To me this article (frontend developer imagining healthcare data simplification) reads sort of like if a personal trainer wrote a few pages on why we should all stop using plastics. A laudable goal, yes, but the post does not shed light on the extremely entrenched problems or offer a glimpse of a path forward.
HL7 v2, v3, FHIR, LOINC, ICD, SNOMED, RxNorm, J-Codes, NCPDP, Z-Index/G-Standard... Epic, Cerner, Allscripts, etc., all amount to a curious bag of standards and companies that might not be very interested in commoditizing themselves.
The majority of healthcare interfaces I've seen run on creakingly old--and much-abused by idiots--HL7 v2.3. And I've even seen some signed overpunch codes within the past year. https://en.wikipedia.org/wiki/Signed_overpunch#The_codes
Have you worked with any of the things you listed?
I ask mostly because this is largely a list of words associated with healthcare tech rather than any coherent statement about the complexity.
For example - LOINC, ICD9/10, SNOMED, and RX-Norm are all methods of coding various things. A few of them have overlaps or translations and I’ll agree they are a bit of a pain. But using the same tool to express that a patient had their left hand amputated and to describe the dosage and form of the pain meds they are given doesn’t make sense.
Similarly - Epic, Cerner, All Scripts are companies. They compete. That’s like saying that we only need one news company.
And you’re attacking hl7 v2, honestly outside of FHIR (which my professional involvement largely stopped at dstu2) it’s the least awful. Sure it’s funky and a hair old but it’s fairly easy to work with. I will say I much prefer FHIR but it has its own warts (extensions can be a pain)
There are certainly problems in healthcare but blaming just the tech is akin to blaming the hammer when your house is framed wrong. Sure maybe it was the hammer but maybe the fact that the contractors were drunk was a bigger issue than they tried to use a banana as a hammer.
> Have you worked with any of the things you listed?
Yes, and I offer the (indeed rather scattered) list as a sampling of the formats and taxonomies one encounters--sometimes by essential need to code wildly different concepts like "struck on head by tortoise" vs "lower back pain, moderate" vs "Lisinopril 5mg oral tablet", sometimes by separate efforts across time or geography or organization to do sort of the same thing. From the outside it might seem a lot more tractable space than it is after you get into the weeds.
Although my time with HL7 v2 over the years amounted to probably less than 100 focused hours, it didn't take long to encounter folks mangling the spec in ways HAPI rejects, mashing giant base64-encoded PDFs into ORU records, etc., and worst of all, using the "flexibility" of the records/segments to ram through rubbish not aligned to any nomenclature. Ugly pipehat serialization without schema or validation. I'm sure someone with 10 years at MGH would have a very different impression, but this one is mine.
There's not really any substance in this article and this person doesn't have enough insight into the healthcare industry.
I disagree with the premise that the issue lies in accurate diagnoses from inaccurate data. There is value in second opinions and maybe some very specifically applied machine learning, but modern medicine is far too primitive benefit from large amounts of data per patient. The bigger issue is the complete corruption of American healthcare. The whole system's gotta go.
The only good thing in the current system is that HIPAA allows you as a patient to access every single of your medical records from any provider you go to. You won't be able to do much with these records, but it's still nice.
Problem: EHR companies like Epic have corrupt leadership and are purely profit driven and intentionally make it harder to share data between provider networks. The Epic CEO Judith Faulkner is evil and directly profits from this. Same with Cerner and Brent Shafer. They have no desire to make data easier to share.
Problem: Same for hospitals, many of which are corrupt and make millions despite maintaining non profit statuses. They have an incentive to make it harder for doctors not in the hospital network to access their records. Forget about record sharing and easy healthcare data access.
Problem: Most hospitals only care about profits, not patient outcomes and efficiency. This runs deep and leads to inefficient hospitals, including reduced doctor availability. Hospitals also profit from things like billing errors or intentionally sloppy billing. Optimizing for outcomes is irrelevant to hospitals.
Problem: Doctors have conflicting evidence of what to do for health outcomes. For example, some doctors still incorrectly think lowering cholesterol will improve cardiovascular disease, and think people with high cholesterol should be on statins. Giving them more data isn't going to help fix their poor training.
Problem: There's corruption among doctors who care more about profit motives than patient outcomes. They have figured out how to game the system (there are many ways).
