How Insurance Works

Quality Assurance For Drug Therapy – Module 7, Session 8

>>William Douglas Figg:
We are honored to have Dr. Charles Daniels provide our next lecture. Chuck is pharmacist in chief at the University
of California, San Diego, Associate Dean for Professional Practice, and Clinical Professor
in the Skaggs School of Pharmacy at the University of California, San Diego. Chuck received his Bachelor’s of Science degree
in pharmacy from the University of Arizona and completed a residency and hospital pharmacy
at the NIH. And earned a Ph.D. in social and administrative
pharmacy at the University of Minnesota. He has held pharmacy leadership positions
at the University of Minnesota and Pharmacist in Chief at the NIH Clinical Center in Bethesda. I’m confident you will enjoy today’s lecture.>>Charles Daniels:
Hello, I’m pleased to be with you here today. And I’d like to introduce myself. My name is Charles Daniels. I’m the Associate Dean for Professional Practice
here at University of California, San Diego. Skaggs School of Pharmacy and Pharmaceutical
Sciences. I’m also a Clinical Professor of Pharmacy
and the pharmacist-in-chief here at U.C. San Diego Health. I’m pleased to be able to share my ideas and
thoughts with regard to medication use and medication use quality. So I will be presenting this with a focus
on certain trends and changes. But with the basic element around how medication
use quality is tracked and measured. I will talk about process and tools for monitoring
and improving medication use quality and outcomes. And I’ll just kind of launch into the topic
by chatting a little bit about the medication use process. First of all, it’s a complex system. We’ll talk about that in a moment. Because it’s complicated, there are opportunities
for error and error could mean mistakes or it could mean opportunities to improve medication
use by better drug selection and prescribing. In the end it impacts patient care and the
outcomes of the patients that we are attempting to reach. The process improvement globally is requires
a lot of focus on systems. It’s data driven. And typically it requires an iterative cycle
process. Let me talk for just a moment about the medication
process as it’s sketched out and then we’ll chat some more about the medication use process. So, when you look at this diagram, essentially
there are multiple general categories. The first of them is gaining history. The second is obtaining and documentation
of the medication history and deciding on what prescribing should be done. After that, there is some method of transferring
that information and that request to the people that will be administering or moving forward
with that order. There’s a relatively complex process in pharmacy
that includes some manual and physical related activities but also some activities that are
related to understanding the patient, figuring out whether or not the drug dose is correct,
and whether or not it needs any adjustment before it’s ready to go. Following that, there is another cycle of
both administrative and clinical activity that includes patient education. All of that happens before the patient gets
their dose. And that’s only on the inpatient setting. If you look in the ambulatory or clinic settings
it can look even more complex. The point of going through this is I want
to demonstrate that because there are many steps it gives multiple opportunities for
analysis and multiple opportunities for improvement of the process. Now, the Shewhart cycle in the quality improvement
activity, which is the slide that I’m looking at now, has really four steps. And this is classic. It’s not related to health care. But Dr. Shewhart created a concept that is
significantly has worked well in multiple industries. But the improvement process looks something
like this. The first step is a planning stage, define
the data which is important and available, define what new data might need to be collected
in order to do this correctly. Plan the change or essentially what the test
is the interaction or the intervention that is going to presumably change the results. Step two is implementation or pilot stage. And during that time period, it’s the — when
the change that is proposed to improve the activity goes in place. Step three is observation. In other words data collection and at that
point in time the ability to look at whether or not the intervention that you’ve created
has done its job. In step four essentially evaluate the data. What’s important about the Shewhart cycle
is not exactly the steps but it’s the fact that it is billed and implemented as a cycle. So if at the end of the first cycle you get
the results that you were looking for, then good. Most times it requires a second cycle through,
with either major or minor adjustments to the original intervention plan. So, in looking at this slide what I really
wanted to share with you is that data is the driver whether or not it’s medication errors
which this set of slides or these two graphics include or whether or not it’s optimal outcome,
data is the critical piece. And different ways to look at the data, some
of which are relatively standard. And if you look at this simple diagram here,
you’ll see that it’s really just looking at what’s going oo not — with no adjustment
for either number of patients involved. It just looks at how many incidents happened
