[Inaudible Remark] You've been here before. The comment was, yes, CPTs are level 1 of the HCPCS.

Transcription

1 >> Okay. So, first I want to go through the base claim file portion and talk about what useful information there might be there. Well, first of all, the encrypted BENE_ID which you will use to link across records for the same individual. Then, the beneficiary demographics are in the carrier file up until a few years ago, they were not. They were only in the beneficiary summary file, and so you had to do the link to get the demographics into the carrier file. But now, all of this demographic information is in the base claim file portion of the carrier file. There's information about the claim because in addition to those dates I showed you for each line item, a claim may spend more than one day. So, there's a claim from-date and a claim through date, and we'll see this in the exercise. Not every claim is for one day, and obviously not only for one service. I mentioned earlier the claim diagnosis codes but now, after--beginning in 2007, there are up to eight diagnoses. Thank goodness. They're still ICD-9-CM codes. I really do plan to retire before the data have the ICD-10 codes. I just can't handle it. I don't know about you. How many of you were--how shall I phrase it? The whole world is using ICD-10 codes, right? I mean, who uses ICD-10 codes in the United States? What large government organization? Who? The VA? No. The VA is still ICD-9. The CDC and the National Center for Health Statistics, all the death data has been coded in ICD-10 since The world uses ICD-10, okay? So CMS is only now 14 years behind, but I'm not saying bad things because the change is so radical, I mean, you're used to seeing if you know the ICD-9 codes to seeing numbers, now they have alphanumerics. So when I look up pneumonia, my favorite 481, I don't know where it is. It's I--I don't even remember. But, yeah. So the transition is going to be, you know, difficult and I really do plan to retire before the CMS data are available for I was going to mention, one or the--oh, I know, a political story. I was thinking not to do any but I just remembered, the change from ICD-9 to ICD-10 will cost money, right? I mean, there will be a--there's going to be a big disruption. There was a period I would say like six months ago where there were some news stories where the Republicans, bless their heart, were blaming Obama for adopting ICD-10. Now, the whole world, you know, Zimbabwe, Albania, everybody uses ICD-10, and parts of our own government do. But Obama was getting criticized for causing additional expenditures by adopting ICD-10. The world gets crazy. Okay, the final thing I would note here because you will see this frequently is there's this funny diagnosis of XX000, and that's for a laboratory test. So, again, don't get thrown by that if you see it today or when you go home. Yes, a good question. The question is if a claim from and the claim through date are different, is it possible to find out actually when the test was done? Yes, because the test will be a line item. And the line item will have a claim from and a claim through date and those will be the same, okay? And we'll see this afternoon how infrequently they are different. And there are times when they may be different, but I won't tell you why because I want you to tell me during the exercise. But yes, you could find a precise date of a laboratory test when you've got a claim that spans a period, okay? Any other comments or questions?

