Open Source / Software As Medicine


I believe that closed-source, proprietary clinical software is fundamentally unethical, going beyond even the pharmaceutical industry in how closed source clinical software subverts the duty to share medical knowledge, and protects intellectual property to the detriment of patients.

The Chamberlen Forceps via

When developing healthcare software it is important to recognise the additional ethical dimension that medicine brings with it. Medicine affects all human life in such profound ways that we need to consider the moral dimensions of its technological developments in a completely different way to that of other industries:

In other areas of human endeavour, for example, there is usually more choice available to the consumer — including, often, the choice to opt out of use of that technology. People cannot ‘choose’ not to need healthcare.

Further, the consequences to an individual of being denied access to the best available treatment may well include physical harm, early death, unnecessary pain, avoidable disability, and many other unpleasant and potentially permanent disadvantages.

Medical Software now IS Medicine

Not long ago, clinical software consisted of simple systems for patient administration, databases for recording clinical information, and messaging. Nothing particularly exciting, and nothing particularly likely to make huge differences to clinical outcomes. One can easily see how these dull systems weren’t considered to be ‘part of medicine’ and not thereby subject to the scientific process and moral obligations of a medical innovation.

Now, however, we have a range of complex interactive clinical systems which have become so integral to the delivery of care that it’s likely that a good system could positively influence clinical outcomes (and conversely, a bad system could cause harm!). We have disease-specific scoring systems, clinical algorithm implementations, clinical decision support, patient-facing and clinician-facing mobile apps, triage engines and much more. These things can certainly influence clinical outcome. Yet they are in most cases closed-source, proprietary, non-peer-reviewed, and not independently testable.

Looking to the future, we’re promised that ‘just around the corner’ (I’m sceptical, can you tell?) is an #AIrevolution in which machine learning algorithms (not actually AI, because that doesn’t yet exist) will be able to make clinical decisions. The business model of these algorithms is always based around a closed-source, proprietary, profit-making model. Often there’s a venture capital-based funding model which further distorts thinking around the intellectual property.

But, with these closed-source proprietary systems, there’s no way to independently test their claims, there’s no scientific process, and the need for proper clinical peer review has been completely over-ridden by the wish of the company to make profit.

As software and technology have become more and more part of medicine itself — neither distinguishable or separable from the practice of medicine — we must consider that these software artifacts now have to become part of the corpus of medical knowledge, and thus there is an ethical duty to share these technological artifacts openly, as Open Source software and Open Hardware, in order for the maximum benefit to humankind.

Increasingly, software is being developed which — through whatever mechanism such as reducing error, improving clinician information presentation, decision support etc — has the inherent capability to reduce suffering and save lives, and in such circumstances it is unethical and immoral not to be able to openly share such innovation with the world. The model of open-source software provides a clear platform for such sharing.

Indeed it is hard to imagine how clinicians involved in healthcare innovation could argue against the open-sourcing of this work without it coming into conflict with their responsibilities as a medical professional. If you are a clinician involved in software development, let me ask you: how are you justifying the restriction of that intellectual property to only those patients who can pay?

Medicine Was Open First

Medicine has historically always been an ‘open-source’ profession, even before the term open-source came into being. As an ancient profession, doctors have had ample opportunity over the millennia to experience the dangers of quackery, alchemy, and witchcraft which are the unavoidable sequelae of a lack of open sharing and open peer-review of developments in medicine.

The earliest medical pledge, the Hippocratic Oath, contains in its second paragraph a clear knowledge-sharing commitment — to teach the next generation of doctors ‘without fee or indenture’ (although it stops short of recommending openly sharing the secrets of the medical arts such as sharing with non-doctors)

The Declaration of Geneva, which is the internationally-ratified successor to the Hippocratic Oath, makes this knowledge-sharing responsibility even clearer:

‘I will share my medical knowledge for the benefit of the patient and the advancement of healthcare’. — Declaration of Geneva

So it’s clear in historical precedent that medical professionals have a duty to share medical knowledge with their peers (I’m including all clinicians under this ethical umbrella — even ones who do not take the Declaration of Geneva or the Hippocratic Oath)

Closed Source subverts the Scientific Process

Let’s try a thought experiment. If I was to announce that I had developed a new medicine, or surgical technique, or lab test, which I claimed had clinical benefits, then in order to be taken seriously I’d have to be able to prove that.

Imagine I had a new surgical technique which I say definitely saves lives. But I’m not telling you how. I offer to come to your hospital and perform the operation, the anaesthetist will be blindfolded, the patient will be asleep, and all other staff are sent out. Nobody can observe my secret surgery. I perform the operation, take my fee and leave.

I would be laughed out of the profession for this. Without publishing my new surgical technique nobody will believe my claims of ‘saving lives’. It would be dismissed a pure quackery. Without publication there can be no peer review. And without sharing this knowledge I deny this benefit to the majority of the world. I cause suffering and early death to those denied my proprietary surgical technique. In medicine this would and should be completely unacceptable.

But if you replace ‘surgical technique’ with ‘clinical software’, I’m fine.

Using closed-source software subverts the scientific process by preventing researchers from fully inspecting the intervention they are trying to study.

Medicine nowadays is a relatively conservative profession, deliberately slow to jump onto the latest exciting technological trend, for very good reasons — much of the harms of the latest tech is not immediately apparent, and may take decades to be obvious. After radiation was discovered in the early 20th century, an incredible array of quack cures sprang up — radium and radon as a general tonic, X-rays to treat TB — only for the appalling harms of radiation to become apparent in the decades to come.