or how many activities events happened. And that typically is a launch point. But this slide which really is the Shewhart
run chart is designed to actually start bringing applied statistics into your quality metrics. And in this particular case, using an upper
and lower control limit helps define when particular counts are statistically significant
or whether or not it happens to be just the normal variation that you would see in data. So, with that in mind, this becomes the beginning
of how to implement a statistically driven quality process. You know, I’m going to take one quick look
at a couple of documents slides that are related to computerization of the medication order
and administration process. So, one of the things that came out of some
of the early studies into medication safety and quality were designed to look at ways
to reduce variation by forcing the process to have fewer choices. One of the presumed saviors of that was computerizing
the order entry process design and we’ll talk more about this later but designed to improve
the process, reduce the — improve the standardization and reduce the variations that could be avoided. So, there were multiple studies, I’m not going
to go through all of these, but there were multiple studies that looked significantly
at medication errors, specifically related to computer order entry. And in fact, a particular study that goes
back to 2005 really identified that while there may be some improvement associated with
the use of computerized order entry as a medication quality improvement activity, that it also
generated new types of errors. And this study was a reminder to all of us
that sometimes there are unintended consequences and the more times that you look at that cycle
and ways to improve it, the more likely you are to be better. Now, there was an interesting corollary to
this one and that’s this study that was about the simulation of technology impact. Now, that was again a prospective modeling
type of approach. But based on the data that they had available
at the time, they identified that computer implementation as part of the medication order
entry process was likely to save approximately 1,200 days of excess hospitalization for this
particular study site and they identified as 1.4 million in associated costs. And again this is for one particular site. So, the important point is that there are
opportunities to make things better. The computer process does help in some of
the variation but it’s also creates some opportunities for failure. So I’d like to speak for a short while about
medication use evaluation. The reason I use this as a central part of
quality of medication use is because it stands as the performance improvement method of choice
to focus on evaluating improving medication use processes and improving patient outcomes. So there’s a long list of categories of things
that might trigger a medication use evaluation. This slide gives a list of those. I’ll just point out that some of the examples
might be new drugs things that have been added that may be associated with the disease states
that are prone to problems if it impacts a large number of patients. Those are all categories. And last but certainly not least is the cost
or the expense of the medication to you, the patient, or to the system. So, with that in mind, I’ll just mention that
there are clearly some opportunities by looking at pretty basic information that tells you
that there might be a change in what’s going on or how the drug is being used. Change may be good or it may be legitimate
or it may represent a breakdown somewhere. But in these particular cases on this slide,
if you look at either antifungal antibiotics and the change between those two years, the
last two years on the chart, and antivirals and the change there, those are the areas
where you might be inclined to say with a change like that something’s different. So, remember what we’re really looking for
are items that would highlight us to particular areas of interest. The ready access to evidence based guidelines
is an important element that has changed a little bit of the landscape of being able
to do medication use evaluation. The reason is that evidence based guidelines
provide the foundation for whether or not medication is being used effectively and whatever
the organization is, large, small, or very small. The question is are you using it the way the
clinical evidence supports? So, this is national guideline clearinghouse
within AHRQ is an important source. And they capture not just government based
guidelines but things done by many of the major clinical academic units, chest surgeons,
and internal medicine from around the country and around the world. So, this is an example of one of the existing
evidence based guidelines that is available online right now. And the reason I bring that up is not because
I’m going to speak more about VTE and non-surgical patients but just because it’s a typical kind
of an item that frequently begins to create the foundation for what is perceived, what
is expected to be the best evidence based use of a medication. Those become the criteria that can be used
by any organization that wishes to do a medication use evaluation and it provides a source that
can be used to create the criteria. So, I will go now into a couple of specific
examples of MUE activities from different organizations. And I’ve started with this slide because this
