>> Desney Tan: So it's my pleasure to introduce Liviu Klein who is a cardiologist down
at UCSF in the Heart and Vascular Department. He is also director of the Mechanical
Circulatory Support and Heart Failure Device Program and specializes in heart failure as
well as arrhythmias in his clinical practice. In his research life, the other half of his life,
Liviu has done a bunch of work looking at epidemiology especially for heart failure in
women. I did a little bit of digging online and found two interesting nuggets: one, he was
born in Romania a couple of minutes away from Dracula's castle -- That's kind of an
interesting nugget -- but two, that his parents were engineers which probably accounts
for the engineering bit that he takes in much of his work, working with sensors and
devices for remote monitoring which he's going to tell us about today. So without further
adieu, take it away.
>> Liviu Klein: Thanks so much, Desney. Yeah, so interesting nuggets. Both my parents
are engineers. And in Romania you go from high school to graduate school; they don't
have college. And when I was trying to choose a high school, it was 1988 so computers
were sort of just coming out at least in Romania and my parents wanted to me go into
computer programming and that sort of engineering. So we went to kind of visit high
schools and we always stopped at the computer high school on the way home.
"Wouldn't it be nice if you actually go here?" So I ended up going there at the IT high
school in Brasov, actually. I think 50 percent of my class actually works at Microsoft, and
they were working here when I was in med school and they're laughing at me. But I
always sort of had this combined interest. And I used to program decades ago in C++.
That was many, many years ago.
But, anyway, thank you for the invitation. Hopefully I will try to introduce you to remote
monitoring which is something that you guys think about every day and then sort of try to
emphasize the other side of the story, my clinical view on this concept and why we really
need this in healthcare. So as you probably know healthcare is one of the most
expensive industries in the world, and US tops that with over 8,000 dollars per year per
patient. And if you look at the civilized world, they're around 3,000 dollars and Canada is
about 4,000 dollars. So we really spend a lot more money that any other civilized
country, especially if you think about Medicare. Obviously everybody age 65 and over
are covered by Medicare. It’s a huge in the Medicare cost in the last 30-40 years or so
and it continues to rise even further with really no end in sight. It has plateaued for the
last two or three years, probably because the economy is so bad nobody can spend any
more money but not a real decrease.
And it's also actually very interesting, we always hear about the one percent, this is the
one percent that nobody really talks about. So twenty percent of Medicare costs are
really spent by the one percent of the population, and then half of the Medicare costs are
really spent by the top 5 percent of the population. So a lot of chronic disease in this
country and a lot of money being thrown away at it. And I'll make an argument, actually,
it's not all for providing great treatment; it's a lot for doing things that we shouldn't maybe
be doing. So how do we actually spend your money? Well, if you look at the National
Health Expenditure in 2010, 2.6 trillion dollars. If people don't know what that means, it's
actually Apple, Microsoft, Google, Berkshire Industries, J and J, Proctor and Gamble and
Chevron; this is what we spend every year. Okay? So imagine this piece of the pie which
is hospital care, this is Apple, this is Google, this is Microsoft; in one year, boof. It's gone.
And then, we start again next year.
So if you really look at what we spend on, half of the money is spent for monitoring and
testing and some treatment but mostly it's monitoring and testing. And you think if we
spend so much money, you want to actually maybe fix the problem and spend
something on prevention. Sure we do. 3 percent of our healthcare costs every year are
spent on prevention, so a very, very small amount of money which actually leads to what
the problem is. So what did you get for you money? Not much really. So if you look at
survival across the world, here's where Japan and South Korea are, about 80 years of
age. Western Europe and then US, 78 years of age; that's our life expectancy at birth
which is actually close to Cuba which is not on this graph. But it's actually close to Cuba.
The difference is we spend 8,000 per year; Cuba spends about 400 dollars per year so
not a lot of profit for our investment.
And what actually leads these healthcare costs? Well it turns out cardiovascular disease.
So if you look at the top 22 diagnoses that account for the majority of healthcare costs,
one is heart problems which are mostly myocardial infarction and heart failure,
hypertension, other circulatory conditions or peripheral vascular disease and stroke. So
these actually are the top – if you combine them together as cardiovascular disease, this
is the top reason for spending money in healthcare in this country. And if you actually
distribute this by age, you can see the fast-arising segment is the 65 to 80 years of age.
So people who have just retired and are healthy, otherwise, enough to live longer lives
but they spend a lot of money. And if you break it down, where all this money goes: the
red bar is hospitalizations. So a large amount of money is spent in the hospital.
And again if you think about where people live, they live at home. So we should really be
spending this money at home where they are trying to prevent them from getting to the
hospital. Yet the paradigm is actually the other way around. And I think the main reasons
for what I call the Broken Care is what happens is the patient usually has some
symptoms but they don't recognize the symptoms early enough. At some point they get
worsening of their symptoms. They go to see a physician in the office. The average
office visit is about 13 minutes, right, so that's less than 15 minutes, so limited time.
There's limited staff to get all these symptoms. Limited diagnostics. If you've ever been
to a physician's office, there's really not much there, a couple of blood pressure cuffs
and a scale maybe. So not a lot of tools to diagnose anything. There's really no
monitoring of any kind and then, there's limited intervention. You really can't do anything
in the office, right, just write the prescription. And then, there's really no education about
diseases and management.
>>: Have there been studies that have shown that if you were to recognize symptoms
earlier and more correctly that in fact the cost would change dramatically?