In order for the benefits and harms of a new intervention to become apparent it takes careful adherence to the scientific process, repeated studies, replication and extension of existing studies, and most importantly: complete access to the thing you’re studying. It’s hard to study and understand a ‘black box’.

So, what happens when medical innovations are kept secret?

History gives us a particularly appalling demonstration of the human consequences of secret medical treatments — the Chamberlen family of obstetricians — who for around 150 years kept secret their development of obstetric forceps. During the time that the innovation was kept hidden from the rest of the world, it is likely that hundreds of thousands of mothers and babies diedas a result of the lack of this treatment.

Their secret forceps were found by accident hundreds of years later, and it was only then that it was realised how the Chamberlens had been delivering babies, using their proprietary technology, for the exceedingly rich and Europe’s royal families only.

It is because of such precedents that modern healthcare interventions are required to be openly published in peer-reviewed journals before they are taken seriously, indeed it would be unthinkable in the medical profession to offer a ‘secret’, unpublished intervention which was not openly published and its principles understood. To do so in the modern medical world would risk professional ridicule and the wrath of medical regulatory bodies, and would possibly even be a basis for criminal charges in some circumstances.

Open source for quality, safety, and speed.

Sharing knowledge means reduced duplication of effort — if knowledge is freely available to me as a clinician I don’t need to use my resources to create it all over again. Instead I can use my resources to create new knowledge which I can then also share, meaning we all benefit from two interventions rather than one. We all progress faster.

Open publication of medical innovations is the only way to obtain independent scrutiny of experimental results, which helps us ensure that assertions of clinical efficacy are supported by evidence. Without this, we simply have to rely on the assertions of the owners of the technology that the intervention is clinically safe and efficacious.

Open publication allows for continuous, iterative development of an existing technology —for example, another team’s refinement of a published surgical technique in order to reduce risks or improve outcomes — but in the case of closed-source, proprietary software this iterative development simply cannot happen. The copyright owners have exclusive control of the direction of development, and indeed whether development happens at all.

Public Money, Public Code

Quite a large proportion of medical research is publicly funded. This is taxpayers’ money which is used for developing medicine. Where this medical research results in development of software, data sets, algorithms, or any other technological feature, it’s imperative that these be open sourced, so that the the maximum taxpayer benefit is extracted from the investment.

The Free Software Foundation of Europe’s excellent Public Money, Public Code campaign extends this beyond clinical software to all publicly funded software. I agree.

In writing this section, it made me think about the parallel silly situation in clinical research: content is given to medical journals for free, which they then hide behind a paywall, denying access from the very people whose taxes paid for the research. That’s another story, but it’s related and highly pertinent in terms of the revolutionary effect that the Internet has had on traditional publishing and software development models.

Crimes Against Humanity

Ok, here we come to perhaps the bit where I’ve gone too far with my hyperbole. You decide. But I believe that based on all of the above argument, closed source clinical software is a crime against humanity.

(Bear with me)

Think: If one developed a clinical innovation that provably saves lives, then as a doctor one is duty-bound to share that knowledge, such that as many people as possible can benefit from that worldwide.

By denying it to people who can’t afford it, forever, I’m harming those people who can’t access that technological wonder.

“Aha”, you’re about to say, “what about drugs? — the pharma industry denies access to its latest drugs, patents them, charges for them, denies them to those that can’t afford them. And the medical profession goes along with it, even joins in with it!”

You’d be right — it’s deeply unethical — but even in the pharmaceutical industry, their limited patent or exclusivity right has a relatively short life, a fixed duration after which the drug becomes generic, and anyone can make it, meaning expensive proprietary drugs become cheap and widely available with a decade or so of their inception. (And as an aside, you probably shouldn’t be using “well the pharmaceutical industry does it” as a defence of a moral standpoint under any circumstances.)

The pharmaceutical industry has yet to come up with any alternative business models that allow them to ‘recoup’ research and development costs (plus several thousand percent) — without having a period of exclusive production.

But in the case of software, we do have alternative business models: SaaS (Software as a Service), provision of core services around the free software itself, development of custom functionality, integrations, data migrations, feature plugins, apps, security testing, hosting, warranty provision, user support and much more. There’s plenty of ways to make a perfectly good business around an open source product. And because of the collaborative development, the R&D costs can be shared across a huge open source community, rather than being a burden on a single entity.

With software, replication of the original costs almost nothing, so there’s simply no reason not to share these developments with everyone that wants them, as libre software.


As I make all these assertions about how clinical software should be developed and shared, and very hyperbolically state that who impede open sharing of clinical software may be committing crimes against humanity, I’m also aware that there are some among the medical profession who are complicit in this process of closed-sourcing medicine and developing proprietary clinical interventions.

As I asked earlier in this article, I ask again, how do those clinicians justify their involvement in the closed-source, proprietary software that they are developing? Without independent peer review it’s impossible for them to know if they are actually doing good or harm with their intervention, and — if their intervention IS actually doing good — then by keeping the intervention secret they are just as guilty as the Chamberlen family.

I propose that the clinicians and the medical community should actively take steps to prevent the further ‘proprietorization’ of medicine, and going forward we should insist on Open Source software for all of medicine.

We must educate other clinicians in open source software, and advocate openness in all areas — open governance, open algorithms, open standards, open data, and open organisations.

Marcus Baw, MD is a physician and software developer based in Yorkshire.  He can be followed on Twitter at @MarcusBaw

39 thoughts on “Open Source / Software As Medicine

  1. Good read.

    Theoretically speaking, I am all for open everything in a just, egalitarian and prosperous world, like they have in Star Trek. It is however difficult, or perhaps unwise even, to pull selective pieces from that future reality and apply them to a world ruled by greed and avarice, mostly because selective pieces without the whole end up serving and reinforcing the current rulers.

