DRF 29: Improving Pharmacotherapy in Pediatric Critical Care & Heart Disease and End of Life Care

DRF 29: Improving Pharmacotherapy in Pediatric Critical Care & Heart Disease and End of Life Care


(upbeat music) – Thanks for coming, everyone. We’re gonna go ahead and get started. My name is Samantha Dallefeld. I have had the pleasure of being a part of the Fellowship here at the
DCRI for the past couple years, and had the opportunity
to serve this past year as one of the chief fellows. It’s certainly been a great experience and I wanna present today some of the work that I’ve been doing over the past year or so. I don’t know a lot of you
in the room, (chuckles) so a little bit about me. I’m originally from Texas. I came here for my pediatric
critical care Fellowship. I’ll be finishing that
up in about a month, which is crazy to think of. And then, I will be headed back to Texas. I took a job in the pediatric ICU in Austin at Dell Children’s Hospital, so I’ll be there starting in
August and hope to continue some of the work that I’ve
done here and working with the amazing group that
I’ve had the opportunity to work with here at the DCRI. So, today I want to present my work. Improving pharmacotherapy
in pediatric critical care. Specifically, dexmedetomidine on ECMO. I have nothing to disclose… Yet (chuckles). So the outline for today, I’m gonna start talking about some of the gaps in
pediatric pharmacotherapy, address the impact of critical
illness on pharmacotherapy, and then talk about two
studies of dexmedetomidine pharmacokinetics, or PK,
in children supported by extracorporeal membrane
oxygenation, or ECMO. So let’s talk a little bit about pharmacotherapy in pediatrics. Does anybody here have a,
either have a child with asthma, or know a child that has asthma? Can I have a show of hands
for all those in the room? Probably most of us, right? We know somebody who had
been affected by this. Or maybe, it was you
when you were a child. Well, the gold standard therapy for that, if you have a child that’s
having an asthma exacerbation, they show up to the emergency
department, they’re gonna get what’s called a DuoNeb, which
is an inhaled solution of a couple of different
drugs. That’s standard. That drug is actually not approved for use in kids (chuckles) by the FDA, which is kind of
interesting to think about. A very common illness, a very common drug, safety and efficacy has not actually been proven in all ages of children. So off-label drug therapy
refers to drugs that are used outside of FDA approval. That could be outside of the age, it could be a different age indication or maybe even formulation. And this is important because
these drugs lack safety, efficacy and dosing data in children and that leads to the potential for therapeutic failure and drug toxicity. So therapeutic failure is
the result of doses that are too low to assure the concentration necessary for adequate treatment. Toxicity is the flip side of that, so levels that are too high, potentially putting the
patient, or the child, at risk for unnecessary adverse events. So why are there so many drugs
used off-label in pediatrics? Well, there’s a lot of challenges
to drug studies in kids. For the parents in the
room, how many of you, if I were to say: “So I wanna do this drug
trial, can I use your child?” “It’s on a drug that, well we
haven’t really proven safety “and efficacy in children
yet, but I wanna poke them, “I wanna give them this
drug, and then poke them with “needles several times
to draw their blood, “and then I’m gonna monitor
them in case they have any “expected or unexpected side effects.” Like, anybody signing up for that? No, so I’ve recruited no one
yet for my study, in the room. So reduced sample size is an issue. There’s lots of barriers to consent. And kids are small, so they
have limited blood volume and that reduces the quantity
of blood that we can get and the number of samples that
we can obtain in children. And then, there’s a wide range of ages across the pediatric spectrum, so a study that I do in
one age group may not be representative of all pediatric ages. So this represents, let’s
talk a little bit about that last point, the different age ranges. There’s a lot of challenges
to pediatric drug development and all of these charts
here are showing age across the horizontal axis, but
they’re showing changes in metabolic capacity,
changes in GI function, changes in the skin, all of
these things that happen from childhood to adulthood. Changes in metabolic
capacity is one of them. Again, age on the horizontal
axis and percentage of adult activity on the Y axis. The colored bars represent
the activity of different enzymes that metabolize drugs. So we can see, if you
look at the blue bar, this is CYP3A4, which is
one of the most common metabolizing enzymes for
drugs, and it doesn’t reach adult activity until
around year one of life, sometimes later than that. The green bar is CYP2D6
that’s only present in the first three days of life. So these things are changing
all the time over the course of sort of, the pediatric spectrum. Acquisition of renal function, so again, age on the horizontal axis. The leftmost vertical axis is para-aminohippuric acid clearance. That’s a measure of kidney plasma flow. And then the rightmost vertical axis is glomerular filtration rate. And what this graph is
showing is that renal function doesn’t mature to adult
levels until probably, six months to a year of
age, and that can certainly affect drug clearance. All right, so we’re already
getting difficult, and then you add critical
illness on top of that. So critical illness also
alters drug disposition. So, a common patient that I
may admit to the pediatric ICU we’ll say, a two year old with sepsis. Overwhelming infection,
overwhelming inflammation in the body, this patient usually presents to the emergency department and
the blood pressure’s low, they need tons of fluids to
help get that blood pressure up. They may get a liter of
fluid in 20 to 30 minutes. What happens over the next 24-36 hours is, that inflammation in the body
causes their blood vessels to be really leaky and that
fluid starts to leak out into their tissues and
you end up with this like, pitiful little puff ball,
they’re just super swollen and edematous, and that
increases the volume of distribution of a lot of drugs. The volume of distribution is really the concentration of drug
in the plasma to the total amount of drug in the body,
and that’s really important for drug pharmacokinetics. Additionally, these patients with this overwhelming inflammation,
usually have kidney dysfunction, they could have liver dysfunction, and that’s gonna impact the
clearance of these drugs. So we have big spectrum of ages, all of these things changing, adding critical illness on top of that, and then we have therapeutic
