Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Why is medical software so hard to use?
49 points by flerovium on Nov 2, 2021 | hide | past | favorite | 43 comments
-

This is coverage of electronic medical records by The Times, in reverse-chronological time, at the clip of about an article a year:

-

"Do Electronic Records Help or Hinder Patient Care?" [1]

"Our Hospital’s New Software Frets About My ‘Deficiencies’" [2]

"Broken, wasteful, inhuman, expensive, deadly" [3]

"Why Health Care Tech Is Still So Bad" [4]

"Tech Rivalries Impede Digital Medical Record Sharing" [5]

"Doctors Find Barriers to Sharing Digital Medical Records" [6]

"Doctors complain that the electronic systems are clunky and time-consuming" [7]

"The Cost of Electronic Medical Records" [8]

"Usability is the single greatest impediment" [9]

"An Unforeseen Complication of Electronic Medical Records" [10]

"Most Doctors Aren’t Using Electronic Health Records" [11]

"Doctor-Patient-Computer Relationships" [12]

"The Computer Will See You Now" [13]

"There’s no way small practices can effectively implement electronic health records" [14]

[1] https://www.nytimes.com/2020/01/20/opinion/letters/electronic-medical-records.html

[2] https://www.nytimes.com/2019/11/01/health/epic-electronic-health-records.html

[3] https://www.nytimes.com/2019/12/31/opinion/doctors-nurses-and-the-paperwork-crisis-that-could-unite-them.html

[4] https://www.nytimes.com/2015/03/22/opinion/sunday/why-health-care-tech-is-still-so-bad.html

[5] https://www.nytimes.com/2015/05/27/us/electronic-medical-record-sharing-is-hurt-by-business-rivalries.html

[6] https://www.nytimes.com/2014/10/01/business/digital-medical-records-become-common-but-sharing-remains-challenging.html

[7] https://www.nytimes.com/2012/10/09/health/the-ups-and-downs-of-electronic-medical-records-the-digital-doctor.html

[8] https://www.nytimes.com/2012/09/27/opinion/the-cost-of-electronic-medical-records.html

[9] https://www.nytimes.com/2011/07/17/technology/assessing-the-effect-of-standards-in-digital-health-records-on-innovation.html

[10] https://www.nytimes.com/2010/04/22/health/22chen.html

[11] https://www.nytimes.com/2008/06/19/technology/19patient.html

[12] https://www.nytimes.com/2009/03/11/opinion/l11medical.html

[13] https://www.nytimes.com/2009/03/06/opinion/06coben.html

[14] https://www.nytimes.com/2009/03/01/business/01unbox.html

My favorite is "An Unforseen Complication of Medical Records", despite the several articles before it about the complications of medical records.




I wrote Hacking Healthcare for O'Reilly Media, created ClearHealth/HealthCloud, WebVista, yada yada...

There are several factors that collide to make everything in software associated with health exponentially more difficult than in other areas. First, people's lives are actually on the line, the stakes are very high. Second, government regulation plays a significant role in how things can and cannot be done, this is a double edged sword in preventing some abuses but also making iteration and innovation more difficult.

The other most signifcant point is that there are often very large gaps in vision between different layers of a very large and complex human system. The needs of keeping the lights on and billing are often in direct tension to the needs of optimal care, this often manifests in software as a chimera that pleases no one.

One other tidbit I will add is well covered in my book's chapter "The incredible bandwidth of paper". When you actually sit in a room and see how fast and how complex the information recorded by a group of medical professionals with nothing but pen and paper form is, it is daunting to actually deliver that level of performance and reliability with any technology that exists today.


I like the concept of 'paper bandwith'. That rings so true with my experience. I've lived through a number of transitions from paper to Cerner.

At one hospital, they were going live across the entire 600 bed facility in a 2 week period. It was also the start of the new interns - I was there as a surgical registrar.

We went from having our 4 person team (Senior Registrar, myself, 2 Juniors) pushing the cart with the files around, taking 4 minutes per patient, Juniors carefully watching and observing the examination, writing useful shorthand, to a situation where both Juniors were pushing dual-screen COWs frantically trying to keep up with reported findings and plans.

Later in that transition period I was on nights and called to review a patient with SOB. They had been in for 10 days - typical chest that could be APO/could be Pneumonia. The last time the chest findings were documented in Powerchart was 10 days ago on ED admission!! Of course, poor documentation is part of the problem but every day that patient had had their chest examined, there just wasn't a useful way to document it in a shorthand way and the juniors were totally overwhelmed, requiring 2 of them to do the job of half of one of them previously


Thanks for writing this fantastic book. I think it is a must read for any software PM or engineer looking to "disrupt" healthcare. That chapter about paper-based workflows was really eye opening for me as someone who has never worked in that kind of environment.

