Institute for Child Health Policy
Betsy: Okay. I'm Betsy Shenkman and I'm from the Institute for Child Health Policy at the University of Florida. And Donna Wegener, who is the project director, was able to accompany me on the trip.
So I’d like to just provide really the highlights of what we've done over the past four years, and there’s not really sufficient time to go into a lot of detail about some of the approaches on methods that we use. So I’ll rely on you to ask questions. Also, if there’s additional information you want we’ll be glad to send it to you. I did need to take a moment just to acknowledge our project team. Bruce Vogel, who is an economist, worked very hard with us throughout the four years. And Donna, as I said, was our project director. We also had a very good working relationship with our Title V Agency. Phyllis Sloyer is in charge of their children’s medical services network. And then we had a great project advisory committee that was comprised of parents, health plan representatives, and physicians. And so we’re really excited to head their representation as well. And there were also numerous physicians at the University of Florida that we worked closely with. And as I mentioned our project partners, we’re the Title V Agency primarily, but also we worked closely with representatives for the Florida KidCare Program. And the KidCare Program is very broad and comprises Medicaid, Healthy Kids, MediKids, and Children’s Medical Services. So the Healthy Kids and MediKids component are our Title XXI or SCHIP component of the program.
Now, our project really had, I think, two major areas of focus and the first two years really were devoted to identification of children with special healthcare needs. And we didn’t really want to develop anything new. We thought there were a lot of very good tools out there. So we wanted to use existing strategies and consider how to use them within the context of operational issues for state agencies like Title V and like Medicaid. And also consider it with in the context of the managed care plans.
Now, when I talk about identifying children special healthcare needs, I'm really not going to focus on that in great detail. I'm really going to give you just real key highlights of what we found there because we've done other presentations I believe to this group about that issue in greater depth. And during the last two years of the project, the focus really was looking at quality of care for children and what kinds of measures could be put into place at the state level, and hopefully at the heath plan level. And there was a huge interest in measuring the medical home for children. But more importantly, and I think different speakers that I was able to hear late yesterday made some comments about, you know, if you provide the medical home will plans reimburse for it. So not only did we want to measure the medical home, but we really wanted to see what difference did it make in the care that children receive across areas that would be meaningful to health plans where you can use this information to say, "Hey, health plans, please reimburse us for this." Or "Hey, state agency, please reimburse us."
We also looked at other issues that we knew were critical from a parent standpoint and a family standpoint, unmet need and out-of-pocket spending. We looked at healthcare use patterns, and we did look at how status measurers. Although, it’s very difficult to measure health status in children and you'll see that when I show you some of the results.
Now, in terms of the identification I’ll just go over that first. Again, our focus really was to look at things through the eyes of a state agency in a health plan. And if you look at things through the eyes of a state agency and health plans, one of the things that jumps out very quickly is that they want to use data that are readily available to them. And that usually means relying on administrative data, claims and encounter data. Enrollment files are things that they already have available to them and that are relatively cost-effective for them to use. But we know that survey strategies are extremely important. We know about the CSHCN screener and other tools that are out there that are very well developed and accepted. And so we wanted to look at what the issues were of using administrative versus survey data to identify children with special healthcare needs, and then in particular to recommend these strategies to our Title V program.
And I need to give you a little bit of background. Florida has across the years developed an increasingly aggressive program to identify children with special needs that are applying for Medicaid or applying for our SCHIP or Title XXI program. And they're very aggressive about wanting to move most of those children that have special needs into Title V. And that program, as I said, over the past several years have stepped up in its intensity. And more and more and more children are being moved into Title V in their specialized network. So overall, there’s a huge interest in identifying these kids.
Now, in the terms of the goals and objectives for the quality again, as I said, it was to look at issues and areas that could be implemented with state agencies and health plans. And then to look at how that related to healthcare use and charges, which is of huge interest to both the state agencies and the managed care companies, and then to recommend indicators that could be used.
In terms of our evaluation design, we used a variety of strategies both for identification and quality of care. We did telephone survey administration to families and we gathered for the identification portion of it, we interviewed I think close to 4,000 families using the CSHCN screener. And also at the time we used the questionnaire for identifying children with chronic conditions because at least early on this study the CSHCN screener was still undergoing development. We have the good fortune to warehouse claims data for Medicaid Title V and our Title XXI programs. So we have an extensive claims data dataset, and also in enrollment files. And the same for quality of care. We did interviews with almost 2,000 families looking at the quality of care that children receive in Title V, and also in our Title XXI program focused on only the kids with special needs. And we have claims data for all those children.