2 Okay, [inaudible]. Okay, this bi-rule. I know, Beth, yesterday said, "In the MedPAR file in the inpatient information, there are no rule outs," okay? And everybody kind of accepts that. However, even though the rule is the same for the carrier file, we sometimes say, "Oh, go on. It may not always be true." But I used to check before every presentation of CMS 101 with the coders at CMS and say "Is this really true?" And they would say, "Yes, Marsh, it is really true." But think of the situation where someone is trying to determine if someone has, again, diabetes or benign prosthetic hypertrophy, and there may be a sequence of drawing blood, taking test to see whether the person has the diagnosis. Well, if the person has no other disease at the time they first see the physician; the physician has got to put to some code in there. You know, we don't like "well patient" as a diagnosis. There's got to be a diagnosis. So there will be some times when the physician is really trying to rule in or rule out a disease even though that's not allowed. Then I mentioned earlier that the diagnoses that are found in the line items should truly be in the claim file portion. We'll see an example or two this afternoon where that doesn't occur, but most, most, most, most often, it does. And I put this bullet in here, and sometimes I wonder why I did, because I'm going to talk about comorbidities after break. But just thinking about comorbidities, again, if you have up to eight diagnoses, the one that's listed first is probably the reason that person came to see the physician, and then you can consider all the other as comorbidities, if you will, but you don't know how important they are to the patient at that moment. We'll talk more about comorbidities later. Okay, so now in the line item portion, one of the things that's changed is, first of all, to note that the line item may occur up to 13 times, I said that earlier. And then, in the old days, there was a count variable. So you went along in the claim--in the base claim file portion and it said seven. And you knew, "Aha, there will be seven line items." And that was wonderful for the database managers and others who were trying to make sure they loaded this variable block record or variable block file correctly. Now, that count variable isn't there, but that's just life. Then again, for the third time, the line diagnosis codes or line diagnoses codes. Okay, so now we jump to the line file portion. So any questions about the claim or the base claim file portion? Okay, first thing I wanted to say is to remind myself really that there are three variables useful for linking the carrier claims to the MedPAR hospital file or to the outpatient claims. So, if you've got a person who spent--excuse me, hospitalized and you would like to know all of the amount paid by Medicare for the--or to the hospital, and then to the physicians that attended the beneficiary, you can link, and the most useful variables are place of service because in the carrier line file, we will have place of service hospital, and then you've got the dates. So, very commonly, people will look at the date of admission and date of discharge in the MedPAR file and then associate with that all the carrier claims for which the first expenditure or the last expenditure date lie within that range. In practicality, when you do that, I think when

3 many of us do it, we'll say, dates of hospitalization plus or minus three, or maybe plus or minus seven. Again, the convention rather than perhaps, missing something by limiting it to the precise dates of hospitalization. So I have here a couple of reasons why you might want to link carrier claims to MedPAR hospitalization claims. One is to sum the amount reimbursed for care which I just mentioned. The second one, perhaps to validate the occurrence of a procedure. You know, I think we generally trust hospitals to know what was done. But if you're looking at, let's say, urologic procedures, or any other procedure that might be specific to a certain specialty, wouldn't it be nice to look in the carrier file to see if there were a claim for that same procedure? Now, there won't be the same numbers, right? There won't be the same coding because the ICD-9 codes are used for the MedPAR procedures, and these CPT HCPCS that I've mentioned a couple of times are used in the carrier file. So there isn't a perfect correspondence, but, again, we frequently will do that. We'll make sure that what's in the hospital claim that says a person had a certain procedure that there is in fact a physician claim, place of service hospital, with the dates, and it is the same procedure, or pretty darn close. ^M00:10:22 Comments or questions? Okay. The other thing is it's good to link so you don't double count. If someone has--there are certain kinds of prostate procedures that could be done in the hospital or in the office. And so by linking using place of service and dates, you make sure you don't double count. Is there any other reason that you might want to link these files? One of you mentioned bundled payments, right? So if you're bundling payments, that's only a good time to think about linking, not only information that occurs in different files for the hospitalization, but then you're linking around the hospitalization, pre and post hospital care. These are some of the place of service codes. None of them is particularly surprising. Physicians do things in different areas. And, again, remember that because this is the physician supplier file, there are places of service like in ambulances and other places that physicians might not go. So, it isn't just related to where a physician might show up but it's other kinds of care places to work or care types also. Other codes. Now here we're finally getting to the HCPCS codes. And every time up until now that I've said, HCPC, I've said CPT HCPS. Why do I do that? First of all, what is a CPT code? What does CPT mean, yes? Yes, they are codes developed by the AMA, American Medical Association. A good point to make is that, they change every year. The actual CPT acronym is Current Procedural Terminology. And you can get those. You have to pay money for them because they're copyrighted. But you can or your facility will have those codes. Does anybody know the relationship between the CPT codes and these things called HCPCS codes, Healthcare Common Procedure Coding System codes? [Inaudible Remark] You've been here before. The comment was, yes, CPTs are level 1 of the HCPCS.

