I asked provider IT professionals to weigh in on this question, whose genesis was an outsider’s assessment via an HIStalk comment: do we health system IT people automatically dismiss potentially disruptive technologies (such as AI) because we are too entrenched or too well rewarded by the status quo?
The edited responses are below. Thanks to those who offered their excellent insights.
Outsiders have trouble with understanding the speed at which physicians want to move. They are one of few occupations still paid by the piece, not by the hour or a salary. Every second they are waiting for information to populate on a screen, the screen to flip, or the log-in sequence is a lot of money to them. Until you have a database faster than MUMPS, don’t waste my time. For that reason alone, blockchain is a non-starter.
I’ve been involved in minimally transformative ideas and projects that are shot down due to the (over) regulatory environment. A lot of industries are heavily regulated (airlines and nuclear power come to mind), but their regulations are generally around safety. In healthcare, the regulations are around both safety and the prevention of entities over-profiting (Stark). Sometimes these regulations contradict each other and the outsider only understands some of them.
I believe any technology that puts information and decision-making into the hands of the patient has the potential to be disruptive. This will not be a get-rich-quick application, Most people requiring our services are older and even less receptive to change than outsiders would perceive us to be.
They are correct. We actually look to see if incentives are aligned, like everyone else. We rapidly adopted pagers, cell phones, MRIs, and stem cell treatment when we were paid for it or it makes our lives easier. We didn’t rapidly adopt EMRs or other IT solutions when they made our lives harder and cost us money. The issue is not doing something new, it’s doing something reasonable. Make sure incentives are aligned with realistic business models before introducing anything new. One of my favorite quotes: “Incentives matter, whether you think they do or not.”
I think we probably are resistant to AI, but not solely because of incentives. There’s too much vaporware out there and it takes a lot of time to weed through the good and bad, with the good often being no better than the best systems already provide at the added cost of an AI system and/or of nominal value. If they want to blame someone, blame IBM, who taught us that you need to spend a year training your commercial software only to have it continue to provide inaccurate info. To be fair, it is providers that control the data that makes AI training work and our reluctance to share is probably an issue. On the other hand, Google did manage to wrestle millions of records away from the NHS and they still have nothing to show for it.
It’s the pot calling the kettle black. Everyone wants into the healthcare cash cow, but no one wants skin in the game when it comes to actual outcomes, and that includes providers.
Take a look at the technology adoption lifecycle. We’re still in the innovator phase and they’re not yet screaming from the rooftops to get on board. Technology adoption takes 10-20 years, even for consumer products (many of which in recent years had the benefit of being “free”). Why should they be expecting instant results?
We’re dragging our feet on AI when it comes to digital imaging so we don’t tick off providers. We provide a lot of exceptions where it may not work, yet we don’t fire the entire medical community when we have a misdiagnosis rate of 10%.
What disruptors don’t understand is that their solutions are typically unaffordable in the long term for health systems that like to spend more money building buildings than they do to support their existing IT infrastructure. The new shiny object may get some attention and might even get an executive to bite, but at the end of the day, it falls on IT to implement, support, and maintain that disruptive solution over time, all while our budgets shrink due to “cost controls.” The disruptors must demonstrate real-world (not hypothetical) ROI and in reality be at minimum a budget-neutral solution in order for us to take them seriously.
Treating people, while doing no harm, is an art in addition to science. Humans are not machines made to exacting specs that benefit solely from repeatable process. The chance of patient harm or malpractice is real with bleeding edge technology.
Everyone I know on the health IT side is very aware of our limitations and looking for any way we can help out the providers. AI/ML, although promising, so far has limited proven use cases. That, coupled with a very high barrier to entry due to the skills required, means that AI/ML often gets lumped into the “maybe, if we have money left over” part of the budget. Not a lot of healthcare organizations ever get to make it to funding that portion of the budget. Trust me, if you proved your ML model could improve clinical care and/or save lots of money, organizations would adopt it in a heartbeat. If you haven’t proven its value, then why would you expect us to adopt it?
Honestly, it sounds like a comment from someone who runs an ML-centric company and can’t find a partner to provide the training for their model. That’s a risk and investment for the healthcare organization, and typically the vendor gets most of the benefit if it works even though they tapped the provider’s knowledge and training to make the product. If you really want us to do that, show up with a fully-funded project, including our expenses, and we’ll consider it if you give us partial ownership of the successful project. At this stage of the game, that’s the only deal that makes sense.
You can kill people with the wrong tech, bad tech, or badly-implemented tech. As a clinician who supported clinical decision support, it is easy to talk it, but harder to prevent the medical misadventures that may happen to said Heath IT Outsider’s child.
Speaking as a provider who works in the vendor space, we prefer to wait and see what works in other industries before taking a risk and sinking big development dollars into expensive new solutions. Exhibit A: cloud computing, which went mainstream with Amazon Web Services in 2006, but only in the last few years have we seen this model take off in our industry with web-hosted EHRs. That’s why we’re always 15 years behind. None of the established players wants to spend $500m to develop a buzzword concept (remember “big data”?) that will fold or go out of fashion next year.
Next time you are sick, open your AI program get a diagnosis, prescription, and any blood tests. There is a place in healthcare for AI, but it is not replacing trained medical professionals
I’m guessing that comment came from a former Elizabeth Holmes devotee. Health IT outsiders have a long history of declaring the US health system stupid, launching a startup, then quietly giving up a year later. If our outsider had any real ideas, they’d have products in the marketplace making money. Optum, Health Catalyst, Arcadia, and many more aren’t waiting around for provider permission. They are innovating, pushing the quality-cost envelope, and growing. I don’t know if AI will truly move the needle positively in healthcare any time soon, but I’d have to hear a great conspiracy theory to believe provider IT people are protecting their EHR vendor from AI, open APIs, or any other technologies that would make the customers happier and their jobs more fulfilling.