>> Liviu Klein: Yeah, so actually I will show...
>>: [Inaudible] respond to what their doctor [inaudible].
>> Liviu Klein: Right. So I'll actually show some data in heart failure which is a subset of
the cardiovascular problem that just deals exactly with that. And then, the symptoms
worsen so what do people do is they go to the emergency room. And then, the ER
physicians who have no relationship with the patient, they've never seen the patient and
legally they have to send the patient to the hospital. So the patient ends up in the
hospital and of course the first person that you meet is a case manager that tells you
that you have to discharge the patient. And then, you sign papers that as long as you're
physical able to go then you have to go. So, again pressure to decrease the length of
stay. There is not enough – You know, if you try a new therapy or a new diagnostic, you
don’t have enough time to see if it works. And then, you send the patient home. A bunch
of people come back so get re-admitted and the cycle kind of perpetuates over and over
again. So I think this is just reactive care, okay; we react to something.
So we really have to change that to be proactive. And I'll use heart failure which is my
specialty to try to emphasize this point. What is heart failure? It's a very complex
syndrome that can result from any structural or functional impairment in the cardiac
function that prevents the heart for either accepting the blood or ejecting the blood. And
the cardinal symptoms that people have are shortness of breath and fatigue, which limits
exercise tolerance, or fluid build up either in the lungs leading to lung congestion and
people becoming more short of breath. Or fluid builds up in the legs or the belly and
these features really impair the functional capacity of the patients and make them
symptomatic.
But the main physiological abnormality, if you had one way to define this from a
physiological standpoint, what happens in heart failure we have actually an increase in
the intracardiac pressures. In an ideal world if we all have this pressure sensing in the
heart that can determine the pressures 24-7 then we know how to manage these
patients. Obviously, we don't have that. So what do the patient's look like? Kind of like
this. So they're really short of breath and they're puffed and then, you can see the blood
fluid builds up in the lungs and the patients get short of breath. The pressures in the
heart increase. There is heart muscle weakening and the heart muscles become stiff.
This is, again, another picture of a patient. So you have the swelling in the belly or in the
legs. They're coughing all the time because they have fluid in the lungs. Hopefully not
you. And then, they are tired and short of breath.
So this is what it looks like. And heart failure, again, it's sort of a [inaudible] basket of
everything: if you have coronary artery, heart attacks, if you have high blood pressure,
diabetes, atrial fibrillation, valvular disease or cardiomyopathic factors so people who
have cancer and had chemotherapy at some point in their life that can affect the heart
muscle. They all lead to some sort of injury in the heart muscle itself, and eventually
some of the patients actually have a decrease pumping function which we call
decreased ejection fraction with systolic heart failure. And a lot of patients will die, some
suddenly from deeper arrhythmia; some will just die from progressive heart failure.
Well, how common is it? It's quite common. So if you look at all adults in the US, about 4
percent of the adult population has heart failure. But it's a really a disease of the elderly.
Actually, 1 in 10 people over age 75 has heart failure. It's a very, very common disease.
And again it kind of brings the Medicare point into perspective: a lot of these people are
actually on Medicare.
>>: Can I ask something?
>> Liviu Klein: Yes.
>>: There is constant trend with men being more than women, and 75 plus goes
inverse. Is there a reason for that?
>> Liviu Klein: Right. Yeah, so actually it's a very good reason. It turns out that in the
younger ages men have more coronary disease or heart attacks leading to heart failure.
In the older ages women have more high blood pressure leading to heart failure. And it's
just manifesting over age 70, actually.
>>: Is there an exact definition – I'm actually surprised in some ways that the number is
as low as 10 percent. My intuition is that some people would have a little bit of heart
failure and a little bit of [inaudible]. Is there a precise definition in terms of pressure or
ejection fraction? Or is there a simple definition that you can give now or is it very...
>> Liviu Klein: Right. Again, the simple definition is the pressure. So if you have a
catheter and you measure the pressure in the heart, normal pressure should be below
14 millimeters of mercury. So anything above 15 is abnormal. So these patients will have
heart failure. So I think all in all it's about six million people but there are a bunch of
people, actually, who don't know they have heart failure but they have what we call
premorbid conditions. And that's another ten million people but they don't have the
symptoms yet.
>>: But that is sort of the – if you had to boil it down to one definition, it's there is a...
>> Liviu Klein: The pressure.
>>: ...threshold for intracardiac pressure that you...
>> Liviu Klein: Correct.
>>: ...define as heart failure.
>> Liviu Klein: Correct. And that's the six million people. And then, the ejection fraction:
again, about half of the people have a low ejection fraction, half have normal ejection
fraction. But actually it doesn't really matter because if you look at the prognostic value
of that survival-wise it's abysmal. Half of the people actually would be dead within in five
years. And this data comes from Rochester, Minnesota, from the Mayo Clinic and they
looked at trends over time. And there's really not a lot of change despite all these good
medical therapies that we have and electrical devices. The prognosis has changed a
little bit but not too much, so still about half of the people will be dead within five years.
So a very morbid condition.
And not only do people die but actually come to the hospital quite often. So we actually
have over a million hospitalizations every year just for heart failure alone. And this data
shows from the American Heart Association statistically – It was published a couple of
months ago – if you look at data from 10 years and 2010, there's no change. All we've
done is kind of shifted this quarters as people got older but that's about it. There's still
over a million hospitalizations for heart failure. Again, despite all these technologies we
have, we have not been able to alter that. So heart failure is really the only
cardiovascular condition for which the numbers are going up actually as opposed to
heart attacks or strokes which are going down. And if you think about cost, again, it's
very, very expensive, probably about 30 billion dollars in 2012. And again, if you look at
the majority of that it's just room and board really. It's just hospital care: twenty-some
billion dollars.
