Years ago, Peter Tippett, MD, Ph.D., then a presidential advisor to George W. Bush on health IT, was on the President’s Information Technology Advisory Committee (PITAC), a group whose strategic plans resulted in billions of dollars being eventually spent on bringing healthcare into the digital age. Indeed, Dr. Tippett readily admits he was a “cheerleader” behind the transformation towards electronic health records (EHRs), though he says even at the time he had concerns that putting all structured in data in one place would leave out the direct messaging element already happening between care team members.
Now, Tippett, chairman of Florham Park, N.J.-based electronic delivery services and solutions company DataMotion, CEO of HealthCelerate, and just as of last year, the former chief medical officer and vice president of industry solutions at Verizon, is continuing to push the idea of having both structured and unstructured data in healthcare. Dr. Tippett, creator of the first commercial anti-virus product, later known as Norton AntiVirus, and well-known authority on information security, medicine and healthcare technology, recently spoke to Healthcare Informatics about his views on the industry. Below are excerpts of that interview.
In your view, how does the health IT world sit today compared to years ago?
The health IT world is in a shamble, and the clinicians of the world are nearly in revolt. Maybe that’s a bit extreme, but if you took a survey of doctors and how they like their EHRs [electronic health records], the majority of them would say they don’t like them and they don’t serve their needs. But if you ask people how they like their email, of course they’d say that it is fine. The same holds true for their word processors, computers and phones. There just is no question about email being a productivity tool for the rest of our lives—it’s so beyond obvious. But, we’re still here 20 years into it trying to figure out if EHRs have productivity benefits for clinicians. Studies say they don’t except in narrow areas like e-prescribing. For the majority of doctors, they don’t’ make their days better. They don’t make patients’ days better either. So whose day do they make better?
I was on the PITAC which led to meaningful use, with David Brailer, M.D., Ph.D., the first National Coordinator for Health IT, 14 years ago. We had subpoena power, the ability to figure out how to bend the cost curve, and we asked what can we do with IT? The basic output of that, and I am oversimplifying it, was that if we could get the IT usage in healthcare that we have in other industries, three things would happen: people would get healthier and live longer, costs would go down, and we would have an entirely new idea of science where you can just look answers up. Here we are, 14 years later, and HIEs [health information exchanges] still don’t work. People can’t share information in a meaningful way. The majority of doctors sharing information with other doctors is done with paper or fax.
Peter Tippett, MD, Ph.D.
How can some of these issues be fixed?
I was on PITAC, so I am one of the culprits. We have let perfection get in the way of good enough. We haven’t allowed for big, solid, iterative changes to be made. Instead, we have said that if everything was structured, in one place and available to everyone whenever they needed it, wow that would be great. So we have painted this picture of centralized data storage stuff, be it an HIE or EHR, in which all records will be structured for every patient, and if someone needs it they can access it. But does your boss dip into your hard drive when he needs something from you? We have set up this structure that doesn’t work anywhere else on the entire planet. When we do fetch data repository-oriented methods, we do it for structured, highly defined areas. For other things, people put together the answer and send it to you. It’s a mixture of push and pull.
Can you explain this push/pull belief a little further?
We have taken this notion of fetch, and are overwhelmed by the notion of push, but both are required in the world of computing. Back in the day of PITAC, when at the time, there was $10 million a year of medical records being created by dictation, if only we started by making it easy for those to be sent to the other person, make it simple enough, store them, search them and find a way to send them. If you could just send the stuff you already have in whatever format it’s in, and make it available, you’d be way better off than we are today. This doesn’t have to be as hard as we have made it. There is a fundamental question in notion of push/pull, and we have decided that a single repository made bigger and bigger, giving the right people the access to those repositories is a great idea, but that’s a 20- year vision. We also need to include the ability to send stuff.
Separately, we have decided that structured data is required for everything, but that’s not the way humans work. If someone sends you a CCD [continuity of care document] extract from an EHR, the doctor hates it, since it’s useless. You have to go searching through the whole thing for one little nugget, which might be a photograph of a note that summarizes what’s going on. Doctors call this the bullet—tell me what we are up to, what’s the problem we are trying to solve? This stems from the idea that data scientists tend to think about things in which you can put a field around it, with structure. But most of clinical medicine is organized around a nuance, meaning my situation is different than someone else’s even with the same four diagnoses taking the same four medicines. My response might be different.
That is subtlety, which doesn’t exist in pick list medicine. I think the world will change from doctor-patient relationship to brand-patient relationship, just like the travel agent has been replaced with Expedia. Most of what we care about as patients are questions we can find the answer to. The majority of these questions could be answered by a machine. We do need people involved of course, but for much of interaction, there is value you can put into a brand relationship with patients. The subtlety problem makes it so that the push on structure was premature. With transitions of care and care coordination, it’s all about the subtlety.
You’re certainly an authority on IT security. What can be done to slow down the disturbing trend that we’re currently seeing in healthcare?
My mother has a saying I love: “We have been putting the emphasis on the wrong syllable.” Two-thirds or more of what we spend our money on for “security” has zero or minimal benefit to anyone. The things that have had strong benefit have not had strong traction, even those things might not cost a lot. Most of my data is coming from the Data Breach Investigation Report (DBIR) at Verizon. What actually goes wrong is this notion of identity. If you could know who it was that’s logging in, well over 85 percent of all cases of computer crime in all industries, boils down to the fact that passwords aren’t enough. There is no difference between strong passwords and weak ones. The bad guy will put something on your system to capture your password; it doesn’t matter if it’s 40 characters, all data types, or random, they will get the whole thing. Two-factor authentication would stop the vast majority of computer crime. And the same holds true for ransomware—it wouldn’t happen. We put energy into intrusion detection, yet for the vast majority of cases, less than 2 to 3 percent of crime wouldn’t be affected by that. It’s not even marginally valuable. We are building meter protectors for our automobiles instead of brakes.
What do you see unfolding in the future?
There’s a huge interest in people who have a personal story in health. Every time you talk to anyone in about some problem in healthcare, everyone has a story about how long it took for something to get done to coordinate care. At the same time the utilization rate for patient portals is less than 10 percent in the best organizations, and less than 5 percent in most. Most people’s data isn’t there since they don’t go to a single hospital or have outpatient EHR data that’s connected to the hospital. There are some exceptions, but for the majority of people it doesn’t work that way.
I have been an advocate for getting push to complement all the pull, and in this sense I think DataMotion comes in as it can take any data from any format so that the people who need it can work with it. I would like to see permission, if not emphasis, for unstructured data. Take a photo of your vaccination card and send it. Take the handwritten note and send it. There was a study a few years ago in which someone did alerting off the Department of Veteran Affairs (VA) dictation files, which are digital but unstructured. They looked through those files, parsed them, and created alerts back to doctors and medical staff, and did the same with the EHR data which was structured. The study showed they got better, timelier, and more appropriate alerts out of the unstructured data compared with the structured. Accept that humans are just fine with poorly structured data. Maybe machines aren’t, but let humans get their jobs done.