Using Hospital Discharge Data to Improve Hypertension Control

Using Hospital Discharge Data to Improve Hypertension Control


Julia Howland, is the epidemiologist with
the Illinois Department of Health and she, as we said in our introductions, is joined
by Mary Driscoll and Sandy Anton. So we’re very glad to have you all on this call. And
so Julia, I’ll go ahead and turn it over to you. Okay, thanks, Emily. So thank you, everybody,
for joining us this afternoon. This presentation that Sandy and I will be giving today is really
comes out of the work that we’ve been doing with our community as part of the ASTHO Million
Hearts project and then also, as Emily mentioned, in collaboration with many other states, some
of whom are on this call, to develop an algorithm and a way of thinking about discharge data
that is effective and useful for communities in identifying high risk hypertensive groups
and controlling that hypertension. So we’re very pleased to be here with you today. We
also just want to recognize Geoff Dougherty who is an intern who worked with us quite
a bit on this project. So we appreciate his efforts on this as well. So just as a bit of background, we came to
this project and we view this data analysis as part of a collaborative effort between
our ASTHO Million Hearts project – obviously that’s why we’re here talking with you today
– and also as part of this community transformation grant Healthy Hearts project, which is what
Sandy and I started with working in these two communities. So as part of the Healthy Hearts project,
we’re actually pulling data from the primary care clinics’ electronic health records to
identify hypertensive patients from the clinical record. So we viewed this use of discharge
data as a way to complement the data we were pulling from primary care clinics’ electronic
health records. So we’re working in Macon and Peoria counties
– for those of you who aren’t familiar with Illinois, those are counties in the central
part of the state. And one thing, when we were planning for our grant and planning for
our ASTHO Million Hearts work, one thing we thought that we could bring is Illinois Department
of Public Health, and in particular the Division of Patient Safety and Quality, was active
to the hospital discharge data. So we here at the Division of Patient Safety
and Quality are the folks who collect and analyze and release that data. So we thought
that perhaps our access to that data and use of that data would be an asset to the communities
and so it’s something that we sought very early on about how to use that in our Million
Hearts work. I think one thing that the discharge data
brings that other data sources maybe don’t is that it is current data. So we receive
data with a delay of approximately six to nine months. And it’s not sample data. So
that’s something that sets it apart from data sources like UDS data or behavioral risk factor
surveillance system data, or even the data that we’re pulling from primary care clinics.
I mean, those are all subsets of a population, whereas the data that we receive from the
discharge data, it’s every single discharge that happens in a community or in a state. And I think in particular when we’re looking
at counties like Macon and Peoria that aren’t huge population centers. They both have mid-sized
cities in those counties, but they’re not huge cities. I think the use of non-sample
data is really excellent, because we’re very used to seeing unstable estimates of burden
of conditions like hypertension in smaller cities and I think it can be frustrating for
communities to try to plan programming or assess effective of programming when they
don’t have stable, current estimates of burden or a way to evaluate efforts that they’ve
been undertaking. So I think especially when we’re looking at smaller population centers,
non-sample data has a real place. And then another reason that we were interested
in looking at the discharge data in particular was that we think of hypertension as a condition
that could and should be controlled in the out-patient setting. So when a patient is
being seen in an emergency room, or worse yet, in a hospital, in an in-patient hospital
setting for a condition like hypertension, it really represents probably a failure of
the primary care system or access to the primary care system in controlling that patient’s
hypertension. So it means that we have patients who are
really sick with a condition they shouldn’t be really sick with, and also it means that
we’re devoting a lot of resources in terms of healthcare spending and also in terms of
patients being out of work, being unable to participate in their normal activities, because
of a condition like hypertension. So it really represents the far end of the spectrum and,
in my mind, an opportunity for the patients who need our help the most. So those are some
of the reasons we got interested in looking at discharge data in the first place. Just a brief word on our methodology: We look
at the 2012 Illinois hospital discharge data records as part of this analysis. I assume
most people on the call are familiar with the discharge data. But we here in Illinois
receive basically a complete record of the patient’s hospitalization. This includes demographic
information about the patient, information about the patient’s payer, information about
how much was charged to the patient’s payer, some information about the sort of procedures
the patient underwent as part of the hospitalization. And also information about the diagnoses the
patient received. We receive up to 25 diagnosis codes per patient.