    If the Chamberlen family would have had some form of legal protection for their invention, perhaps they would have brought it to market immediately. They would have offered it for sale to all obstetricians and would have become rich and famous selling forceps. Instead, they had to guard their secret sauce and become rich and famous delivering babies for the very rich who could afford their unique services. Granting patent protections to inventors assures that inventions will be disseminated and that dissemination is indeed more profitable than secrecy. That’s a good thing. And as is the case with all good things, it can and it is given to abuse in some cases. This is why we must periodically retool our laws based on experience.

    Open source software is a wonderful thing. It’s a wonderful thing for modules and basic pieces of functionality. It is not really a way to bring finished (and polished) software products to market. It never has been (with few exceptions). There really is no way to build a good solid business around open source software for any length of time. Sooner or later, those maintenance of freebies shops either go bust or take the code in house, build something much better than the open source version and try to sell that new and enhanced product.

    My biggest beef against the open source model, when used by large corporations, is the inherent devaluation of programmers’ work. People are volunteering and contributing with zero compensation, and then IBM takes their stuff, changes and adds proprietary layers to it, and sells Watson for small fortunes. How is that fair? How is that benefiting anyone but corporations? And yes, I know IBM donates software from time to time. We put perfectly good cheese in mousetraps too….

    That said, I think every piece of software used in medicine should be subject to FDA approval, just like any other device or drug. That does not require that the software be made open source. It should not require that the software is granted a patent either (software patents should not even exist). There may need to be a new process defined for FDA approvals in this field, which is fine, but to me, that’s what is missing now in the emerging “medical” software gold rush. Most of it is just snake oil anyway, or actually intended to hurt people and maximize profits using “evidence based” algorithmic rationing of medical care. Hopefully the FDA will not be corrupted to approve this garbage…. not holding my breath.

    Sorry for the long rant….

  2. I disagree with Margalit and think John’s analogy is weak:

    The disagreement I have with Margalit is easily understood in terms of her point about FDA. Much, maybe most, innovation in medicine comes from licensed physicians acting locally under risk of lawsuit or loss of privileges. The FDA is almost never involved. As clinical medicine incorporates software, a switch to centralized corporate regulation via the FDA becomes a form of censorship. It’s a form of regulatory capture. I’m surprised to see Margalit advocate for bureaucracy, centralization, and regulation.

    1. Adrian,

      In a perfect world I would agree with you. However reality is what it is. We were all complaining about the Big Boys in the EMR industry (i.e. Cerner, Epic, etc.) in the olden days. Little did we realize that in a decade or two, we will be dealing with Gargantuan Godzillas (i.e. Google, Apple, Amazon, IBM), all overlaid by a few humongous insurers, including the public payer, in cahoots with PBMs that double as convenience stores and drive-by clinics.

      Whatever comes our way next will come from these guys. The algorithms for who is worthy and who is not, who is amenable to intervention and who should be tossed aside, and all sorts of other “evidence-based” “precision medicine” “interventions” will be massive. Maybe the day will come when doctors can assert their leadership again. Maybe (real) direct pay will be an option for some. In the meantime I want at least some sort of exposure to light. The FDA for better or worse is all we have. Without it we’re all toast. We’re probably toast anyway, but it’s worth a try, if only to slow the damn thing down a bit.

      1. Regardless of how much of a pessimist one might be, I just don’t understand why a doctor would not favor Free / Open Source Software every chance they get. That doesn’t make them extreme or non-pragmatic. It’s just a declaration of preference, like not eating food off the floor. Is there some short-term benefit to the doctor or patient for avoiding a statement on Free software? The long-term harm to the profession and to society seems pretty clear.

      2. All,

        There is another factor that was touched upon, and then ignored. More, it centrally is involved in all of these discussions. That is: the FDA no longer thinks it has the time, money or inclination to regulate healthcare IT. Rather, their new rules say that if the vendor/company is deemed “responsible” (as defined by past sales/actions), then there will be no oversight whatsoever. The justification for this new carte blanche approach is that because software is so malleable, always emerging and always responding, it’s foolish for the FDA to try to regulate it. Rather, they assume “responsible” firms will behave well.

        My response: the horror, the horror.

        Yes, I see their point that software ain’t like the 10 year old design of a toaster. But does that mean the government must give up entirely? My gosh, we have 45 years of user-hostile, error-generating EHRs. Yes, they have improved. Yes, I do believe they are wonderful in many ways…and better than paper. Yes, they pay lip service to user-centric design. But they still are primitive by comparison to other software. And they are time, error enhancing, clinician frustrating behemoths. They should be better than they are

        Oh, for a big medical center they cost $400,000,000 to buy and 3 times that much to implement.

      3. You were Adrian. I thought I was too. I think we agree more than we disagree. Open Source is great. Free is great. If you find and it meets your needs and you are the decision maker, use it in good health.

        I have no objection to Open Source, BUT unlike the OP, I have no objection to non-Open Source either. That’s all…. 🙂

      4. I think from the start, the FDA had no mandate to regulate EMRs. There was some talk at some point, but nothing came of it, except a very narrow definition of what they can regulate (I believe it was mobile and diagnostic in nature). When the EHR bonanza started in earnest, ONC decided that they should play the make believe regulator role. Maybe it was a turf war…. I don’t know.

        I would venture a guess that the industry, which had a seat at every table everywhere was largely responsible for this position. The usual drivel about stifling innovation, increasing costs, etc etc. etc.