And we know that the cost per hospitalization is actually increasing, so the longer the
stay the more money you spend. So if you end up – The average is about seven days in
the hospital: about 11,000 dollars. If you spend more than two weeks in the hospital then
you end up costing about 40,000 dollars. And most hospitals on average get reimbursed
about 20-some thousand dollars so actually most hospitals lose money on heart failure
admissions. So not only is it a very morbid condition but actually what's even more
striking is that a lot of patients that are discharged from the hospital actually come back
within a month. So this is actually in an article from the Wall Street Journal a few years
ago where I quoting the study which was done on the Medicare population. So they
basically looked at what are the top ten conditions for which people come back, get
readmitted in the hospital? And the top was actually heart failure. More than one in four
people being discharged from the hospital for heart failure come back within 30 days
with heart failure. So we're not doing a good job at keeping the patients – treating the
patients and keeping them out of the hospital.
More data about that: again, looking at Medicare data from the last five years, we know
that patients come back within the first 30 days. And Harlan Krumholz' group from Yale
tried to identify, is there any time period that you can say, "Okay, if you can monitor the
patients for the first five days, we'll prevent most of the readmissions?" And it turns out
there is really not a whole lot. So if you look at a display over 30 days, it's a pretty
constant number of patients coming back every single day. Maybe in the first couple of
weeks you get about 60 percent of the admissions but really it's across the first 30 days.
So really the so-called vulnerable period, you can't really come up with any definition
because it's a constant flux of patients coming back to the hospital.
>>: But the reason they come back is because the same symptoms...
>> Liviu Klein: Exactly.
>>: ...from the first [inaudible]...
>> Liviu Klein: Exactly. Exactly. Same symptoms, same heart exacerbation. And
actually what's even more interesting: they looked at the first three conditions that
patient's were admitted into the hospital for. The three most common conditions that
Medicare patients come to the hospital for are heart failure, heart attack and pneumonia.
So if you look at these conditions: patients go back home and 40 percent of the patients
that have heart failure come back with heart failure. But look at heart attacks: 20 percent
of the patients that came in initially with a heart attack actually come back in a month
with heart failure. And the patients with pneumonia: 10 percent of the patients that have
pneumonia come back within a month with heart failure. So really a lot of these patients
come back with heart failure and the question is why? Well, let me show you the
standard of care for heart failure monitoring.
It's a blood pressure cuff and a scale. That's all we have in 2014. Patients go home. We
tell them, "Weigh yourself every day. If your weight goes up more than two pounds, call
us." Okay? That's the standard of care that we have. So no wonder that the Broken
Care, it's still there because it's reactive care.
>>: Do patients do that? Do they...
>> Liviu Klein: They do.
>>: They weigh themselves and they call you?
>> Liviu Klein: Most of the patients they do. The problem is, the weight is so unreliable.
But, again, that's what we have. So a lot of hospitals now -- Because Medicare became
smart and said, "There's no more money so let's spread the wealth." So now Medicare,
what it does is it says, "Well, if you have readmissions for heart failure, more than the
national average, we'll actually penalize you. So in 2013 you had to pay a 1 percent
penalty on all your Medicare diagnoses. In 2014, it's 2 percent. Next year it's 3 percent."
Imagine a hospital that has a very active orthopedic program. What do old people need?
Hips and knees, right, which are very lucrative. So imagine now that if your heart failure
patients come back. Let's say you have ten heart failure patients a year and 5 come
back within 30 days; that's more than the national average. Suddenly, you get the 3
percent pay cut from all your orthopedic business. So of course the hospital is starting to
pay attention when it's big money. So they try to implement team approaches to prevent
heart failure.
Actually UCSF is no different. So we went from this reactive care to active care. What
does it mean? Well, we've looked at our program and said, "Okay, what are the
elements that we need to prevent heart failure patients from coming to the hospital?" We
need to identify why they're coming. We need to teach them about their medications,
about diet and try to coordinate appointments and things like that. So, establish a really
good team of nurses so that when the patient gets admitted to the hospital, they come,
they assess the patient, they do a so-called Teach Back where they actually tell the
patient, "Okay, can you tell us what we've told you," so trying to assess how much they
understood. And you discover that the patient didn't really get what they told them so go
back and teach them again.
And we have to set up an appointment. So within seven days of discharge -- I just
showed you that the readmission time is pretty constant in the first 30 days but it's
probably concentrated the most over the first couple of weeks. So having an
appointment to see a clinician in the office in the first week, we think it prevents
readmissions.
>>: What happens at that followup appointment?
>> Liviu Klein: You just sort of assess the patient: make sure that they have no reaccumulated fluid, make sure that the medications are in order. So not much.
>>: What do you do then to measure fluid at that followup appointment?
>> Liviu Klein: Just physical exam.
>>: Physicals.
>> Liviu Klein: Right.
>>: There is no other medical examination, like MR, get some flow on the [inaudible]?
Nothing?
>> Liviu Klein: Nothing. Just really blood pressure, weight and physical examination.
And then, again we have a lot of referrals for dietician, physical therapy, home health.
