Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner

Chronic Disease Cost Calculator Web Conference: Transcript (Part 2)

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

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 2 of 2 (files split due to size of full-length recording) (MP3 Audio File, 40.5 minutes, 19.0 MB)

Return to Transcript, Part 1

Amanda Brodt: Great. Thank you for the demonstration, Susan. We could have used any State as an example State for this analysis. We purposely picked Kansas to do a walk-through of their data because the greatest number of participants on today's call are from Kansas. We've also asked the State's Medicaid Medical Director, Dr. Margaret Smith, to provide some reactions to the numbers as they were just presented. Margaret, did any of the prevalence rates or associated costs surprise you? What did you learn from the analysis?

Margaret Smith: This is Margaret Smith from Kansas. I thought it was interesting. I didn't expect that the prevalence rates would be lower in Kansas than they were in the U.S. as a whole. But they were pretty close. So not really surprised here. I was interested to see that, with each of the diseases that you looked at, we're pretty close to the U.S. average on cost per beneficiary. So that was an interesting thing to note, especially since we have been noticing that our cost overall for our chronic disease population has increased rapidly over the past few years. But of course, this data is from 2004 so it wouldn't capture that increase.

I did have a couple of questions that I wanted to ask the presenters. First of all, correct me if I'm wrong, my understanding is that this data includes both the managed care population as well as the fee-for-service population in the State in the estimates.

Susan Haber: That's right. Yes. That's because we were using the MEPS data rather than the Medicaid claims data for the main source of our cost estimates. So MEPS, they go out and survey everybody or a sample of the population regardless of whether they are in managed care or fee-for-service Medicaid. And then they ask people to keep track of what services they're using and then they also go out and validate that against medical records and provider offices and things like that. They're not relying on claims data for getting their estimates. That's one of the advantages of using the MEPS data rather than Medicaid claims data. More and more Medicaid beneficiaries are enrolled in managed care, which is a great thing, but then we lose their claims data, and you can't do these types of analyses in something like the MAX data or the MSIS data that States report to CMS. That's the great advantage of using the MEPS data. Obviously, certain populations are enrolled in managed care rather than fee-for-service Medicaid. So if you only look for Medicaid at the fee-for-service population, you get a very biased picture of what Medicaid is spending and what the prevalence is and things.

That was a long answer to your question. Sorry.

Diane Orenstein: If I could just shortly add to that, Susan…one of the reasons why we wanted to use the MEPS data is to make it easy for States that don't have access to the data or don't have clean data or have the resources, being staff and financial resources, to actually do their own calculations. This allows every State to have the ability to use this calculator and use a consistent methodology as well as data source across all States.

Dr. Margaret Smith: The other question I have is, the cost per beneficiary that you present here, is that the cost that's attributable just to that disease? That's not going to be our total cost for that beneficiary, is that correct?

Susan Haber: That's correct. That's one of the ways that our cost estimates differ sometimes from other estimates that you may see out there in the world. They may look at who are the people that have these diseases and what's the total amount that Medicaid is spending for them. Whereas we're looking at just the amount that's attributable to that disease. There are reasons for doing each way, there are pros and cons of each. But obviously even if you control hypertension or you find less costly ways of strokes--for example, keeping people out of the nursing home--that's not going to get rid of all of the medical costs that people have. So that's why we wanted to focus in on specifically the costs that are attributable to these diseases.

Dr. Margaret Smith: Do we know if those people that have the chronic diseases have higher costs otherwise--say, for their preventive care or whatever--beyond the cost that is attributable to the disease?

Susan Haber: Well, we didn't look at that in this. I couldn't speak to preventive care in particular, but certainly lots of other people have looked at what are the costs of people with chronic diseases--the Center for Health Care Strategies has done this, there's some work from people out of John Hopkins--and all of that shows that overall the costs for people with chronic diseases are much higher than costs for people without chronic diseases. But we didn't address that in this calculator.