interventions that we do. So we may need to use a
ventricular assist device. Those are devices that are used in patients that are in heart failure. Essentially, it does
the work of one or both, of the ventricles, so the lower pumping
chambers in the heart. Dialysis, if their kidneys
aren’t working properly. And sometimes, we have
to put these kids on extracorporeal membrane
oxygenation, or ECMO. So let’s talk about that. So, ECMO is complicated. Here’s a schematic here. So essentially what it
is when you boil it down, we put a big canula in the
body, typically in our patients it’s in the neck, we pump
blood out of the body through a number of circuit components, including an oxygenator, and
then we pump it back into the body either through
an artery or a vein. And it’s used in patients with refractory heart or lung failure, or both. So does anybody know how
much blood volume is in say, a full-term infant? About 300 milliliters. Any idea how much blood
volume is required to fill all of this tubing and all
of these circuit components? About 300 milliliters
(laughs), so we have to fill this circuit with blood
products before we attach it to the patient, but you’re
essentially doubling, sometimes more than
doubling, the volume of distribution for a lot of these drugs. If these patients haven’t
already had kidney dysfunction, or liver dysfunction,
before going on ECMO, certainly ECMO can make things worse altering the clearance of drugs. So getting more, and more complicated, and then the circuit itself
can actually extract a drug. So extraction is really when the tubing, or the components of the
circuit pull drug out of circulation and it binds
to the tubing, the plastic and all of this supportive
material and then that drug isn’t available to go to
binding sites to be effective. And that is really drug-specific
and circuit-specific. Over the last couple years
I had the opportunity to write a book chapter
on ECMO and sedatives, with one of my mentors,
and this really highlighted some of the knowledge gaps
that exist in this area. So sedation is really important
in these kids on ECMO. As you can imagine, we wanna
minimize the distress and the pain of all of these procedures, and having a big canula in their neck, but we also wanna decrease
the risk of complications such as, dislodgement of
said life-saving canula. So we always have these
patients on sedation, and we don’t know a lot about
dosing of these medications in patients on ECMO, varying
ages, critical illness, ECMO. It gets really complicated. Some of the things that
we have found in studies, both in the lab and in patients, is that drugs that are highly lipophilic, like Fentanyl, drugs that
are highly protein-bound, like Midazolam, tend to be more extracted by the ECMO circuit. So what that tells us is,
usually those medicines require higher doses, loading
doses and maintenance doses, in order to obtain the
same therapeutic levels as a patient that’s not on ECMO. But dosing data is really limited. Dexmedetomidine, there’s
essentially no information for, which was a big surprise to
me because dexmedetomidine is sort of a new, hot, sexy
drug in critical care that we are using all the time. I think it’s twofold. One, it has a relatively
benign safety profile, but it also may have some
neuroprotective effects, which is different than
some of the opioids and benzodiazepines that
we’ve historically used. But we just don’t have a
lot of information about it. Now, dexmedetomidine
just like anything else, doses that are too high, bad. Doses that are too low, not effective. So we need to be able to dose
these medications effectively. This is a drug that’s used
off-label in children on ECMO. Safety and efficacy hasn’t been proven, but we’re using it in essentially
everyone that is on ECMO and many of the kids that are mechanically ventilated as well. And we expect, because of
its high lipophilicity, and it’s high protein-binding, that it would be highly
extracted by the ECMO circuit. Now, there are two prior
ex vivo studies that have been performed of dexmedetomidine. Now, ex vivo is essentially,
when you do a study in an isolated circuit,
not attached to a human, and so if you lose any
drug in that circuit over time, you know that
it either went somewhere in the circuit, or it was
bound-up in one of the other products that are in that
circuit prime, in the blood. But those studies are really limited. One of them is actually,
sort of done in test tubes, it’s not even really
done in a whole circuit. The results are conflicting,
and there’s just a lot of limitations between the two studies. So I wanted to look at
this more in detail and figure out how we can
effectively dose this medication in our patients, to effectively
sedate our patients, and to decrease the risk of drug toxicity. So I designed a two-part study. The first part was doing
this ex vivo study to really quantify drug extraction by
these circuit components. And then the second part,
being to do an in vivo study, actually in-patient, to
characterize volume of distribution and clearance
in infants on ECMO, the two of which could be
combined to recommend some dosing data to help
intensivists across the world use this medication more effectively. So I hypothesized that
dexmedetomidine does undergo substantial extraction
based on its lipophilic and protein-binding characteristics. That the volume of distribution
would be higher and clearance would be lower
in children on ECMO compared to those not, for reasons that we talked about earlier. And that higher dosing of
dexmedetomidine would be needed in children on ECMO to achieve similar sedation levels as children not on ECMO. So let’s talk about the
ex vivo study design. So big picture what we did is, we took an isolated circuit,
we inject a drug and then we measure concentrations over time. That’s basically what we did. But then, we one-by-one,
took out different circuit components to try to isolate where the drug was being bound to. So I’ll talk about that a
little bit more in detail. So we started with this complete circuit. So what this is, is tubing
that goes to a pump, continues around to an oxygenator. We attach the hemofilter in there, which sometimes we’ll use for patients whose kidneys aren’t working very well and we need to pull off some extra fluid. And then we put the tubing back and we just attached it to a bag, essentially. We primed this circuit
with what we would use for any baby that’s going
on ECMO in the hospital. So blood, plasma, we use
an anticoagulant, calcium. We used bicarb to achieve
physiologic Ph in temperature. And then before we dosed the drug, we took 30 Mls out into a control sample. And the reason we did that
is ’cause I wanted to know if this drug just gets degraded over time, just by light, or
natural drug degradation. The we dosed both the circuit and the control with