I think another big area here is that regulations concerning medical software (such as the design controls needed for FDA Software-as-a-medical-device) are simply outdated and do not reflect software development best practices. Instead of things like testing, fast deploys, monitoring, redundancy, scalability, etc the regulations strongly push people to emphasize up-front specification and documentation, which takes an incredible amount of time and energy out of a project.


A friend used to work ER, any time the computers went down, the backlog in the waiting room would get cleared up. Paper works so much better when you're just trying to save someone's life, and don't know which of the hundreds of different forms you'll eventually have to fill out in the computer.


Wow. This is eye-opening.

I'm surprised this isn't at the top.

Is paper better? Why do hospitals use electronic records then?


Because paper record absolutely sucks to retrieve/interpret later, and billing/coding would be even more arduous from paper than it already is with EMRs.


A LOT of money is tied up in reporting that is all but impossible to do without comprehesive electronic versions of medical record information.


Then why isn't more of it done on an electronic paper? If it served in the old world, why can't the software equivalent work?


Because various processes still need to extract structured information out of it to send to other organizations (insurance, billing) and that part doesn't work automatically without someone converting the scribbles on e-paper to verified entries in a specific form.


How did the system work before? My sense is that doctors spend more time on administrative work now than they did before. How is the new system better?


Is there any electronic paper that works well and is reasonably priced?


Isn't it the classical enterprise software situation where the user is not the buyer? These are bought by administrators from sales reps based on features that appeal to "leadership". Its the same with any SAP or other major vendor implementation, the software itself is a steaming pile of crap, but it ticks all the boxes a non-user wants to see (compliance and reporting and whatnot). There is no incentive for the vendor to "waste" money on usability.


Really? I know medical professionals waste 50%+ of their highly-compensated time because of poor usability.

If that isn't a cost to the buyer, I don't know what is.


I think the point that version_five was trying to make is that the decision makers have no visibility on how usable a system is in advance - they only see the checkboxes. At best, decision makers if they're curious, will learn how unusable and inefficient the system is years after they're locked in to their multi-million dollar commitment to implementing and training people on the system. And upon this discovery the vendors are of course there to say, that the hospital just need some more customizations, user training, or a new expensive module. Which just furthers the hospital's lock-in. The lather-rinse-repeat of enterprise software.


Same reason every other awful UI is awful: egos, opinions, hubris, turf, ignorance, neglect, incompetence, resource constraints, NIH, etc, etc.

I designed and implemented the back half of a caregiver facing "portal", working hand in glove with our people making the front half. And I prototyped what became our patient facing portal (think MyChart). And I wrote a medical image viewer.

We were first to market, had customers that loved us, making money.

Then our product team was sold to MedPlus (Quest Diagnostics). Who promptly destroyed all of our work.

I'd already been thru a handful of acquisitions beforehand (both sides), so I knew what to expect. The Golden Rule. He who has the gold makes the rules.

Even so, as a patient, stuff like Epic, Cerner, MyChart, etc. fills me something like despair, rage, hopelessness.

Just today, reviewing my scripts, during a visit, yet again... The hospital's list of scripts for me is wrong. Again. Despite being manually corrected every time I visit.

I know PRECISELY how and why it's wrong. Because I fixed those exact queries myself, back when the hospital had been one of our customers. Over a decade ago.

So aggravating.

As you can guess, a new CTO had come in with grand ideas and ambition to match, threw away their predecessor's work (including our product), and brought in Epic.

Same as it ever was, I suppose.


As a physician I would add:

-The UI of Cerner's product is inconsistent and maddening. A constant experience is that on some screens one must right click to open up extra data fields for entry but elsewhere they open on their own.

-The UI is often randomly changed and one must figure it out again. We would really like reliable tools.

-Latency is substantial.

-Simply reading data is frustrating because popups occur everywhere. We were told that a way to use Cerner is to hover to get a popup with more info. This is rarely useful. What happens is that popups get in the way of what you are looking for.

-Between Cerner Powerchart, the desktop login software, and dictation software there are random roadblocks several times daily. Something won't load or fails, or has a moment of 10x latency, or crashes. This is frustrating since we are trying to help seriously ill people and these tools are random/chaotic.