Dr. Margaret Smith: Okay. The other thing I thought was interesting is that if you look at our prevalence data and you look at our cost per beneficiary…if you just look at the cost per beneficiary, you would tend to focus on, say, congestive heart failure or diabetes. Obviously, diabetes is one that is very big and it still has one of the higher prevalences of these diseases. But something like hypertension, which has an attributable cost that isn't as high as those other things, certainly not as high as stroke, has the total overall greatest amount of costs because of the high prevalence.

Susan Haber: Absolutely.

Dr. Margaret Smith: Just a technical question--so how small a population can I introduce into the calculator before I start getting lots of insufficient boxes in my estimates?

Diane Orenstein: The insufficient boxes are based on our analysis of the MEPS data. We were analyzing the MEPS data--the prevalence of these conditions that was surveyed in MEPS was not large enough. If you have your own data, if you have some estimates yourself of prevalence conditions at the county level or something like that that you want to use, you can always enter those in. You won't get any insufficient. You just need to use your own judgment that your sources of data are adequate to support analyses at that level. We're just saying that based on the data we use to create our default estimates, we couldn't do it any finer than what we show there.

Dr. Margaret Smith: If I'm trying to develop a care management program that focuses on diabetes and I've got a pilot and I only have 200 people in that pilot, I could still get some estimates even though I only have a small number of people?

Susan Haber: Sure. First of all you've got what the per-person cost is. You know what the per-person cost is. So you could just take that number there and say, "Here is the per-person cost, I have 200 people in this pilot, I think I can reduce per person costs by 10%" You can use that number and the 200 people you've got in your pilot, reduce that per-person cost by 10%, multiply by your 200 people and see what your savings are.

Diane Orenstein: This is something that Susan and I have presented at other meetings when we are there in person. You can look at this as, if you feel that you have a program that would reduce the prevalences, what would the costs look like? You can enter it the other way as well--if you can reduce the cost by X amount, what would that look like?

Once you go in and download the tool and use it, you can play with that and say, If we have--whether disease management programs or other programs through our health departments, etc.--and we were able to reduce the prevalence, what change would we see in those costs? When you mentioned about hypertension, this is a wonderful way to look at this. If you are going to use this in aiding you in having discussions about disease management, you can see that the focus of hypertension is so high, if you can reduce it, then you know that you probably are going to be able to reduce stroke and heart failure in the future. Even though this is a burden as well as a cost, it still gives you some of the information when you are thinking about how to focus your programs or which programs you would like to create and hopefully reduce those burdens.

Dr. Margaret Smith: That's what I see is the most benefit from the calculator. It could be used to either help you figure out where you want to focus your efforts or it could be used to help you decide if you really have made a difference, sort of if you bent the curve in terms of the increase in costs for the chronic disease population.

Amanda Brodt: Thank you, Margaret, Susan, and Diane, for some of your answers. I know we have a lot of people in the queue who also have some questions to ask. I wanted to open the call up to all participants. Feel free to either continue to ask questions electronically from the Q&A panel or we're now going to open lines such that you can use the raise hands function that is on the participant list. If you click on that option, our moderator will unmute your line and you'll be able to ask your question orally. For those who have submitted electronic questions, if you prefer, please do feel free to ask any of those orally or we'll continue to go through the list that has been generated as we were in discussion.

Question: The first question that we're going to address from that list goes directly to the tool demonstration. We had a request to show how exactly one goes about saving the outputs in a text and in an MCC file.

Susan Haber: Gee, I'd have to go back into my desktop to do that. Actually, if you want to, I can just tell you. It's quite simple. Let's say you wanted to save it to the text file. Click on the button that says save and it's going to ask you where you want to save it to, what folder in your computer you want to save it to. It's going to ask you to give a name to it. Once you do that, it will save the information for you. The output is going to be saved in something called a CSV file. It looks like an Excel spreadsheet. It basically has that type of format. Same thing, if you want to save the input, again you just click on the button and it will ask you where you want to save it to and give the file a name and you're done. It's quite simple.