dexmedetomidine at a level to achieve therapeutic concentrations per what’s documented in the literature. And we did this three times
for this complete circuit for 24 hours and we drew blood
at various time points, 11 different time points, over
that 24 hour period of time. Then we completed the
exact same experiment with an oxygenator circuit. The only difference between
the oxygenator circuit and the complete circuit was that
the hemofilter was removed. So in this way, if there
was any difference between the concentrations recovered
in the plasma between the oxygenator circuit
and the complete circuit, then we would say it had to have been taken out by the hemofilter. That’s the only thing that’s different. And then we did this
another time with the pump, what we called our pump circuit, same as the oxygenator circuit except the oxygenator was removed. So again, if there was
any difference between these two circuits in the recovered drug over those time points, it had to have been taken
out by the oxygenator. After we had all of our blood
samples we used a validated assay to measure the concentration
of drug in those samples. We calculated recovery of
drug using this formula. So essentially, this was
the percentage of drug that was recovered over time. So if I dosed a drug to a
certain concentration and had that same concentration
at the end of 24 hours, has a 100% recovery; no drug
was lost, got it all back, circuit isn’t doing anything. And then we sorta
subtracted that from a 100 to see how much was extracted. So this was sort of the big point. And again, we kind of already talked about this example below. Any difference between extraction
in the complete circuit versus the oxygenator circuit, it had to have been in the hemofilter. And this is sort of what
we used in our analysis. So here’s our results. I think they’re pretty exciting. So I’ll draw your attention first to this dotted line at the top. So this was our control sample. So all those samples that were in the test tubes just in a water bath, pretty much 100% of
the drug was recovered. Great, so no extraction,
no natural drug degradation over time, the light doesn’t do anything, drug can be recovered. Now, the pump circuit, well, let me actually start
with these two down here. So the complete circuit
and the oxygenator circuit over 24 hours, only about
30 to 40% of the drug was recovered at the end of the 24 hours. Meaning, 60 to 70% of the drug was lost. But where did it go? So the big difference here was between these two circuits and the pump circuit. The big difference being the oxygenator. So this allowed us to
conclude that the oxygenator was really the place where
all the drug is going. Great, okay, now we know. Potentially you could
create an oxygenator that would extract less drug,
that’s probably hard to do. So maybe we just need
to to dose differently, knowing that most of the drug is being lost to the oxygenator. A few limitations to this. So this was the lab study. And our patients that we treat on ECMO, we’re able to continuously
monitor their asset based status. We couldn’t do that in
the lab, so potentially, that may play a role. Saturable and reversible
binding was not evaluated. So there’s some theory
in the literature that if drug is bound-up to the
tubing that at some point it could leach back out of the tubing into circulation and then
become available again. Certainly, we didn’t see any of that. The drug concentrations
were continuing to decline at 24 hours, so we didn’t
see any of that in our study, but maybe if we had done it longer, we would have seen concentrations kind of start to rise, or level off. And then, dexmedetomidine
is typically given as a continuous infusion with
extra doses on top of that. So in our study we just gave one big dose. Now, it’s not being metabolized, it’s not attached to a baby, so it’s gonna hang around
there for a long time. But we didn’t evaluate the full effect of a continuous drug infusion. But nonetheless, we know that
the oxygenator extraction, likely it’s gonna effect
efficacy of this drug, if 70% of your drug is being lost. So we expect drug exposures
to be decreased in patients and that higher
dexmedetomidine loading doses may improve sedation
efficacy in these patients. And we think maybe a maintenance dose, a higher maintenance dose
may be required as well. So the next step was to look at the volume distribution and
clearance in infants on ECMO. So this is still ongoing. We’ll talk a little bit about
how we’ve set up this study. So I looked at patients just
less than two years of age. The reason I did that is
just to limit variability. We talked a lot about
the characteristics of the physiologic parameters that change in children over time. So just really wanted to pick a subset and limit variability. And then we looked for
patients who were already receiving dexmedetomidine per
their standard of care and who were already on ECMO. So essentially then I got
to go to a parent and say: “Your kid’s already getting this drug.” “They already have a line
where I can draw blood.” “They’re already on ECMO.” “Can I take a little bit
extra whenever they’re “drawing blood for other things?” And then we excluded some
really young preemies and then any other concomitant conditions that we thought would make
things more difficult, like CRRT, which is essentially dialysis. So what I did was, each
morning I would look at the unit to see who all the patients were. And I would find the patients
who were on dexmedetomidine and getting ECMO and approach
those patients for consent. Once I did that, I sort of collected a ton of demographic data. So I collected how long
they’d been on ECMO. I collected laboratory
data that they may have, knowing that this drug is
metabolized by the liver. So if their liver enzymes
are more amebic, would that affect things? All the dexmedetomidine dosing information any other medications they
were on that could affect the dosing and then, adverse events. And I basically just
waited for them to need another extra dose of that
medicine per standard of care. So maybe they’re starting
to get a little wild, crying a lot, maybe you
think they’re agitated. The nurse would call me and say: “Hey, I’m about to give an
extra dose of this medicine.” And I would rush in and collect
some extra blood samples. So we collected a max of
10 samples per patient in pre-specified windows of time, after an extra dose of the medication. We were able to do this study
with a very small amount, like a twenty fifth of
a teaspoon of blood. And then all of these samples are currently being stored for analysis. So just like in our ex vivo study, once everything is complete
we’ll send the samples off to a lab and evaluate the concentration of drug in those samples. And then the fancy part
comes in where we’ll develop a pharmacokinetic model. We’ll use some fancy