-Anyone using EHRs know that the EHRs make mistakes, delete things. I've caught the EHR deleting signed orders in the background that I had to reorder. Initially I thought I must be starting to forget to order things on patients or was going crazy. But, when our quality people played back my actions in the EHR (at my request) they could see that I had used it properly and the EHR had deleted orders. We have to overwatch the EHR for safety reasons.

-EHR is a recruitment issue. Some practitioners have bypassed us because of the EHR we are required to use.

-Productivity is at least 3x lower with the EHRs compared to paper flows.

-The EHR contributes to practitioner burnout for many reasons. It is frustrating and exhausting to use. It is random and chaotic.

If anyone from the EHR vendors, login vendors, dictation vendors, or overwatching government agencies are reading (and that might care) I plead with you to improve the nearly uniformly terrible EHR/EMR end-user situation.


> EHRs make mistakes, delete things.

Wow. How do they let this happen? Google never deletes an email.

> Productivity is at least 3x lower with the EHRs compared to paper flows.

If paper flows are better, why haven't they reverted to paper.


>Wow. How do they let this happen? Google never deletes an email.

I'll add that the deleting of signed orders in the background I witnessed occurred a few EHR upgrades ago but after we had been using the EHR for 3-4 years. However, in a high-risk setting like a hospital, once you witness that happen, you forever feel like you have to be on-guard with the EHR for the sake of patient safety. Trust but verify, etc.

>If paper flows are better, why haven't they reverted to paper.

Most of us would love a return to paper but this is not allowed in the US by government regulations unless the EHR has gone offline or is failing in real-time.


I work at a software company that works with medical data. We used to be much more agile before we were bought. After we were bought we went for government validation of our product. Once our system was validated, it really hindered our ability to change it. Every change has to be validated, and the system it touches has to be retested. It has really slowed down how quickly we get out updates, and I can see why software just stops after awhile once it is good enough


If you're particularly interested in this topic, I'd encourage you to look at primary research and not just NY Times. The academic field for technology in healthcare is generally "biomedical informatics" or "health informatics". AMIA (amia.org) is the main association I'd look at, although industry conferences (like HIMSS) might also be interesting for a look at what companies are pitching these days.

I think the main issue comes down to pressures for many different stakeholders. The large EHR vendors (like Epic) were originally designed to replace paper processes, and so they're customizable to a fault. Then, obviously, you gotta pay for the software (Epic has historically not dealt with contracts that are less than $100M, or at least that's what the rumor has always been in informatics circles), and so billing optimization gets tacked on. Issues like usability get studied regularly, but it seems like a more common topic in academic circles than in industry circles. EHRs have historically been closed systems, so you can only really make improvements if your vendor makes improvements (Epic has also, at times, contractually prohibited tools from being added on to their software) -- so, you can't rely on third parties to improve.

Silicon Valley enters the chat So, entrepreneurs see this disaster and think they can disrupt and fix the problem. Only, they know very little about the actual space. Look at the backgrounds of the founding teams of EHR startups (e.g., Practice Fusion, DrChrono) or healthcare startups with home grown EHRs (e.g., Carbon Health). There's very sparse prior healthcare experience. And, while it's totally possible to learn the space, it's hard to learn the space without repeating other's mistakes first.

So, there's been over a decade's (err, actually over 3 decades) worth of effort put in to building standards for interoperability, but the main players at the table have been the Epics... the startup scene has mostly sat it out (either because they think they can build better standards, or they're simply not aware of these efforts existing). So that just creates more tech debt, but without many of the healthcare startup CTOs realizing they've incurred more tech debt. If you've ever wondered why One Medical doesn't integrate with Apple Health, this is your answer ;)

There's a ton of really good work going on in this space though. It just gets overshadowed by 20 years of crappy user experiences with EHRs. Check out companies like Canvas Medical, or researchers like Julia Adler-Milstein or Genevieve Melton-Meaux.


In short, healthcare is very complex and there is no incentive structure that prevents the software from managing that complexity poorly, primarily because the users of the software are often not the ones making the purchasing decisions. Many are optimized for billing, and others are optimized for health systems who want lots of data points and jam the software with data collection forms.

There's lots of _better_ healthcare software out there. It's still complex, while remaining fairly user-friendly, but it tends to live inside healthcare startups and healthcare tech, and not deployed as broadly as the big EMRs like Epic, Cerner, eClinicalWorks, etc. These established players tend to be the only ones with the feature sets that big hospitals and health systems want/need.

