[nradov]

Standardize what? With a complex medical condition like that every patient is different so it's tough to apply evidence-based medicine practices.

The technology for sharing patient records is gradually being standardized. But that won't help if the doctors don't actually take time for a detailed review of the patient's records.

[trowawee]

Personally, from my observations as a software engineer married to a doctor and friendly with lots of doctors, and as an occasional participant in the American healthcare system, the general inadequacy of electronic medical systems is one of the most maddening aspects of our system. Epic is a monstrosity and none of the other systems are much better. The tech is outdated, the interfaces are awful, they're incredibly difficult to get to communicate with each other, and it sometimes feels like every single provider is using a different system that won't integrate with any of the other systems. It makes me angry as person whose doctors are using those systems (and it kinda scares me), and it makes me angry as a professional, because whoever the members of our profession are who are building those systems are doing a crap job.

[maxander]

> whoever the members of our profession are who are building those systems are doing a crap job.

They're doing an excellent job, probably producing a level of functionality comparable to Amazon web developers or Oracle database writers. They're doing an excellent job at navigating the unholy mess of archaic regulations, mismatching institutional requirements, and hostile corporate interests, without getting sued or convicted. User experience, or even user usability, is a secondary concern, since being difficult to use (even if this leads to multiple deaths per day) is in no way illegal.

Medical informatics is definitely a field where smart and motivated people can make a huge difference to the world, and perhaps even get filthy rich while doing it. But it will take much more than a hotshot UX designer to work out.

[pnutjam]

American EMR systems are designed around billing systems. This is almost entirely the problem.

[knightofmars]

Yes, this too.

[riverstones]

Our situation is almost identical. I refuse to even go to the doctor unless I have something I suspect is going to be life threatening. My wife thinks I'm nuts, but she understands. I grew up in Europe under the "socialist model" and I miss it dearly.

If I need to go because I have an infection, I go to the indigent clinic, and pay the $25 for anything they need to do. I then go around the corner to the indigent pharmacy and get my $4 RX.

Yes, we have insurance. But it's still so freaking expensive to use it I never go. Insurance premiums, co-pays, RX costs. Why even go unless you HAVE to go. My wife takes the children to see the doctor when they need it. I refuse to go. I've not had a physical since the early 90s and I don't plan on going. I refuse to enrich the system. Now that the mandate has been undone, I may drop myself and cover my family alone. I can always go to the indigent clinic. Anything more serious and I just cannot afford to pay. I do not want to leave my family in medical bankruptcy or massive debt. It really sucks that in America, one has to first consider whether one can afford to visit the doctor in the first place. And I cannot convince my wife to consider living in Europe or Australia, both places we could easily move to and adapt with our relative skill sets.

[abawany]

While I admire your principles, it seems that by not getting physicals, you appear to be trading off your health for your principles. It is a choice for you to make but might I suggest that if the insurance/healthcare that you get is too expensive, it might be worth making an annual trip to a suitable country to get the physical done. Alternatively, consider paying a doctor with cash for a pre-negotiated price.

These are unsolicited suggestions on my part but my concern is that by not getting these physicals, which arguably have a significant impact in early detection of preventable diseases such as high cholesterol, you are endangering your life. Just my opinion.

[sli]

> the interfaces are awful

I honestly have no clue how my eye doctor fills their stuff out. It looks like someone discovered Visual Basic's visual design tools, went nuts, then never evaluated whether or not it was a good idea. No alignment on the inputs, they're just placed willy nilly with seemingly no rhyme or reason beyond maaaaybe being contained inside a labeled box. Maybe.

I'm convinced that whoever designed it thought that inputs in the middle of sentences was a super clever idea. Well, "designed."

[knightofmars]

Maxander is correct. As they say, "Don't judge a person till you've walk a mile in his shoes." As a someone who has been writing healthcare software (not EPIC but close enough) for nearly 10 years I can tell you it's not the software engineers.

The regulations (often vague and open to interpretation by the customer) often play a part in creating the monstrosities that power our healthcare systems. Because of the subject to interpretation aspect customers often say, "No. This is how it has to work because our processes say this is what we do to meet the regulation." Inevitably it's implemented to be configurable because that's what's required.

Another culprit is the institutions and lack of standards surrounding process. There's a reason EPIC software is customized for every institution it is installed in. It's because every institution wants to do things differently. Even in the space which I work, it's the same. Every institution wants "some specific change" that they can't live without and won't go live until it's available. I'm saying this is neither a good nor a bad thing. It's just a reality.

And the ever present legacy, take EPIC as the example, it was founded in 1979. I'm not saying that their code is all from 1979 but there's definitely a fingerprint of what was in their modern day applications. There are layers upon layers of data from mergers and acquisitions translated into various codes and mapped to various databases for any number of uses. Any day of the week your state code may be two letters, three letters, full name, a custom internal legacy code, you name it you'll see it.