I hope that answers the question. If necessary, I can go back to the desktop.

Moderator: At this time we have a live question request from Mark Mailer. Mark, your line is unmuted and you can go ahead with your question.

Question: Thank for that response. I guess my question along that line had to do with, number one, I'm not aware that I have an MCC file on my computer. So I just don't know how to make that connection. I don't have MCC software and so I don't know how it would go into an MCC file. The second question is, You said to import it into a text file and you can do manipulations on the data. I don't know how to do manipulations on text file. I can do manipulations on an Excel spreadsheet but that would not be a text file. So I was just kind of curious as to what kind of files these things are going to go into. Are they in my computer and how can I manipulate the data if it's in text rather than actual data?

Susan Haber: Right. So I don't believe you need any special software to do this because I don't have MCC software on my computer either. It's just a format it will save the file as. You don't need anything special to do that when you save it. That's the format it will be saved as. If you want to manipulate the output file, it will be saved as a CSV file, but then you can open that up in Excel, for example. It will go into Excel perfectly and you can manipulate in Excel.

Question: We have had a number of questions about whether you'd be able to select site-specific cancer estimates, e.g., colorectal cancer only?

Susan Haber: Yeah. We did not do that for this project because the sample size in MEPS just wasn't large enough to support site-specific cancer estimates. So we don't have that in this, unfortunately. This is just an average across all cancers and it reflects the mix of cancers in your State Medicaid program.

Diane Orenstein: That's actually listed in the technical user guides, which cancers.

Question: Great. We've had another question come in that refers to someone mentioning at the beginning of the presentation that one could download the code for the cost calculations. They would like some clarification on how that would be done.

Susan Haber: Sure, actually Diane said I would talk about that and I didn't, so I apologize. The calculations in this are based on the MEPS data, but we also wrote parallel code where you can do the same type of calculation using your Medicaid claims data. We've written sample SAS code that you can use, and we've got instructions about what you need to do. Basically, we describe what variables you need to create in your data to do these analyses, what you should call the variable, how to set up the file to do the analysis. You can run the data against the SAS code that we provided using your own State Medicaid data. It means you are going to have to be able to manipulate your Medicaid data. For example, the basic variable that you need is a variable for what the total Medicaid expenditures are for a person during the year. You need information on what diseases the person has. So basically, you need to go through your claims data for a person to find claims that have different diagnoses on them. We have written code that will do that calculation for you, but you have to have the Medicaid data set up in the right way to run through it in that way. We have provided that sample SAS code. So you've got to have SAS and you have to have some ability to manipulate your own Medicaid data.

Diane Orenstein: It's not anywhere in the format where the tool is, but it has all been for you using the MEPS data. You have to have that information and manipulate it and do all that yourself. We just provide how to do it--the codes to do it or the program to do it.

Question: Great. Thanks. We have a quick clarification to that question and some followup. We're wondering where to go to download the information and the sample code. Is there a link on the Web site? How do you access that information?

Susan Haber: It's on the CDC Web site. There's a link to the code and to some documentation how to use the code. It's all there on the CDC Web site.

Question: Then we have had a couple of questions come in that are related to trending and whether there's any way to use the tool to see changes and cost over time.

Susan Haber: Version 2. Do you agree Diane?

Diane Orenstein: Yes, version 2. That's not something we could do here.

Susan Haber: I was going to say, the reason we included that in version two is this question came up many, times when we presented this and we realized that this is something that people wanted.

Question: Great. We have a couple of questions related to cost estimates as well. One is asking whether there's any inflation factor that can be used to project expenditures based on today's costs?

Diane Orenstein: Actually, Susan, you may want to speak to this more. These costs are inflated to 2007 dollars. Even though the data are from 2001 to 2005, they have been inflated. So what you're seeing is the costs inflated to 2007 dollars, and version two will be inflated to 2008 dollars. There's because when the MEPS data become available, there's a lag time. So version two will be 2006. That's the latest, so we will inflate that to 2008 dollars.