They are in ICD-9 format. And we apply this discharge data to PQI #7, which is the hypertension
admission rate. I think the reason we were interested in using this rate in particular
– and it goes back really to what Emily was saying earlier about our work with other states
in the ASTHO Million Hearts collaborative – is that we wanted to be able to share the
data with our communities in a way that they could compare to other communities in the
country. So we wanted a nationally recognized measure,
a nationally recognized algorithm that would allow them to compare across communities and
to other states in other communities elsewhere. So that was the reason that we tried to identify
something that had been vetted and used elsewhere. But we did modify this measure slightly. The
most important modification is that we decided to apply this measure to out-patient cases
as well. So the cases in which a patient was treated in an emergency room and then released
without being admitted to an in-patient facility. So that was the most important modification. We also chose to include patients who were
transferred from one hospital to another. In the original algorithm those are excluded,
and we won’t get into the reasons there unless we have a question about it. But that was
a choice we made. So I’m just gonna now show some examples of
how we shared the data with the communities. This example here shows a graph of the rates
of in-patient admissions for hypertension, where hypertension was the primary diagnosis
in the patient’s record, over the four quarters in 2012. So you can see here that it’s just a rate
for the entire county over the course of the year and allows the county to see that the
rate is rising, although probably not significantly. So this was the first kind of graphic representation
of what we were able to share with the communities. We also shared these sorts of charts with
them. These charts obviously show the incident rate, the standard air, the lower confidence
interval, and the upper confidence interval for our two target counties, and then also
for the state. I thought that this was an important chart to share with the counties,
because it allows them to start to understand some of the instability inherent in this data
so they can see the wider confidence intervals and they can also compare their rate in a
quarter to the state, for example, and see if it is or is not significantly different
given the confidence intervals. And I think this is an important thing to
share, if people are interested in doing this kind of work in their communities, because
there is some risk in using something like discharge data for a condition like hypertension,
because the data is relatively unstable unless you have a large population center your working
in. But I think sharing this kind of thing was
very effective, because counties could see the instability and understand the limitations
of the data, but still I think they found a lot value and worth in it. But this I think,
in light of being a responsible steward of the data and sharing this data responsibly
and clearly, I think this is important to share. This graph just shows the rate similar to
the first graph I showed among out-patient visits. So these again are treat-and-release
emergency room discharges for the same county. We found that counties were very interested
in understanding both in-patient and out-patient discharges. So I think showing these separately
was useful for our communities. They really felt like these were different patient populations
and that they could be approached in a different manner. We found quite a bit of interest in the communities
we’ve been working in to demonstrate the payers that were the primary payers on discharges
for hypertension. And I think this is important for a couple reasons. The first is that of
course payer serves as kind of a proxy for socioeconomic status and allows communities
to target particular socioeconomic groups or age groups talking about Medicaid, self-pay,
and Medicare patients. So I think that that’s interesting. And then also if they’re talking to legislatures
or they’re talking to payer groups, this really allows them to demonstrate some true impact
or need. They can say 38 percent of the patients who have a primary diagnosis of hypertension
have Medicare as a primary payer, so Medicare would have real stock in interventions to
help decrease hospitalizations for hypertension. So this sort of graphic I think is really
useful for our communities to consider. This shows the same information, but again
by in-patient. And the reason I’m showing you this is because in my mind if we – I’ll
just slip back here to out-patient. You can see that we have Medicare rate of 38 percent,
and then for the in-patient visits Medicare jumps up to 55 percent. So in my mind this
is a good demonstration of why showing in-patient and out-patient data separately is useful,
because the experience of the patients, the demographics of the patients affected really
are different. And then this, I think it’s probably the most