Spend a year working for a company with a regulated legacy healthcare product and significant user base. You'll have to become proficient at security, regulations, data standards (HL7, FHIR, etc), legacy data migration, and any other number of skills. If you're lucky enough to have all of those in place then you're still going to spend time coming up to speed within the specific healthcare domain you're working in and where it touches other healthcare (and non-healthcare eg financials) domains. Oh, and often you won't be allowed access to production instances to troubleshoot issues and a copy of the production instance isn't available because HIPPA and the customer is uncomfortable giving access to engineers. You get really good recreating problems purely via error logs and staring at the code where the issue "might" have occurred.

[byzin]

I've worked as a healthcare professional provider in a major medical center in a major metro area. My parents are both physicians and my wife is also a professional provider. I've used EPIC during an initial rollout, and my wife has used it during two rollouts. I've also used other EHRs in smaller clinics since then, and have used the VA system.

My sense is that EHR mandates were colossal screwup. They should have never happened. No matter how good they seem in the ideal, mandating them should have never been the case.

The reason why is because each hospital had a very well-tuned staff with a system that was designed for that hospital, in-house, over years. Implementation of EHRs should have been done the same way, ground-up, on a site-by-site basis, in a way that allowed for more gradual, flexible adoption with complete autonomy by each site. If they wanted to buy into something like EPIC, great. If they wanted to develop something in-house, great. If they wanted to contribute to an open source project, great. That sort of system would have been much better in the long run.

As it happened, EHRs were just sort of slapped on, top down, with the providers being forced to adapt to them rather than vice versa. It was horrible, and a perfect example of government regulations fucking things up. I'm very pro-public sector, nonprofit, etc. but also think that regulations (in terms of restrictive licensing laws, FDA nonsense, things like EHR mandates, etc.) are the unrecognized disease in American healthcare systems.

EHR mandates at each hospital system I or my spouse worked at to resulted in cost overruns of billions of dollars. Those are just two systems in the US, and believe me, neither of those hospital systems--which were very successful, well-run enterprises, without EHRs--would have never implemented them when they did without the mandates.

The most egregiously stupid thing about the mandate is that EHRs would have been implemented in both these hospitals relatively soon anyway, but it would have happened on a much better timeline, in a much more sane way.

[nradov]

The EHR mandates were very loose and had long deadlines. Hospitals had plenty of time to buy or build whatever they wanted. But it would be ridiculous for almost any hospital to develop something in house; they don't have the engineering resources or a core competency in software development, and it would be a huge duplication in efforts. Some hospital administrators like to think their institutions are unique and special snowflakes, but the reality is that most of them are the same and would operate more efficiently with standardized tools and processes instead of something customized.

While a single hospital might be able to operate reasonably well without an EHR that's just no longer sustainable in the broader healthcare ecosystem. Hospitals can't be islands. They have to be able to share data with payers, public health agencies, other hospitals, outpatient clinics, researchers, etc. Doing that requires an EHR now.

[knightofmars]

Agreed. Just standards in general. I'm all for continuing efficiency improvement through pilot programs but 80% of the function and business process of healthcare should be standardized. However, is it going to happen? No, probably not. I see it similar to the metric system. The US decided not to adopt what is a (better and universal) standard because "That's not how we do things here."

[nradov]

The EMR usability problem is tough and expensive to solve. Medicine is extremely complex and every specialty has a different workflow. Build a system that pediatricians love and oncologists will hate it, or vice versa. Instead of just complaining that Epic developers are doing a crap job, what specifically should they change?

The communications and integration problem is gradually improving. In order to comply with government mandates, most EMRs now include HL7 interfaces that comply reasonably well with current standards and no longer charge extra for that feature. But every system still has a different internal data model so something is always lost in translation.

[epicthrwaway]

Everyone is saying "don't blame the engineers", so I felt the need to post this.

I am a software developer at Epic. I've been here long enough to have gone on sabbatical.

This place has some of the most outdated software engineering practices I've ever seen. There are nearly no automated tests; a team of ~1000 people manually tests everything, including things like scaled pub-sub systems. The company actively maintains about 25 million lines of code. 1000 people couldn't possibly manually cover all of that each year. The majority of known bugs get released.

Internal builds are broken more often than not; there is no CI system exposing this fact. The code is mostly giant, unmaintainable monoliths. I saw a single class that was 50,000 lines long. Imagine trying to modify that monster with no automated tests to support you.

Nearly all of the developers that Epic hires are fresh out of college. Life-critical systems are regularly designed, built, and manually tested entirely by developers with <3 years of experience.

I've seen more than one bug get released that could have killed someone. It wouldn't surprise me if one has.

There is no incentive to change. We make so much money that the execs do not care. It makes me depressed.

[mr_overalls]

ICD 10 is a medical classification system by the World Health Organization. It contains thousands of extremely detailed codes for diseases, signs and symptoms, abnormal findings, complaints, social circumstances, and external causes of injury or diseases. It is used by all parties in the healthcare system.

https://en.wikipedia.org/wiki/ICD-10

[nradov]

Everyone in the industry is well aware of ICD-10 and we've been using it extensively for years. It's only one small piece of the puzzle and not nearly sufficient on its own. Just within the terminology / ontology space most clinical systems will have to support a wide range of others that are mostly orthogonal to ICD-10 such as CPT, RxNorm, LOINC, SNOMED-CT, etc. And that's not even getting into issues with message and document formats.