Susan Haber: The reason data are only inflated to 2007 is that basically we were using estimates of the Medical Care Consumer Price Index [CPI] to inflate the data from 2001 to 2005 up to 2007. 2007 was the most current Medical CPI estimate that was available at the time. If you've got some estimates of how costs in your Medicaid program have been increasing between 2007 and currently and you wanted to apply that inflator, you certainly could, although I caution you to make sure it's being used appropriately. That's the only reason we go up to 2007.

Diane Orenstein: One thing--I didn't know if it would be useful to address--is if there are any questions about the cost, how you looked at the regional costs to get the costs for services.

Susan Haber: Right. Well what we did is--basically, the costs are adjusted for many States, and the adjustment represents differences in medical costs between your State and other States in the country, or between that and the national average. There are some States where the sample isn't large enough in the MEPS data to actually get a good State-level estimate. So for some States, we're using regional estimates of what that region's costs are to adjust the medical spending. But basically, these are all being adjusted to reflect variation and costs within the States or within regions of the country. Is that what you were referring to Diane?

Diane Orenstein: Yes. You don't want to use the cost for New York to represent all the States or the cost for South Carolina to represent all the States.

Susan Haber: Yes. Absolutely.

Question: Another quick question about version two…will that allow grouping by year of data?

Susan Haber: The way version two will work is there will be a base estimate, which will be, maybe, 2008, and then we're going to show for the next 10 years. I believe it will be year by year. I won't swear to that. I don't know if it's going to be year by year or if it's just going to be 10 years out, what the costs will be. That's how that's going to work. It's going to drive off of changes in population demographics and the prevalence of these different conditions within different population groups. I don't know if that answered the question. I don't remember right now or I don't know if we've decided right now if we're going to show each of the 10 years or just 10 years out.

Diane Orenstein: I think we're working on that. Trying to figure that one out.

Susan Haber: Version two is still in process.

Diane Orenstein: It's a lot more complicated because, instead of just Medicaid, we have Medicare and other payers. So that makes it even more complicated for our development. Hopefully not for your use.

Question: Speaking to demographics, we had a question raised. Will calculations be available by race and ethnicity in version 2?

Susan Haber: Unfortunately, that's a MEPS sample size limitation and the sample wasn't large enough to support estimates by race.

Question: There is a question about the per-person costs included in the direct costs. So we're including outpatient, inpatient, medication, long-term care, etc., but we're not including indirect costs in the calculator, right?

Diane Orenstein: In the current calculator, we are not. In version two, we are doing indirect costs but only by defining them by data on absenteeism. Indirect costs can include so many different costs, and after looking at the literature, we believe this is the best way to capture indirect costs for lost productivity. So we'll also, in the user guide, be very clear to explain what absenteeism is, what that represents, and what the data source is.

Queston/Comment: Great. As a break in questions, we've received a comment, a request for version two to please include the total costs per member rather than disease-attributable costs in future versions of the tool.

Susan Haber: I think what this person is asking is, What's the average cost per person with diabetes or with hypertension, and not parsing it out to the costs attributable specifically to those diseases.

Question: There was also another question raised about costs in terms of the next version. Are you looking to break down the costs by buckets, including pharmacy costs or inpatient costs associated with a given disease? That would be a powerful feature if you could.

Moderator: At this time, we do have an open line request from Lisa. Your line is open. Go ahead with your question.

Question: Hi. Actually I have two questions. One is whether the cost estimates that you are providing are annual estimates. That's pretty simple. And the second question is, since you are going to be including cost estimates for children with asthma in version two, what are you going to be using for absenteeism if that's going be included?

Susan Haber: The answer to the first question is yes. These are annual costs. And the answer to the second one, in terms of absenteeism for children with asthma, it's going be based on children missing time from school and then it's assuming a parent has to miss time from work when a child is absent from school. So the absenteeism will be the parent's lost time from work due to the child's illness.