And home health, again, is a nurse goes into the patient's home within 48 hours of
discharge to make sure that all the [inaudible] medications are in order, throws away the
old ones, gets the new ones, throws the pizza out of the fridge. You know, that kind of
stuff. So we did that. Our nurses got a grant from the Betty Moore Foundation to do that.
We established the color coding system as we talked about earlier: green, yellow and
red. And then, whenever something happens then you have a plan to do something, call
a physician or the physician knows about it and changes medications.
So when we started our readmission was 22.5 percent for heart failure within 30 days.
That was our, sort of, target. And we wanted to get it down 16 percent. So this shows
our progress in 2009, '10 and '11. And actually in 2013 we're down to about 11 percent.
So we've done really well, decreased the heart failure readmissions, decreased for all
causes of readmissions and 90-day readmissions. We increased the number of people
who were actually referred to home health and increased the number of patients that
had a followup appointment. So everything was great and, again, this is data going to
2012. So good heart failure readmissions data: 30 days, 90 days. And I think everybody
was happy. So we actually get reimbursed about 24,000 dollars per case, so with our
decreasing number of readmissions compared to the national average which is about 25
percent we probably saved about a million dollars to Medicare. It's not a lot but it was
great.
But what are the costs of that? Well, have you ever seen this movie, Saving Private
Ryan? The Mission is a Man? That's all we have. So I showed you the team approach,
right? So we have three nurses, a pharmacist, a physical therapist, two physicians,
home health, so like ten people literally trying to prevent one readmission. This is really
not sustainable. We've done well and it's great, but it's not sustainable. Also the hospital
is interested because they're not losing the 3 percent on all their Medicare diagnoses.
But I think in the future I'm sure that somebody will draw the line and say, "Hey, why is
this red hole in the budget here?" So we need to actually come up with smarter ways.
And actually there are smarter ways. If you look at how heart failure progresses, you
have patients that are normal patients. And then, you have patients that are
asymptomatic so they have an abnormality in their heart function but they don't have
symptoms yet. And then, as they progress they develop heart failure. What's
compensated initially then becomes decompensated and then refractory and they end
up in the hospital. Well, it turns out that the body sends physiological signals even in this
stage here.
So when you have an asymptomatic patient, all these pathways in the body get
reactivated. And all this actually can be captured. There are physiological signals that
can be captured. So even in the early stages of heart failure you can actually capture
these signals and you can use them to prevent something bad from happening. And it
turns out that intracardiac devices are able to capture this. So the intracardiac devices
were initially designed to monitor and treat arrhythmia, so ventricular fibrillation or
ventricular tachycardia. And being inside the heart, you have electrodes, you can
capture a bunch of things. So engineers actually came up with a bunch of things to be
captures. Initially nobody was using them but finally people have started to use them.
And I'll show you some data that these devices are actually so powerful that if used right
they can actually prevent readmissions and intervene before patients develop
symptoms.
And I'll just show some of these features here. So we know that congestion, which is
fluid build-up, precedes hospitalization. And this is one of the patient's clinical trial we
had three years ago, and we actually had a sensor in the heart, a tiny sensor, about this
big that you put into the pulmonary artery and you just leave it there. And then, it can
send information twice a day and it can actually send the pressures in the heart. It's an
invasive procedure but once it's there, it's there and it can send this information. And we
can see, actually, the hospitalization. We know why the patient came into the hospital.
And then, when the study was terminated and we un-blind the data, we actually saw that
the pressures were high every time he came into the hospital. So it was very useful to
see that.
The point I'm trying to make here is look at the dates of the hospitalizations here. The
first one is actually Labor Day and this is Easter and Christmas. It's very clear what
drives him to come to the hospital: bad diet. And the patient actually stayed after the trial
was completed, he stayed in the study. And my nurse used to see his pressures every
morning and would call the guy and say, "Hey, did you eat out again last night? You
have to change your medications. Don't eat so much salt." So it works. Another very
good physiological sign that is very powerful is actually patient activity. So we know that
as you get heart failure you don't get enough blood to your muscles and patients
become less and less active. And you have a capacity limitation in heart failure. They
actually have very good studies to show that the rate of heart failure correlates with how
much you can walk. And intracardiac devices actually have the ability to capture how
much you walk because you [inaudible], and not only how many hours a day you can
walk but actually walking intensity, how high your heart rate goes. So it can actually
capture all this data.
And it turns out that it is a very powerful data set. So if you have a patient that has a low
activity level compared to someone who has a high activity level, they have more than
five-fold increase in the change of being admitted to the hospital. So very, very power.
Imagine if you have this patient, for instance. He's a healthy patient at the time of
implant, when they got the device implanted. And you can see very high activity level.
And this is a patient who was admitted to the hospital a bunch of times. Again, looking at
the time of implant: doing okay for a while but not going up, sort of going down. And here
clearly the activity level is going down. And you can actually see that even before the
patient went into the hospital, you see at least 30 days or so of lower activity level. So
had we actually intervened here, we could actually prevent his hospitalization.
And you can see the 30-day readmission, right? The patient gets out of the hospital.
Immediately within a month, he ends up back into the hospital. Okay? So clearly even
more so the activity level decreased even more. So this was a somewhat high risk which
could have been – if you use this data – actually prevented from returning to the
hospital.
>>: If you catch that early on that curve, so you're still sort of a few weeks away from
what would've been [inaudible], what's the treatment? And how reliably can you actually
prevent that hospitalization?