simulation software to develop a model for this drug
and then be able to run some simulations with all
of this data that we have to determine effective
dosing of dexmedetomidine in these really vulnerable patients. So as I mentioned, this
study is currently ongoing, although I’ve closed enrollment. So I have 14 patients
which totaled 101 samples. These patients are quite
young, age at time of ECMO was 43 days, so on average,
just over a month old. Pretty low birth weight, I mean, I guess a normal birth weight overall, although, we did collect a
baby that was two kilograms when they were born. Mostly male, although
a fair representation of females, as well. Mostly non-Hispanic, Latino, mostly white. And then up to 10 samples
were collected per patient, but a median of eight. And so, that’s sort where we
sit with this study right now. Everything’s currently in
contracts with the lab to analyze these samples, so I hope to have some really exciting data within
the next couple of months, that I think, not only is
gonna help in our ICU with dosing these patients
better and helping maintain comfort in our patients
better, but across the globe, as dexmedetomidine is really
being used everywhere. So future steps is, I’d like to do this, expand this study to patients
older than two years old. I’d like to do some data, I’d
like to analyze the data from this study compared to data
in other patient populations. So some of you may know the
pediatrics group has a big, sorta mega-trial,
Pediatric Trials Network, if you’ve heard of that. Well, one of the studies we
do within that is called POPS. And it’s essentially this,
on a much bigger scale. So it’s a multi-center study. We have a list of drugs that
we’re enrolling patients on. And all these centers can contribute blood samples for these different drugs. And dexmedetomidine is actually one of the drugs within that trial. So they’re not enrolling
a lot of patients on ECMO, but they’re enrolling a lot
of patients just on dex, for something else, maybe
they’re mechanically ventilated, or maybe they’re getting a procedure and need sedation for that. So hopefully, I’ll be
able to compare my data to their data in these patients not on ECMO, to really make a nice comparison of what are the dosing differences
between these babies, patients on ECMO to the patients not. And then, I’d like to
evaluate patients undergoing continuous renal replacement
therapy, or dialysis. And there’s actually another
Fellow who has started to look more at pharmacokinetics
in our patients on dialysis in conjunction with Kevin Watt, which I think is another
whole separate field. You’ve got ECMO, and you’ve got dialysis, all of which is definitely
affecting our drug dosing. But a lot more information, I think, will be coming in the next few years. So I have lots of acknowledgements because I couldn’t have done this alone. My primary mentors, Dr.
Zimmerman and Dr. Watt, and a lot of the other ICU
team, and Dr. Greenberg who helped with this protocol. My colleague who helped me
in the lab, and then I was supported through an NIHT32,
Clinical Pharmacology T32. So I definitely thank them, Dr. Benjamin, Dr. Brouwer
for that opportunity. And then, the samples are mostly being, running the samples is
mostly being funded by Eric Peterson’s Executive
Director’s Pathway to Supplemental Funding, so
I’m indebted to him as well. So thank you for your
attention and for listening. I guess, should we wait
til the end for questions, or does anybody have questions? Can take a few. Okay (laughs). I’ll pass it along to Haider. (applause) – [Haider] That was a great presentation. It’s really interesting. You know, our first presentation
was about challenges that patients have really at
the beginning of life in the first critical few years. And I’ll be going all the
way to the other end of the spectrum and talking
about challenges that patients face at the end of life. If any of you were expecting
to see Alex Fanerov today, I’m sorry to disappoint you. He’s actually away for
a pretty good reason, so I’ll be filling in
for him and showing you some of the research that
I’ve been able to do at DCRI during my one year here. So I went to medical school in Pakistan, and then when I came to the
United States I felt like there was a lot of things
that were very similar to how patients received
care in Pakistan and here. If you had a heart attack in Pakistan, we did the exact same thing. You’d get coronary
angioplasty, you’d get aspirin, you’d get the medications,
everything felt pretty similar. But you know, when it
came to the end of life, there was a lot of stuff, how
end of life care was delivered was very, very different. Things like, you know,
that you hear or that you may have experienced
by taking care of you know, family members, et cetera, like DNR, DNI, or people being on life
support for a very long time. I mean, these were all
things that were very, very foreign to me, and I
became very, very interested. And many of these encounters had a very powerful effect on me as a person. And one of the things that
I did was, I actually, I did something that I’d
done previously as well, which is to process
things, I wrote about them. And I remember this one case particularly that I wrote about back
when I was a resident, and I will share that story with you, just to tell you how I got
interested in this whole area. I met her in 2011 doing
during my first few weeks as an intern in internal medicine. She was not the kind of patient who had appeared in my textbooks. She had been admitted from
rehab after expressing an intent to take her own life. And I soon learned that her body was as troubled as her mind. She had heart failure
from a condition called hypertrophic obstructive cardiomyopathy in which an overgrown
muscular wall keeps blood from properly flowing out of the heart. She might have been born with it, or acquired it from years of
neglected high blood pressure. Her blood pooled in her
swelling legs and seeped into her lungs, causing her
to wheeze constantly. She was in her late 50’s,
but seemed much older. While learning her medical
history, I also got to know her. A former artist derailed by addiction, she had become alienated from
her career and her family. She had let the problems
in her heart fester and had never received consistent treatment. At this point, she wasn’t
a suitable candidate for heart surgery, but I felt
that there was still hope. A procedure called alcohol septal ablation could potentially reduce her symptoms. It involved injecting alcohol
into a blood vessel to strangle the blood supply to
the muscular part of the heart obstructing blood flow. The patient was desperate to
have the procedure performed. She hoped to experience a
life outside of hospitals, to reconnect with family,
to start painting again. With my sense of righteous