is a snapshot from the results, or the presentation of the results in this one. And what is important is that this guideline
starts out by identifying, this MUE starts out by identifying that the guideline recommendations
from CDC, from the World Health Organization, and from the Infectious Disease Society of
America have provided the foundation for what represents quality or appropriate use of this
medication in this patient group. The second point I’ll make right now is that
it includes two elements. One is who it should be used in and the other
is what should be the correct dosing that goes with that. And as we get, you know, a couple of slides
up we’ll see the implications of knowing that there are multiple criteria that go with this
particular set of guidelines. So, without going through all of these documents
I will make the case that the objectives for an MUE are probably similarly designed, set
up, and the study design appropriately created to test the questions in the particular population. So, frequently, maybe almost all the time,
these MUEs are not large, large databases, they’re frequently what would be seen as a
smaller or well-defined database. That could be defined by the number of patients
in a given health system, it could be the number of patients that were hospitalized;
it could be the number of patients that were seen in a clinic over a particular period
of time. But it typically is not very large numbers. They’re typically smaller. So, in terms of setting the objectives and
designing the study itself, it requires appropriate rigor and it should be done in a way that
allows the organization to be comfortable with the results. But they may or may not be as fully scientifically
grounded as one might like to see for a double blind study. So, I’m going to just go through a couple
of the findings from this particular study because I want to talk about it. I want you to understand the reason that these
studies imply ways to improve the quality of the prescribing. So, in this particular slide you’ll see a
pie diagram that basically says that seven percent of 149 patients that they looked at,
seven percent had empiric therapy. There was not a test, a diagnostic. And seven percent had targeted therapy because
they were diagnosed with — they had diagnostic data that said that this patient had truly
had influenza. The remaining large percentage, 86 percent
of the study where patients that would have qualified for Oseltamivir therapy did not
receive it during their hospitalization. So, that, I think, will help set up the conversation
about what’s next. This second slide here from the same data
set really looked at the question of, if they used Oseltamivir, did they use it correctly? And I think again it looks like from the data
that you’ll see here on this part of the chart that largely the answer was, when they used
it, they used it pretty well. On the other hand, there were still a fairly
large population, as pointed out in the last slide, that may have qualified that did not
get it. So, this becomes for an organization that
wants to improve the quality of their medication use, this becomes sort of a go to point of
the slide. You’ve studied the data. You’ve looked at the results. You’re now in a spot of saying, so what? And that’s what this type of slide is designed
to do. And if you look here in this corner you’ll
see strategies to improve and that becomes the next steps for the organization in order
to be able to be successful in their process or their activity to improve the use. So, I’m going to go through one more example,
an MUE related example, again to give you a sense of how this can be used to set up
improved medication use activity. So, this was a study looking at liposomal
bupivacaine, Exparel. And this has been a somewhat controversial
product across the country. And I think the question is so if it’s controversial
then why don’t you study it and look at what you’ve done with it? So, the study objectives that were presented
here really focused on whether or not if you used this particular product, the Exparel
versus the alternative therapies that have been in standard use for a while, whether
or not you improved post-operative opioid requirements, improve pain scores, or impacted
will make the stay after surgery all of which are important in terms of whether or not you
wish to use this drug which adds more to the cost of the hospitalization. So, I won’t go through this slide in a lot
of detail. I’ll just point out that the key point on
this slide is this duration both of the other more traditional local anesthetics, bupivacaine
or ropivacaine have a substantially shorter duration. And the rationale for use of this more expensive
agent is that you — the duration of the anesthesia provided after the surgery is — provides
longer duration, less pain, the ability for patients to get up and move faster, less need
to use opioids to treat pain during that therapy. So, that yellow or red highlighted box up
to 72 hours versus two to eight for the other alternatives becomes sort of the critical
question. So, again this study design, while with limitations
based on the number of patients, was designed to try and answer the questions with the level
of statistical confidence that can be generated out of the data. I’ll go through briefly just a couple of the
results slide because I think they really speak to the question. So, we’ve looked on this slide — if we look