Aside:

A lot of people in this thread are complaining about "regulations" making things difficult, but HIPAA, HITRUST, Meaningful Use/Promoting Interoperability, SCRIPT, etc. are not _why_ EMRs suck. They make it harder to start from scratch, but they do not prevent you from building easy-to-use software. The software sucks because there's generally no incentive to make it not suck and a whole lot of legacy suckage with a lot of momentum, money, and influence.


Isn't this problem with inefficient and expensive healthcare systems another example of poorly aligned incentives? The vendors have financial incentive to build and sell systems that require long term, expensive, support contracts. Hospital administrators are incentivized to install turnkey systems (that don't require lots of their involvement), and to choose systems that provide reasonable assurance of regulatory compliance. It is simplest to accept vendors claims of improved productivity and overall long term return on investment. When the cost turns out to be higher than anticipated due to reduced productivity of staff and growing software customization services, those costs can be passed on to patients. From the perspective of the Hospital admins, and the vendors, all is well.


Most of the major medical records systems are iterations of 1980-1990s platforms that (particularly in the case of cerner, or which I have an ungodly quantity of experience with) were adapted from accounting software and are focused on billing (which is not really important where I practice but is a impact of the US medical system and incentives).

The systems have horrific UI/UX with no consistency and every time a new feature was decided to be added a new button was added (ie imagine Word 97 with every toolbar enabled).

As a clinician, you lose the ability to use nice shorthand (ie on morning rounds, the fastest way to convey your examination findings is to stick figure a person and draw lines/ticks/circles/x es through parts of the body to indicate your findings).

You lose all of this with an emr.

The long term contracts and aggressive vendor lock-in (cerner operates a proprietary database and doesn’t play nice with letting other people have access) means we are stuck with these systems possibly forever.

If they were to be redesigned from the ground up it would be totally different but because they have microsoftjuggernaut style BD power the situation is really unlikely to be resolved in anyone’s natural lifespan.

The impact on clinician well-being is immense. The impact on patient flow is immense - a deloitte study of Cerner Firstnet[0] in New South Wales emergency departments indicated a significant decrease in patient throughout - not surprising given the increased burden of administrative responsibilities imposed on clinicians, however something that has not ever been resolved - I work in NSW emergency departments and in places that still do not use first net I can cover 10-12 patients in a shift; in a first net hospital I am lucky to break 8

[0] https://www.google.com.au/amp/s/docplayer.net/amp/3816353-Do...

Ultimately they are not fit for purpose, they are not designed for humans, they create patient safety risks (whilst now I think potentially decreasing some other risks due to automated systems monitoring observations etc) - medication management is horrific and makes it easy to make errors - and the situation is unlikely to change

[0] https://www.google.com.au/amp/s/docplayer.net/amp/3816353-Do...


It used to be the patient's charts were literally hanging from the foot of the bed and the doctor doing the rounds would flip through them.

Is there any current system that provides that level of frictionless experience for doctors? Or observability that the doctor is actually looking at the charts thoroughly for patients and patient advocates? In my observation there isn't. Although I will say that the ICU doctors I've observed are generally quite conscientious with the digital systems, even though it's clearly about as frictionless for them as JIRA is for us.


OP here: Every medical professional I've spoken to, including relatives, has nothing but vitriol for medical records software.

I got blood drawn recently and the nurse spent 20min helping someone enter medical codes into EPIC.


It’s billing that’s the problem. The software has to account for all sorts of billing codes to make sure it’s billed correctly.

I’ve said it before but I think medical billing is the problem with healthcare in the US.


As a patient I've given tech support to nurses that are stuck with the unbelievable piece of shit software known as epic

I wonder how much fraud and abuse is done because of that software. The lack of interoperability between different providers and even different locations from the same provider

I've never committed fraud, but I've used the systems crappyness to my advantage many times

It's a complicated ui built upon a shitty data model that wildly varies. I think the developers must have heard "ui and schema" well fuck all that noise


I have a related question...

The two big EHR systems (Epic and Cerner) are infamously complicated, user-unfriendly monoliths. And they're not interoperable, so once you pick one, it's very hard and expensive to switch your hospital to the other.

My question is, could EHR be "refactored" into smaller pieces? So your billing system is provided by Vendor X, and your imaging system is provided by Vendor Y, and they all talk to each other over an open protocol? That would make it easier for hospitals to switch away from individual systems / vendors that they hate.


I work for a medical startup and most of our clients' staff at small to mid-sized practices are poorly trained on the software they use. Even when it's a single EMR software they stay in all day long, helping their patients, a large part of their staff won't understand how it actually works. It's like clinics buy whatever software to handle the breadth and volume of work they want to produce, then neglect to train their staff on it and get frustrated when it doesn't work out. Sometimes repeatedly.