optimism intact from my medical school days, I
convinced the other doctors that in spite of the risks, we should try it. The procedure was uneventful,
but when I was updated on the patient’s condition the
next morning, I was stunned. Overnight she had gone
into complete heart block, a failure of the electrical
transmissions that cause the heart to
beat; a rare, but known, complication of the procedure. She temporarily lost her pulse, underwent CPR and
electrical defibrillation. A pacemaker had to be inserted, she was transferred to the ICU. Later in the day the
complications cascaded and she found herself buried again
under the flexed wrists of medical residents and nurses pushing her sternum into doughy pulp. After half an hour of
frenzied CPR, infusions, electrical shocks and
other assorted hail marys, resuscitation was called off
and she was pronounced dead. This was not the way the
story was supposed to end. Certainly, not how I imagined it would. So during residency, I was
exposed to so many cases in which patients with heart disease, we just really struggled
to take care of them at the end of life. We had no good way of knowing if they were truly at the end of life. We had no way of assessing
what their prognosis was. And all of these things kind of coalesced, and led me to basically
pursue this interest further. And I ended up writing
a book about it called Modern Death: How Medicine
Changed the End of Life, in which I track some of
the history surrounding how the end of life has changed; how technology has affected
patient’s experience at the end of life and what do
we need to do to go forward. But what I wanted to do is, I wanted to go a step further and use the
opportunity I had at DCRI and at Duke, to really
understand what is in fact, the landscape of patients with heart disease at the end of life. And this is an important question
because of a few reasons. First of all, if you’ll
see in the graph here, this is data we used from the CDC. The number of patients who
are dying with heart failure, which is a terminal, which is
basically terminal pathway for all heart diseases, really
growing as our population ages. If you look at this purple line, this is people above the age of 85. And you can see that it’s really skyrocketing in recent years. And the other thing that’s
going on at the same time is that costs related to healthcare, especially Medicare money, is also going up pretty exponentially. This is a paper I wrote with
Kevin Schulman, who is here, which shows that Medicare
expenses, and Medicare, is again, the insurer that
takes care of eight out of 10 people who die in this
country, are really going up, with no, sort of, substantial
increase in revenue. So this is a huge issue
not only because it affects a lot of patients, but it is
also placing a huge burden on our health system and our economy. So first I’ll show you some
work that I was able to do which just paints a landscape
of what end of life care in heart disease really looks like. Again, this is data we used from the CDC. It’s publicly available data which shows, which looks at where people, what locations people died
that had heart disease. And if you look at this top
line, which is a blue line, you can see a trend. This is people who died
in the hospitals and you can see that there is a
reduction in the number of people dying in hospitals, over
time, who had heart disease. And similarly, you can
see a similar trend in people who are dying in nursing homes. And these are both good trends
because what we know from surveys of patients is
that patients by-and-large, prefer to be able to die
at home, around sort of, a familiar environment,
surrounded by family, rather than in a hospital. And at the same time, we have
seen that the number of people who are able to pass away
at home, has increased. So this is really a
positive trend which shows, which reflects the sort of,
increased patient centeredness in medicine, and helping
patients achieve their goals. Lastly, this is the number
of people dying in hospice. And for those people who don’t know, hospice is a service
that’s provided by Medicare in which, for patients who
are truly at the end of life, in which services can be provided either at home, or in a facility. It’s a covered benefit which allows for nursing providers to come
visit patients at home and be able to take care of them there. And while you can see that
there’s been a gradual uptrend, that number still remains pretty small. Then the next thing I
wanted to do was really try and add some context
to what this means. And so what I did was,
along with co-investigators, was compare cardiovascular
disease to patients with cancer. Cancer is a great reference group because not only is it the second
most common cause of death, after heart disease, but
really, if you look at how end of life care has been
designed in the United States and around the world, it’s
been designed really to take care of patients with cancer. But there are important differences. Patients with cancer, which
is the green bars here, are much less likely to die in a facility, or either a hospital or a nursing home, and are much more likely to die at home. This reflects a few different things. One is that patients with cancer do have a more predictable survival. Patients, for example, who
have metastatic lung cancer, or metastatic pancreatic cancer, we know that their survival’s limited, but sometimes that can
be really hard to tell in patients with heart disease
because it’s just so hard for us to know when are they
truly at the end of life. And the next thing we found was, and this is on there on your left,
is overall causes of death, and on your right is, that’s in hospice. And what you see is that
even though heart disease accounts for the majority of
deaths in the United States, it continues to be underrepresented in patients dying in hospice. So clearly, there is a big discrepancy. And what you can see
here is that cancer is overrepresented in this population. And what this means could be
couple of different things. One could be that perhaps, hospice is not the right
place for many patients. Perhaps, because hospice has
been designed so specifically for patients with cancer,
it isn’t able to fulfill the needs for many of our
patients with heart disease. So having established this,
sort of, broad overview of what really is going on with
patients with heart disease, the next thing that I wanted to do was actually look at what is, in fact the quality of care
that these patients receive. And they do in fact go and receive end-of-life care services in hospice. So this was a project that we did from the Get With the Guidelines
Heart Failure Registry, which is housed here at DCRI. And the question really was, one is, how many patients were hospitalized with heart failure in the US? We only looked at Medicare