4 So as it says here in the slide, the HCPCS codes come in three flavors, vanilla, chocolate, strawberry. Three different levels. First of all, level 1 are 5 position numeric codes. They are exactly the same as the CPT codes. No difference, nada, nada. And here's an example of which is one of those office visits, or as we will talk again in the few minutes, evaluation and management codes which is the term generally used for visits to physicians' offices, care of patients in hospital, or other institutions. The level 2 codes, again, I have said this probably 15 times, Medicare works with people other than physicians. So the CPT codes--i'm sorry, the level 2 codes are needed to be developed to cover other things. So hear this one example, maybe a little silly, is an injection, so it is a J code, all of the level 2 codes are alphanumerics. They start with a letter. Again, 5 position. We'll show some more of these in a little while. And then there are the level 3 codes, also 5 position alphanumeric, beginning with the letter W, X, Y, or Z. And those are developed by the local areas, now the MACs. So if you think of some new innovative procedure, physician has applied it. There isn't yet a national code, or the AMA hasn't gotten together to formulate their codes, although they will issue supplements during the year. Then the carrier--excuse me, I said that on purpose, DME MAC will issue its own code, and it will start with a W, X, Y, or Z. So those are the three. And here's that XX000 again, okay? Okay, so here are some examples of the CPT codes. And they generally follow physicians' specialty areas. So, anesthesia, surgery, radiology. The medicine codes are the 90,000 series. And within the group and we have these evaluation and management codes, okay? So what is actually an E&M or Evaluation and Management code? Is anybody a coding expert here or have you waded through these? But--office visit? Yeah. Well, I could almost get away with anything than if not many people raise their hands. But these are physician--let's call them physician attended visits if you will. It can be to office. It can be to hospital. So when the person is admitted to a hospital that first admission note if you will is--has it's own code. Subsequent progress note visits have their codes. And then there's the code for discharge. If the person is seen in a skilled nursing facility, there also are codes. The codes generally are time based. So, for example, another type of example is for a new patient versus an established patient. You get more money seeing a new patient. And then within those, there are times, and roughly 20 minutes intervals I believe, maybe 15, but I think it's 20, that the physician says, "This was a really difficult patient. It took me 40 minutes, so I'll charge this code, whereas a simpler patient, 20 minutes, I'll charge him based on this code." The problem though is you don't know exactly what the physician did. Now, did they actually have the person take off their shirt and listen to their heart, or did they say "Hmm, you look fine." So the details, and for those of us, myself included, who've studied

5 diabetes, the question is did that intern has actually looked at the eyeball, okay? Did they dilate the eye? And so these are the kinds of things that you're left wondering about because at the current moment we don't have that precision, okay? Okay, here are some other--or some examples of level 2 codes. They're kind of fun because, you know, they're not that the things you normally think about. You may not be studying them so you'll never get the HCPCS level 2 code book. But if somebody has it in your building, it's just fun to look at different kinds of things that Medicare pays for. Here are some more. It's always been a thought of mine to see how--for people who have diabetes, how many times they get their strips, you know? They get the free monitor, but then they got to buy the strips, okay? So are they getting those strips? Are they the lancets to stick their fingers? That's pretty old fashioned because they don't that anymore. They have the magic stickers that come with the machines. But, no, these are kinds of things to think about. And I put the chiropractic code in there just to remind us that chiropractors are also paid by Medicare. Here are some of my favorite codes, the Preventive Services codes. You'll notice that most of them start with the letter G, so they are level 2 HCPCS codes. However, influenza vaccination still is a level 1. And there's a note here because when I last updated the slide last year, now, I started checking the influenza vaccination codes and they were changing. And so just like you said a few minutes ago that the CPT codes can change, the level 2 codes can change, or may change. And you need to keep track of that. If you're doing a