at the four categories. The amount of 24 opioid use, the stretch between
24 and 48, and the median 48 hour opioid use, I think that if the question is does addition
of liposomal bupivacaine to the post-surgical procedure reduce the amount of opioids that
need to be used, it does not appear to be a statistically significant difference. And in fact, as you can see there, they are
essentially the same from our data but certainly no statistical significance. So, that’s one question that we felt comfortable
being able to resolve. The second category slide set that I wanted
to share is related to pain scores. And if you look here across there with the
darker blue and the more aqua color bar as being the liposomal bupivacaine and the alternative
standards that we talked about earlier, you can see that there is no data to support the
fact that this particular liposomal product provides better pain score, if post-operative
pain scores are what are used. And there’s no statistical significance to
this difference on any of these. Now, this is the one that actually caught
a lot of attention as it was presented and that is associated with hospital length of
stay. And so, if you go down to this final road
here and look at median post-surgical length of stay, there is a difference between the
liposomal bupivacaine group and the standard of care. Now, it’s not statistically significant. So, in this particular case, the MUE design
would not be allow us to answer that question but it does provide an opportunity for the
organization to look at it and say maybe we need to look a little deeper into that particular
question. So, in looking at the results from a liposomal
bupivacaine MUE, one of the important things that we want to draw out of it is that if
you define the objectives well up front and you follow a lot of data to be able to drive
your conclusions, that you can make the opportunity a useful way to improve the use of decisions
about whether or not particular drugs are elements that are important for patient care
or whether or not they have a different role in therapy. So, it would be — it’s important to make
sure that I share the concept of the formulary. It’s a highly misunderstood process. And essentially the definition that you see
here also has not changed for some long period of time and it’s continuously updated list
of medications and related information representing the clinical judgment of physicians, pharmacists,
and other experts. The reason that’s important to keep in mind
is because formulary is an important way to be able to apply quality decisions whether
it’s about evidence that supports the use of a particular product or whether or not
it’s about the results of particular MUEs that are being concluded. And so, effective use of the formulary is
an important tool to implement quality medication use within whatever size organization you’re
talking about. So, I won’t go through the list, but I wanted
to share with you a longstanding list of, it’s called the safeguards against errors
on high risk methods. And while not all of these items apply to
every situation, one of the things that is important is that as you learn from your medication
use activity, as you learn where the risks might be, wrong decisions, unknown, or holes
in the data, it allows you to be able to start filling in how medications are used in those
setting. And you might be able to, for instance, change
the order entry process for a prescriber or you may be able to change the required information
that has to be submitted to an information system before you can give a patient their
medications at whether it’s hospitalized or in the community pharmacy. So, this list is an opportunity to look at
risks and it works well along with both an MUE as well as the information system, the
ordering process used by prescribers to get the best optimal use out of it. So, I want to talk now for the rest of the
slides that we’re going to go through today, about the changes that have occurred. So, if the last several examples of use of
localized data to try to improve quality, the expanded use and availability of EMR,
electronic medical record related data, is opening some new horizons for improvements
in medication use. So the examples that I’ll use here, I’ll talk
about each of them specifically, but they represent essentially benchmarking not only
across a given organization but to look more carefully at how organizations are using medications
compared to other similar types of organizations. So, in this particular example that we have
here, this is really a measure of, very high level measure, of observed mortality versus
expected mortality. And it’s pretty high level. But as it relates to medication use or anything
else, in the end the question is, what are the important outcomes? I’ve underlined here this particular category,
general surgery. And if you run across this list what you’ll
see is that for this particular observed mortality in this particular organization 1.67 percent
compares to a observed or expected of 0.73 and that puts the — and that compares the
UHC which is a comparative group in this case, median of 0.87. So, for this particular organization looking
at the results of this particular multi-hospital comparator shows them to be in the thirty
sixth rank out of 131 that reported. So, in that, this particular measure is not