patients actually end up being discharged to hospice. And then, what happens to them
after they are discharged? So this analysis which is
currently being reviewed, but was presented at QCOR, showed some interesting
and important findings. One is, we found that only
5% of Medicare patients in this national registry
of patients who were hospitalized with heart failure, were actually discharged to hospice. And these are data between 2005 and 2014. And even though this has
increased from about 2% to 5%, so there has been improvement, it still remains pretty small compared to patients with other diseases like cancer. This is perhaps, the most
important thing that we found. We found that the median
survival of patients who were hospitalized
with heart failure and then were discharged to
hospice, was only 11 days. And what that means is, that patients, these patients who are at the end of life who are sent to hospice,
they’re only living on an average of only 11 days. Hospice is a service that is indicated for patients with a survival
of about six months and lasts for about two years, so really, we’re leaving a lot of this benefit without being used for patients. And perhaps, we are too late in when we send patients to hospice care. And this is actually
represented very importantly in this number, in which we
found that 23% of our patients actually died within the first 48 hours. And what does that mean? What that means is that
you’re a family member, or you’re a patient,
you are critically ill, you’re at the end of life,
you wanna be at home, you wanna receive comfort management, but you’re really in the throes of death. And within the first 48 hours
of you leaving a hospital, going home or to a facility, getting everything set up, you pass away. And one of the things that
we found was that patients who were sent home in this state, also had worsening symptoms. So really, even though we are
seeing this gradual increase in patients going to hospice,
we are also seeing that perhaps, how these
patients are being managed is perhaps, not optimal. And that in some ways we are actually burdening families
and caregivers more, by discharging really, really
sick patients to a setting in which we may not be able
to manage them appropriately. Here are some time-to-event
curves for these patients. So on the left, you see mortality. The red line is hospice
and then this blue line is advanced heart failure and
then, the black line is all patients with heart failure. What you see is that
patients who go to hospice are much, much, much more likely to die, even than patients who
have advance heart failure, who are otherwise very sick. But what we do see is
something that’s important, is that only about 10% of
patients who are sent to hospice, are alive at one year. And this is another thing that’s important when we’re looking at hospice. We don’t want to send people
who are too well, to hospice; who are going to outlive the benefit. And what this shows is that
this rate is very similar to patients with diseases like cancer. The other challenges that
are very specific to patients with heart failure and heart disease, is that they have a very high risk of readmission to the hospital. The whole idea of hospice
is that people will forego hospital-based treatment
so that they can be at home and be managed there, so that they can be surrounded with family members. But what we find out is that
patients with heart disease, sometimes they just are so symptomatic, they’re having so much
difficulty breathing, or they have so much
extra fluid in their body, that hospice just can’t take care of them. Or that family members are just not, have not accepted the fact that
this is a terminal disease, so many patients will actually give up their hospice benefit and
come back to the hospital because that’s what they’ve
always done when they’ve had, when they were troubled
with heart disease. And we find that about 5% of patients who are discharged to hospice, actually do come back to the hospital. This is more than patients who have other disease, like cancer. And that the younger age,
and African-American race, are the biggest risk
factors for this happening. We went a step further and
actually looked at differences in patients who went to home hospice, versus getting hospice in a facility, like a nursing home or a hospital. And we found that patients who went to a hospice facility were even more sicker. That 35% of them died
within the first 48 hours. Not only is this terrible
for the patients and the family members, it also can cause a lot of financial distress
for these organizations. In hospice you get paid
on a per day basis. But most of the facilities,
they’re spending a lot of money upfront, on getting these patients in, and getting their care set up. But if patients die
within the first 48 hours, they end up spending a lot
of money getting them into hospice, but are not able to
accrue any benefit over time. And then lastly in this analysis, we found that there is a lot
of variation by hospital level in how many patients are
discharged to hospice. And this is an area we
want to understand further. What are the differences
between hospitals that are sending many more patients to hospice versus those that barely do so? So now, let’s show you
couple of other issues that are very specific to this heart failure and
heart disease population. That I had the opportunity,
with great collaborations, to be able to look at. So one of the things that we see a lot in patients that have heart
failure, which is different from other disease, that they
have a lot of comorbidity. So if you look at patients
who have Medicare and have heart failure, they have on average, five comorbidities. So heart failure is just one
piece of a complex puzzle. They might have COPD,
they might have diabetes, they may have renal disease,
et cetera, et cetera. So these patients have
a lot of comorbidities. And one of the things that has emerged as a potential therapy for
patients with heart disease, with advanced heart failure, is something called an LVAD. Now, what’s an LVAD? So an LVAD is basically
a mechanical pump that is surgically sewn into a patient’s heart. And what it does is, it
basically augments flow. It’s like a pump, it does what
your heart is unable to do. It pumps blood to the rest of your body. And this is something that’s
become a viable option for patients who may have
not been able to receive something else that’s definitive,
like a heart transplant. But many of these
patients have comorbidity, so we wanted to understand, well, what is the effect
of comorbidities on how patients are able, how
patients process this decision, and then kind of, go with it. This was an analysis done
from the Decide LVAD trial, in which Duke enrolled patients as well. And so, the green line, is
patients who had comorbidities. Blue line, is patients who