Problem: No one in America wants to see a doctor because of cost. And doctors and hospitals and everyone else in healthcare has gamed the system to make everything as expensive as possible, so the desire not to see doctors is more than warranted. You probably know health insurance companies are totally corrupt and the cause of this, but it's also doctors, hospitals (the biggest enemy), the government, medical device manufacturers, non profit health awareness groups, consulting groups like Deloitte, physical therapists, there is corruption at every level of healthcare.
> You probably know health insurance companies are totally corrupt and the cause of this, but it's also doctors, hospitals (the biggest enemy)
How are health insurance companies the cause of this when I can’t get any healthcare provider (aka doctors) to spit out how much they will charge me?
No one is stopping doctors from telling me they will charge me $100 per 15 min, or give me the codes they are going to charge. Yet when I call various doctors offices to schedule a routine procedure such as an ultrasound or a physical or X-ray, they all tell me to sign a blank check.
I can figure out how much I need to charge to develop real estate, but a doctor can’t tell me how much (or what codes they are going to bill) for a physical? And if you so happen to start talking about something, you might be charged a different charge code that costs you hundreds extra?
Doctors who don't run their own practices usually don't know how much their services will cost, because those prices are determined by private negotiations between practices/hospitals and health insurance companies. The rates vary depending on location, provider, insurance carrier and other factors.
Health insurance companies pay their executive staff huge salaries while working as hard as they can to deny medical claims to patients. And the ACA mandating the minimum medical loss ratio for insurance companies had the unintended side effect of incentivizing health insurance companies to pay more expensive medical bills, because they can only pay their exec staff a fixed % of what they spend on health claims. The more they spend on claims, the more they can pay their executive staff.
Doctors can be corrupt in many ways, but not being transparent about what they'll bill isn't an important one.
>Doctors who don't run their own practices usually don't know how much their services will cost, because those prices are determined by private negotiations between practices/hospitals and health insurance companies.
It's other doctors who own the practices and companies. As far as I can see, the golden years for doctors were 80s to 00s, but now they are having to band together as it's big insurance company vs big doctor group vs big hospital group, and if you're not big, you get crushed. Sucks for the younger doctors, but the decision makers are many times still older doctors.
>And the ACA mandating the minimum medical loss ratio for insurance companies had the unintended side effect of incentivizing health insurance companies to pay more expensive medical bills, because they can only pay their exec staff a fixed % of what they spend on health claims. The more they spend on claims, the more they can pay their executive staff.
Yet the rate of increase in healthcare costs went down after ACA went into effect.
>Doctors can be corrupt in many ways, but not being transparent about what they'll bill isn't an important one.
That is an incredible claim to me. Literally, the only time someone asks me to hand over an infinite amount of money is when I purchase healthcare, and if one can't see the moral hazard there, then I don't know what to say. We're not talking about a person in the ER with unknown ailments and required work, we're talking about completely routine procedures that healthcare providers are refusing to give information about.
They could provide me the codes so that I can ask my insurance company what it will cost. They could take those codes and look up what they negotiated with my insurance company. But they do neither. Because why not get a blank check from everyone.
There's no guarantee a doctor will know what their procedures will get coded as. Doctors who join hospital networks are also often surprised when their learn what their patients get billed. You can literally get a degree in medical coding, and practices employ medical coders to extract as much money from medicare and insurance as possible by exploiting the system. The salaries of these coders might appear on your hospital bills. Consulting groups, CMOs and billing departments might change codes depending on what insurance or medicare will accept. Like trying to bill one surgery as three separate codes. There's corrupt back room work that doctors may never see depending on the practice.
There are some corrupt doctors who work in specialist networks to suggest how to code the most profitably for procedures, but I consider that different than doctors who do regular day to day practice for a larger entity and know less about how their work gets billed (and might not have a direct say in it either unless they're on the board).
Those criticisms of Epic and Cerner are rather unfair. Both companies participate actively in HL7 interoperability standards development and expose fairly complete APIs for accessing patient charts.
IME these efforts towards interop are rather superficial. They may still be implemented in wildly incompatible fashion, and you will need to open up the ol' wallet wide even to get access to figure out whether it is feasible to build something--anything--to work with Epic.