>> Liviu Klein: So you can actually prevent it. The treatments are really: one, controlling
fluid level that the patient builds up, so we can use diuretics, water pills to control that;
and then, some sort of vasodilator, so blood pressure medications, to lower the filling
pressures in the heart. Now...
>>: Those are medications that they're not on already?
>> Liviu Klein: So some of them they're not on; some of them they are but lower doses.
We can actually change the dose to prevent that.
>>: Do you have any idea of cause and effect here? If the patient is not feeling well and,
therefore, the activity is going down? Or is the activity going down and that's causing the
patient not to feel well?
>> Liviu Klein: So the patient is actually asymptomatic; they have no symptoms here.
And that's why we think it's cause and effect. I think something is going on so the activity
level is going down.
>>: [Inaudible] right time for intervention so you're not now – Before that you were
talking about after the first event. Now you're trying to be [inaudible] before the event.
But how about trying to treat even before that which may involve things like exercise, the
diet, stress and so forth. When would be in your mind the right time to...
>> Liviu Klein: So that's an excellent question. I think that's sort of the whole purpose of
prevention, right? So the easiest way to treat a disease is actually to prevent it from
happening. So we try to address these messages way before the patients get the
disease. Once they get it, we always emphasize that they should follow the right diet,
exercise and try to stay, otherwise, healthy but they're usually not. So I don't think this
information would change my message. They'll be exactly the same, but it will make me
have some other form of intervention either medication-wise or other treatment to
prevent them from coming to the hospital.
>>: So just trying to make sure that I understand what you said: so what you're saying
that if you try [inaudible] ahead of time, you know, years ahead of time, people would
just not respond to this and this is why this is the right time for the intervention?
>> Liviu Klein: Right. So we've delivered that message already and they're not following
it, basically. Yeah, something that the American Heart Association calls ideal
cardiovascular health: there are seven steps to that. There is your diet, not having
diabetes, having enough exercise, not smoking, having controlled blood pressure and
controller cholesterol. So there are seven steps basically. If you put them together,
people that have all these seven characteristics their change of developing
cardiovascular disease by age 85 is close to 0; it's really, really, really low. But if you
look at how many people in the U.S., for instance, have those seven parameters, it's less
than 2 percent of the population. So a very small number of people actually have the
ideal cardiovascular health. We know it works it's just, unfortunately, people don't follow
that.
Another parameter that we can follow with intracardiac devices is actually heart rate and
heart variability. So we know that the heart rate is governed by sympathetic and
parasympathetic forces, and whenever we have heart failure, that balance changes. So
you can actually asses the heart rate from these devices and the heart variability which
is very easy. So we have two different heart rates and then by actually averaging those
you get the variability. And imagine you have these for 24 hours, you get a 24-hour
variability there. And if you look at actual cases, the heart variability here is
hospitalization. So we can actually see that we think probably a normal value would be
around 80 seconds and you see less than 40 is at a very high risk for hospitalization.
And about 20 days prior to hospitalization, you can see that threshold being met.
So, again, had someone followed these parameters 2, 3 or 4 weeks before the
hospitalization, they could intervene and patients could hopefully be prevented from
coming to the hospital. We also know at nighttime the heart rate is supposed to go down
as your parasympathetic system is more active. Actually if it goes up, it means your
sympathetic system is more active; it means your body is sort of in the compensated
state. And that's exactly what you see here. So you can see about 3 or 4 weeks prior to
coming to the hospital, the heart rate at nighttime is actually going up, signaling
something abnormal with this patient.
>>: You sleep poorly then also, is that – Do patient's have sleep problems then when
they're in this state?
>> Liviu Klein: They could have sleep apnea as well, and they could wake up because
they're short of breath. Yeah. If you have an intracardiac device, again, the major goal is
monitor arrhythmia, so if you have something like that you can actually monitor for atrial
fibrillation or ventricular fibrillation. You can see whenever this happens. The patients are
more likely to decompensate. And if you integrate all this information together, again,
you see the heart rate at night and day. Normally you should really have a nice
separation but whenever they are actually not separated, that could mean that there is a
problem with this patient. And also you can see the activity level kind of up and down. So
you can integrate all this information and draw some conclusions.
Finally one of the last parameters that people have used is intra-thoracic fluid
accumulation. So we know that when the pressures in the heart build up you have a lot
of fluid in the chest, a lot of fluid in the lungs. So if you actually measure this impedence,
you can actually see when the fluid accumulates, impeding decompensation can occur.
So this the impedence. Lower impedence means more fluid. And this is the so-called
fluid index for physicians who are not really physics-sensitive; they need cues. So
whenever the fluid level goes up then there's more fluid. If you look at the impedence
that translates into actually impedence being lower. This is a patient that has very little
swings but I actually see patients with lots of swings in their fluid level so a very power
sign of decompensation.
And I think the last one I'm going to show you is actually respiratory rate. So this is data
from an implantable cardiac device and it can measure the actual respiratory rate. And
you have patients that have little variation throughout their day and patients that have a
lot of variations. And you can see that people that have a lot of variations in their
respiration rate actually end up in the hospital. Again, this is very powerful predictor and
it comes from a study that was presented last year at one of the EP meetings. Again you
have a [inaudible] at about 5 compared to people who have normal respiratory rates with
no variation versus people that have wide variation during the day. And the thought is
that, you know, when you have fluid in the lungs you have to breathe more shallow and
faster. And then again, if you sit upright and your breathing gets better. But if you cannot
sleep then your breathing gets faster again. So having these wide variations is what
makes this happen.