Now, moving back, and I’ll talk about the identification strategies now in greater depth. I'm just going to talk about the CSHCN screener and what we learned with it because this, again, is really the tool that I think has been mostly widely accepted. All the tools are very good, but the CSHCN screener has a distinct advantage of being brief and easy to administer. The claims and encounter data, as I said, plans really like to use that. State agencies like to use that their claims data. Specially, to look at diagnoses that might be indicative of special needs or to look at children that have high healthcare use or high healthcare charges. Also, sometimes to look at kids that have inpatient admissions. And use that to targets kids that might have special needs. We also use the CRGs and I assume John yesterday talked about how that was organized and the different components of the CRGs. So, we did use that in our analyses as well to classify the children.
And probably the four biggest things that we found--in the first part of the study what we did was we identified children who had diagnoses that might indicate special needs in the claims data. And what we found is roughly eight to nine percent of the children in a Title XXI program had a special healthcare need, a diagnosis recorded that might indicate a special need. And then what we did is after we identified these children then we selected a random sample of kids that didn’t have any diagnoses in the claims data that might indicate a special need. And we administered the CSHCN screener and the questionnaire for identifying children with chronic conditions. And what we found was not terribly surprising. That if you rely on claims data alone, you run the risk of missing about 22 percent or 30 percent of the children that do have special needs. And I think that, you know, everybody’s pretty well aware of what those issues are. A child can have a chronic condition and the diagnosis is just not in the claims data. They could be being seen for acute conditions and their chronic diagnosis wouldn’t be recorded. Maybe they're very well controlled and not using.
One of the things that we subsequently did find is that for the children that we do miss some percentage of them if we had reviewed the pharmacy data, we would have picked them up from pharmacy information. And I know that most of those developers that are working on these classification systems and these risk adjustment systems are trying to incorporate pharmacy information much more aggressively. So you will get children and improve your identification from pharmacy claims.
Now, the other thing that we did, if you look at the second bullet point, is we turned it on its head. And we said, "Okay, what would happen if we only relied on survey-based approaches to identify the children, and didn’t look at the claims data?" So because we had this constant influx of enrollment data from both Florida and Texas, we had a lot of flexibility to this. Now, this second bullet point relates to work that we did in Texas. We went in and we used survey-based approaches, in collaboration with Texas, to identify children and their Medicaid program that might have a special need. And they used a screener that was very much like the CSHCN screener just minor variations. And then after these children were identified, we looked at the claims data and used the CRGs to see who we might have missed. And what we found is if you exclude children that were in the healthy category of the CRGs or the acute category of the CRGs, and only looked at children that had a single chronic condition all the way through to catastrophic conditions, that we would have missed about 24 percent of the children by relying on survey-based approaches alone. And some of the reasons for that I think are complicated.
We have found some association with people with lower education having some trouble responding to these surveys. And we might miss children that way. We also think there’s probably privacy issues. We know that whenever we field a survey, especially with people that have kids with special needs, we get phone calls saying "Are you sure you're from the University of Florida? Are you sure you're not a health plan calling me?" We also have had woman call and say they're going through a horrible divorce and they think that this is a trick by the husband to get information. So there’s a lot of privacy issues I think that go on that could influence who you pick up on a survey. And Ardis and I had a very interesting conversation yesterday about some of these issues. An Ardis was saying that sometimes when families have children with malignancies because there actually were quite a few children with malignancies in that 24 percent. But the focus is on we’re going to cure; this is not a chronic condition. And so some of it also may be how families are choosing to perceive the condition and how their doctors are helping them perceive it. So the bottom line here for this is, you know, you can’t really rely on any one approach.
The other issue--and we do have a paper that’s in Healthcare Financing Review--we did find some differences in identification for black and Hispanic children where we do see somewhat of an under identification of black and Hispanic children using survey-based tools relative to white and non-Hispanic children. And what we suspect some of the issues are they certainly maybe cultural and language issues going on here. But there also maybe access to care issues because if you look at the survey tools, they also measure access to care. They ask about the need for and use of elevated healthcare services and so. So there also maybe some issues here related to race and ethnicity.
So, we really strongly recommend and have recommended to Florida and Texas a combination of approaches to identify kids with special needs that you want to use the survey-based approaches to find out who these kids are, when they're coming into the program. You don’t have a claims history on them. You may even want to periodically re-administer that survey. But for financial reasons and because you're going to miss some children you also want to do some searches of your claims data. Yes.
Lynda: Question on the use of pharmacy data. Since obviously almost all kids at some point use pharmaceuticals--antibiotics, et cetera--would the pharmacy focus on certain types of drugs? I mean, I know back to my AIDS work that we use pharmacy quite a bit, and that there were still a lot of false positives. Even using, you know, tracking certain drugs that primarily are used with people in HIV diagnosis. And even with that there was a lot of false positives.