As a health system CIO, Individuals who are flabbergasted by the risk-averse nature of the healthcare industry as a whole do not fully understand nor appreciate the current healthcare system business model. It has a customer (patient) market that is shrinking. It is becoming more segmented, with alternative specialized scope limited services. The net revenue opportunity per patient is shrinking as operating costs (especially labor and regulatory related) continue to increase.
Entrepreneurs by their very nature take financial risks if they see an opportunity for a high financial return when no one else does. There is a ton of cash flow within the healthcare industry, but no new opportunities for significant cash infusion The customers (patients) do not have any opportunity to shift their spending from one source to another. The industry players are protecting their revenue stream as best they can. Most healthcare providers and hospitals do not have an entrepreneurial spirit, nor do they have the financial reserves to take on the financial risks.
It is also important to note that the financial industry and venture capitalists do not invest in healthcare providers nor hospitals. The risk is just too high for no foreseeable reward. Thus, it is not surprising at all to me that “health IT outsiders” looking to be disruptors are disappointed when they are not embraced with open arms. I predict that someday there will be disruptors who will change the business model itself with a better SYSTEM of mousetraps rather than just one highly effective mousetrap.
I don’t see provider IT people as being entrenched or particularly well rewarded. Rather, we insiders are pragmatic. Too often we’ve been sucked in by the breathless exuberance of the purveyor of the next big thing that will revolutionize healthcare, only to realize that it’s not nearly what it’s cracked up to be. Or worse, we take the blame for it not turning out to be what it was purported to be.
Technology is evolutionary, not revolutionary. Incremental advances by potentially disruptive technologies – once field tested – make their way into the mainstream. Let’s not forget that a mere 15 years ago, EHRs seemed revolutionary.
Look at FHIR. The bright shiny object du jour which will solve all problems in the delivery of healthcare. Will this technology magically address every issue? Absolutely not. Or will it even address any of the issues better than some long-existing technology? I’m on the fence. Is FHIR really even disruptive? Nope. Interfaces have been around since there was more than one computer. But by being a pragmatist and viewing FHIR as an incremental improvement, I get painted as a curmudgeon.
I think there is an extreme sense of being jaded from a long list of previous failures. People often don’t understand the complexities of healthcare, the countless variations, the messy data, the fickle users mixed with the extreme regulations of privacy and billing. Add all of that to hospital bureaucracy, understaffed IT departments, and low-salaried (and therefore often mediocre) IT staff and you have more sub-optimal systems than you can count.
Healthcare doesn’t operate financially as other industries. I’ve spent the majority of my career in community hospitals and it is difficult for them to sink money into disruptive technologies when you’re payer mix is 40-60% government. We would love to invest in disruptive technologies, but when replacing an EMR originally installed in the mid-90’s causes a financial burden, what’s a girl to do?
What’s the evidence of benefit to (a) patients and their caregivers first; (b) physicians, nurses, and other bedside technical caregivers second; and (c) then everyone else? As a 40+ year emergency physician and 20+ year medical informaticist, let’s see the evidence that AI and other disruptive technologies deal with the chaos of patient variability and sensitivity to initial conditions better than the competent, compassionate physician.
This is healthcare. Ultimately, people’s live are literally on the line. There is no room for alpha or even beta level products in a production environment. If AI can do my job and help save lives, so be it. But that is not now and it is not anytime soon.
The workplace dynamics of provider-based healthcare are different than any other industry. Who is the customer? Is it the patient, doctor, nurse, CFO, payer, government, or someone else? Or all of the above? Outsiders have not not been able to solve that riddle yet, although things may be changing with consumerism on the rise.
Also, in my long experience as a CIO (25+ years), it is rarely the CIO who calls the shots. Hospital CEOs are notoriously risk averse with a huge herd mentality when it comes to IT. The history of the industry is littered with multiple failures of so-called IT solutions. In addition, CFOs control the purse strings, and if they do not control IT, are out to hamstring it.
I have seen several outside CIOs try to “fix healthcare” and they have all failed to recognize the unique cultural characteristics.
I’m not worried about protecting my paycheck. There are always positions available in my particular medical specialty and my current income isn’t that great anyway. What I am worried about are the costs and potential negative consequences of inadequately designed and tested “disruptive technology.”
Healthcare technology is not like trying a bunch of free or cheap apps on your personal IPhone to see if any generate major or disruptive improvements. Instead, with healthcare technology, there are significant upfront costs (often with no guarantees of benefit or acceptance), significant personnel costs for installation and training, significant changes to workflow, and potential for unintended consequences, including inefficiencies, lost revenue, and actual harm to patients if it doesn’t work correctly. Indeed, I’d be concerned that anyone who jumps on the bandwagon too quickly is impulsive and reckless.
Add to that all of the half-baked snow jobs that we’ve been sold over the years and it’s no wonder that HIT providers (and users) are cautious and skeptical.
I welcome the challenge, but I am more often than not faced with those who do not want to accept that they don’t know the extent of what they DO NOT KNOW about the unique specifications of the industry. If only the industry was established in the status quo. Most who make such proclamations dismiss the history to the why and how we are where we are. Case in point — the Jim Cramer declaration and folks like Chrissy Farr who just pass along without doing the basic journalistic research on Epic. If they came about it with some sort of due diligence, it might be a different story.
Reader Comments