longitudinal study, please get the code books. Follow them. Make sure that you don't miss something because a code was dropped or a new code was added. The good news is that the people who keep these codes, be the AMA or CMS, are very cognizant of the issue of we using codes. I mean, there'd be nothing worse than dropping code, you know, for some test of the liver. And the next day use it for some test of the gallbladder, okay? They don't do that. They usually retire codes for about three years. So, there's not much chance that you'll be coding liver when you want to be coding spleen or something like that. So--but do pay attention because things keep changing. As it says here, the level 1 and 2 HCPCS may change annually. The level 3 may change more frequently. And I put this third bullet in here and I keep saying that CMS is trying to get rid of level 3 codes, but they will never will. ^M00:20:11 They try to form committees and the DME MACs get together, and they try to get a national code for every new procedure as quickly as they can. But there are times when these level 3 HCPCS do creep in. What are we saying here? Blah, blah, blah, okay. I've already said all of this. Okay. So, couple of things to think about in terms of HCPCS. I've already mentioned the E&M codes with their inherent vagueness. A second point to think about is not all items that you would think of as a physicians service, or something that might be coded as a level 1 or a level 2 HCPCS, will appear in the carrier file. So, if you're reading ahead here codes--level 2 codes

6 for injections and chemotherapy, which we also call J Codes because they all begin with J, are sent to the Durable Medical Equipment MACs. So if you're interested in chemotherapy for cancer, Beth may have mentioned this yesterday, I've forgotten whether she does or not, or some of the fancy drugs for rheumatoid arthritis. Some of these injectables may appear in the DME file. And if you have a question about that, go to the level two HCPCS code book, find your codes because you'll know the drugs that you're interested in, and it will tell you whether they are to be sent to the DME MAC and then appear in the Durable Medical Equipment file. I have a story about level 3 codes. Jonathan Govette [phonetic] was a fellow at that time at Hopkins when I was back in Baltimore, and we were looking at cataract surgery which was changing quite a bit at the time. And he, in fact because there were these level 3 HCPCS codes, he called every single carrier, now MAC, to find out which codes they were using for cataract surgery. And the positive part of the story is they were very friendly, okay? I mean, think about this job, you know, you're sitting in Harrisburg, Pennsylvania, Montpelier, Vermont, I don't know, Madison, Wisconsin, Boise, Idaho, you know? You go to work, you go home. Here is somebody calling you, you know, and asking you about your job. So, not to make fun of any of these people, but to say people are willing to help. If you have coding questions both at CMS and at the MACs, I can't think of a time when people haven't been friendly. Okay. Nothing--HCPCS modifiers. I had that in an earlier slide, but I jumped over it. There are these modifier codes. They're important for some of your work. Some of you can ignore them. Some of you might need to pay attention. So there are examples, and it's just like the HCPCS themselves. There are level 1, level 2, and level 3 HCPCS modifiers. So level 1 are numeric, just like the CPT codes. Level 2 are alphanumeric. And level 3s can be either one. So, here are some examples. And the one I have here, 26, and the one below technical component, are I think particularly noteworthy. So that--as an example yesterday, I kind of mumbled if you have an X-ray in an out-patient department, you'll have two claims, you know? And they go to these different routes, and you got to be careful. Well, when things get coded using HCPCS, that if the person is a radiologist who is reading the X-ray, and they all are getting paid for that, not for doing X-ray and reading it, then they may only bill for the professional component. So, again, if an X-ray being done in it's entirely cost 80 dollars, well, there'll be a claim from the physician for perhaps 30 dollars, and it will have the X-ray, chest X-ray whatever that CPT code is, and then we'll have as a modifier, modifier 26, indicate "He or she is only billing for that professional component," okay? Same thing can be true if someone is billing for the actual machine it would have the modifier TC. And one day a long time ago I went around looking for things that were billed without any of these modifiers, and then those that were broken into two pieces, and always the two pieces equal the whole thing. So, you don't make any money by doing piece work with CMS.