        As a result EMRs were allowed to make claims that were not supported by reality and to knowingly install faulty products with zero consequences, because time was of essence and we needed an EMR in every practice and hospital, just so HHS can plaster its success stories all over the place.

        Maybe now, someone should reevaluate these policies of haste.

      5. I think the issue is partly a communications challenge. Open source evangelists and other technologists often speak their own language and are not adept at translating their concepts for others. It’s sometimes hard for them to understand that the technical solutions that drive them are less fascinating for others, and are often even off putting

        Doctors are barraged by messaging from tech vendors, bureaucrats, consultants, administrators, pharma + device reps … Each one demanding time and selling their solution as the most important new thing. It’s easy to see why so many doctors throw up their hands and head home for the day.

        The key is explaining WHY doctors – and the rest of us – should care. How do you do that without using buzzwords and language that gets tuned out? Opening this up …

      6. Allow me to supply the specific history of the FDA and EHR regulation. (I’ve written extensively on this)
        Technically, the FDA does have regulatory control over EHRs. They are class one devices–which need only report post hoc on known problems. But in 1997, the EHR industry and its true believers convinced the FDA to allow a regulatory free zone for EHRs. Two key questionable forms of “logic” were used in this effort: “regulation will prevent innovation” and “they too safe to regulate.”

        Since then, we’ve certainly had innovations. But, my gosh, think about how the rest of software has moved since 1997. And we have increasingly documented the vast harms EHRs cause–harms the industry has sought to ignore or downplay. Even the recent tidal wave of evidence is contested. At HIMSS right now in Orlando, only happy stories are being told.

        It’s unclear if the innovations we’ve observed are more or less than if there were regulations. Certainly, most clinicians regard EHRs as user hostile and clunky. (And, I recognize there were errors with paper systems. Let’s not have the Luddite argument. EHRs are complex integrated systems that are stunningly useful. They just ain’t as good as they should be or could be.)

        The new rules that say software from “reliable” vendors is defined as safe….sans testing–a new form of carte blanche for regulation, IMHO, are additional capitulations to the industry. We don’t say to the airline industry that because their planes’ software is always evolving we shall not test it if it comes from a reputable vendor. But consider the risk differentials: There are about 100,000 flights each day in the USA. In contrast, there are about 900,000 hospital beds. Almost everyone in those beds is sick. Each patient is getting at least 14 medications a day (counting repeated doses). Plus there are about 3.6 billion prescriptions per year. EHRs affect more people than flights.

      7. Thanks, Ross. Wasn’t there another push around 2009-2010 (if I’m not mistaken) to have the FDA get involved? I vaguely remember some sort of committee concluding that there is no need for FDA oversight, with one very strong dissent opinion. There was massive lobbying around that…

      8. Good memory, Margalit.

        In 2009 I published the linked article in JAMA. It was on the hold harmless clause and the non-disclosure clause. (Bravo JAMA here.) BTW, everyone incorrectly calls the latter a “gag” clause but it’s not that. In a sense it’s worse: it prevents clinicians from disclosing problems they find in the EHRs…or even publicly saying something nasty about EHRs. It also prevents posting screenshots.

        Initially AMIA tried to subvert the article. Four AMIA leaders (who were often CMIOs) claimed all the faults were due to the CMIOs and implementations.

        But the firestorm from AMIA’s leaders’ efforts resulted in an AMIA task force with a published report that called those two clauses unethical. Later, two IOM reports called them unethical. Alas, neither the ONC nor the industry demanded their removal.

        Vendors are not clear about it.

        I’ve worked with healthcare lawyers who tell me the contracts have a “joint and several” clause that makes it almost impossible to ensure those clauses ain’t there. Certainly Epic is still seeking to stop anyone from publishing screenshots….and some industry leader just this week repeated the nonsense that showing a screenshot — even one that led to the death of a patient — is a violation of IP.

      9. Ross: Thanks for this update, dear buddy. As you know I must be a Luddite. I can’t find my 2000 passwords. Any other Luddites out there?

        I realize you, as a giant in the HIT field, emphasize all the substantial under-documented harms of EHRs and still call them “stunningly useful” in comparison with paper. I think you are referring to an idealized future in which EHRs are inter-operable and less clunky among many other current failings. But we have written that almost all of the ~50,000 studies of EHR effects in the last half century have concluded that overall they don’t improve safety, health outcomes or lower costs IN COMPARISON TO PAPER! They do improve process quality, but the major promises have not been fulfilled. When we publish the anti-Luddite remark without evidence we have been dinged by clever commenters.

  3. Dear Steve

    While EHRs are primitive compared to what they should be, they are still amazingly useful.

    Yes, they cause many errors.
    Yes, they eat time by the barrels (mixed metaphor of the week award).
    Yes, they cost tens of millions more than they should.
    Yes, they are freighted with everyone’s good and dumb idea of what to add.
    Yes, they should have similar layouts across vendors and they don’t.
    Yes, they take 3 or 4 or more years to implement.
    Yes, the vendors have fought data standards for 40 year.
    Yes, the vendors psychotically insisted usability was entirely in the eyes/hands of the users.

    But modern medicine is not possible without them. As it is, healthcare is still stuck with FAXES.

    This is not a case of my hating the sin but loving the sinners. The sinners are the vendors who fought usability, standards, etc etc. But the product is needed, IMO.


    1. Dear Ross: Of course some EHR systems are useful. But “amazingly useful?” Where is the body of evidence that says “amazing.”

      HITECH is just one more example– like ACOs and and readmission penalties and genomic medicine– of corporate lobbying and wishful thinking sans evidence of amazing heath and economic miracles which never came to pass. At a cost of trillions.