didn’t have comorbidities. And these are patients
who were considering LVAD. Some of these patients
ended up getting an LVAD, some people didn’t. And what we find is, that at the start, when people are considering this decision, people with comorbidities had
much higher decision conflict than people who did
not have comorbidities. And this makes sense. If you have comorbidities
and you’re considering a big, surgical operation,
you’re worried about so many other things, not
just your heart failure, but all these other things. But what we find is, that over time, that difference actually disappears. Meaning, that either people just get, that few people actually
regret their decision whether they had comorbidities or not. But the thing that we do find,
is that patients who have comorbidities, they end up having very high levels of struggle with illness. So, I mean, if you look at
people without comorbidities, their struggle goes down,
while the patients who had comorbidities, how stressful
their illness is to them actually stays up. And this also makes sense in retrospect. If you have diabetes, and
if you have renal disease, and you have COPD, and you get an LVAD, your LVAD can make your
heart failure better, but can’t do anything
for these other issues. So we feel like these data are important when we’re counseling
patients for these procedures, is to take into account the level of comorbidities they might have. The other thing that I was interested in, was also assessment of prognosis. Which is to say, how good are doctors at predicting survival in patients. If you see a patient, how good are you at getting a sense of how long
does this patient have to live? If you’re a patient or a family member, that’s the single most
important thing that your physician can tell you. You’re constantly asked:
“How much does dad have?” “How much does mom have?” Because they use that information to make all these decisions. So we performed this
survey in which we gave standardized cases. We used a lung cancer case
as really as a control, and then, two different
types of heart failure. One in which the heart-pumping
function isn’t weak, it’s just stiff. While in the other one,
in which the heart-pumping function is actually weaker, and we looked at it by specialty. So we found that cardiologists
were less accurate in assessing prognosis in
patients with lung cancer. This is not surprising compared to cancer doctors and
general medicine doctors, Interestingly, we found
that cardiologists were actually no different
than general medicine or oncologists, when it
came to the HFpEF case, which is the sort of, heart failure preserved ejection faction, where in fact were more
accurate when it came to heart failure with
reduced ejection faction, in which the heart-pumping
function is reduced. But here is the more interesting finding. So these are just raw estimates
that everyone gave for these specific cases. And if you look at the blue bars, which is the cardiologists,
they’re lower every single time, for every single estimate. Meaning, that when they see a patient, they’re much less likely to think that this patient’s gonna die,
than any other specialty, perhaps suggesting that as a specialty, we’re just optimists, we just
see and we hope for the best. And that may be because
there is something to be said about survivor bias in
the sense that people who you thought were gonna
die, but then didn’t die, are people you’re gonna
see much more likely, back in clinic, back in the wards. You’re exposed to them a lot more. As opposed to people who die sooner, your exposure to them is limited. But it may also reflect the nature of what we do in cardiology. A lot of times when patients are sick, there are still chances that
you can actually recover them. This may not be true if you, for example, have like, metastatic pancreatic cancer, or something like that. And lastly what we found is, this is Likert scale, so lower is better, is that overall physicians
are just more comfortable discussing palliative care in cancer, as opposed to heart
failure, so really this comes up as a big limitation
in whatever happens to these patients downstream. And lastly, one of the
things that I wanted to do is actually go to the hospice providers, and nurses and administrators. The people on the front
lines who take care of these sick heart failure
patients in the community, and really ask them: “What are the challenges
that you’re having with “taking care of these patients?” So first, one of the questions
that I asked them was in two different ways. First I asked them: “What are the most common
symptoms that your patients have?” And the distribution of these
symptoms is very similar to what you know, as a
cardiologist, I’m used to seeing. So almost everyone had
shortness of breath. You saw a lot of leg
swelling, which is because fluid collects in the lungs in a lot of heart failure patients, fatigue, anxiety, and then there’s a drop off. But then I rephrased the
question, and I asked them: “Well, what is the most challenging, “what is the most difficult
to manage, symptom?” And then, the distribution
is very different. In fact, what we classically think of as a common symptom of dyspnea, which is shortness of breath, was in fact, not thought to be very challenging. But other symptoms like anxiety, fatigue, ascites, which is accumulation
of fluid in the belly, depression and confusion,
were actually much more represented when it came to how difficult these symptoms were to manage. Showing perhaps, that these
are areas that we can focus more on, with regards to
providing resources and also educating our hospice staff. The one thing that’s
notable here is that pain, which is really a sort
of, fundamental symptom in patients with cancer, is actually not seen that commonly in our population. I asked them what are the most
fierce challenges faced by caregivers for these patients, and difficulty managing symptoms,
was the most common one, followed by hospitalizations. Again, something that’s very
unique to this population, that we just don’t see enough in patients with heart failure, with cancer. And the last thing that we asked them was about something called
palliative inotropes. There’s a type of
medication called inotropes which increase how hard
the heart squeezes. And in this population
we found that only 42% had ever used these medications, which can sometimes be a
very, very important thing for these patients at the end of life. And then we looked at what
is the most common reason why inotropes are not used. And most common reason was in fact, that the agency just didn’t
cover these medications. So we knew, even though we spend a lot of money on healthcare, overall
when it comes to hospice care, the difference between losing money and gaining money is so
little, that many give up what could be potentially,