medication specific. I’d like to go more into some of the examples
that are out there that are more focused on medication related issues. Now, this is data from a stroke category within
this particular database. And in this particular case I’ve underlined
discharge on statin medications. So this is a measure of whether or not at
hospital discharge, patients had been asked or had been prescribed a statin medication
following their M.I. And this is came again from standardized evidence
based criteria that say that patients are better served. And so, if the intent is to make sure that
all of your patients at discharge after M.I. started with a statin, then having this benchmark
to be able to compare this particular organization’s effectiveness in doing that compared to a
larger group that is — has a similar type of character is an important way to find out
whether or not this organization is meeting its target. In this particular case, it’s, obviously,
it’s better than the average. And I think the idea is look and see if you’re
not where you want to be, then you need to figure out what the best practice is and start
implementing change. This is — there’s two versions of this one,
and I’ll just — this is what the slide looks like as it is and this is benchmarking medication
use associated with kidney transplant BRG. And in this particular hospital, it allows
one to look at the high level numbers but also how they have implementation or implemented
use of specific drugs along with the important piece that is available now that historically
was not and that is medication used in comparison to patient outcomes. And so, this is the same slide but if you
look here at the top I’ve got, I’ve just taken that piece that was at the top of the previous
slide and I’ve blown it up a little bit. So, in this particular case, the target hospital
is compared to Vizient benchmark group which are identified as hospitals that have the
same profile in terms of patient care et cetera and then also all Vizient participants in
this particular database. So, if you look across at the observed mean
length of stay of 35 days, the expected, which is to say some variety of the risk adjusted
number for all that reporting hospitals was 35. And when you look at this, you get a mortality
index that is reported here. But critically what you’re looking at is that
the defined daily dose cost per case, which is a standardizing tool to find a daily dose,
looks like you’re getting about the same length of stay, not as strong a performance on mortality,
and the price of this therapy for these patients is substantially higher than the benchmark
group. So, this becomes an opportunity to look at
it and say, here is something that we need to have a look at and if you drill down into
the bottom half of this chart, you see some opportunities that are out there. But this is the intent of doing this type
of benchmarking is to look at actual data on not only how much medications were used
but impact on outcomes like the stay, mortality, there are multiple others. So, here’s another type of an example that
is now available out there. So, if you are — if you identified a hospital
that you’re interested in within this database, you can even get a list of patients that are
identified in your medication use activity that had, in this case, potentially inappropriate
for patients over 65 years of age. And so the column on the left hand side is
related to the Beers criteria and I’ve underlined benzodiazepines. And in this particular institution, you can
see that there is, would appear to be an opportunity to use less benzodiazepines compared to the
Vizient which is the reference group target in this one. Again, the intent is to look at this — these
types of data and say, is there an opportunity for us to improve? On this particular case, if this data is presumed
to be relevant and appropriate then one might want to go back and have a look at criteria
for use of benzodiazepines in geriatric patients and decide whether or not the additional risk
is well-suited or whether or not there ought to be an intervention to try to reduce that. A couple more examples from different globally
available or some cases proprietary databases look at opportunities. In this particular case, it’s a liver transplant. And I’ve underlined rabbit HEG. And if you look across there you can have
a look at the HDO is the individual hospital, the UHC benchmark group, and then which would
be those that are have best practices, and then the all UHC. And if you look across there again you’ll
see that there are some differences in the frequency of the use of that particular drug. And if you run down through the rest of the
table you’ll see more cases where there is variability. And while variability isn’t always bad, it’s
always something to have you look at it especially if you have a target group, in this case the
benchmark groups, that really is expected to have performed higher at a higher level
than the average. So, a couple of real quick points that I wanted
to make. So, if the question is knowing that you have
the kind of data that we’ve looked at in the last few slides, is this something you can
act on? This is a summary from a study that was done
a few years back by some of our colleagues at the University of Kansas Medical Center. And the point of this slide is that the intervention
that they used, if you will, is they started providing feedback of the types of similar