      There were actually a number of (cluster) randomized controlled trials in both Europe and the US before HITECH was passed by Congress. EHRs and decision support improved processes, but failed to improve health outcomes and improve efficiency. However, a Rand study said HITECH would greatly reduce the cost of national health expenditures. Until it debunked its own study years later after it was too late to stop the train. Here is an essay by Ross et al.

      Ross: you and your enlightened colleagues found they harmed sometimes. Like the other policies above, they did not save money, they did not improve health, but they sometimes increased adverse outcomes including mortality.

      I still think of your case example of the EHR system that changed from weight in kilos to weight in pounds in a single record; it resulted in a high dose of a toxic drug to a young child, resulting in death. This is the terrific work you do to identify the causes of harms and fix them. But how many harms are being fixed?

      The RCTs show no benefit comparing health IT with “paper.” I know you have devoted your life to fixing these errors, but there is no evidence that in totality all that national investment improved costs and health any better than paper. Paper is getting a bad rap. Nether paper nor electronic records follow patients to the many providers they see. The result is that seriously ill mental health patients’ acute admissions are almost never in the EHRs of disconnected PCPs. They have no idea what drugs they got or are getting after the hospital admission.

      We should not have built health policy empires before we had some evidence of benefits and harms. This nightmare is recurring over and over again with no end in sight. Well, on the optimistic side, CMS, after almost 10 years of vocal concerns, is finally considering what to do with the early readmission program after multiple studies show it isn’t saving money and may be increasing deaths among non-admitted seniors with heart failure and pneumonia. My colleagues at Harvard Medical School were instrumental in influencing the adoption of this failed policy and then discovering that it didn’t work almost 10 years later.

      The lesson is we should err on the side of caution and public health when we have “amazing” policy ideas about how to transform the health system. In the last 10 years, and perhaps beyond, we may have harmed more than we helped. This is especially true in cost containment and health IT policies.

      1. Steve,

        While it’s true that the promises of EHR wonderfulness were built on sand, aided by semi-secret payoffs to consultants/authors, and assisted by unbound and uninformed technophilia by policy makers and law makers, it’s also true that studies conducted 10 to 15 years ago (including mine) are not a fair representation of today’s EHRs. Today’s EHRs are both more terrible (72 opening tabs, mindlessly different formats, fonts, standards, etc, etc) but, at the same time, far more powerful and offer far more functions.

        As you well know, I’m among the the authors of the most damning research on EHRs. Perhaps I’m still the lead author attacking them. And the companies that make them have been beyond vicious in demanding my head (and firing). But one cannot use a 14 year old study–or a collection of old studies — to say current EHRs are no better than paper. Heck, you can no longer even find a paper system with which to compare EHRs nowadays.

        No one condemns the vendors more than me for their myriad sins…and monstrous continued lies. But we must use contemporary scholarship to be fair. A head-to-head comparison of EHRs to paper nowadays ain’t possible. And it would be silly.

        Thanks to murderously ill-informed parents, we can compare children with and without measles vaxx. It’s horrific, but we can do it.

        We can point out the horrors of current EHRs, but we can’t say they are less safe than paper. They simply are a hell of a lot safe than they could be and should be.

      2. To Steve’s points, that’s what I had in mind when discussing the FDA. EMRs were deployed based on all the “lovely” things you mentioned, but there should have been something at least mildly resembling clinical trials, efficacy studies, and comparison to what is already out there, i.e. paper. I don’t think it’s too late to ask for that, particularly since clinical decision making is becoming much more dependent on these software products.

        As to comparison to paper, we can still do that too. Certainly not in hospitals of any size, but in small practice. Not sure what the purpose would be though, since for better or worse, nobody is going to uninstall these things now.

        But as they keep adding algorithmic functionality, and this goes back to the original post, maybe we can do a little bit of proper oversight for those portions. For example is the IBM Watson thing undergoing any kind of FDA approval before it advertises that it can treat cancer?

        Most important though, we need to apply this logic not just to EMRs or software products, but also to policy prescriptions. Experimenting on unsuspecting patients with all sorts of penalty/incentives schemes should not be acceptable, particularly in view of past (and current) epic failures.

        There needs to be an independent framework for evaluating changes to the system, a framework that includes trials and evaluation and informed consent. Yes, it will slow things down, but it will also save us many of those wasted trillions of dollars and most likely avoid harm as well. I also think it would be more ethical. I am getting this uncomfortable feeling that most everything we do today is driven by either money, or panic/fear. Either way, this is bad.

      3. Fully agree it was monstrously wrong to demand these billion dollar investments sans trials.
        Fully agree there are parts of EHRs (e.g., CDS) that must be tested….but/and ain’t now.
        Doubt we can find paper based US system for A/B test because CMS loaded the dice: the docs either use EHRs or suffer horrible pay losses.
        But, yes, there may be pockets of functions (more than CDS) that can–and should– be tested… including reliance on all kinds of algorithms and protocols.

        I’ll add that the avg EHR connects to a few hundred devices (or more) in a hospital. It’s a dongle jungle with no testing and much vulnerabilities. Vulnerabilities of pt. safety, cyber safety, unknown myriad connections perhaps even before the many devices reach the EHR. And there are connections with labs, outside labs, pharmacy IT, outside pharmacies, inventory, suppliers etc. Each one is an opportunity for confusion and mischief.

