Submitted by The U.S. government claimed that turning American medical charts into electronic records would make health care better, safer, and cheaper. Ten years and $36 billion later, the system is an unholy mess: inside a digital revolution gone wrong.
It seems to be a recurring theme all over the world that governments are absolutely terrible at doing anything related to the digital world. I’m sure insane bidding requirements set by special interests play a huge role in this problem, but that doesn’t mean politicians tend to be terrible at properly understanding the digital world.
Disclosure: I work for a combined teaching/hospital organization that has a very large Epic install– but I’m not an expert on Epic, so take whatever I say with appropriate amount of NaCl.
You don’t “buy” Epic– You buy an infrastructure, and a number of analysts, employed locally, and by Epic, and your system is a one-of-a-kind system that’s written, built and configured for your organization. Integrating with existing systems is a massive headache, as most pre-existing systems are either homegrown, written by a company that folded years ago, or written by a company that has no incentive to work with Epic. There are Epic modules that can replace lab, pharmacy, even label printing systems– but that means retraining, converting of existing systems, etc..
There are standards (HL7), but implementing it requires more home-grown code, and some form of interface engine– Our old one was essentially discontinued and the new one has difficulty keeping up with the number of transactions per second. And not every existing piece of medical software speaks HL7 natively.
There is no “standard UI”– the UI is customized based on each organizations needs, which sounds great, but the people designing the UI aren’t UI experts– they’re doctors, nurses, or technicians.
When the system works, it’s fantastic– as a patient, my records are available to myself, and whichever doctors within the organization need access. Unfortunately, my primary isn’t a part of this organization, so he has his *own* EHR system, which doesn’t talk to the Epic install at my place of work.
When I asked my doctor about his system, he said after 4-6 months of hard work, he was now at the point where he could see patients as quickly as he could before implementing the system.
All in all, EHR’s aren’t a bad idea, but the medical industry has hundreds of largely incompatible standards, and unifying them is non-trivial.
I too have worked with Epic. And, yes, you’re right that its not one system but a customized ecosystem.
I also trained in the time of paper charts. Every doctor had their own way of documenting the same thing. There was no standarization in nomeclature, vocabulary, style, handwriting, sentence structure (or lack-there-of) or anything similar. Paper was a disaster. For any particular patient, maybe somewhere between 10% and 90% of the information documented would be unintelligible (unless you knew what to expect, then you might be able to see that the scribble said MI vs something else). (That’s illegal by the way — charging Medicare requires legible notes and signatures otherwise the documentation doesn’t justify the bill/charge and its considered fraud.)
But it was “good enough” at the time and a necessary evil — it made seeing patients fast enough to not let someone die because you were still working on documenting on the previous patient. In the end, it slowed down clinics and rounds because it took a while to “interpret” and put a story together about what had gone before.
Now imagine all of those hundreds of thousands of doctors are told that they have to do it on a computer. And some can’t type. The base system needs to be everything to everyone (and is sold as such). So its going to be a UI nightmare. Each doctor or team of nurses will try to convince those in power that they need it “customized”. But that costs money and so won’t happen. Each department will try to convince those in power that they need it “customized”. Often they’ll have some budget and so it will be done to an extent. Each hospital will do the same. They set the budget and buy the ecosystem, so they’ll make their unique EMR based on a product like Epic.
Hospitals aren’t going to pay for the systems to inter-operate. Any part of their tight budget (I’ve no clue how much these migrations cost but it must be in the millions of dollars) will go to making the ecosystem work with their environment — their insurance carriers, their pharmacies, their labs, their imaging centers, their call centers, there transcription, paging, phone, security, HR, calendar, email, . . .
No money can be earmarked to make it easier for the hospital’s patients to pull their information out and move to another provider. This isn’t out of malificence, just budgets.
None of this is unique to healthcare, the system is just much more massive. How many legacy software companies can easily pivot to a different source control system, build system, security system? Not many. It’s a mythical company that has the ability to use a git monorepo where everything is standardized, continuous integration tests, streamlined build processes for multiple targets, continuous deployment, and have dependency tracking and incident response. Most software companies probably don’t even know how many servers and services they have running, let alone keep them up to date.
Now, suppose their field had the potential to kill someone quickly and it was mandated that they deploy CI/CD with formal verification. Yeah right, that’s not going to end well.
The other problem is semantic interoperability. Even if you have a well functioning HL7 2.x interface (which is still limited in what you can send between systems), there still isn’t conformance necessarily on what meaning to derive from the data from the originating system (by the destination system). Additionally, so much of healthcare is still reliant on “documents” – often TIFF of PDF files that it is difficult to discretely transmit data from them across systems. Attempts to develop stronger standards by document types – e.g., the CCDAs – haven’t borne out as well as expected either because EMR vendors either 1.) charge insane amounts to implement the interfaces or 2.) they only implemented the bare minimum so that if I were to send you a referral with a summary of care, you could only discretely import labs, medications, and problem list.
I run a small health information exchange and maintain the analytics for two ACOs in Midwest, and from my experience (15 years in healthcare IT), this was a relatively measured article. I would venture to say that the current state is probably a lot worse. The government basically bankrolled an EMR business model that allowed vendors to charge high costs for interfacing under the guise that the providers and health systems could make up the money with incentives. EMRs could have been a good thing if they were designed for their intended user (i.e., physicians and other clinical staff). The government’s incentive model made the government – and to a lesser extent private payers who hounded providers for more data to bump up their HEDIS and STARs ratings – the customer, not the actual users. This is a classic case of a third-party payer economic problem. Regulation won’t make it any better, it will likely just strengthen the problem with EMR vendors developing for government diktat, and then blaming the government when they produce poor quality code.