7 There is another alphanumeric, or a level two code, the left and right, which I've always thought might be fun looking at people with diabetes who are having amputations, you know, are they continually amputating on one leg, or are they going back and forth? And those codes are used. I can't tell you whether they're used or how frequently they are used, by they are certainly are in the claims data. Level 3 again from the MACs and they may change. Okay, any comments on CPTs, HCPCS or their modifiers? All right, I want to mention another variable that's in the line file portion which is called the line NCH BETOS code right here. And the way I try to introduce it is that there are roughly 10,000 HCPC codes, what's a poor researcher to do if you're trying to organize things, okay? I mean, you're saying "I'm really interested in the eye and I've got all those codes organized." Well, what's going on with the person's liver, or what's going on with the procedure to other parts of the body? Did they have different preventative services? Whatever. And so the line NCH BETOS code, or we just call it the BETOS code, is a way of aggregating the various HCPCS. And so here are examples, the actual code values and what they stand for. They're actually useful in some sense because they've got some things here for psychiatry and ophthalmology, that if I wanted to find or get started looking at psychiatric visits, boy, that's where I would go. I'd work backwards. And I say Paul Eggers, and John, I think was--berenson put these codes together. They thought about this. So they've got all the psychiatric services codes right there. Let s go find them and then start working from there. But for the most of you, it is a way to organize all those HCPCS codes. And here's just for a little fun to see--i've forgotten the year, but how many services and the amount paid for each of these different categories. Some of my best friends are ophthalmologists and they have the nicest boats at the Marina. So, ophthalmology still is one of the best subspecialties to be in if you didn't already know. Okay. Next thing I want to talk about in the line file portion is money. And this for me has always been--i wouldn't say boring. That would be a total misconception. But something that I deal with at the vanilla level, okay? When I do studies, it's a personal choice, I'm interested in the amount paid by CMS. And that's generally good enough for me. And what I usually therefore say in this workshop which is CMS 101, not 607, or some advance course, that if you're beginning and you're trying to get some idea of the cost burden, maybe you don't want to worry about the cost burden to society that you just want to worry about the burden to CMS. And so the variable that's in the line file portion that you would pay attention to is the amount paid by CMS. So the payment amount. You also can know the allowed charged amount, and that's what CMS--you know, that's the thing. The 40 percent of what the physician really wants, amount, okay? So the physician bills a 100 dollars, CMS says the allowed charge is 40 dollars, and then you get 80 percent of 40 dollars, 32. So, you could work with the allowed charged amount if you wish, or, again, the amount paid. I don't think Beth harangs you about never using the charges.

8 Did she harang you about that? Okay, never use charges, okay? Particularly for hospitalizations because they are a way out to lunch. And we've done some empirical studies. A number of years ago we had a wonderful--i'm going to digress for a second. We had a wonderful chap here from Indiana, one of the prestigious institutions in Indiana, and I said "Well, you know, the charges just don't make sense." And he said, "Not in my institution. We never make a mistake." And so I went back and I could find his institution. ^M00:30:07 And I looked at the chargers and they were negative numbers up to 10 or 15,000, then positive numbers up to 100,000. But the point was that those charge values in the hospital file were pretty bogus, or unreliable, sorry not bogus. And, well, Beth likes validity and validity. They were unvalid--invalid, okay? They might have been reliable cause they kept resubmitting some of these charges, but they were certainly not valid. But, again, when you're looking at dollar amounts in the carrier file, I would stick primarily with these two. Okay, so here're two questions, and we'll do this again in the exercise. Why is the payment amount generally 80 percent of the allowed charge? Co-insurance. Yes. Someone said co-insurance, yes. For laboratory--laboratory services, the two values are the same, why? There is no co-insurance, okay? Medicare pays a 100 percent of laboratory services if they're allowed. Okay, here are some other fun stuff I think in the line item which I introduce by saying "Don't over count the count." I mentioned that briefly when