The Virtual Reality of Health Care

Kaiser Permanente digitized everything from medical records to doctor-patient visits. CEO George Halvorson describes the process and how it transcends technology.
  • Transcript


Question: What is the history of the multibillion-dollar effort to institute electronic medical records at Kaiser?

George Halvorson:  Well, Kaiser, as you said earlier, is a vertically integrated system, so we have the largest private medical groups in the world.  We have a hospital system that we own and operate.  We own our own labs and clinics and have all the pieces of the care delivery system.  What we didn't have was an integrated database about all those patients.  So we also, even though we had integrated care delivery, had information silos, didn't have the connectivity, didn't have real-time data.  And we said, if we're going to take this very nice model, which we're very fond of, of vertically integrated care system -- if we're going to optimize this model, if we're really going to perfect it, we need information.  We have to have the data.  We have to have all of the data pieces available in real time to the doctor in the exam room.  So we went down that path, and we said we're going to computerize the entire database.  And we did.

And so right now, doctors in Kaiser Permanente clinics see no paper; totally electronic.  They get all of their information about the patients electronically.  And they have the information in real time, and if they want to do a consult with another Kaiser physician, they can do that electronically.  So you can actually have the patient and the doctor and the specialist hooked up in a virtual model to do the consults because the data flows from site to site seamlessly.  And that makes a big difference in care delivery.  So we took that model and put it up first in Colorado.  We took the data and extracted from the computer system -- once we had all the patient in the record, we created care plans for all the patients with heart conditions and made sure that we delivered on the care plan so that it was a team of primary care doctor, cardiologist, pharmacists, nurses, receptionist -- I mean a full care team relating to the patient, with care plans for each patient that we tracked through the computer, supported through the computer, and reminded people of the next thing to do through the computer.  And we cut the death rate from both major forms of heart disease by 73 percent in two years.

So it's a cut in the death rate just by getting care right.  No new science, no new drugs, no new procedures, no new surgeries; just consistency, absolute consistency of care delivery.  And if a patient didn't fill a prescription -- and right now a doctor writes a prescription, the patient walks out with the prescription, you have no idea whether or not that patient is filling that prescription.  And if they do fill it, you have no idea if they've refilled it.  And so the number of prescriptions that don't get refilled in this country is huge; it's a majority of chronic care prescriptions don't get refilled.  In our model, what happens is, the patient doesn't refill the prescription, it's our pharmacy system.  A bell goes off, the primary care doctor is notified, and the care team goes into gear, and then typically a pharmacist will sit down and talk to the patient and say, that particular drug, you didn't refill it.  Well, actually first the doctor will and then the pharmacist.  But that drug didn't work for you.  Here's two other drugs that do the same thing.  Let's try one of these.

And so we end up with a care team approach, and a follow-up drug, a follow-up drug.  And the consistency -- by creating consistency, creating follow-up and compliance with the care patterns, we've cut the death rate.  And we've done a similar thing in cutting broken bones.  We've created the same kind of consistency, teams of caregivers, and had a reduction in broken bones of 37 percent.  Now, if you think about that, that's a huge reduction.  It makes sense for us because we're prepaid; we get a premium to take care of the patient.  And because we get a premium to take care of the patient, we don't have to bill by the piece for each piece of care.  We're about the total package of care, and if we had been a fee-for-service hospital system and lost 37 percent of our patient admissions for broken bones, you'd have a committee up in arms and alarmed and wondering how do you fill that space.  For us, it's about the patient; it's not about the opportunity to bill for a hospital admission, because we're prepaid for the entire amount.

Question: What are the barriers to achieving health reform?

George Halvorson:  Well, there aren't any providers out there that are deliberately doing bad care.  But if you are not within Kaiser Permanente and you're a primary care doctor, you don't have a relationship with an orthopedic support team.  You don’t have a relationship with a nurse practitioner who's working with the patient.  You don't have the database about the patient to identify that this patient is high-risk of having broken bones.  So they don't have the pieces.  I mean, one of the reasons I say health care will not reform itself in the book is that health care doesn't have the toolkit to actually achieve reform.  So if you get two doctors, three doctors who want to coordinate care for a patient in America right now, it's actually very hard.  There's no mechanism, there's no data flow, there's no interchange.  There's even privacy barriers, the HIPAA sorts of things, that keep the doctors from sharing information with each other.