types of presentations of the last few slides that I presented to you. They shared them with the service leaders
and as a result of that they were able to change the pattern with the UHC line, the
red line being sort of the global benchmark that for all people reporting and the blue
line representing this particular medication use in that target hospital, the University
of Kansas. So, I want to make the case that once you
know where the opportunities are which is where some of the big data can help as well
as the MUE data that we talked about earlier can help, this is — there’s data out there
that says that you can use that information to change behaviors. Just a couple of last examples that I wanted
to share with you. The value of some of the databases that are
out there allow you to be able to look exactly at a small group of hospitals, a cluster that
are of particular interest to the group. And so over on the far left hand column that
you can’t see has been removed for, to protect the privacy of the individual hospitals, but
each of these rows down across here represents a different hospital that has similar characteristics. And if you look into the red box area I think
what you can see is again using that defined daily dose standardizing tool, what you can
see is that not only is the cost of the drugs used, these high impact drugs which is defined
by this particular group, Vizient, and if you look at that, you can see that there are
differences in the amount of drugs used for this particular diagnosis group. So these are all for the same patient group
and I don’t recall right now what the disease state was. But you can see that there was a substantial
amount of variability in terms of the cost of the drugs that were used and some variability
in terms of length of stay. Now somebody pointed out to me a few days
ago that length of stay is influenced by many things and it’s true but it may also be influenced
in part by medication use. So, I’m not insensitive to the fact that length
of stay is a pretty crude measure but I will say it’s one of the measures that are being
widely used and we need to look at that as well as some others. Now there are — this is a different metric
that for a study that was done really some years ago. But the reason I pointed out is because when
you look on here what you’ll see is the purpose of this study and this particular diagram
that you’ve got in front of you the chart is to look at the number of patients that
are — that were using NovoSeven, a high cost, high risk drug for prevention of bleeding
for which it is not indicated, there is minimal data on that, versus those where it’s used
for treatment of active bleed for which there is good data on. And so this allows each of the hospitals in
this compare group to be able to decide if, where they sit. And so these are all, you know, specifically
numbered so that no names of the institutions are there. But this kind of data would tell you if you
happen to be the hospital that is right here number one, on the use of prevention of bleed
that you may have an opportunity to improve your use of that drug which impacts not only
outcomes and patient results but also impacts the cost. Lest you think that cost might not be a fair
item to include in medication use quality, all I will say is that it is widely incorporated
into payer metrics now and failure to be able to provide cost effective care including results
on patient outcomes, length of stay, failure to be able to be competitive on that one does
impact reimbursement rates so organizations typically, whether or not they’re accountable
care organizations, or HMOs, or hospital systems are looking carefully at that data. So, and again, there is, this is another example
of pay for performance related material. So, just a couple of last items I think we’re
down to the last few slides now that I wanted to share and that is when you get into this
question of how much can you draw out of some of the larger data sets that are now available,
large frequently and widely available to proprietary or group settings, the ability to look at
data and the flexibility to dig into this becomes substantially broader than we’ve ever
had before. So, let me summarize real briefly right now
what I wanted to share with you. I’ve hopefully given you some pointers toward. Medication use and the quality associated
with medication use is a fairly complicated process. It is error prone. It’s been shown over the years to not be easy
in spite of the fact that it seems that it should be. Medication use evaluation helps direct toward
opportunities to improve the use of that medication and drug use can be improved and we’ve demonstrated
that through some of the slides and studies that have been presented to you today. There are substantially more pieces of data
that are out there that are transparent, visible to the public. I don’t — I think in the interest of time,
we’ll not talk about those today. What I will say is this, it’s been my pleasure
to share these several minutes with you. I hope you’ve enjoyed and learned from the
lecture. If you have questions about this or any of
the content please go to the course coordinator and be very confident that feedback will get
to me if there are some things that you saw or opportunities that you would suggest for
improvement. Thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *