Monday, July 8, 2024

Healthcare Needs a 21st Century Infrastructure

Matthew Holt is going to tell me I’ve been thinking about infrastructure too much lately (e.g., cybersecurity of them, backup plans for them), but if you don’t have infrastructure right, you don’t have anything right.

Healthcare infrastructure can be like that. Credit: Bing Image Creator

And healthcare most definitely does not have its infrastructure right.

We’re spending between 15-30% of our healthcare dollar on administration, and no one views our healthcare system as efficient or even particularly effective. We have numerous intermediaries like PBMs, billing services, revenue cycle management vendors, and all sorts of digital health solutions. There are layers upon layers upon layers, each adding its costs and complications.

In some ways, healthcare’s infrastructure has changed remarkably in the last two to three decades. Most transactions – e.g., claims or eligibility – are sent, and often processed, electronically. Most physicians, hospitals, and other health care clinicians/organizations have electronic health records. You can find out the expected cost for prescription drugs at point-of-sale. You can do a virtual visit with your doctor. There are vast amounts of health information available online. AI is coming to health care, and, in some cases, is already here.

But: we’re still sending faxes. We’re still filling out paper forms, repeatedly. We still make innumerable phone calls, usually spending long waits in queue. Everyone hates provider directories, which are never up-to-date and often inaccurate. Talk of interoperability notwithstanding, there are far too many data silos, leading to at best us lugging around disks with our downloaded records to at worst physicians acting with incomplete information for us. Healthcare has had far too many data breaches, and cyberattacks have held patient data hostage (e.g., Ascension) or put a halt to those electronic transaction (e.g., Change Healthcare). And we’re not at all sure how to govern AI.

The amount of medical literature has been growing exponentially for decades, and the volume of health care data is growing much, much faster. Physicians once guarded health information like the guild they are, but the Internet has democratized health information – while doing the same for misinformation. If anything, we have too much information; we just can’t use it as effectively as we should (e.g., it can take 17 years for evidence to change physician practice).

This is not an infrastructure that is not coping well with the 21st century.

I recently read Deb Chachra’s How Infrastructure Works, and among its many insights I was struck by her calls for reshaping infrastructure for the future. Infrastructure, she argues, needs to build in more robustness, redundancy and resiliency, through more diversity and decentralization. She predicts that infrastructure systems of the 21st century will be more diffuse, diverse, and distributed.

That’s what healthcare needs to be preparing for – now.   

I’ll suggest a few basic building blocks for healthcare’s 21st century infrastructure:

Information: we’re drowning in health care information; some of it good, some of it bad, some of it groundbreaking, some of it conflicting. We need mechanisms that help curate and make available that information, so both patients and clinicians can easily discover what the latest, most credible information is. No more doing Google searches and hoping for the best. No more waiting weeks, months, years for the best information.

Such information should have a layered presentation, with a high level one accessible to all and more detailed ones for those with more clinical knowledge.  In line with Professor Chachra’s thinking, this should not be a single database/website; it needs to be robust, redundant, and resilient, so that it can survive cyberattacks and not become polluted with false information.

Data: The data apocalypse is upon us. We’re all generating massive amounts of health data, both from “traditional” sources as well as from newer ones. Yet we have limited ability to understand anyone’s health history, much less detect a community outbreak in real time. We can’t tell which clinicians are treating who for what, or how well they’re doing that. We can’t even link a person to all the sources of data that exist about them.

We need unique person identifiers. We need unique health care practitioner identifiers. We need to have consolidated views of patients and clinicians, and the ability to analyze at an individual, health system, community, regional, and national levels. Again, not a single database, but data that can be queried and analyzed in real time, while remaining – you guessed it -- robust, redundant, and resilient. And strong privacy protections must be baked in at every level.

Insurers: I worked in the health insurance industry for many years, and I still believe it has a critical role. But it’s insane that each company has its own interfaces, its own credentialling, its own provider networks, its own medical necessity and preauthorization requirements, its own set of exclusions and limitations. ACA standardized some elements of health plans, as did HIPAA before that (certain transactions). But the hundreds of insurers/third party administrator/self-funded plans are a drag on the system that cannot be tolerated.

I’m not calling them to be abolished, but if they can’t agree on common infrastructures that alleviate pain points for everyone else, they will be.  

Artificial intelligence: AI is the technology of the 21st century. It is going to impact every job, every industry, every business interaction. Healthcare will not be an exception.   

Right now there are many independent efforts to develop and use A.I. in health care. Right now we’re still exploring point solutions for A.I. in health care. Right now there are no overarching rules for how to train or deploy A.I. in health care. Right now there is no common vision about how A.I. can or should transform health care.

All that must change. We need to thoughtfully build A.I. into healthcare’s infrastructure, making them more effective, efficient – and robust, redundant, and resilient.  

AI will be part of healthcare's infrastructure. Credit: Bing Image Creator

I’m sure there are several more components I’m missing, but let’s start the discussion with those.

-------------

We’ve seen healthcare systems be overwhelmed by climate change, in the form of hurricanes or power outages. We’ve seen healthcare organizations brought to a halt by cyberattacks. We’ve all gotten caught up in healthcare red tape.  We know our healthcare system is way too expensive while being way too ineffective.

These are failures of infrastructure. These are failures of imagination. Changing infrastructure is a long, expensive, and complex effort, but it’s like the old proverb about planting a tree: the best time is twenty years ago. The second best time is now.

It’s 2024, and if we don’t start reworking our healthcare infrastructure for the 21st century now, it will soon be the 22nd century.

No comments:

Post a Comment