It would be nice if everyone had control of their own personal health record but so far no one has really figured out a viable business model. Microsoft and Google tried and failed. Now Apple and Ciitizen are trying again, with Apple using it as a value-added feature to sell more iPhones and Ciitizen selling data to researchers. Hopefully someone will succeed. But in general most consumers just don't care; the only segments that actively engage are educated, affluent people with chronic medical conditions and the "worried well".
And you do already have the freedom to seek second opinions. No one is going to stop you, although time and cost prevent obstacles.
If you actually want to live a long time then preventing chronic diseases is far more effective than early detection and treatment. For most people, prevention doesn't require any medical intervention at all.
So there's a couple of things that are encouraging and a couple of things that are worrying in terms of interoperability. The government is putting its thumb on the scale with (meaningful use/promoting interoperability program) so that facilities adopt systems that can interoperate. So now everything is moving/moved to HL7 and theoretically clinics can transfer your records to one another.
You are never going to get your "full" medical record. The law only gives you access to the "Designated Record Set" as defined in § 45 CFR 164.501. There are also shenanigans going on at the big boys where they are giving patients summary visits, instead of the full note.
I'm not sure if anyone will crack the nut of having your health records in one place. I'm highly doubtful of big tech
solving these problems, since they are not revenue producing and take time to get right. They are also dealing in an area that is highly regulated and not their main line of business. If someone does accomplish it, they will be an EMR or your insurance company.
I also doubt that this is the right target to aim for. The value proposition is that you can understand what is going on with you and easily transfer information to your physician. It's unlikely that you will be able to fully understand what is going on in your medical records. Most records need to be interpreted by a physician due to the information asymmetry in medicine. Lab readouts, x-rays, ekgs, are useless in the hands of a layperson. The speed of getting the information could be useful, but it's generally not practical. Most information can wait 3-7 days (and boy does it take that long) before it gets transferred. If you knew you had someone's complete medical record, you may be able to save on testing, or catch an error before it got serious. However, data sources get forgotten along the way, patients are unreliable with their own medical history, and medicine is shockingly unstandardized.
Reading someone else's medical records can be like stepping into a new codebase for the first time. You will see weird and different patterns people do, and you will often question what the other person was thinking.
I would rather focus on efficiency, there's a lot more low hanging fruit there.
(I founded an EMR company and we work on a lot of these problems)
Pretty much every EMR and health insurance insurance company already offers a captive PHR. But they are never complete and aren't portable as the patient moves to other providers or insurers.
Australia has the PCEHR (Personally Controllable E-Health Record), which all health providers are (slowly) being required to upload their results. This includes flags to see who has accessed your records and at what location.
From what I can see, there isn't a huge uptake in patients actively reviewing their data. On the other side of the coin, not everything within scope is being uploaded yet so we can't see if this is the start of a patient-driven revolution.
> It would be nice if everyone had control of their own personal health record but so far no one has really figured out a viable business model.
I think if you couch health outcomes in terms of viable business model, you lost from the get-go. A personal health record is a utility function, it doesn't need a business model to justify it: it's not done for-profit, its done for the social utility.
> This article is my reflection on how we could improve healthcare and why it is important to counter uncertainty with regards to health.
I like the article for the point that it raises (i.e. Diagnosis procedures are performed according to a cost\benefits relationship as outlined by protocols and are prone to error or omission), but I am not sure by what mechanism increased exposure to or control over your own healthcare data will improve the diagnostic process. What are your thoughts?
While I appreciate the idea, the reality is that very few patients will actually actively maintain such a thing on their own. Sharing information among multiple providers is definitely the goal for the past many decades, and there are various approaches to this eg hl7 and ccda, but all are limited by the adoption of standards by existing software. We are just now standardizing on an international standard for labs, procedures, and diagnoses (loinc, snomed, icd 10). But often this is powered by mappings between systems that might lose details. Moving actual medical records and commentary among systems is even more lossy and sometimes hard to detect!
HL7 v2, v3, FHIR, LOINC, ICD, SNOMED, RxNorm, J-Codes, NCPDP, Z-Index/G-Standard... Epic, Cerner, Allscripts, etc., all amount to a curious bag of standards and companies that might not be very interested in commoditizing themselves.
The majority of healthcare interfaces I've seen run on creakingly old--and much-abused by idiots--HL7 v2.3. And I've even seen some signed overpunch codes within the past year. https://en.wikipedia.org/wiki/Signed_overpunch#The_codes