And again, it's a very powerful prognostic sign. And how do you put it all together? Well,
actually there was a study conducted a few years ago where several of these
parameters where investigated in a clinical trial. And all these things were collected. And
what it showed was that if you have positive diagnostics, which in the trial the ideal they
find is two – it doesn't matter which two but two of them were abnormal compared to
everything being normal, you actually have a pretty high chance of being admitted to the
hospital. And it turns out that they sampled this at two weeks, three weeks, and four
weeks; somewhere between three and four weeks is actually the optimal time to kind of
[inaudible] collected data. And I shared with you before that we did this in Chicago. I was
there a few years ago at Northwestern and we had about 1,200 patients that had
devices. We put them on remote. Every three weeks we interrogated their data. And
then, if three parameters were abnormal, actually my nurses called the patient and
identified the actual problem. Brought them to the clinic. Have some intervention. And
within six months we actually decreased the hospitalization by 50 percent compared to
the six months prior to initiating this program.
So very simple steps that we did and very, sort of, back of the hand analytics and that
was really, really powerful. So I think we have a lot of data that physiological signaling
works if you get these signals processed. And this is what the usual care is today:
basically, daily monitoring weight and symptoms which we know are very, very
unreliable. And then, you have a schedule assessment, seen in-clinic every three
months or, "If something happens, come into the clinic and I'll see you today." And there
are some – The visits are not very comprehensive, maybe a few times a year. And then,
the therapy is mostly reactive. So I think what the ideal monitoring should be is really
having a proactive monitoring of hemodynamics or physiological parameters alongside
with weight and symptoms maybe.
But actually having less in-person visits but more data being transmitted remotely and
the therapy being actively transmitted remotely back to the patient in real time not every
three months when the patient sees a physician in the office. Well the problems with the
implantable devices: They are great but they have to be implanted. So if you look at the
current devices, the indications for them are really arrhythmias. And only about 10
percent of the heart failure patients will qualify for that diagnosis. Also, it's an implantable
device so we have complications associated with the implant. And it's not trivial; it's 5 to
10 percent. They are not life-threatening complications but, nonetheless, complications:
infections, bleeding, things like that. And they are pretty costly, about 40,000 dollars per
device. So, again, a lot of money and a lot of problems.
There are newer implantable devices now that the electronic industry is trying to come
up with. The companies that make these defibrillators are trying to come up with newer
implantable devices that are smaller or just featuring diagnostic features. But again we
don't know how much they'll cost and, again, they're still implantable so there could be
complications. And lastly would you really want to have an implant, right? There are not
a lot of people that voluntarily say, "Oh, sure. Give me an implant now." So I think the
proposed solutions are non-implantable. But if you look at these solutions out there, they
are pretty funny. I mean, would you wear this a half an hour a day to monitor your heart
health? It looks pretty fancy at a Star Trek convention maybe but not in daily life.
And if you have a patch, would you really wear a big, bulky patch every day? You have
straps. For some reason everybody wants straps. And then some other [inaudible] that
are kind of looking puffy under the skin. So the idea is to send all this information to the
cloud, something gets processed and then the physician gets notified and the patient
gets notified. The problem is that all these solutions they -- because of the amount of
data that's required and the archaic technology they use – can only sample 10 minutes a
day or 20 minutes a day. Again, who knows what those best 10 minutes are, right? And
where's the patient compliance? Who's going to remember, "Oh, it's time to put my collar
on," 10 minutes a day? With the fitness solutions on the other hand they are cool;
gadgets that everybody wears, FitBit and all that stuff. But they really don't do much. I
mean, they monitor your heart rate and maybe respiration and how much you walk. So
they're really not very helpful for healthcare.
So if you think about remote monitoring, this is sort of the standard of care today where
you have these big, bulky devices sending the information. But that's what I really want
to have, right, something between cumbersome versus awesome. This is what we need.
We need a very good device that's easy to use that the patients can use very readily.
And this is really where the paradigm should be changed. So you have the patients that
are older now; they have more comorbidities. I showed you that the top 5 percent of the
patients spend 50 percent of the budget just because of comorbidities. They are frail.
If you look at admissions, only about 40 percent, for instance, in heart failure are due to
heart failure. The rest are due to not even cardiac reasons. Older people that come from
the hospital, they go home. They fall, break a hip; they come back to the hospital
because they have a hip fracture. And they're actually doing a study at Berkley in heart
failure patients to see if we can predict from their gait if they're at risk for
decompensating after hospital admission. More and more patients are actually tech
savvy. So we did a study at USCF and we found out that 67 percent of the patients
actually had a smart phone, even patients as old as 80. And they're [inaudible] in using
social medial and things like that. And a lot of people now want their own data I think
partly fueled by this sort of health-fitness industry that people can see their heart rate,
they can see how much they run and how many calories they consume and things like
that. People really want their data to improve their health. You just have to give them the
right data to do that.
And of course the payers are very interested, right? Imagine if you have a device that
will save Medicare 10 percent; that's 2 billion dollars. That's a lot of money. And the
technology, I think, is there. We have cheaper and better diagnostics. We have faster
processing. We have predictive analytics. We have cloud computing. And I think these
things can, in the right frame, be put together to actually deliver proactive care. And then
the health system, it's really very interesting because there is a very limited pot of
dollars. Right now we have [inaudible]. We're moving into paying episodes of care. And
in the future we'll actually get paid per disease. So we have 1,000 heart failures – With
the ACOs, right, you have 1,000 heart failure patients. You'll be given 5,000 dollars per
patient per year and what you do is your business. You can put a defibrillator or you can
keep them away from McDonalds or give them a device; it's really up to you what you do
with that money.