popular or useful thing we did with the data. This is a map of Macon County and it is broken
down by ZIP Code. And we are showing here the rate of in-patient visits by ZIP Code
in each of the ZIP Codes in Macon County. And people were very excited by this, very
interested in this. And in some cases, in some of the counties, very surprised by the
information we found. So I think that when counties and communities
are planning for targeting interventions or targeting outreach efforts, this kind of information
is very helpful to them. They can see where the biggest problems are in their county and
really target efforts to those communities in particular. Another note on this is if people look down
at the lower right-hand corner of this map, you can see the legend. And you’ll see that
especially that third category, the highest rate category there, it has a very wide range
of 109 to 540 cases per 100,000 population. And so that’s a huge range and that speaks
again to the instability, especially when we’re dividing the data by ZIP Code. The rates
come to be very unstable. But I do think that there’s a responsible
way to report this and so I was very careful to group all of those highest rate ZIP Codes
together in the same color so that we weren’t identifying one ZIP Code as the absolute highest
rate when perhaps there weren’t actually very many cases there and it was just an unstable
estimate. So I think doing things like only including
three groups in a map and trying to keep highest rate groups together and not draw attention
to one ZIP Code in particular is a good way to address some of the instability issues. So we’ve had quite a bit of interest and success
around the use of the discharge data for our communities with the ASTHO Million Hearts
work. There are some, I think, exciting and important next stops we have planned. The
first is that we have – you might have noticed that I have some slides from a different county
included in this presentation. So we do have many more communities that are interested
in this data and excited about seeing this data for their counties and their communities. So that I think is the thing that speaks most
to our success is that we have a lot of interested people are wanting to get engaged and work
with this data in a similar way. So that’s the biggest next step. The other thing that we are working on right
now is, as I mentioned all the way through here, we have only applied this algorithm
to the primary diagnosis spot in the discharge record. We are interested in applying this
algorithm to all 25 of the diagnosis spots in our discharge record and seeing how that
changes things. Obviously it would increase the stability
of some of our estimates, give us some more cases, but there are some problems that perhaps
we are not getting an accurate sample of all hypertensive cases if we include all discharge
codes. So we’re interested in looking at that and seeing how it compares. As I mentioned, people are very interested
in the issue of payers and charges. So we are considering how to best and most accurately
and usefully demonstrate and use the charge data we have available to us. We are working now to learn and apply some
new software tools here at the State Health Department to help us identify patients who
are being seen in the emergency room frequently or more than once in a short period of time.
And we would like to work with our communities to identify these patients and make sure that
they have good access to primary care and try to address some of the reasons why they
might be frequently visiting the emergency room. So that’s another thing that came out of these
conversations is that our communities are very interested in trying to reduce the utilization
of the emergency room for conditions like hypertension, and so we hope to be able to
use the discharge data to identify frequent uses of the emergency room and connect with
appropriate primary care. And then I think the next step that we would
like to do is we here in the Division of Patient Safety and Quality are very concerned about
social determinates of health and identifying disparities and addressing disparities in
our communities. And so we’re thinking all the time about how to capture social determinates
of health in our data analysis. So we are working now on calculating rates by payer,
rather than just showing percentage of payer. We would like to be able to identify a rate
by payer to find the highest risk group by payer. And then also having an overlay of poverty
measure from the census data that appears on the map as well might help to identify
some of those social determinates of health. So these are some of the things we’re interested
in doing in the future with this analysis. So I think now I’m gonna turn it over to my
colleague Sandy Anton, who’s our project lead, and allow her to just finish up the last couple
of slides here. Thanks, Julia. And it is just quickly two
slides. It’s been interesting. We wanted to share with everybody the impact of this data.