      4. With the benefit of hindsight and due respect to Ross for sounding the alarm, the premise of epic EHR as a clinical tool, no matter how or if it’s regulated is broken from the start. Broken, primarily for business reasons and regulatory capture during the HITECH era.

        epic EHRs are primarily an institutional, material resource planning system akin, as someone mentioned, to air traffic control or running an auto plant. Treating such a complex multi-stakeholder system like a drug or medical device is misguided, if not truly hopeless. The reason for hopelessness is regulatory capture which ensures that patient and physician interest is drowned out by “stakeholder” interests, including intellectual property, that have little to do with clinical medicine. I’m not saying the stakeholders are all wrong, just that their legitimate interests don’t usually favor the individual patients and physicians.

        The clearest example of how epic EHR is broken is inability to create a longitudinal patient record. Imagine how we would get such a record? Would it be tied to a government or other authority? Would it be proprietary software? Would there be mandated standards from some authority? Who would be the customers for such a thing? Would we call it an EHR?

        If we have some idea about these questions, the ways to regulate it might be more obvious.

      5. From someone who truly doesn’t know:

        1. Are you saying that EMRs cannot give you a longitudinal medical record (what I understand by that term is one that tells you the full spectrum of all treatment of the patient, including meds, and maybe even including mental health).
        2. Why hasn’t some enterprising entity given docs what they truly want? Or is it impossible to tell what docs truly want. I mean, in this day and age, apps are designed to be truly responsive to user’ needs. Why doesn’t this happen in healthcare?

        I truly thought EMRs were the way of the future, along with ACOs. Why don’t what seem to be really good ideas not work in healthcare?

      6. Because, Jim, whether sick or well, clinician or administrator, human beings are multivariate systems that are more idiosyncratic than normative. Yet we have the hubris to design ACOs and EHRs as if that was not the case. Multivariate stochastic volatility models don’t work reproducibly to predict the behavior of stocks on the stock market, why should they do better with people. Artificial patients could be well served by artificial intelligence. Most of us need wise human beings who read poetry.

      7. In less poetic words: In health care, nothing, absolutely nothing, scales as predicted on paper.

      8. Sigh…not quite the response I was looking for. I realize humans are variable. Goodness. But bitching about what IS doesn’t get us to what should be. I’m hoping for more.

      9. A willingness on the part of physicians to change how they do things. And a technology platform that allows them to maximize their face time with patients and yet have a full picture of what their patients are experiencing. Is this impossible technologically?

      10. Not at all. It will come. If they left well enough alone, it would have come sooner…. and so we wait…

      11. It’s not at all impossible technologically but it is impossible economically when neither the physicians nor the patients have market power in a privatized system.

        HIMSS is this week. I’m not there because there’s nothing at HIMSS that physicians or patients can buy. Whatever we get is given to us, like Facebook and Google and Epic, under a surveillance capitalism model. The hospital EHR model and the regulators that fertilized it with HITECH leveraged the simple idea that physicians won’t pay for IT and patients are forced to trust authority when sick. As a result, the medical profession will pay ten times over in terms of professional respect and long-term income. The cost to patients and civil society is already an excess 10% of GDP compared to other rich economies.

      12. The fact that humans are multivariate systems and idiosyncratic has a lot of implication for their care. But, I don’t think that it has the same implications for the systematic recording of that care nor the orders to address those issues. Coding restraints may impair one’s ability to address complexity, but EMR has not, in my experience.

        Although my roots are deep in the paper era (having missed clay tablets and stylus narrowly…) I generally find myself defending the electronic record. It reminds me of Henny Youngman’s “How’s Your Wife” question. The answer, of course, is, “Compared to what?” I never see those complaining about the EMR frankly acknowledge the true horrors of the paper era.

        I couldn’t stand it during the paper era when consultants would leave illegible notes with illegible justifications for their illegible (and sometimes mis-transcribed) orders. Or, waiting two weeks for a consultant’s note and find out that he ordered a drug that conflicted with the one that I had ordered. Or, having colleague ask me about the chart of a public figure because he had taken the liberty to look at the chart on a rack. (Privacy would be a separate topic, but all the concern about privacy of EMRs tends to ignore the fact that electronic tracking has made charts much more secure than the paper charts that used to sit around for anyone to open and read. )

        The data gathering capacity of EMRs offers significant hope for faster and better assessment of therapies. Kaiser could never have discovered that Asian women have a risk of atypical femur fracture (a fairly rare complication of bisphosphonate therapy for osteoporosis) six times higher than other women without the mega-data crunching capacity of their EPIC system.

        Yes, EMRs are still clunky (though improving) and the implementation missed the mark in many ways. But, lots of people disliked the nasty, noisy, smelly motor cars that scared the horses drawing their buggies. Change is just messy

  4. Jim: In answer to this question:

    1. “Are you saying that EMRs cannot give you a longitudinal medical record (what I understand by that term is one that tells you the full spectrum of all treatment of the patient, including meds, and maybe even including mental health).”

    Here ( is the abstract of that paper showing a typical EMR (not unusually integrated Kaiser- a tiny fraction of the US) captures about 10% of acute ER and hospital admissions of mentally ill patients. Interoperability has improved only modestly since this time I think but Ross would know better.

    1. ability to show good longitudinal record is dependent on:
      –was all care provided by that one facility and/or by fully integrated facilities. (p = low b/c you’d want all Rx, all outpatient, all hospitalizations, all vaxx, etc)
      –how the implementation was developed.
      –when the implementation was developed.
      –how well even that one facility had its acts together (plural of “act” intentional)
      –use of standard metrics/standards etc
      –when pt was first and last seen AND age of EHR in relation to age of pt. Medical records on cuneiform tablets seldom incorporated….even though they are actually on a hard disc.
      Sigh….and also try for many paper notes, etc.
      –age and condition of servers, backups, ransomwear protections, staff, etc
      –kindness of strangers, sun spots, cut T1 lines,
      –kindness of medical office staff of former clinicians
      –availability of HIE that actually worked
      –quality of documentation in the record

      probably more.