And I think the provider, now, is really aligned with the health system. It used to be that
most of the providers were sort of private practices or doing their thing, seeing the
patient in the office. Patient goes to the hospital and they can go to the hospital. They
can bill there. But now they can't survive any more, so actually big hospitals are
acquiring most of the practices. So in fact in primary care I think 80 percent of the
practices are actually acquired by hospitals. In cardiology, 65 percent are acquired. And
it'll probably be 100 percent within a few years because no one can survive on their own.
So we really have an alignment now between the providers and the health system and
the payers.
Because payers will have – Aetna will go to say, "Well, UCSF you're going to be ACO in
the Bay Area. And your heart failure care model is going to be what we want to pay for."
So four million people will be just flying to UCSF to get their heart failure care there. So
there is really a lot of alignment in this. But what we need is better monitoring tools. First
of all we need tools that collect relevant data, okay, not just body weight, relevant
physiological data. And I think they need to be collected continuously. It's a 24-hour
continuous or 10 hours? I don't know. But they need to be more than 10 minutes a day.
And I think we need to collect multiple physiological signals. The problem with the
current existing non-invasive tools, they all want to collect one signal. The problem is the
body is just a very complex network; it just doesn't have one signal.
We need multiple physiological signals to be able to detect changes. There is no magic
bullet. And it has to be passive. The moment you ask the patient to do something, that's
the end of your product. No one wants to do anything. And it has to be easy. It has to be
live transmission to the physician office, processed in the cloud with a predictive
analytics and immediate diagnosis. The physicians don't want and don't have time to
spend to try to identify what the signals mean. You have to give them, "This is green.
This is yellow. This is red." And then there has to be a feedback loop with the patient
what to do now not in three months when he sees the physician. And I think it needs to
be noninvasive and unobtrusive with a very easy to use system. You can ship it in a box.
Everyone can put it on their wrist or whatever and can use it right out of the box.
So I think, just to conclude my presentation, the admissions for heart failure are really
common and costly. There is a rapid increase in the number of patients; again, the aging
population is expanding and 1 in 10 people over age 75 will have heart failure. And I
think we need to change from the unidirectional passive remote monitoring to the
bidirectional information exchange between physician and patients. And I think it's a very
unique opportunity to decrease the fastest growing cost in the Medicare system and be
rewarded for it. And I think it's a very unique opportunity for a technology company to
actually combine the existing engineering skills, the data mining skills and the
predicative analytics to deliver this product in the 21st century. And I think the nice thing
is I actually can take these lessons from heart failure and apply it upstream to all the
cardiovascular problems: apply it to myocardial infarction, apply it to hypertension. I think
that heart failure is the case example because it's so morbid and so deadly and so
prevalent.
So if we can actually learn something from this and then take it upstream and actually
make huge indents into primary prevention and preventing cardiovascular disease.
Thank you. And this is my contact information if you want to email me or call me. Thank
you.
[Applause]
>>: So you're generally at the beginning trying to collect many signals and use analytics
to infer risk from those signals. Someone could have given a similar talk that said the
thing we want to measure in this case, for example, is intracardiac pressure. Let's put all
of our effort into noninvasive measuring. I'm interested that you didn't talk about
emerging technologies for noninvasive pressure assessment or things like that. Are
there specific technologies emerging, let's say, either for portable [inaudible] impedence
or whatever might be practical, for measuring the specific signals that you believe to be
interesting? Or have you sort of seen what's out there and it's not that promising?
>> Liviu Klein: Yeah. So there's really no good noninvasive way to measure intracardiac
pressure, so most of the tools nowadays are miniaturized invasive. So you put them in
and kind of leave them there and then, they transmit some data. But I think, again, just
measuring the pressure alone is not going to tell the whole story.
>>: Even if you had reliable...
>> Liviu Klein: Even if you have reliable – Exactly. Because you can have pressure
changes and none of the other parameters are abnormal. So the best example, again, is
you have this heart that's failing so the pressure is already high. But let's say you have
this patient in an experimental state in the cath lab, and you infuse 2 liters of fluid in this
patient. The pressure will go from 20 to 40. But the patient is not in heart failure because
nothing has changed. And that's why I think the change doesn't happen. And that's one
of the good things about cardiovascular disease is doesn't happen overnight. Actually it
happens over time, so you have time to react.
>>: So you're focused on this kind of multi-variable approach isn't just because we can't
measure the right thing yet it's because there's not...
>> Liviu Klein: We don't know the right...
>>: ...[inaudible].
>> Liviu Klein: Right. We don't know what the right thing is, that's the problem. And I
don't think there's only one right thing because the body is so complex and there are so
many [inaudible]. Again, if you think about pressure, part of the problem with pressure is
that the kidneys don't work. They cannot eliminate fluid, right? Because if I give you 10
liters of fluid, you're just going to eliminate it. But someone who has an abnormal heart
may have abnormal kidneys and then they cannot eliminate the fluid. There are so many
interactions.
>>: So you talk about this thirst, I guess; I can see we're just starting to have [inaudible]
data and the bidirectional information flow between patients and providers. But we've
noticed a dichotomy in providers where some providers are really amenable to this and
really think it helps them do their jobs, and others are – for lack of a better way of
describing it – deadly afraid of it, claiming it adds a lot of load. We wrestle with this
>> Liviu Klein: Yeah.