we were looking at the specific CMS-1500 forms earlier. And the question was, these were for ambulance services, did the beneficiary use 40 ambulances? No. They used 40 miles. And so what you have here is--and they're organized backwards actually in the file, but you have an MTUS, Miles Times Unit Services Indicator Code that will tell you "Heads up. We're now going to have information about an ambulance, about anesthesia services, basic services, oxygen or blood." The predominant--predominant, predominant, predominant value that you will see is three services. So, one office visit, six office visits, two flu shot--no, one flu shot. Usually the unit is services, and usually its one. But, again, in the line item you'll see which of these kinds of services it is and then how many--i shouldn't say kinds of services, well they are services, okay, services is a kind of service. Anyway, okay, there are two things we're going to talk about, one is the [inaudible]--or two groups of things. One is the PIN and UPIN. And then there's the NPI. So, the PIN, Provider Identification Number. So, the variable is the carrier line performing PIN, Provider Identification Number, carrier line performing UPIN, Unique Provider Identification Number. So, the first one is not unique, second one is unique, then from that we derive specialty codes, that's a diversion there for a second. And then we have now the line performing NPI or the National Provider Identification Number.

9 And over the past six or seven years, there's been a transition from the PIN and UPIN to the Tax ID and the NPI. And the reason there was the transition is that these Unique Provider Identification Numbers weren't unique, okay? So, finally they just gave up and we said "We'll have a new system." But, up until 2006 and '07, if you wanted to know who the provider was you needed to work with the UPIN and the PIN. And I'll tell you a little story about that in a minute. But since that time, and there was a transition period where there were both PINs and UPINs, now, all providers are coded using the NPI. Up until a year and half ago, we had to say there was no crosswalk between the UPINs and the NPIs. Now there is. So, if you're working in that difficult time of 2006 and '07 where you really do want to be able to determine if the UPIN is also for a particular provider if the UPIN and NPI are indicating that person, and if you're doing a longitudinal study where you have some UPINs from 2005 and some NPIs from 2008 and forward you can have this crosswalk file. So, here's the old story, and I felt pretty comfortable telling this story. One of the requirements was that in order to be paid you had to submit a PIN. So, CMS paid the PIN. Now this PIN could be for a person, solo practice, or it could be for a group practice. And you as a researcher wouldn't know where or couldn't know. Then the carrier back then in order to complete the form picked a UPIN out of the basket of UPINs associated with that PIN and put it on the claim, or put it on the line item. And again, if you had people from Abbot to Zawadski [phonetic] who were the providers in that thoracic surgery group, they might choose Abbot all the time, okay? And that's why I never work with these variables because I was afraid of that. Colleagues, many and rich and famous colleagues have work with them even back then and said there wasn't that big of a problem. But you never quite knew whether the UPIN was the actual person providing the service. And that was a little bit of difficulty. On the other hand, if you only cared about the practice, you could take all that PIN--all of the services that were provided by that PIN and you could feel very certain that that was accurate data. Okay, so the UPIN for me was a little squashy, again, for a lot of other people it never was. Then finally because I had mentioned the Provider Specialty Code, CMS uses the UPIN or used to use this UPIN to then put on a Physician Specialty Code which I find actually quite useful. Now, here are the next couple of slides that talk about the transition. I should really take these slides out and just say its been transitioned. And, you know, once we got to 2007 there were no problems, but if you want to look at these notes when you go on the airplane going home it tells you the sequence. Here are some empirical data showing it took a while for people to get their act together. But by the second half of 2007, when they were supposed to be using the NPI, if you sum the last two rows there, roughly 87 percent of all carrier line items had the NPI on it. So, certainly, beginning in 2008 and 2009, as it says in the second bullet, you only have the NPI, the 2006, 2000 data I mentioned earlier, will have a mixture. And then the thing that I want to restate because it may have slipped by