So when you want to share information and do a better job of delivering care for a patient, instead of facilitating that, we actually create barriers to it.  And we create functional barriers and legal barriers.  And that's not good.  That's not good for the patient because the patients need team care and best care.  But there are other issues like that.  Hospital safety is a huge example.  If you look at the infection rates in hospitals -- in California we looked at some numbers a while ago; 2 percent of patients in California hospitals were ending up with sepsis.  Okay, 23 percent of the deaths were from patients with sepsis.  Now, that's one out of five patients who are dying.  If you do the right job up front, if you identify the people immediately, if you intervene immediately, if you have a team of caregivers ready to jump on each sepsis case and prevent it and roll the condition back, you can cut the sepsis deaths in half.  Now, you think about that as a patient:  if you're going in as a patient, do you want to be in a setting that has twice as many people dying of sepsis, or do you want to be in a setting that has half as many people dying of sepsis?  Hospitals have no financial incentive right now in this country to prevent the sepsis deaths, and the sepsis cases are huge bills.  And so there's a lot of revenue that flows from those cases.  And nobody would ever deliberately go down that path; no hospital in American would ever deliberately create sepsis.

But what we need is an agenda that cuts those deaths in half.  We really need that in this country, and we need to get all the hospitals on board with that agenda, and we need to get all of the major buyers on board with that agenda, because we need to make hospitals safer in this country and have better outcomes.  And in the process we'll end up reducing costs.  There was a study that came out about two weeks ago.  And it was a fascinating study, because they analyzed each of several thousand hospitals, and they analyzed the hospitals based on their process.  And they rated them from one star, which was very weak internal process, to five stars, which was very solid internal process, doing continuous improvement, tracking care and then doing follow-up care.  So they rated them based on process, then they went back and looked at what's the difference in survival rate from a one-star hospital to a five-star hospital?  And it was amazing.  If you went to a five-star hospital instead of a three-star hospital, you were 52 percent less likely to die of anything you were admitted for -- anything you were admitted for.  And if you went to a one-star hospital, you were 79 percent more likely to die than if you went to a five-star hospital.

That's in the United States of America today, ratings that consumers could have in front of them.  We should have that, but also we should have an agenda in this country to have nothing but five-star hospitals.  We spend way too much money -- I mean, we spend twice as much money as anyone in the world on health care.  We should have nothing but five-star hospitals.  And to do that, we have to think systematically.  We have to have the people in the hospitals thinking systematically about best care.  We have to have each hospital creating an agenda that puts in place the systematic thinking and the team training and doing all the right things for the patients, to make sure -- things as simple as that everyone washes their hands while they're moving between patients and treating patients.  There's some basic stuff.

And making sure that there are response teams.  I mean, the number of hospitals in America who don't even train people in Code Blue crash teams is stunning.  I mean, if you think about it, you're in the hospital, you're about to die.  The hospital -- you know, Code Blue -- you know, you watch all the TV shows; you know what it looks like.  And they keep inventing that over and over again on the spot.  If you invent Code Blue on the spot with whoever happens to be there at the time, that's not anywhere near as good a process as if you have drilled and rehearsed and choreographed, and you knew that Code Blue means that you've got to get a doctor in the room now, and you've got to get a nurse who understands crisis in the room right now, and you've got to -- if you do a Code Blue that's based on getting the right people there at the right time and doing the right thing, you should save a lot of lives.  And if you have a Code Blue that's based on "Quick, Code Blue!  Is anybody around?" -- you know, I mean, what's sad is the TV shows about the people kind of inventing this on the fly are more accurate than you as a patient should want them to be.

So the best hospitals and the best systems are going to be doing things systematically.  And if you're in a hospital, ask them how many Code Blue rehearsals they've done.  And do they do simulations?  I mean, we have actually -- at the Kaiser Permanente -- we have mannequins.  We have mannequins that have heart attacks, have strokes, give birth, whatever.  And we actually drill with the mannequins, and so our people can -- instead of drilling on live patients, they can have a stroke crisis and deal with it with a dummy that actually responds.  One little interesting thing -- you might find interesting; I found it fascinating -- was that on the mannequins, a couple years ago they programmed them to stop dying, because caregivers go into caregiving professions because they're caring people.  And some of the people on the training, if they lost a mannequin, would actually have emotional problems.  Some folks actually would take that experience out and feel really bad about it because they're really nice people, and it was a bad thing.  So they now go into kind of a comatose situation, and don't go all the way to demise.  But every hospital in America should be training with mannequins.  Every hospital in America should be making sure the sepsis patients get an immediate response.

Question: The change you speak of seems to transcend technology. How did Kaiser approach doing things differently?

George Halvorson:  Well, one of the things we did was, we started collecting data.  And before we collected data about infection rates in hospitals, we were assuming that all of our hospitals were doing really well because everyone knows the right things to do, and everybody felt we were doing things really well.  And if you asked the people running each of our hospitals to put their hand on a lie detector and take a test on how well they're doing, they all would have passed.  And then we collected the data and discovered that some were doing exceptionally well, and some were not doing well at all.  And then we started tracking and providing internal feedback.  And the rest of the world needs to be doing that as well.  You need a database, and that's part of the issue.