Julia mentioned earlier the importance that we’ve seen of integrating this data with other
data systems. This is not intended to be a standalone data source. We are truly looking
at it as part of our tool kit of data-driven change and it’s important to realize just
that. It’s not going to drive everything. It has to be part of the package. What we’re seeing though in addition to that
is that the use of this data and our epidemiologic expertise is driving an ongoing relationship
with the Department of Public Health kind of cementing that relationship and it’s actually
encouraging other counties to participate with us. Once they understand the depth and
the richness of this resource, they’re very interested in how this can actually support
the work they are currently doing. This is helping all counties that we’re talking
to identify new groups or geographic areas to target. It’s helped a couple of our counties
actually drill down to specific ZIP Codes in which they want to conduct some focus interventions
on hypertension screening management and maintenance. It’s been very helpful from the big-picture
perspective. And as Julia said, everyone is very accepting
of the data instability. The fact that it’s a small numbers in some cases, but they’re
nonetheless finding it very, very meaningful to their local application. Julia, can you
forward for me, please? Sure. And then a few lessons: Very high level here.
Everyone wants more data. The minute we show them data, they want more analyses. They want
more overlays. They want a different angle. And it just shows how hungry people are and
the value that the Department of Public Health can bring to the table. We’ve been kind of pleasantly surprised at
how the presentation does matter. Julia mentioned that maps have been very impactful, has both
validated and surprised some of the counties with what the data showed against their perception.
So the presentation is very, very important. And we’ve been pleasantly surprised at how
much the local health departments and the clinicians, the clinical providers, do value
the state level data. They see it as an adjunct to the data they have with their particular
population they serve or their particular local service area. And as mentioned, how important it is that
this discharge data can either refute or validate local assumptions. It has indeed created quite
a few discussions and actually stimulated some work in the direction of how do we target
populations? And again, using this data has helped us be an objective driver for local
collaboration. Health departments and the providers are looking at this data as outside
of their local systems and therefore more objective and is helping to bring people around
the table. So with that, I will turn it back over, Julia.
Any conclusions from your side? There’s our contact information. We’re very glad to hear
from anybody. Julia, anything you’d like to add? No, I don’t have anything at this time. I
do see we have some questions. Emily, would you like us to address those now or wait until
after both presentations are completed? I think you can go ahead and address those
now. Sure. So we have two questions in the chat
box there that I see from Robin. So Robin, so one of the questions is “When you say out-patient
on your side, so you mean just ED? If not, what else is included in out-patient?” So we here in Illinois do receive data from
several different sources in an out-patient setting. So we receive data from out-patient
imaging centers, out-patient diagnostic centers, ambulatory surgery centers, and from the emergency
rooms. So we don’t receive data from out-patient clinics or out-patient physician offices at
all in our discharge data. So in this case we really do mean emergency
room, but we say out-patient, because there are other sorts of services that are being
reported in the out-patient discharge system. But to answer your question, yeah, we really
do mean emergency room in this case. And then how would we capture the mapping
of poverty? So if people are interested, we actually have done quite a bit of work with
mapping here in Illinois. There’s the Illinois public health map, if people want to Google
that and take a look at some of the other work we’ve done. But we often just lay an overlay of rate of
poverty over a community. So for example, showing the rate of poverty in a county, and
then under that showing rate of a disease or burden of a disease in the county. So it
helps to sort of think about both issues at the same time and the same map with hashing
or something like that. You can show both poverty and rate of disease, and of course
you see that there’s quite a bit of overlapping and co-occurrence of high rates of poverty
and high rates of disease. And so we aren’t saying that the disease is
occurring within the same patients as are being seen for hypertension, but just that
we’re able to look at both where there are high rates of poverty and where there are
high rates of disease at the same time. I don’t know if that answers your question. Yeah, it does. Thanks a lot. Yeah, sure. Yeah, we get that information
from census and other sources like that. Okay, thank you, yep. Okay, great. Well, thank you so much, Julia
and Sandy. That was a really great presentation and it’s very clear that there’s a lot of
interest around this type of data, both at your local partners and at the state and national
level. So I think with that we’ll move on to Maryland’s presentation. So I think Vanessa and Sarah, if you’re presenting
– Vanessa Walker Harris is the medical director for the Maryland Department of Health and
Mental Hygiene Center for Chronic Disease Prevention and Control. And Sara Barra is
an epidemiologist for the Family Health Administration at the Maryland Department of Health and Mental
Hygiene. So I will turn it over to both of you. And just let me know when you want me
to advance your slides. Sure, thank you. So we are conducting Million
Hearts in Maryland. And if you’ll advance to the next slide, please, just to give you
sort of an overview of what we’re doing in Maryland as far as the ASTHO project is concerned,
we are coordinating with our internal programs and external stakeholders to maximize the
region impact that Million Hearts really were looking to identify control and improve hypertension. We are engaging multi-sector partners in our
target jurisdictions, and I say target because for this project we couldn’t work with all
the states, but we chose some very select areas in very different geographies and populations
to really see the impact of Million Hearts in different areas. When we did that, we wanted to enhance community
clinical linkages and facilitate quality improvement projects. We implemented a learning collaborative
[beep] which has been very successful to date. We really had a lot of participation and that’s
been fantastic. And finally, we’ve been doing a lot of PDSAs,
including those focused on data, and so that’s actually going to be our focus for this presentation,
is really taking it through a particular PDSA for our data. So can you go to the next slide,
please. Just real fast, these are our implementation
sites in Maryland in red. So we have Washington County in western Maryland, Cecil in the upper
right-hand corner, Baltimore City in the center there, and then Saint Mary’s County, in the
southern part of Maryland. So you can see very different geographies, different populations.