therapies that could help patients be more comfortable. So this is something that’s an area for future policy reform and payment reform. And here’s some of the open-ended. So one of the things we did
at the end of the survey was because we feel like so
many of these challenges, you just can’t put them,
you can’t check a box to show what they mean,
and I really wanted to get a sense of what
are the real challenges that providers are having,
that the hospice nurses are having when they take
care of these patients, so we had some open-ended fields. And they gave some very
sobering comments there. So the first one here,
one of the nurses wrote “Heart failure patients
tend to have ups and downs “which makes it challenging
to the patient and family to “understand what to expect, what’s next.” And then she wrote: “When
is this flare ‘the one’ that “he or she doesn’t recover from.” And I felt like this was
such an interesting and insightful question, because really, we don’t have the answer to this. These patients have so
many ups and downs that we just never know which down is gonna be the one that takes them. Another one wrote: “Our biggest
challenge is getting them “off the ‘go to the ER’ train and “accept their terminal nature.” And think about it, these
patients have heart failure, whenever they get sick, what do they do? They call 911, they go
to the emergency room. And now suddenly, you’re
asking them that you know, now that may not do more
good and now you really have to be able to manage
your symptoms at home. So that can be really,
really tough as well. And we in the hospital,
can do a better job of preparing our patients
for this transition. And then, this comment was interesting: “Our rural patients are
so afraid of morphine.” “The patients that actually
need them are too terrified to “use them, so they go
back to the emergency room “when they have an exacerbation.” And this is again, another side effect of the opioid epidemic. That even though we know
that so many people die of opioids, there’s a lot of
people who actually will benefit from them. And yet at this point, there’s
so much confusion about what their role is and
what their benefit is, in this population, it can be really, a magic drug in helping
patients be comfortable. So this was an interesting comment. So I’ll skip this, and
then one of the other things that we’ve looked at, as well, is looking at racial disparities. So this was an analysis
that is working on, that looked at location of death. And this was from the PAL-HF trial, which was also done here at DCRI. We found that the only
factor that was related to more people dying in the
hospital or in the facilities, actually African-American race. And this is very similar
to other patients with other disease states like
cancer, and certainly something that we can do better at. So the question is, we
know what’s going on, well, what is the way forward? And one of the things
that I’ve thought about is perhaps, changing how
we look at end of life care for patients with heart disease. Instead of looking at
the last few months or years of life, we start from
the beginning and look at how things like race and gender
can cause you disparities that actually end up having an
effect at the end of life. We know, for example, that
women, even though they live longer, they have more
years with disability, so we need to broaden our lens. And then, if you go into early adult age, what we know is that early
onset of chronic diseases due to poor lifestyle
choices, obesity, et cetera, can have this lifelong effect
of increased disability and sickness down the road. And then furthermore,
looking at things like comorbidities, disability and frailty, again, if you just focus
at the end of life, we’re seeing what has been
the result of decades of risk factors and diseases. So really, I think one
of the key things that makes this different from
cancer, which is an acute thing that starts and then that’s your journey, the journey with heart disease really starts from the very beginning. So couple of things that
I think are important is developing more quality
metrics at the end of life, addressing disparities
and how people receive end-of-life care and
how they can access it. Improve delivery by having
payment reform by giving hospices more resources
so that they can do what needs to be done to keep
these patients comfortable. To improve education, not
just of hospice nurses and providers, but also
physicians, so that we can do a better job of
transitioning these patients when the end is near. And then lastly, to really
increase research in palliative care and to really prioritize patient-centered research, specially when it comes
to the end of life. Thank you. (applause) Questions? – That was great. So your story in the beginning,
about your optimism and the untimely death, or
the more rapid death. Link that back to us on what you’re, how does that fit in with
your overall philosophy now, on the topic? – I mean, there are so
many things that the, I think one of the things that’s clear, you’re in the CCU, we
can’t avoid bad outcomes. That’s not possible. People will still die. Things that we do,
especially for sick patients, will have adverse events consequences. I just felt like as a team,
we just didn’t prepare this patient well enough. I don’t think we were prepared. I can’t say for everyone,
but I personally, felt like I just didn’t get a good
sense of the different ways that this patient’s outcome could be. I was so personally fixated on, this person has this problem, if only we can fix this problem,
everything will be okay. That I didn’t think about that
things may not go perfectly. We never talked about palliative care. We never talked about
what her wishes might be if things didn’t go the right way. And I felt like it was really a, I felt like her story was cut short and I felt like we, as a team, I
personally, as her resident, and who is in some ways
very passionate about this, was just not prepared for that in any way. And I think that’s another aspect of this, that is important is that I
think a lot of end of life care, is leading to a lot of physicians, it affects physicians, it affects how we think of what we’re doing. I think, and there’s
data that suggests that it leads to a lot of burn
out amongst providers, especially in the ICU and nurses. So. you know, I think that we,
I think before anything, I think that we could
have been better prepared. We could have been better
prepared as a team, and we could have done a
better job in preparing her for any type of outcome. And we just never had, we
just had this one narrative in which you do this
procedure and you’ll do fine. We never talked about, well,
what if this doesn’t work out. What if this isn’t a bad idea? What are the other alternatives to this? All right, thank you, everyone.