution.

  22. Makes a person want to go back to Telnet. What fun that was. This was before even TCP/IP. It was a hacker’s seive and was replaced by SSH. There is an Open SSH also. This allows one to use passwords and encryption and take over a target OS and computer. I think the weakness today would be with images.

  23. “Perhaps machine interoperability on a national scale is a wonderful thing, but so is having arugula in every fridge. There is absolutely no evidence that either one will improve health and/or reduce the price of care. ”

    Well said! Somehow we need find a way to pull the plug on the ehr mandates/incentives and penalties if the digitization of health records is ever going to evolve in a way that adds value to doctors and patients.

  24. Thank you for the great explanation, but I still have some questions. How often do people end up in the ER unconscious and with no identification, other than a piece of paper with a name and address in their pocket, which may or may not be their name and address? And for all those unfortunate folks, how often do they die or get harmed because no medical history was available to ER staff upon arrival?

    I think the ROI issue cannot be easily dismissed if we are supposed to use our “finite resources” in a judicious manner. You could make your interoperability infrastructure full proof, by the way, if you add biometrics and/or face recognition, but how much money should we be spending on exotic use cases, while ignoring simple things that can immensely improve the lives of millions of people? Pinning discharge summaries to nursing home patients robes, costs nothing and could prevent much suffering. Just because something is not representable in 0 and 1 formats, doesn’t automatically means it is useless.

    I do agree that building that grand machine interoperability network is a very difficult task. I just don’t agree that it is either necessary or sufficient for providing better patient care.

  25. Clinical Support Systems are generally accepted as having the power to improve the quality of healthcare. This article raises valid questions of cost benefit analysis although it does so with a lot of whining and complaining. People are treated by multiple providers in multiple locations. Doctors need clinical data from many sources pulled together to make accurate diagnoses.

    Let me now toss out some problems that an interoperability healthcare IT data architect works with on a daily basis and see how well you do at solving them. Let’s start with patient identity. A patient arrives at an emergency room in an unconscious state. His name is John Smith and he lives in New York City. Sometimes the date of birth is not enough to identify someone but it is commonly used. You are the ER doctor. Does he have drug allergies? You can’t ask him. What is his blood type? You don’t know. Has he been in this hospital before? Which John Smith is he? You look up the addresses for all your John Smiths none of which match the one on an item in his wallet. Is he an existing patient? Did he move? Is he an organ donor? He has no drivers license. All you have is questions. How can you be sure of who he is? If there was a state or a national Health Information Exchange, chances are you could find this name and this address. But he may never may have been at this hospital. Is there a cost benefit here? Yes! I could go on and on. The vast majority of people have no understanding of the difficulties of Healthcare IT.

    Patient data exists in various vendor silos which are not compatible with each other. This is because the medical data has been coded differently (blood type A+ = “A+” in one but blood type A+ = 10289 in another) . If one exchanged the patient data, it would be meaningless to the other system. So standard international sets of codes are used when exporting data outside an EHR system. Everyone translates to the international code set and a meaningful exchange of data is possible. What is the problem? Well it takes years and many subject matter experts to do these translations and create these international healthcare data standards. It is an on going effort.

  26. These are very good questions being asked here. To understand the real issues here, one needs to delve deep into the workings of Healthcare IT data architecture. Please see healthcareITINteroperability dot com for explanations of these issues. This site was written to explain the complexities of medical terminology, data exchange betwen EHRs, and interoperability.

  27. We’ll “solve” interoperability by defining it down into operational meaninglessness (i.e., we’ll simply give in to more complex workflows).


    Gasoline powered automobiles are “interoperable” is two senses.

    [1] They all run on gas available from ANY gas station (“standard data”), and,

    [2] The “user interfaces” are functionally “standard” as well — both the info “dashboards” and the physical manipulation “interface” tools.