So we’re getting a lot of different impacts with [beep]. If you’ll go to the next slide,
please. So as I said, we thought we’d take you through
one of our PDSAs on data just to show how we’re using this process to help shape our
data and our decision-making process. So we agree with Illinois that hypertension should
be controlled in an out-patient setting. So we are looking at ED cases only. We really
didn’t want to get into the in-patients, because we felt that was sort of too far down the
road for the impact of this particular project, given the short timeframe of the project. So we’re looking just to ED cases and we need
to determine a methodology for measuring Maryland-resident ED cases reporting hypertension. And just
a caveat, we’re able to talk about Maryland residents in Maryland hospitals. We weren’t
able to get data from other states, surrounding states, so if they go to another jurisdiction,
we wouldn’t be able to capture that. But given that these are ED cases and not in-patient,
we felt that was okay. And then we also wanted to obtain a baseline
number of resident cases reporting hypertension both as a primary diagnosis and as any diagnosis.
As Julia said for Illinois, they had 25 diagnoses. In Maryland we have 30. So we really wanted
to see it both as the primary diagnosis as the main reason why they’re at the ED, but
also as any diagnosis to capture those coming in for other things but also having hypertension,
which may be contributing to their primary diagnosis. And so in addition to doing this for the state
of Maryland to really have sort of a state baseline, we are doing these analysis for
partner jurisdictions, which are the four counties in red that we highlighted earlier.
Could you go to the next slide, please? So as far as our doing, we actually did an
initiate ED data analysis. We did determine baselines for Maryland and all of our counties,
but then if you go to the next slide, we found some things that were a little funny. And
so part of really going through our data to understand what’s happening – and again, this
is 2012 baseline data – we found an issue where not all coders were correctly coding
county of residence, which was rather interesting, because not only does that determine the county
you’re in, but actually determined if we considered you a state of Maryland resident or not. So
that was a rather interesting finding. And what we ended up doing is working with
our partners at the HSCRC, which are primary owners of these data, to redefine resident
based on a ZIP Code algorithm, which was found to more reliably code. It was more reliably
coded in the beginning. You’re more likely to have ZIP Code than you were a county entered
in actually. And then using the algorithm, we could place them within a specific county,
even though ZIP Codes crossed lines. By using that algorithm, we were able to get them into
the county where they belonged, and hence giving them the correct county of residence. So if we ___ Act – that’s kind of where I
was leading to. Can we go to the next slide, please? Thank you. So we redefined the definition
of residence and we adopted that, so we now are moving forward with the data owners to
make sure that there are more quality assurance protocols in place to make sure that data
that we are receiving are data that are both accurate and reliable. So that was a rather interesting process for
us as far as a PDSA, and I think was very helpful in making sure that we had good data
to present for this project. And not only this project, for other projects within our
health department. So if we could go to the next slide, please.
So progress to date: So we have our brand new data set with more accurate residence
variables, which has to make at least at the state very happy and also those at the county.
So we have created baselines using 2012 data, and that’s something that we are working on
for 2013 for the quarters leading up to this project, and then with the remaining quarters
in 2013 actually starting to look at what does ED data look like within the intervention
period. And that’s something that we’ll continue to work on through this project. We also have created a hypertension algorithm.
It is actually slightly different than what Illinois did. We also started with the PQI
7, but we decided to make it more broad. The idea being that the PDI 7 excludes those who
have other conditions such as heart failure, and we felt that they should actually be included
in our ED definition, because we felt that providers can intervene with all patients,
even those with heart failure. And that was a decision made with our clinical staff as
well. We did adhere to some of the exclusions of
the PQI 7, including those who are under the age of 18. We excluded our transfers from
other acute hospitals, or EDs, just because we felt like you should be counted once and
not twice. We also excluded non-Maryland residents. As we said, we were solely concerned with
Maryland residents, because that’s our population with whom we’re intervening. And just as a data point, we had talked about
the county of residence being a little hanky, and so we got that clarified. But we were
amazed to see that the records had age gender. Almost of them had race. So a lot of those
basic demographic characteristics were all entered, which is fantastic, because if you
have a lot of missing data then it really makes you wonder about what analyses you can
and cannot do. Could you go to the next slide, please? So
finally some next steps: We are looking at 2013 data currently. We are really interested
in seeing the trend chart to show what our baselines, including those quarters, looked
like prior to this intervention before this program and what it looks like during an act
of the intervention to really be able to tell the story of Million Hearts and if it did
in fact have an impact on the ED data. We are also interested in mapping hot spots
for ED visits within counties. This would be done at the ZIP Code level and it’s something
that we’d want to do in terms of doing it in both in race, because you do want to be
able to compare. So it would be age adjusted rates. But we’re also wanting to be sure that
we have some actual number numbers behind them just to ensure that people who many are
not in their data realm understand that these cases, they can actually go and count the
number of cases. They can actually go and recount and touch them, and so I think that
makes it more real for them if the rates seem a little fuzzy or out of reach, or that sort
of thing, the actually having numbers to go along with rates makes it much more applicable. So finally, just some lessons learned: We
felt that the PDSA process is really a great way to frame our data discussions and to document
how we were able to move through this process and to show some of the changes and actions
that we did. And I think we’re going to continue to work with our partners and keep on sharing
what we have and guess taking in their input in being able to provide specific data as
they need. So does any of the rest of the Maryland team
wish to add anything? Okay, I think that’s all we have for Maryland. Great, okay, thank you so much. I think we
can open it up for questions now for either the Illinois or the Maryland team. So if you
have any questions, feel free. Don’t forget to unmute yourself, and then feel free to
chime in with any questions. Well, let’s see. Paula, I don’t know if you’re
on the call at this point. But I was wondering – I am. Can you hear me? Yes, yes. We can hear you. Okay, I’m sorry. No, that’s okay. I just wanted to invite you
at this point if you have any questions or thoughts on these presentations. Sure, yeah, one question that I do have is
actually around – I actually have a few questions, but this one is around workforce. And I’m
wondering in Illinois and Maryland, as well as in other states, if you have the epi and
biostats staff who really understand the hospital discharge data and then in Maryland and Illinois,
did you have to do any kind of training or really delve into the data to begin to understand
it? Or is this something that you’ve been doing for some time now and you really understood
it already? So for Maryland, we’ve been working with these
data for a number of years. So we were able to really dive in, though we, like I said,
with the PDSA, we were rather surprised with some of the findings and possible key weigh
issues that we found. But generally speaking, we’d been working with these data for a number
of years. This is Julia from Illinois. I would agree
with what Maryland said. Our division’s, Division of Patient Safety and Quality, is actually
the division that is responsible for the discharge data here in Illinois. So it’s a data set
that we’re familiar with and work with for many different projects. ____ have you had any challenges around getting
access to this data? This is Vermont. We used hospital discharge
data on a really regular basis. Our biggest challenge is when we have very poor ___ comes
to hospital care there and some of our neighboring states are a little bit behind in getting
us data. So sometimes our biggest problem is that we’re limited to only putting in-state
hospitals and we can’t take care of the thousand or so people that every year leave and go
to a hospital on the other side of the river. But for the most part we’re pretty well versed
in the use of discharge data. Great. And you don’t have any barriers accessing
it, other than the interstate issue? Or do you have other challenges as well accessing
the data? So when we get our data files, we get all
of the hospitals at once. So I think Illinois was saying they got a quicker turnaround.
Our hospitals are a little bit slower. Also the way the data, the way the mandate is set
up, they have to get it to us within six to eight months or something. And by the time
I can actually get my hands on it, it’s usual about a year. So that would be one barrier,
but for the most part there’s been a system in place for us for a while that seems to
be working. Just to be clear – this is Julia in Illinois
– we don’t get interstate data either. It’s just that we have – I’m sure we have some
patients who are leaving the state, but we don’t consider that to be a huge loss of patients
here in Illinois. And this analysis that we presented here today was really only for Macon
and Peoria counties and I don’t think residents from those two counties would be going to
another state to receive care. Yeah, and here in Vermont we know that at
least 20 percent of our population goes to a hospital in New Hampshire, let alone the
number that go to ____. So for us we know that that’s a barrier we’re always addressing. Yep. I’m wondering how Maryland feels with that.
I live here in Maryland and I know we’re bordered with five states, so how do you deal with
that crossover to other states? So we do have access to some other surrounding
states’ data, and so that is very helpful. I think someone made the comment that it’s
oftentimes behind, and so we found that’s the case for us. We only have access to the
in-patient from other states. We have not been able to access out-patient data from
other states. Really we have an idea of where they’re going,
particularly with those that surround D.C. We know that there’s a fair amount of traffic
going from those two counties specifically into D.C. We know that for other counties
that are on the border with other states that they will go to larger cities, such as in
West Virginia and Delaware, especially for in-patient, but we also know that the way
that the ambulance runs are, they’ll most likely be staying within Maryland if they’re
being picked up by an ambulance. So for ED data, we’re figuring the ED data,
it means it’s an emergency that you’re traveling to the nearest hospital. And generally the
nearest hospital is within the state, except for those counties that surround D.C. However,
none of those counties that are directly touching D.C. are in our Million Hearts collaborative. So we understand it’s there and we understand
it’s a limitation, but for this particular project, it’ll be less of a limitation than
others. It’s how we are thinking that it’s point in time. Well, I guess I’ll just ask one more. The
issue of timeliness have come up to how quickly data is available and I’m wondering for your
PDSA cycle for hospital discharge data, what is kind of your best-case definition of timely
data? Is it six months, nine months, a year? What are you considering good enough? I mean,
not good enough, but gold standard, rather, for your PDSA cycle? So in Maryland we are very close to having
access to 2013 data, and once we have that, that last quarter, that quarter four, is the
start of our intervention period. So we’re very happy to have that. And then from there
we are still working with the data provider to see exactly how timely it can be. Is it
six months after it’s collected? Or that sort of thing. We certainly expect no more than a year. But
we are also cognizant of the timelines of this grant and so we are working to the best
of our ability to meet those timelines as we can. And here in Illinois the data I presented
today is 2012 data. We would also be happy to work with 2013 data to show our communities,
but the way that we thought about this data analysis was less as a way to assess or evaluate
the ASTHO Million Hearts work and more as a way to direct it. So to give a baseline
to communities and to find populations or ZIP Codes or areas as a county that are particularly
high burden. So I think that we’re dealing with, like I
said, a nine-month delay, I would say, with data. So I don’t know that it’s such a good
data source to evaluate such a short project as the ASTHO Million Hearts project. In my
mind a better data source for us, for that kind of purpose, is the data we receive from
the electronic health records at the clinic, because that can be as fast as yesterday’s
data. And so that’s a much more responsive data system. I mean, in general do you feel like you need
a more responsive data set to measure interventions, not just this interventions, but interventions
that other interventions you have going or ones that you may have in the future? I don’t know about need, but it’s certainly
helpful and a few more can be something that the clinics and communities really respond
to. Okay. I’m asking really from a personal point
of interest, because I was involved in a conversation with CDC on Friday and they were saying, “Well,
hospital discharge data is good enough and a survey data is good enough. We don’t need
immediate data for chronic disease.” And I was just wondering to hear other people’s
perspective other than CDC, people who are actually working boots on the ground. Yeah, well, I think my perspective on that,
something I feel like I’m always saying on my EPI soapbox here, is that we have to get
to a place where we’re using existing data systems better. So I think one advantage of
a data system like a hospital discharge data is that it’s data that clinics are already
reporting. So rather than asking them to report it to another system, fill out another form,
it’s data that’s already existing. So in my mind that’s another huge advantage of it,
although if there is a delay, it’s not a duplicative effort for them. Great point. And we Maryland certainly agree with that
point. And we’re also looking to other ways to find data that are more instantaneous than
these ED data, because we do understand limitations of that. We are looking at local health information
exchanges and we’re really – part of our Million Hearts efforts are looking at working with
clinics, working with EMRs, getting those EMRs connected into health information exchanges,
which then be connected into our state information exchange, which already has hospital discharge
data in it and hospital data are starting to become more and more available. We, the
state, are trying to get more clinicians and systems plugged into this so that we can have
that more breadth and depth of data in a much quicker timeframe. Right, thank you. I appreciate that a lot. Do we have any other questions coming up in
a dialogue box? No other questions in the chat box right now. Okay, then any other questions
or any corrections that people have but just
like to ask? This is Emily with ASTHO. One question I had
for both Illinois and Maryland, and then of course for anyone else on the call, Julia
with Illinois had touched on this briefly, but sharing the data with partners, it sounds
like, at least in Illinois, there’s been a lot of interest with the local partners. And
you had also mentioned the possibility of sharing it with different payers and Medicare,
Medicaid as being useful information for them to have. Has anyone thought about other partners that
might find the information useful or helpful to them? And the second part of that question
is have you reached out to them or talked to them to find out in what format that data
would be useful? This is Julia in Illinois. So I mentioned
that we have a lot of interest from community groups from local health departments, from
clinics. I think the other group that it’s probably Sandy can probably speak to it more,
but I think we had quite a bit of interest from the hospitals that are in the target
communities we’re working in. The limitation we have there is that we do have strict rules
about sharing hospital-specific data outside of the agency. So that’s been a challenge, I think. But I
think that would be another group that we know is interested. And I think probably the
format they’re interested in is seeing as much patient level data as possible, so seeing
patient names, or dates of discharge, or other reasons the patient was seen, that kind of
thing. I have had a couple conversations with one
of our large payers in Illinois – actually with the Blues. And they’re interested. I
don’t know where that interest is going to go as far as data presentation or activities,
or anything, but started discussions at that level. Great. Well, Paula, I don’t know if you have
any other questions or thoughts to the group before we wrap up. Does anybody have any last
thoughts or comments? I think I’ve asked all my questions. Great. All right, well, we will wrap up just
a couple of minutes early. Thank you so much again to the Illinois and Maryland team for
presenting today. And thank you, Paula, for facilitating the discussion. That’s really
I think rich and interesting to hear the perspectives from everyone. I thank, too, all of the rest of you for participating.
So following this call, we’ll send out the notes, along with a recording of the call
and the webinar slides to all the attendees, so you can look for an e-mail with those materials.
And then we’ll also be posting a recording of the presentation, as well as slides, on
our new peer group call webpage. So Katie Potestio has done a great of creating a web
page as part of our ASTHO Million Hearts web page that lists the archived recordings for
all of the peer group calls to date. So feel free to pass that link on to anyone
who may not have had a chance to register for the call or was unable to attend. And
the links for that webpage is on the screen now. So if there’s nothing else, thank you
so much for your time. I hope you have a wonderful afternoon and I’m sure we’ll all be talking
again soon. Thank you. [End of Audio]