  5. Question (and I really don’t know the answer):
    How important is it for patient care (not research or life insurance) is it to have a complete medical record from birth to current day that includes everything?
    Every cold, every broken bone, every med and every test and every unremarkable physical for say 30-40-50… years?
    Do you guys really need all that, or just major highlights (or lowlights)?
    If you do need this, then fine. But if you don’t, why are we trying to build this right now?

  6. Rainy day musings:


    1. People love to go to doctors. We need to make that event costly and difficult in order to avoid a deluge.
    2. Doctors are trained to overtreat so that opening the floodgate will lead to epidemic iatrogenicity and bankrupt the country.
    3. Doctors can defer the collection of data to impersonal instruments. After all, we are on the cusp of “personalized medicine” which promises that all illness can be reduced to elements for which specific therapies are or will soon be available. Defining those elements in a particular patient is an exercise in data collection, not in history taking.
    4. Empathy and trust are too time consuming and exhausting for physicians to bother with. Physicians and their delegates are the collectors of the fodder for the EMR. The degree to which that collection is comprehensive will determine the degree to which interventions can be personalized. Physicians still sit on a throne and wield certification thanks to their licensure but the throne is unstable and physicians are needed less and less for their wisdom, experience, and moral compass.
    5. Surgeons and other interventionalists wield the sabers and stents of technological advances with the guidance of the all-seeing and all-knowing imagers. The interventional machine is cocked to intervene and the interventionalists are at the ready, and invaluable for being so. The fact that so much has been rendered less stereotypical than the computerized world of gamers only serves to glorify. A decision to forego intervening for whatever reason is neither applauded or rewarded.
    6. “Big Data” are mines containing the nuggets of truth. That pertains to much of modern life: If that’s true for the IRS, Finance, Trade, Actuarial Science, Census, etc.; why not health? The bigger the data, the bigger the nuggets. Hence, the extent of the granularity of the EMR should not be governed by clinical effectiveness but by a quest for a bigger nugget of truth. Clinical effectiveness will follow from the mining of a few of the nuggets that prove relevant to an individual patient, but “Health” requires all the nuggets.
    7. Regulatory capture is efficient. Who can better define the benefit-to-cost ratio than the purveyors who understand the nuances of the interventions? We should increase the reach of regulatory capture into the depths of the FDA, CMS, Guideline Organizations, EBM exercises (like the late Cochrane, Inc.), MOC exercises, and more. In fact, the more the merrier the dividends.
    8. True, the transfer of wealth in the US under the “health care” rubric is in a class by itself. And so it should be. 9. The US shoulders the “peace making” for the resource-advantaged world. It follows that the US should shoulder the “health making”. Vive le oxymore!!!

    And….but cocktail hour is approaching.

    Stay warm and well.


    1. Nortin, after I looked up about five of your words to make sure I knew what they were, I still am not sure you answered my question. But that just might be me.

  7. With respect to longitudinal health records (LHR), I’m with Margalit and Nortin and see what Ross presents as the institutional EHR trap.

    The LHR is patient-centered. That could mean that a DPC physician, a family caregiver, or the patient herself is responsible for what’s there. It also means that any credentialed practitioner (at risk of losing her license) can, by default, access and modify the LHR as long as the patient’s policies allow that class of credentials. The credentials may be issued by licensing boards, hospitals, or community groups anywhere in the world. It’s up to the patient to decide who gets in without a specific invitation. Whoever does get access is held accountable under the laws.

    The LHR does not have to be cheap. The recent HCCI numbers show patients with employer-based health insurance and two chronic conditions are spending over $20K / year. $5-10,000 out of pocket is not unusual. These folks can afford to hire professionals to manage their LHR if we take Nortin’s musings to heart.

    Hospitals can be good centers of excellence for procedures but should be shunned as data brokers.

  8. Luddite here! As if you already didn’t know that. 😊 Thank you for your sane and factual comments Stephen. I am so glad there is someone out there who gets it.

    My favorite quote ever from the Editors thread: “We have written that almost all of the ~50,000 studies of EHR effects in the last half century have concluded that overall they don’t improve safety, health outcomes or lower costs IN COMPARISON TO PAPER!” My answer: Duh?!

    Look, my profession, my patients, and my life are not “an experiment” to see what will work for patients – that is how patients die. I will be damned if I am going to incorporate clunky anything that misconverts a simple weight in lbs to kg. and then causes me to kill a child by overdosing medication. That is not even remotely funny.

    I am an extremely busy practice – well past capacity. I must talk with patients, diagnose them, order necessary lab tests and studies, make decisions based on those studies, arrange consultations and follow ups…. oh and dose medications properly, give immunizations, reassurance and not screw up these children all at the same time. It really helps that I took care of most of their parents and know these patients like the back of my hand, but seriously, how much more do you want to add to my plate?

    How dare policy makers insist on use of a technology that is not an improvement over paper and then force us to use it when it takes twice the time. THIS is the real tragedy.

    And to answer your question Jim, YES it IS impossible technologically. It hasn’t happened because a technology platform that allows physicians to maximize their face time with patients and yet have a full picture of what their patients are experiencing is a PIPE DREAM.

    The problem is until you actually practice medicine on the front lines in a rural, underserved town, you just don’t get it. If you could see the challenges that primary care physicians face on a single given day – well, you would just run like your hair is on fire because it will scare the wits out of you.