Because the primary concern of all projects is compliance. Building the thing is an afterthought and it's gonna be expensive and take for ever to deliver anyway so everyone just throws their hands up and accepts it.

Our whole system is designed around making Healthcare this highly moated bonanza for insiders who know how to gameplay beaurocracy so I'd argue its not much of a surprise that that's what we get: a bloated beaurocracy


Is this the case?

Transportation and logistics (especially national and international) is highly regulated, but the software is highly usable compared to medical records.

Hedge fund automated trading software needs to pass innumerable compliance requirements. But the software is superb.

Low-latency arbitrage software has incredible compliance burdens, but is some of the best engineered code on the planet.


People who own small logistics firms are culturally far more likely to fly in the face of regulation than those in the AMA.

This could explain the adoption/demand of more-cutting-edge SaaS etc.


Why do medical records cause providers to spend _more_ time than before? If the old system was faster for doctors, why not make the new system like the old system but electronic (electronic paper)?

The opinion of most providers (and comments below) is that before medical records, doctors were able to treat more patients and give each a greater level of attention and quality of care.


our company does M&A tech evaluation for companies and every now and then we run into medical software companies, the results more often than not are sad for everyone. Our experience has been that the companies are generally very well tooled up and organized to have conversations with HIPAA auditors where if we weren’t part of the evaluation they would be free and clear. But because we are in the conversation and because of our evaluation methods (most of which are hands on, fun fact most HIPAA auditors will never touch your system in any way) we generally end up with results that show companies have over invested in business process teams and under invested in best in class engineering to bring that business process to market in an industry best process kind of way.


I think one of the additions of the MDR is the requirement to carry out usability studies in the development of medical products, so hopefully this'll lead to less clunky (and less risky due to user error) medical products


1. FUD 2. Regulations 3. Money, or money for certifications 4. Outdated hardware or underpowered hardware 5. Incompetence


What makes it so hard for innovation is all the rules and regulations like being HIPAA compliant etc.... Regulations are a tax on innovation and a floor on prices dropping (usually leading to prices increasing).


Is this really the case? Finance is highly regulated and there is a lot of superb software.


Many of the fintech companies and successes have been around circumventing rules and regulations using clever tricks and/or loopholes.


How is fintech like healthcare?


Not true. HIPAA was a non-issue for my team. Mid 2000s.

Further, having any kind of regulations, rules, procedures whatsoever would have been a huge boon. Instead, we had to invent everything ourselves.

I actually called up lawyers (and others) involved in court cases, asking for pointers. "Try harder" was the answer.

Sadly, I could never figure out how to code "try harder".


HIPAA, maybe. HITECH and Meaningful Use would be more relevant (2009 onwards).

It's been a while, but I remember thousands of pages of granular user-interface-level requirements (think 'display of telephone number icons') to quality for federal recognition as an electronic medical record system. Hospitals/clinics not using a federally-approved vendor received increasing % penalties to their reimbursements as deadlines passed. Forcing through EMR adoption was meant to improve outcomes.. or something.

Some of the requirements are defined by references to standards prepared by consulting companies, which charge for access to these standards. Additionally, certification of compliance with standards was farmed out to a couple consulting companies.


Ugh. Textbook regulatory capture. Sounds like hell. I've dealt with certification hazing in other industries; worse than useless. Further, my direct experience with CSC, McKesson, others is that they are irredeemably bad, complete bights on the economy. No different than every other big IT & mgmt consultancy.

Our problem with "meaningful use" was figuring out wtf they were even talking about, that there was no "there" there. Like being the boy who saw the emperor was naked.

HITECH was after my time.

During our time on the whipping post, we were grateful for all the standards for UI and reports. Less thrashing with customers. It seemed like every hospital had that one doctor who was an frustrated undiscovered human factors genius hell bent on sharing their vision with the world thru the medium of CRUD. Never mind the opinions of the actual users of ours stuff. What we'd call bikeshedding (or sabotage) today.

I don't know what to say about top down bureaucratic coercion. Short of switching to single payer, I can't imagine any other way to get competitors to play nicely together. That's not to say it's worth doing. But I've never seen outcomes which justify the effort (ROI), in any industry. Hopefully someone has examples, case studies, whatever of success that will prove my cynicism wrong.

As it stands today, all that federal money is just cheddar for campaign donors. Any accidental benefit to either patients or care providers will be quickly snuffed.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: