Chronic Disease Cost Calculator Web Conference: Transcript (Part 1)

---

On May 1, 2009, Diane Orenstein and Susan Haber presented the first of a series of events highlighting the latest releases of several AHRQ tools. This presentation addressed the Centers for Disease Control and Prevention (CDC) Chronic Disease Cost Calculator.

Part 1 of 2 (files split due to size of full-length recording) (MP3 Audio File, 40 minutes, 18.7 MB)

---

May 1, 2009

Margie Shofer: Hello, I'm Margie Shofer with the Agency for Healthcare Research and Quality, also known as AHRQ. Thank you for joining us on this Web conference on the CDC's [Centers for Disease Control and Prevention] Chronic Disease Cost Calculator. AHRQ partnered with CDC to develop the calculator, which uses AHRQ's MEPS [Medical Expenditure Panel Survey] data for some of its calculations. The calculator is a downloadable tool that supports States in estimating State Medicaid expenditures for six chronic diseases--congestive heart failure, heart disease, stroke, hypertension, cancer, and diabetes--and generates estimates of the cost to Medicaid of selected chronic diseases using customized inputs--for example, prevalence rates and treatment costs.

This Web conference is the first of a series of events highlighting the latest releases of several AHRQ tools. Earlier versions of these tools, including an earlier version of the Chronic Disease Cost Calculator, were shared at two workshops for State policymakers and data analysts, in December 2007 and January 2008. These Web conferences are a continuation of the technical assistance opportunities that have grown out of those initial workshops. We are also targeting this event to the Medicaid Medical Directors Learning Network members, as this tool can be quite valuable in their chronic disease efforts. So, if after learning more about this tool today you are interested in further assistance in using it, please let us know.

We encourage your active participation on this Web conference, the primary purpose of which is to help you understand how to use the tool. We will spend some time reviewing the basics of the tool and the data underlying the calculation and will then move on to a live demonstration of the tool.

After the presentation, we'll have an open question/answer period in which we will address as many of your questions as possible. The question and answer period will be moderated by Amanda Brodt and Hilary Kennedy, who work with AHRQ on this project. If you have questions before the Q&A period you can e-mail them to us using the question and answer box on the right hand side of your screen. We will be taking clarifying questions in between presentations, and will take all other questions at the end.

There are a large number of individuals on this call, so if we do not get to your questions, please e-mail it to us at qualitytools@AHRQ.hhs.gov. We have that e-mail address on one of the slides. We'll respond to any questions that we receive after the Web conference.

Today's presentation will be given by Diane Orenstein and Susan Haber. Diane Orenstein is a behavioral economist in the Division for Heart Disease and Stroke at the Centers for Disease Control and Prevention. The focus of her programmatic and applied research is to construct economic evaluations that describe the cost burdens of chronic diseases and the cost effectiveness of interventions, and to develop tools for public health practitioners and other key stakeholders that provide innovative methodologies and guidance for policies and prevention of chronic disease.

Susan Haber is a Senior Economist at RTI International. She has over 20 years of experience in health policy research. Much of her work involves studies of Medicaid and the State Children's Health Insurance Program. Her research includes studies on the costs of chronic conditions, and cancer treatment and screening costs in the State Medicaid program. So with that, I'm going to turn it over to Diane and Susan.

Diane Orenstein: Thank you Margie, and good afternoon everyone and thanks for attending the Webinar, which is an opportunity for us to share the CDC's Chronic Disease Cost Calculator, which I'll be referring to from now on as "the calculator." The calculator was developed by CDC and RTI International, and input from several partners, to quantify the economic burden of these diseases for Medicaid, which is unknown and is likely to be quite substantial. The calculator provides States with an interactive and easy to use tool to quantify and understand the financial impact caused by these diseases. It's our hope that these cost estimates would help inform State decisions on investments in chronic disease prevention and disease management programs, and to begin a dialog about how to limit and even reduce costs in the long run.

As you all know too well, chronic diseases are prevalent among all populations. They're costly and, to a large extent, preventable. We developed the calculator in response to States repeatedly asking for consistent methodology using consistent data that all States can use to generate expenditures for selected chronic disease for their State Medicaid spending. Given the combination of rising costs of health care and the burden on State health budgets, this project has focused on Medicaid because of the growing drain it puts on State budgets. As you see, we have worked with several partners on this project, including both Federal agencies and other States as well. We partner within divisions of CDC as well.

The calculator is designed to be user friendly and yet flexible. We recognize that many States may not have the capacity to conduct their own analysis of disease burden and costs while others may have their own estimates, and we want to accommodate both. As Susan will describe in more detail, we do provide computer codes on the CDC's Web site for States that want to analyze costs using their own Medicaid claims. Susan will now talk to you in detail about how we developed these cost estimates.

Susan Haber: Thank you, Diane. Our cost estimates were derived using three main data sources. The first is the Medical Expenditure Panel Survey, or MEPS, which AHRQ maintains. We also use data from the Centers for Medicare & Medicaid Services [CMS]: the Medicaid Analytic Extract File, or MAX, which is a dataset that CMS creates from the Medicaid claims and eligibility data that States submit to CMS. Then we have the Medicaid Statistical Information System (MSIS).

MEPS is our main data source, and MEPS is a nationally representative survey of the non-institutionalized population. It includes information on people's insurance coverage, their service utilization and costs, and what diseases they have. We supplemented our analyses with MAX data, because the MEPS data do not include the institutionalized population and, for Medicaid, obviously, missing the institutionalized population and missing long-term care data is a big gap. We wanted to supplement the MAX data to capture those costs for the Medicaid program. We use MAX data from four States to get the estimates of the long-term care costs associated with the diseases we're interested in here. MAX is a uniform dataset that is common across all States created by CMS. And then finally we use the MSIS data for information on the Medicaid population counts that we use in the calculator.

The methodology that we use to create our cost estimates is called an econometric approach, or a regression based approach. The reason we use this approach is to avoid the problem of double counting costs across diseases that often arises when people try to estimate disease costs. One of the common ways of estimating disease cost is to look at claims from medical services and look at what the diagnosis is on that claim and then, based on the diagnosis of a claim, decide to allocate those costs to a particular disease. The problem is, as I'm sure many of you are aware, is that claims will often have more than one diagnosis on them, particularly for a population with chronic diseases. You'll often see that. So then you have the problem of either kind of making an arbitrary decision about which disease or how much of the costs on a particular claim to allocate to a disease, or if you attribute the cost to both diseases on a claim, you'll have a problem of double counting disease costs across diseases. So to avoid this problem, we used the regression-based approach to allocate costs, based on the conditions that we identified a person as having. And basically the regression-based approach uses statistical methods to make that apportionment between different diseases. Once we obtained these per-person costs associated with the particular cost using this regression-based approach, we then combined them with estimates of prevalence in each State derived from the MEPS data. We multiplied those two together to generate an estimate of total State Medicaid costs.

Prior to releasing the calculator, we obtained input from a lot of State agencies. We've really tried to make this as interactive a process as possible. Our communication with State agencies has really gone back several years. As you can see here, in November of 2007, we presented the calculator to the National Association of State Medicaid Directors. As Margie mentioned earlier, we presented the calculator at two of the AHRQ State Quality workshops. Prior to releasing the calculator, we did Web conferences with Medicaid directors and also State chronic disease directors. We've done a lot of presentations to different groups over the past few years.

This has been important for two reasons. First of all, we were able to develop a list of Frequently Asked Questions based on the input we received from people during these various presentations and Web conferences. These Frequently Asked Questions are on the CDC's Web site, where the calculator lives. This list responds to some of the issues that have arisen as people have looked at the calculator. But also these different presentations we've done has helped us shape the modifications that we've been making to the calculator, and as Diane will tell you in a little bit, there is going to be a future version of the calculator.

Obviously, you all have the best ideas about how to use the information from this calculator, and hopefully we'll hear some of that today, people's thoughts on how it might be used. But here are some basic ideas about how we thought the information could be used as we in CDC were developing the calculator. First of all, it could provide important information that Medicaid agencies or public health departments or other State policymakers can use to assess the value of investing in disease prevention programs and disease management programs. It's important to know what the cost burdens of these diseases are in order to make the argument for investing in these programs. It can also be used to educate policymakers about the economic cost of chronic diseases in their State, the impact on the State, and to communicate why it's so important to address the problem of chronic diseases. And then last, one of the goals of the calculator was to try to stimulate communication and collaboration between State agencies that may not always communicate as they could. In particular, chronic diseases are a really fruitful place for communication between Medicaid programs, chronic disease directors, health departments, State budget offices, and other policymakers.

As we've been presenting the calculator over the past couple of years, we've heard sometimes from people that they have some estimates, either of the prevalence of these diseases in the Medicaid program or sometimes the costs. The question has come up, what if the State has its own estimates? One of the things we want to emphasize is that the goal of this calculator was not to override any estimates that the State has. It was more to provide information for States that may not have had the resources to develop this information.

If you are a State that does have estimates, of either prevalence or costs of these chronic diseases, it's possible that the estimates in the calculator that we provide for your State may differ from estimates that you already developed yourself. There are a number of reasons this could happen. First of all, you may be using different data sources. You may be using a data estimate methodology rather than the regression-based approach I used. You might be using an approach where you tried to allocate costs based on diagnoses on claims. And you may be using data that come from a different time period. So, it's important to understand why things might be different and then make a judgment on your own about (given these differences and the possible reasons for differences,) what are the most appropriate datasets and modeling assumptions to use given what your needs are.

I'm going to now let Diane talk a little bit about plans for future iterations of the calculator.

Diane Orenstein: Thank you Susan. As Susan described to you, we've had a lot of input from States--from chronic disease directors, Medicaid directors, and health departments--about their interest in trying to address their questions and answers. We have already begun creating version two. Version two will have all the same six diseases that are already in version one but they're going to also expand to asthma, depression, and arthritis. We're also going to expand the burden cost estimates to include not only Medicaid, as in version one, but also Medicare and other payers. We'll again continue using the MEPS data. We're also going to update the calculator, using recently released 2006 data so that we'll have at least another year's worth of more current information. If you look into and actually go into the calculator, you'll see that we have inflated the dollars into 2007 dollars for version one, which we're describing today. Once we do version two we'll also be able to inflate these to 2008 dollars. What you see here also is having lost productivity, having indirect costs. And we've actually defined that. We're using data on absenteeism to describe indirect costs or lost productivity. The indirect costs can be described and data can come from many sources, but our data source will be basically absenteeism. And last, version two will also include a forecast for these costs. We're also working with the American Heart Association, where they are just looking at the four cardiovascular diseases--heart disease, stroke, hypertension, and heart failure. They are going to actually do a more comprehensive forecasting for 20 years out. We're working closely with them. They are using the calculator and the same methodology, so we'll have the same consistency here.

Where is the calculator available? I'm going to leave this up for just a moment and you can see where you can log in and can actually get access to the calculator. You can use it. You can play with it. There's information here that if you need some technical support, where to contact.

On that note, I'm going to hand this back to Amanda, who will take some questions and be ready for some Q&As before we move onto the actual demonstration. Thank you.

Question: The first question asks: are the cost estimates limited by any age groups? Are children included?

Susan Haber: Since these conditions basically don't occur among children or the prevalence is extremely low, the cost estimates are based on adults. As you see in the calculator itself, you can look at cost estimates by different age groups; that's only adult age groups. However, when we look at what the prevalence is in the Medicaid population, we include the total Medicaid population. So the cost estimates are based on adults but it's within the entire Medicaid population.

Diane Orenstein: I think also to answer that question, when we add asthma that will include a younger population.

Question: Great. That comment actually goes to another question we received, which wants to know when version two would be ready?

Diane Orenstein: Our hope, and certainly RTI is working furiously already on this…I hope I'm comfortable, Susan, in saying a year?

Susan Haber: I think so.

Diane Orenstein: I don't think any longer than a year. The structure is already in process now. The analysis--everything is going very strongly.

Question: That's good news and I'm sure people are eagerly awaiting version two. We have another question asking whether the calculator could be used to identify cases of excess cost?

Susan Haber: Yeah. I don't think you could use a calculator in exactly that way, but you'll see that. What we do is provide you with what the average cost of a person with this condition is in your State and the prevalence of the condition in your State. We also provide information on national averages, so you could compare those to the national average to see whether you are high or low. That's the best information, the closest information we have to what is excess. Obviously, there are so many factors that drive the State variation in costs for these different diseases: the cost of medical care in your State, practice patterns. Prevalence is going to be driven by the population characteristics. There's a lot that goes into it. Even in making the judgment of how you stand next to a national average, you certainly want to think about all of those, deciding whether your costs are excessive or not.

Question: Any possibility for use to gauge the effects and cost of obesity?

Susan Haber: The tool as it's set up now, it doesn't really look at a risk factor like obesity. Diane, I'll let you respond also, but from my perspective I would say the answer is no, we're not really looking at the costs associated with risk factors such as obesity.

Diane Orenstein: I would agree with Susan.

Question: Great. What we're going to do is we're going to answer one more question that's come in. We seem to have several in the queue that we think will be appropriately addressed with the demo. Then we'll go back to those questions after the demonstration has been completed. But the question of clarification before that is, just how many States are currently doing this work?

Diane Orenstein: I can answer that from about a month ago. What we've done is basically looked at how many hits we've had and how many downloads we've had. It wouldn't be fair to say how many States. We've had over 400 hits and over 70 downloads, so that's more than 50 States. I think it's being used very broadly, but we have not conducted any type of evaluation to find out who's using it other than the feedback that we've been given when we are having any kind of interactions with the States. So that doesn't break down whether it's a State health department or Medicaid department. We don't know that difference at this point.

Amanda Brodt: Great. Given the level of questions we've had, Susan, I think we have people who are eager to see the tool. Can you do the demo?

Susan Haber: Excellent. Great. I'm going to give you all a view of my desktop, where the calculator is residing. Now that I brought up the desktop sharing, the captioning panel has disappeared from the screen. If there's someone who still wants to use captioning, there's a toolbar in the lower right-hand of the screen. Just click on the icon to the furthest to the right of the toolbar. If you want to increase the size of the captioning panel, drag your mouse across the top edge of the panel until you get a double-sided arrow and click and drag upwards to increase the size of the captioning.

If you were to download the cost calculator, this is what you would see. At the start, this is our welcome page to the cost calculator. I'm going to click on where it says begin here.

You come to a page that has some introductory information on the calculator, just a general description of what's in the calculator and how we're defining some of our terms. It just gives you an overview of the calculator. There's also much more detailed documentation of the calculator that's available for download from the CDC Web site and a more detailed user's guide as well. But as you'll see as I'm walking through it, this is a pretty easy tool to use and pretty intuitive as you get through it.

To move on from the introduction page, click continue. I get our main switchboard. This is where you have choices of various things you can do to input information. This is where you identify what State you want to calculate costs for, what diseases, what subpopulations, things like that. Or if you've already input information, you can click on the bar that says show cost base, current input. If you hadn't entered information before, and want to start over again, you can clear the inputs. You can go back to the calculator introduction. You can also save inputs to what's called an MCC file--like an Excel file. If you saved inputs from there in the past and you want the use them again, you can load them by clicking that button. If you want to leave the calculator, you can click on the button that says exit calculator.

Now I'm going to go ahead and input information. Today, we're going to use Kansas as our sample State. So basically, we have all 50 States and the District of Columbia listed here. So we just click on whatever State it is that you want to calculate the cost for. So I'm going to click on Kansas. You always have the option at any point of going back to the previous screen by clicking on the back button or you can return to that main menu, which is the screen we were at just before.

We're going go ahead and look at data for Kansas. So I'll click save and continue to move to the next screen. This is where you select what diseases you want to calculate costs for. We have our six diseases listed here that are in the calculator. You can click on any subset of these diseases that you want to calculate costs for. Or if you want to calculate all six, just click on the button that says select all diseases. If there's any disease you don't want to calculate on, just unclick it and it goes away. We'll move ahead and calculate cost for all six diseases in the calculator. So again I'll click save and continue to move to the next screen.

Okay. This is the first screen where you really are starting to input information. As you'll see, throughout this, we always provide default values for your State, for whatever State you're calculating costs for. On the other hand, if you have information that you would rather use instead of our default information, then you're always free to do that. I'll show you how to do that.

So the first piece of information is: what's the total number of Medicaid beneficiaries in your State? We have taken data from the 2004 MSIS. That's the most current data available from CMS. If you have more up-to-date numbers on the size of your Medicaid population, you can certainly enter that information. You would just unclick that check mark there where it says use value from MSIS, unclick the checkmark, and you can go ahead and enter your own number. Let's say that Kansas' Medicaid population has gone up to 450,000. You can enter that there if you wanted.

The other thing is that we have these question marks. Those are help or information boxes that we provided. This gives a little more information about the data that we're using for the number of Medicaid beneficiaries in a State. So for the purposes of my demonstration, I'm going to go back and use the default value for Kansas, which was the number for the 2004 MSIS data. Then you'll see a choice of, in addition to calculating costs for the total population, you can do it either by sex or by age group.

So if you wanted to do it by sex, you need to provide what the breakdown of the Medicaid population is between males and females. Just enter the percentages there. If you want to enter your own number rather than use the default that we got from MSIS, you can enter your own number. You'll see that we only require you to enter one number, because basically we're going to force the male and the female percentages to sum to 100% because we need that for our calculation. Alternatively, if you wanted to calculate it by age group, you could do that. Here, we provide the breakdown of the Medicaid population by age group, again using the MSIS data. Again, you can enter whatever numbers you wanted to enter there, but they'll be forced to sum to 100%.

Why don't we go ahead and use an example where we calculate it in total and age group and use the MSIS default data. I'm going to click save and continue. Now this screen asks for prevalence on each of the conditions--by age group, in this case, because we're doing it by age group. You'll see here we're also going to do it by total as well as by age group. The information on the prevalence by age group and the distribution by the Medicaid population by age group is going to be used to calculate the overall prevalence as well. As I mentioned before, going back to one of the questions we got, you'll see here that because the prevalence of these conditions is so low among children we're looking at prevalence and costs among the adult population, and we use 18–44, 45–64, and 65 and over as age groups.

Again, here we're using the default estimates of the prevalence in Kansas that we got from the MEPS data. If you want to enter your own values, you can uncheck the box and enter whatever numbers you think are more accurate for your State, perhaps based on some analyses that you've done. At the top here, the overall prevalence in the Medicaid population in Kansas is estimated to be 5.8%. That's based on the prevalence in these three adult age groups. It basically assumes a prevalence of zero among children.

These are the numbers for heart disease, and you would see a comparable screen for congestive heart failure. Here is one for hypertension, stroke, diabetes, and cancer. They all work in exactly the same way. You either accept the default numbers that we're providing for you or you provide your own numbers of prevalence in your own populations. You can go ahead and do that yourself. You can accept the default numbers for some diseases and not others. It's completely flexible in that way.

Okay. I'm going to go ahead using our default numbers here and I'm going to go back to heart disease. Then we move ahead and save and continue. This is the screen where you enter in the information on the average costs associated with this disease per person in your Medicaid program. Similar to the prevalence screen, there's a separate tab for each of the diseases. In this case, since we're doing it by age group, there's a place where you would enter the average costs per person associated with heart disease for each of the three age groups. We provide default numbers, or you can enter your own. There's an overall population average based on the cost per person and the prevalence by age of each of these diseases.

Obviously, one of the things you'll notice here, which probably won't surprise you given that you all are so familiar with Medicaid programs, is that the per-person cost for individuals 65 and older is actually lower than the cost of the younger age groups. That's of course because essentially everybody over the age of 65 in Medicaid is dually eligible for Medicare, and Medicare is picking up a good chunk of the costs for these dually eligible beneficiaries. If you take a look at something like diabetes, for example, where there's a large outpatient prescription drug component, you see a different pattern. These data, remember, are from prior to implementation of Medicare Part D. This was during the time that Medicaid programs were still paying the outpatient prescription drug costs for dual eligibles. The next version of the calculator is going to use data from 2006, so we'll be able to more accurately reflect the current reality for dual eligibles since Medicare is paying most of the outpatient prescription drug costs for them. We have the costs for stroke. For stroke, you see higher costs than the dual eligible populations because of the long-term care component of their expenditures. That wasn't a good example. I'm sorry.

I'm going to move onto the next screen. This is basically the final screen of the calculator. This is where you see the results of the calculation. You'll see it for Kansas. We provide an estimate for the average of the United States overall. So overall, in the Medicaid population of the United States, the prevalence of heart disease, for example, was 6.5%. That translated into 3.6 million people with heart disease in the Medicaid program. The cost per beneficiary per disease is $1,320, and that translated into over $4.7 billion in Medicaid spending.

Then we show the numbers for Kansas. So we have the overall population numbers for Kansas. Slightly lower prevalence of heart disease than the U.S. average, 5.8% rather than 6.5%. There are about 21,000 Medicaid beneficiaries, we estimate, in Kansas with heart disease. The average cost per person with heart disease in Kansas' Medicaid program is about $1,300. So it's quite similar to the national average. For Kansas that translates into $27 million in Medicaid spending each year.

Then we break it down for each of the age groups. You can see, not surprisingly, the prevalence rises with age. Despite the fact that the per-person costs are lower for the 65 and over population, the total Medicaid spending is substantially highest for that group because of the higher prevalence of heart disease for that age group (among the elderly than among the younger populations).

We have the same type of output page for congestive heart failure. We don't provide estimates on the prevalence in the cost for beneficiary broken out by age group for congestive heart failure. That's because the number of observations in the MEPS data of people who had congestive heart failure isn't large enough to estimate by age group. So we just don't have large enough samples to do that. You'll see insufficient data to estimate.

Here is hypertension. It's a huge cost because of the very high prevalence of the disease. You see almost two-thirds of people age 65 and over in Kansas' Medicaid population were estimated to have hypertension, and that translates into about $74 million in expenditures in Kansas alone, and over $16 billion in the United States' Medicaid program overall.

Here's a similar page for stroke. There's diabetes. Cancer. Then there's a summary page for all the selected diseases, which doesn't provide the information on the prevalence and the per-person costs. It's just the total estimated costs in the Medicaid population. So you can just get a nice summary snapshot by looking at that screen.

Then on this page or on any of these pages on this calculated cost screen, you can print the output if you want to. You can save the outputs to a file you could import into Excel and do some calculations on it if you want to. As I mentioned in the very beginning, you can save the inputs into an MCC file and those inputs--the prevalence, the size of the Medicaid population--can be used and saved in future analyses. You can always be back. That's pretty much the end of the calculator.

I hope it's apparent that it's quite simple to use. It's pretty intuitive. We provide lots of information on the screens, but as I said, you can always go to the documentation on the CDC Web site for a more detailed user's guide and also a more detailed technical explanation of what went into the calculation.

If you want to get out of the calculator, you have to go back to the main menu. Click on return to main and then you would click on this button here that says exit calculator.

But for now, I'm going to exit my desktop demonstration and we'll go back to the main presentation.

(Continued)

Return to Contents
Proceed to Next Section