    Again, build something, make it open source and then test it. If it holds up in randomized controlled trials against paper, then I say we give it a chance. Before then, leave my profession and my beloved patients to our own devices.

  9. Hello to all

    About time for me to add my own thoughts to this string. While I may not have all of the experience, knowledge, and titles as many of you, what I do have is real-world knowledge of my own.

    So with that, here are my two cents:

    Specific to EMR/EHR, the reality is that we bought into a story that if we digitized a medical record that all of our lives would be changed for the better. We would decrease workload, we would improve care delivery, and our cost to do all of it would be well worth the investment.

    Reality is that we didn’t get any of that.

    Heck, I doubt that EMR/EHR has shown or demonstrated any real improvements since the original stone table medical records called the Iamata. Both are big, bulky, difficult to access, hard to document, interfere with patient communication, not very editable, and doesn’t allow you to make edits easily, let alone transfer from one facility to the other.

    The way that we got here was through this thing that we call healthcare. We incentivized a few early adopters and then promoted them to share their best practices to create an echo of how great it was instead of using paper. We highlighted the cost savings, the monetary incentive, and how everyone loved the new technology. We then had the early adopters attend a few conferences to share best practices.

    This allowed us to accelerate to the mean, scale to an average, and normalize to mediocrity. Why? Because that is how we do things in an industrial mindset.

    When it comes to an LHR, the reality is that it will most likely never happen. In order for this to happen, it would have to begin at birth, and then it would need to extend outside of the natural residence of a hospital or healthcare system. It would need to collect the SDoH. It would need to be pliable, interactive, collective, and seamless. All things that healthcare tech does not want to happen, nor do the healthcare systems.

    Hospitals/HC systems tend to believe if the hoard and house the data that they will be able to generate IP, scale business, and make a profitable business. Again, the majority do not have the intellectual capacity let alone the infrastructure to make it happen. Reality is that they are in the business of locking in their share of wallet and market, but people in communities, and the patients they may become do not inherently only go to one physician, let alone the same hospital.

    Now on to the fun part….Nortin, I am going to use your list of 9 as my sounding board. No offense intended, just my own thoughts.

    1. Disagree. People do not naturally want to go see a doctor. If that were the case we would have more people going in for an annual checkup. People only go when they need to see one. The typical it is broke and now I need a fix versus the preventative maintenance.

    2. Disagree. While some doctors may be in the business of milking the system for dollars, I would argue that physicians do not. Semantics maybe, but there is a difference between a doctor and a physician. A physician took an oath, to heal, not to harm, and to try and keep people in a healthy state.

    3. Disagree. Personalized medicine means that not only do you have a longitudinal data set but that you are also taking into account their genomics, radiomics, proteomics, SDoH and other aspects that we haven’t even begun to correlate or understand today. On the flip side, if we actually allowed a physician to sit in a room, in front of a patient, to have a conversation that was open and honest, built on trust, and then come up with the appropriate next steps within their journey, many would say that is the best “personalized medicine” and be happy. We all want to be cared for at the N of 1.

    4. Disagree. We industrialized medicine and removed the MD, Ph.D., MSc form leadership and replaced them with the MBA, BA and bought the BS that in doing so, they would run things more effectively and efficiently. Everyone in this convoluted industry called healthcare believes that value is defined as Porter did in his HBR article, Value = outcomes/cost. Problem is that again, it is an industrial mindset. We live in the time of the connection economy, and value is defined in terms of attention and trust. I, or you, or any of us, as a patient can spend out trust and attention wherever we choose. So, in medicine value is defined by the 4T’s.
    Trust = (Transparency + Transition) / Time Squared. In all honesty, physicians want to do this, we just do not allow them to practice this way. Physicians have the highest rate of suicide outside of our veterans, and it is because we have removed the ability to connect with their patients.

    5. Disagree. Want to fix 80% of the cost of care in America? Focus on radiology and pathology. Instead of having a system that doesn’t allow the expert to call, to question, to educate, or to make a clinical decision and to make a change to the exam being ordered we force radiologists to do as they are asked in fear of losing orders from referrals. If you focus on the right exam, right time, right patient, to answer the diagnostic question you will save dollars. How do I know? Because this was what I did almost a decade ago. I removed all of the barriers and allowed radiologists and pathologists to drive clinical care decisions, best questions to be asked, and to have a “trump” card to use the best exam for the question needing to be answered. They also handed the patient a piece of paper with a summary, a phone number, and knowing the “who, what, where, when” plan before they left.

    6. Maybe…still yet to be proven true. The nugget of truth needs to be driven by longitudinal data, containing as much of the life journey as possible, and then it must align to the expectations of the patient, not come RCT outcome that has no idea what the end user ultimately expects of their care.

    7. Indifferent…perhaps a necessary evil.

    8. Perhaps it is time to get rid of this healthcare mediocrity and revert back to medicine. We have too many modern Henry Fords that collect big dollars for not really bringing anything of value to the table.

    9. Can always get behind this…cheers!

    Thank you
    Andy, the Cancergeek.

    1. Dear Andy, the Cancergeek


      That means they are a rhetorical exercise in which I’m demonstrating the commonly held falsehoods that drive the “health care” dialectic.

      The fact you thought they were meant to be valid assertions and yet you could find nary one to agree with means we are walking the same path.

      Maybe someday we’ll walk that path whistling, rather than gritting our teeth.

      Those of you who are clinicians practicing outside the cloisters, let me disabuse you of any notion that those of us trying to take care of patients in the “academic health centers” are spared the pain you feel and the hassle you decry. Furthermore, we are carrying on in one of the most toxic workplaces I have studied:

Leave a Reply