Question: Thank you. We'd like to return now to the question that we had just asked prior to your question. Looking specifically at the breakdown of costs, in the next version can there be a breakdown of pharmacy costs, inpatient costs, etc., that are specific to the disease? If they had the ability to break down the costs, it would be a powerful feature for the tool.

Susan Haber: You know, well, I guess I partly defer to Diane on this. The current plans for version two do not include that. I believe that, in theory, it's possible. It's just not in our current plans for it.

Diane Orenstein: I think that when we hear these questions, just as we said, they have certainly had input in how we have been designing version two. If we have a summary of these questions, it's something that we do go back as a team and work on, and discuss. Even though it's not in there now--I can't say it will be--but it's certainly a question that will be up for discussion. That doesn't mean we can do it. It means we can debate it, look at whether it's something that can or cannot be done in this version, and what it would entail to do that. I'm not trying to be difficult about this, but every change is a cost to us. We really do appreciate this type of input so we can make sure we create a tool that addresses the issues for our audience, which is you all.

Question: Diane, you may not want another suggestion, but there has been a comment about version two possibly allowing estimates by type of facility--acute care hospital versus critical care hospital, etc. Also when you are considering version two, you may want to break down some of the estimates.

Diane Orenstein: That may depend on whether the MEPS data has that information.

Susan Haber: Yeah. My guess is that it's not in the MEPS data. That's sort of the broader type of categorization, that inpatient versus prescription drugs versus office. That I think we can get to, but that finer breakdown by type of institutional facility, I think, is not in the MEPS data.

Question: We have another question about the current tool. This person wants to know, When seniors can't access health care due to a lack of providers who accept Medicare, how will their health costs be calculated? Would they be included in the tool?

Diane Orenstein: In version one, I think the answer is no. Susan, please correct me. In version two, I don't know if that goes under "other payers."

Susan Haber: I'm assuming that the person means that they've actually paid out of pocket for the services, so they've actually accessed the services, they are just not being paid through Medicare/Medicaid. Version two, the way it's broken down is we have Medicare estimates, Medicaid, private insurance, and then we have total, which would include everything, other types of insurers as well as individual out-of-pocket payments.

Diane Orenstein: Again, I'd hate to say we're limited by what MEPS provides because it provides so much for us, but we're using the MEPS data so we do have some types of restrictions of what we can and cannot do.

Question: A question was raised asking if there was dialog between people that developed this calculator and those who developed the ROI [return on investment] calculator at the Center for Health Care Strategies. Probably a general question is, How does this calculator tool vary from others that are out on the market?

Susan Haber: Certainly, when we started this project, no, there wasn't dialog. But probably about a year or so ago, we were contacted by the people from the Center for Health Care Strategies. I did make a presentation to a group of people through them. Basically, they serve kind of different purposes. Ours is not an ROI calculator, but when I spoke with the people there, they viewed this as something that could feed into their ROI calculator. For their calculator, you need to know what a disease is costing you. Then it goes from that to estimate what the potential savings are from a particular intervention. So I guess, in our discussions, it seemed like it could potentially provide an input that someone might use in their ROI calculator.

We were not in communication to start with for sure.

Diane Orenstein: I think these complement. They're for two different purposes but our data can be used in theirs.

Question: Thanks. We have a quick question about whether risk adjustment has been incorporated in the estimates and whether or not that is applicable?

Susan Haber: No. There's no risk adjustment here. I guess I don't think that's applicable in this case because we're just reflecting what is the cost for this population in your Medicaid program. I'm not sure how risk adjustment would apply here. We're not trying to make people with hypertension more comparable to somebody else. We're trying to reflect what the people with hypertension cost. There's no risk adjustment here.

Amanda Brodt: Are there other questions that people have not had a chance to ask? Please feel free to use the Q&A panel or, if you would like to raise your hand and ask a question orally, our moderator will unmute your line.