I make good money off of healthcare IT, have a lot of experience both on the inpatient and outpatient sides, and while I have found some of it to be rewarding (mostly projects that bypassed EMRs and that have shown a tangible benefit to patients), I am personally at the point where I find the entire industry so disheartening that I am actively working on a plan to exit it because it likely won’t get fixed; it needs to be blown up and rebuilt. I go to my doctor, and I get irritated when he is too busy trying to figure out a poorly designed product.
The one item I will ding this article on is that it doesn’t note the blame that payers – public and private – deserve for this mess. Private payers – especially Medicare Advantage plans – harass providers for documentary evidence to up the risk on patients, and all plans but burdensome documentation requirements on providers; if the excessive data entry is not there, and an audit occurs or there is a billing dispute, the provider and his employer (if he works for a system) lose revenue. That creates a cycle where providers are spending immense amounts of time adding as much documentation as possible or “pushing through” data from previous notes to justify payments. That creates large and unwieldy notes that may have older, inaccurate information. This is only getting worse now that the US is embracing value-based payments and everyone is rushing to get as much health data as possible to perform risk modeling and other analytics exercises.
Governments get shit for overspending, so they are forced to choose the lowest bidder. The lowest bidder doesn’t know what they’re doing, leading to cost blowouts and a higher price tag at the end.
Companies should have no business putting in bids they know they can’t deliver on, and yet public opinion demands they get picked. The fault lies squarely on the companies for not doing their research properly and trying to scam money off the public.
kwan_e,
.
While I agree your logic is sound, Darwinism suggests that the company that gets the contract is more suitable to it’s environment than the companies that might play fair or are honest about the long term costs & performance. So while you can blame the companies involved, so long as the environment is such that the cheapest bids win without much regards to quality, then survival of the fittest will result in less qualified companies surviving because they cut corners. Blaming the company only gets you so far because if they didn’t cut corners, it would leave an opportunity for another company to do so and win the bid.
This actually hits close to home for me because I’ve been loosing clients to offshore teams who come in way below me on price. Sometimes I can’t even believe how low they bid, it’s unreal. Just this week I had a client come back from a 20 or so month hiatus and tell me the offshore teams that they left me for keep screwing up, and they want me to fix it now. Great, but they still want me to come in closer to the price they were paying the other teams. I’m so frustrated by this, but it’s not an isolated incident, outsourcing to india specifically is huge business and I know that I’m loosing business because I’m not matching their prices and outsourcing the work myself.
This is O/T, but in all honesty guys, I could use advice. If I can’t make more money than this, I might have to give up on my consulting and take up a regular 9-5 job, even though I really, really don’t want to. I’ve thought about transforming my company into an outsourcing shop to be more competitive, but even the thought of it doesn’t make me feel right. I like consulting, I just hate the race to the bottom. I’ve decided that I want to work more on AI, but I still have to find a profitable business model as a small company. I’ve considered moving to the tech hubs where there are more opportunities, but the costs of living are so high for a family and with my wife loosing her job I worry that we could be worse off. I don’t have an easy answer, anyways what’s up with you guys? Haha.
The companies are chosen by people, not by natural selection. Darwinism is irrelevant. And the companies may be to blame if they perform poorly. Getting the job is no excuse for doing poor work. Doing sloppy work may have worked out for them, just like robbing old ladies may work out for me if I did it, but it working out for the one doing it is not a good excuse for doing it.
Though none of us really know what happened and none of us know who was to blame. We can only speculate. Maybe the companies who got the job were micro managed. Or maybe they just did not follow up on their promises. I do not know so I am not putting the blame anywhere.
Checker of Types,
There really is no mystery. They always fail the same ways.
They underestimate the complexity of the project. The bidding executives make up some magic numbers out of thin air to beat the competition. As a bonus, the company has probably gone through rounds of layoffs over the previous years to appear streamlined, only to find out they do not have enough experienced people to carry out the project.
Either way, to meet the budget they assign few people to many tasks, or in Alfman’s case, outsource it to people who don’t even live in the country. Then they either are lazy, incompetent or too busy to talk to the people who use the existing system to find out what the problems are and how big the system really is.
And because they’re not a government organization, they develop the new system as though it was a commercial project – fuck the users who get left behind. They don’t understand it has to work for literally everyone, because everyone is a paying customer. They don’t get to buy out a competitor and shut them down.
…because human actions are “special” and not subject to evolutionary mechanisms? Ever heard of memes? (hm, that’s a trick question 😉 )
Alfman,
StackOverflow allows you to search for jobs that offer remote working, and there’s quite a lot the last time I checked. So it won’t need to be a 9-5 office-and-suit job. Most of them are quite interesting projects too.
Either way, taking up a 9-5 for a while will give you time to work on an AI business model.
Had to move from Perth to Melbourne at the end of 2017 just to find a programming job, since Perth is a dead-end. Even for consulting, it’s mostly just boring enterprise .Net or Java or Oracle crap. A smallish 9-5 company is the sweet spot.
Checker of Types,
Textbook darwinism is originally presented to us in terms of sexual selection and culling of species in the wild, but you’re missing the bigger reveal that it actually applies to other kinds of diverse systems too. Economic darwinism has been used to explain a lot of things unrelated to darwin’s finches and biology, even those where human choice plays a factor. I can’t take credit for this observation though:
http://www.geoffrey-hodgson.info/user/image/darwinismecon.pdf
https://en.wikipedia.org/wiki/Evolutionary_economics