10 you is that the tax number, I said tax ID I apologize, the tax num variable has replaced the PIN. So the organization that gets paid is the tax underscore num. And, again, that one you can be sure is accurate. I don't have any stories to tell about whether they now kind of look at the tax num and pick one of several NPIs. I don't know that so I'm not going to say it. So, I'm going to be a believer and say "Work with the NPI. Take it." And if you're looking at pay for performance kind of stuff or trying to identify what an individual physician does, I would go for it. I would use the NPI. If you're only looking at kind of cost I might take the tax num because I know that's solid. And just to complete the story, now the NPI is used by CMS to put in the specialty code, okay? Any comments, questions? Anyone have any experience that's greater than mine that wants to either support or argue with anything I said? Okay, so here, because I do use them and I like using them, here are the Provider Specialty Codes that are generated. They are not the same exactly as the ABMS, American Board of Medical Specialties, specialty areas if you're on--i'm not about ABMS, but they are quite useful. In addition of course there are things that are not physician, so you can have podiatry down here. So a podiatrist is coded. Certified Registered Nurse Anesthetist, Certified Midwife, Optometry. So, there are specialty codes other than medical specialties. Okay, I want to talk about two things, one is counting services provided by physicians and others, and then secondly I want to talk about identifying cohorts of people with chronic conditions, chronic diseases. ^M00:40:06 I'll do that in the next presentation. So let's talk about counting services and the issue of either under or over counting. The example that I choose is mammography, and the data here are a bit old but I think the point can be made nonetheless. And the question would be, and Beth, again, mentioned the good old days before Medicare paid for screening mammograms, you know, the counting of mammograms. You know, how did you count diagnostic or screening, or did you care? So, this is pre 2007 which means for two--means two things, or primarily means the codes have changed. So, if you try to look this up or think about it, the codes that I show you here are not the codes that are used currently for all of those services. So, one of the questions you might ask, and again, Beth posted I think yesterday was, if you're looking at mammography what do you want to measure? Do you want to measure all mammograms? Do you want to measure diagnostic mammogram screening, or whatever? And that's your choice. And then depending on the choice you make, there are HCPCS codes that will help you out. And back then, very nicely, there was a code for screening. So, if you were interested in preventive services, you could say, let's look at this HCPCS for screening. If you're interested in diagnostic services, you would take one of the other codes, or if you wanted all the services, you'd take both. A question would be, we are talking about the carrier file so it's pretty obvious that if you are trying to study mammography you might well start with carrier file data. Physicians who read the X-ray certainly would be billing for those professional services.

11 So I post the question, what additional files might you want to use besides the carrier file that could provide you information? Because, so far, we've talked about the MedPAR file, so you might say "But some women do get mammograms on there in the hospital." But I think you've probably seen by now that that gets rolls up into a radiology revenue center and you're not going to find it, okay? So, if someone had a mammogram in a hospital, we won't know about that. But what about other files that could help you understand mammography? And I should say, if you go to the next bullet, I'm always interested in the number of mammograms or the number of women who had them, but others of you would want to know what were the total expenditures for mammograms. So if we open it up and say if you're looking for anyone of those three kinds of things, what other file would you or want to look at? How about the outpatient file? Okay. Mammograms would be done frequently in a hospital. Outpatient. So, if you wanted to get the complete picture, again, I'm quoting Beth, search widely, and then you narrow it down. If I'm thinking I want to study that universe of mammograms, maybe the women, maybe the cost, maybe the charges--i don't like the word cost, using the amount reimbursed, whatever. Maybe I'd better look widely, and I advice, I think let's look at the hospital. I already said, well, I can't do that 'cause there won't be any useful information. But in the hospital outpatient, I bet there is. So, one would do that. And I want to now show a couple of slides that talk about what you would find if you were looking for these services in the