Susan Haber: I would like to go back to the risk adjustment briefly, as I'm sitting here contemplating it. I guess in a sense, our numbers reflect the different risk associated with these populations, or at least a piece of the risk. So, this is how much more a person with hypertension costs than a person without hypertension. So in that sense, it's reflecting the different risks. Sorry about that.

Hilary Kennedy: We are coming to the end of our time. We have a couple of people with their hands raised. If the moderator could address those, that would be fantastic.

Question: If the data that you are calling costs--I'm assuming it's Medicare expenditure data--what data would you have from private payers that would provide that expenditure data to include them in this cost estimator?

Susan Haber: The MEPS dataset includes everybody, not just people in Medicaid. It includes privately insured people, people in Medicare; it's everyone. What the MEPS people do is ask everyone: What's your source of insurance? So we use that to identify whether a person is a Medicaid beneficiary versus a Medicare recipient versus someone who's privately insured.

Question: How do you know the expenditures that go along with that coverage? It's what they reported paid on their behalf by the payer?

Susan Haber: Well, exactly, but then MEPS is a very impressive dataset. They start out by asking people that question, but then they go out and they do a lot of validation of the information that's reported by the respondents to the survey. I am not a MEPS expert. There are other people in our group here at RTI who are much greater experts than I am, and I'm sure there are people at AHRQ who could speak to this. There are people being asked to keep track of this information and report it in the survey. AHRQ goes out and validates that, for example, by going out to their insurer and their physician's office to make sure they check the accuracy of what people are reporting.

Question: I was just wondering…you've mentioned the sample size being a factor. I'm assuming that also applies to county-specific information, that it's not projected to be available in version two?

Susan Haber: That's right. Yeah.

Question: Thanks. We have one quick clarification question: this is cost data and not charges?

Susan Haber: Right. This is cost. This is what is paid.

Diane Orenstein: This is what Medicaid pays out.

I actually want to go back to the question about counties because we've been asked this before. Again, if you can put in--and Susan, you may need to help me out on this one--if you have some of your own data about the prevalence within your county, I wouldn't say this is an accurate estimate, but it can be probably a better guesstimate for you.

Susan Haber: Absolutely. I would agree with that. If you've got some information on county-level prevalence, for example, you could use that and get an estimate. You would be assuming that the per-person costs aren't different at the county level. But at least you could reflect differences in the prevalence across different counties.

Diane Orenstein: And that's where we were saying in some earlier questions that you can manipulate your output based on what you put in for your county, or for any dataset that you would like to use for the prevalence. You don't have to use our defaults. So maybe the answer to your question is "somewhat."

Susan Haber: If you can provide the information.

Diane Orenstein: Right. It will do the calculations.

Hilary Kennedy: Thank you so much. These are some wonderful questions offered electronically and by raising your hand. We appreciate all of them. We're now going to pass this call to Margie Shofer for some concluding remarks.

Margie Shofer : I want to thank everyone for the great questions. You asked a lot of really good questions. I want to thank you for participating in today's call. We hope this discussion was helpful to you. If you have any more questions either now or after you've had a chance to explore the calculator, please don't hesitate to submit them to the Quality Tools e-mail address. We gave that address to you in the slides. We are also very interested in hearing how you use the calculator. Please let us know about that. Be on the look-out for followup e-mail from us inquiring whether you have used this or have used any of our other tool offerings.

I also want you to know there will be upcoming Web conferences this summer featuring the online version of the Asthma ROI Calculator and the latest version of the State Snapshots. Neither of these tools have been released yet, but we are hopeful that they will be released in the next few months. I'd like you to fill out the evaluation that will show up after we end the Web conference. We really value your feedback and want to get a better idea of how you might use this tool. So thanks again. This concludes the Web conference and we look forward to hearing from you. Bye-bye.

Current as of May 2009

Return to Contents

The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care