>>: Do you have advice on how to think about that and what the right entry point
[inaudible]?
>> Liviu Klein: It's a great question. And I think unfortunately most providers are deadly
afraid of it. And I think there is a reason for that. One is, it's perceived as adding more
work. Someone's already busy and it's adding more work. And I think the existing
technologies have not made the case that it doesn't add more work, it actually saves
work. So I think once that case is going to be made, I think more providers will be open
to the idea of sharing information and being more proactive in that regard. And secondly
I think it's something that's hard to change, the traditional way that medicine was
performed and people thought that, "I am the physician. I am the end-all, be-all," kind of
thing. It's changing with generations.
>>: Some of this -- I think you can almost the characteristics of this device, one that's
not implantable. But do you see the mechanism by which consumers get the device the
same? Like is the implementation the doctor really specifically recommending it? So you
can see...
>> Liviu Klein: So I think the mechanism – It's actually very interesting. So I think if you
look at healthcare as three parties: the patients, the providers and then the payers. So I
think it's going to be probably either patient-driven -- Because they'll say, "Well, I've seen
this data on a commercial or whatever, and it really looks like it's going to improve my
health." And they're going to go to a physician and say, "I want this." – or the payers will
say, "You know what? This actually really saves money, so if you don't get it I'm going to
increase your premiums. And if you don't get it as a physician, I'm going to move my
business to a different institution."
>>: You think there's the data now to support that case?
>> Liviu Klein: Not yet. I mean there are pieces of data but not completely. Otherwise,
we'd probably having some sort of monitoring. But it's not there yet and the major reason
is that people are just focused on one single parameter.
>>: One thing I have seen: the data on obviously a lot of these fitness devices are
oriented towards step counting; FitBit being the most successful story there. And the
CDC is all over FitBit right now, like the data coming out of the CDC is like, "More
walking, better health, FitBit, FitBit, FitBit, FitBit." I actually haven't heard anything to
suggest that it has propagated to payers yet.
>> Liviu Klein: So it actually has.
>>: [Inaudible] heart failure specific.
>> Liviu Klein: So, yeah, it has actually. There are some insurance companies where – I
can't remember. I think maybe BlueCross. I think it's either BlueCross or United where
actually they had a site where you download your data and you get a hundred-dollar
discount a month for something if you show that met the 10,000 steps a day.
>>: And that's not – The providers are not in that loop?
>> Liviu Klein: No. It's patient and payers.
>>: Are providers encouraged or...
>> Liviu Klein: It's patient and payers.
>>: Are providers at all encouraged or reimbursed to get patients to use, let's call it
FitBit...
>> Liviu Klein: No.
>>: Not yet?
>> Liviu Klein: No. I mean, we do it. So a lot of us do it and it kind of depends on what
part of the country we practice in. But at least, I've seen a lot of cardiologists actually
embracing that. And we known the 10,000 steps a day for many, many years but had no
way to actually get it into the population. Now it's very easy. It's like, "Wear this. It's
going to tell you how much you walked."
>>: But is that a sign of a potential problem going forward? That providers aren't –
You're doing that – Maybe it's just a sign of it's time to move away from the fee-forservice-based reimbursement because it's sort of like obviously that's in the best
interest. But it doesn't necessarily enable the next thing to make its way into practice
because you're not being reimbursed to recommend a FitBit to your patient.
>> Liviu Klein: Right. But again it's going to change because people are going to be
reimbursed for that specific thing. And, again, the reimbursement is a very complex
issue in the United States but there are now codes for prevention where people can
actually bill for preventive services. So once that gets mainstream, people will be
encouraged to – a physician will be encouraged to promote that. But I think it's not that
physicians don't want to do that because first of all we are in the business to help
people. I think it's if you make it easier for them. If I have to fill out 20 forms and then get
the data transcribed, it adds 20 minutes of work to my day for every single patient, I just
don't have time to do it. But if you make it seamless and integrative and I don't have to
do anything, then it's going to work.
>>: So you said for cardiovascular diseases, it's not something that happens over night.
An apparently there's not a lot of data yet supporting any further application. Do you
think you could propose some applications, other disease that short term changes would
actually [inaudible] on the data that is available?
>> Liviu Klein: I think the one disease that has the shortest sort of change with outcome
is diabetes. Your blood sugar can change within minutes and then something really bad
happens, right? If your blood sugar drops, you can go into a coma. If your blood sugar
goes up, you can go into a coma. So that's one application where people are actually
working and it's one of the diseases with the most research because it's very
symptomatic. People are interested to monitor their sugars and they're doing better with
technology. Again, cancer is one disease that's a very slow progressive disease but
there're really not a lot of physiological signals; it's more biological signals. And again we
know now that cancers are basically just gene mutations at a different level. And there
are companies coming out now with technology that can test and sample and identify
something. The colon cancer, right? There's a recent publication showing that testing
your stool for mutation in a gene is more sensitive than a colonoscopy. So people are
using that, again, because it's a very slow progressive disease.
And we didn't talk about it but I think in the end, the ultimate sensor should have some
biologic integration, some sort of molecular signaling for heart attacks. We know that
things change into the blood stream, that different cells into the arteries shed. And if you
can detect those [inaudible] before a heart attack happens. So integrating that with your
physiological signals and sort of having the personalized medicine approach that we all
kind of aim for.
>> Desney Tan: Thank you.
[Applause]