Don Berwick from the Institute for Healthcare Improvement, IHI -- Dr. Berwick has been going around to hospitals all over the country and helping them improve process.  And he did a program a couple years ago called the 1000,000 Lives Campaign.  And he identified a couple of very basic process improvements hospitals could make, and if they made them they would save 100,000 lives in America.  And he's brilliant.  So he went out and did his campaign and succeeded, and saved way more than 100,000 lives.  But the most important thing wasn't just that he got hospitals to start doing process improvements and saved 100,000 lives; the important thing was, the same learning they put in place to save 100,000 lives is now in place and can be used for other hospital improvement processes.  So it created a template; it created the beginning of a new culture, because the old culture was, everything is locally invented; everything is done by each care site on its own.  And your job is to be kind of the Lone Ranger, kind of a cowboy.

I mean, they're seeing issues like -- many hospitals in the country, when nurses change shifts, the nurses going out have to brief the nurses coming in.  That is so non-standardized in most settings that it's often not even standardized from shift to shift.  So the nurse who does the one shift will do it differently at the end of the shift, and does it differently at the end of the shift.  And it varies from floor to floor, department to department.  And so many of the information errors that take place in hospitals -- and a lot of them take place -- take place at the change of shift.  So one of the things that we discovered a couple years ago was, if we standardized that process -- if we create a single process that makes sure all of the right information is there and it's shared in the right way -- we actually at KP took the process time from 47 minutes per shift down to 16 minutes and just about eliminated errors, just by systematizing.  And hospitals all over America -- some hospitals are saying hurray, lovely, wonderful and are doing it.  And others are saying, no, no, no.  No, no, no; we will do it our way.  Stay out of our world and don't preach to us, and we're going to do it our own way.  And as a patient that's scary.  As a patient it's quite frightening to know that you get that kind of inconsistency happening in hospitals, and there's such a variation in terms of safety.

Question: How has the continuously evolving technology affected Kaiser’s outlook?

George Halvorson:  There are several things we're doing that are -- we have a place called the Garfield Center, and the Garfield Center looks very California; it looks like a big film studio.  And we have all kinds of mock hospital beds, homes, clinical offices, the whole -- we sit down and we model in the Garfield Center what the care delivery functionality of the future should look like.  So we've got connectivity -- we were talking about the reaching out.  Five years ago, the ability to reach into someone's home and do home monitoring, home connectivity, home visits was relatively light.  It's a much, much, much more robust technology now, and it's going to be much more robust in the future.  So we've got smart people working with those issues, trying to figure out what's the next generation of connectivity, what's the next generation of data flow?  At the technical level we've got issues about data storage on the back end and those types of things, but those are pretty fundamental to anybody that runs a lot of data.

But on the front end we're trying to stay focused on the patient and the care delivery, and figure out what's the next level of interactive technology, and what should we be doing with patients to influence behavior?  One of our pilots that actually worked really well was just a simple pilot that took a bunch of patients who wanted to lose weight and had a computer send them reminder e-mails two or three times a day to not eat, to not snack, to whatever their particular issue was.  They actually got e-mails from us reminding them.  And the people who were getting those reminders over a few months lost about 15 pounds.  I mean, that's a very kind of low-tech, hi-tech type of thing.  But it's the kind of thing that you don't do up front.  And then you start thinking about, what kind of connectivity can we have, and how can we use this equipment in an interesting way?  And again, we've got some very smart people who are doing that.

The other thing that's really fun is the research, because we can do research that nobody else can do.  So we went back and took a look at patient data that we had from 35 years ago.  And we still have the patients.  Because of who we are we keep patients for a very long time.  So we had 35 years of data, and we identified that if you had high cholesterol in your 40s and you had a certain weight issue, you were 360 percent more likely to have Alzheimer's in your 60s; 360 percent more likely.  We identified which population -- we went back; we could take the current database across a broad population and then sort back into the characteristics of those people 30 years ago and identify several very important things about them.  And one of them was that -- and this is very scary for those of us who carry too much weight and have high cholesterol in our 40s -- the likelihood of Alzheimer's is way, way up, which also might explain why there are so many more cases of Alzheimer's now than there used to be.  Because it's -- I mean, it's those kinds of things.

So we're taking that kind of research and learning things that we couldn't have learned without large amounts of data that's current about patients, longitudinal data.  And now we're adding to that DNA.  So we're collecting DNA samples, and we will have the biggest private DNA database in the world in a very short time.  We'll have about 400,000 people who have voluntarily put their DNA into our database by the end of this next year.  So we'll have 400,000 people at the end of the year.  And we'll have all the DNA, and we'll run profiles, and we will then have the medical profiles, so we'll be able to identify all kinds of things. The discovery that's going to come out of that process I think is -- there will be things that we don't even suspect right now.  I mean, who would have thought that cholesterol and Alzheimer's had a direct relationship?  I mean, the things that we'll learn with the DNA database is really exciting.  So our researchers right now are feeling very good about the kind of things they're going to be doing. 

Recorded on November 18, 2009