outpatient file and the carrier file. Sometimes, they get a little tongue-tied for the next couple of minutes but I will try to go slowly and make it as clear as I can. So, here we have the rows for the type of mammogram, unilateral, bilateral, diagnostic screening, all the claims and then persons. Again, I'm always most frequently driven by persons, how many women had mammogram? Although, again, many of you would want to perhaps look at this. So here are the number of blind items that appeared in the carrier file. For each of these types of mammograms and the totals in here for the outpatient file. Also, all of these other records or claims. So, if you add, which hopefully none of you would do, the carrier and the outpatient file, these are the number of records that you would have, okay? So, for all of the claims, about 42,000, almost 40,000 for people. So the question is, is this the number of people? Because, again, I'm always interested in people who had a mammogram. No, because it's the number of claims. So, what you do using the beneficiary encrypted ID, and maybe using dates, you might want to put dates in there, too. You can find that, in fact, there were 26,000 women who had either a carrier or an outpatient claim. And, likewise, if you wanted to do the same thing, have the or as opposed to the and, there were 29,000 unique claims if you will. Now, so the question is, what happens if you over count using both, which is always one of my major concerns that you start counting too many, well, here you can see that you would be over counting by 13,000 the number of women if you just added the claims which of course you would never do. But then, the other interesting question which can be a very practical one is

12 what would the undercount be? I'm sorry, I moved away. What were the undercount be if you've used only the carrier file? You know, as you all know, these files are not free. And so if you were a graduate student, limited budget, or wanted to get a study done quickly, you know, you might say "Well, you know, every person who has an X-ray, or a mammogram sorry, will have a claim appear on the carrier file. So I'll just use that file." So what would the error be? Well, you'd undercount by about 753. This is just clearing that slide up a little bit, cleaning it up. And so you'd undercount by about three percent, okay? So, if someone came to me and said, "Well, I think I'll just use the carrier file to look at mammography," and particularly screening mammogram. I might say, "Okay, if you just wanted to know the number of women." However, if you were trying to sum all the charges, said that word again, I don't mean it. If you wanted to know all of the amount that Medicare paid for a mammogram, then if you use just the carrier file you'd only have 35 percent of the claims. Whether you'd have 35 percent of the amount paid, I don't know. But, you'd be way undercounting which we are looking at. So, again, just a reminder that sometimes, one file, the carrier might be all you need. If you have a certain question, other times, you're going to need multiple files if you have a different question. Any comments or thoughts? Okay. The last thing I want to say and--is about the five percent or the 100 percent carrier file. You have that option. Maybe you've heard about that already. I don't know where that comes up in our presentation. You have that option for all the files but the national carrier file. CMS or Buccaneer. CCW will not give you all of the carrier files. Now, you could ask for a 100 percent of the carrier files for Oklahoma and then Texas, and then Missouri but they'd catch on, you know? So if you have 50 data requests, you know, they'll figure that one out, okay? Plus it will cost a lot of money, okay? But you can, you may get the carrier file very easily for a smaller geographic areas, such as the State of Texas. Pretty big but you can get the 100 percent carrier for that, or as Barbara will talk about tomorrow, based on certain demographic or diagnoses or procedures, you can get the 100 percent. So if you want to study all man who have had radical prostatectomy, okay? And you would probably search, say using the carrier file and the--i'm sorry, the procedure codes and the carrier file and then the hospital file. Yeah, you could get all of the carrier claims. So, it's just you can't get all of that. And to encourage you to get the 100 percent, here's a slide we made back in 1994 looking at immunizations for influenza. I am looking at the county level. I think county level analysis can be fun. Sometimes, I'd like to say really the action unit is the county often, it's not the state. ^M00:50:07 It's the county or even a smaller area, maybe a set of providers, the county medical society. And so working at the--through a local level with a 100 percent carrier data or any of the other file data can be really useful.