Betsy: Yeah. I know that, for example, Rick Kronick with the Chronic Disability Payment System has a pharmacy module. And there are a series of drugs that are targeted. And certainly you could get some false positives there. There, you know, there’s no question. But he has developed an algorithm that he uses and that CDPS. And Rick is actually very good about sharing both the chronic disability payment system and the pharmacy module.
Audience Member: What’s his name again?
Betsy: Rick Kronick. He’s at the University of California at San Diego.
Lynda: So is the idea to focus only on certain types of--
Betsy: That’s what he does.
Audience Member: --drugs that--
AudienceMember: Yeah.
Audience Member: And 3M has been working on a pharmacy module as well.
Betsy: Right.
Audience Member: And decided that they can’t just use the pharmacy. That they have to link it to the ICD-9 codes...
Lynda: Oh, okay.
Audience Member: ...Because of the cross use of specific prescription drugs.
Lynda: Okay.
James: Has Rick [inaudible] so far?
Betsy: No. He does have a paper I believe it was in Medical Care that was published, and he and Todd Gilmore, who came up with the idea, have really worked very closely with us to implement it with our claims data. So we’ll see how that works. We just know descriptively when we went in to look at children that, for example, were in Title V and classified as healthy, or when we went in to look at children that we missed, we saw things like Factor VIII, you know. Certain drugs and chemotherapeutic agents were very common. But, you know, of course, these kids have some chronic condition. Ardis?
James: Our experience in that has been that when you get down to things like hemophilia, those things actually do show up in the claims data without the drugs. And where it’s been most useful to us have generally been in cardiac conditions and in mental health conditions. We've actually not found it, I mean, that in pharmacy claims for kids is not all that helpful beyond those two categories.
Betsy: That’s interesting. Yeah.
John: Well, there are--you're right. But I think if you can combine them with ICD-9 codes, for instance, it can help you define asthmatics or using just a little bit of bronchial dilators versus those that are on it continuously. So if you link it in those areas you can get some better gradations of severity. But it is tricky.
Ardis: The thing that’s often talked about with ICD-9 data is that it’s only who came into the doctor for a year. A few of the big warehouse data. Did you find certain conditions really only showed up if you looked at longer timeframes? Or if you have significant conditions do you really show up?
Betsy: The interesting thing that we found, and this gets to how much do you want to focus some things at a group and plan level versus an individual level. What we find from year to year is the percentage of children assigned to each of these categories changes almost none at all. However, there’s a lot of movement of the individual children. And that poses--
Ardis: --Up and down the CRGs--
Betsy: --Yeah. And that poses a lot of interesting issues because at a group level then you could really I think decide on payment and other kinds of things at a group level. But that’s not helping these individual children that are bouncing around necessarily. So I think there’s a lot of issues there. So yes, there is a fair amount of movement, both up and down within the continuum of these categories, although in the aggregate the percentages are almost identical from one year to the next.
So what’s happening in Florida with this implementation and there’s a lot of work going on right now. Right now the state has finally agreed, and this took years of battling, to include the CSHCN screener into the application form itself. And how this all came about is that CMS, Children Medical Services, Title V, was doing a huge amount of screening of children that were referred to them based on their KidCare application, which would funnel the kids into either Medicaid or Title XXI, depending on eligibility criteria. And they were screening about 18,000, 20,000 children, and they were only capturing or enrolling in the program roughly about 15 percent of those kids. So it was a huge screening effort for very little yield. And so based on some work that we did through this project we thought that they might be resulting in a lot less screening and still identifying children if the screener was used on the application. There’s still further work being done to refine this, and Title V is going to do a great deal of review. And we’re going to continue to work with them. Plus, we've recommended to them some form of ongoing screening of the claims data. And we've not yet decided how to best implement that and we have to look at cost considerations and so on. So there’s been a lot of progress made to implement this approach and Florida, and Texas is also looking at a similar kind of approach.
Okay. I want to shift gears to the quality measures and I'm going to talk a lot about the medical home. We use the primary care assessment tool to measure the medical home dimensions. And the primary care assessment tool was developed at Johns Hopkins. It is a rather complicated tool to administer, but it’s a very good tool that measures experience with care, not simply satisfaction with care. And I think that’s a really important distinction. It asks the family, like, can they get services, for example, in the same day if their child’s sick, not how satisfied they are with those services. And so that is an important distinction and there’s actually a good body of literature that talks about the differences between experiences with care and satisfaction with care.
And we've related the different domains of the primary care assessment tool to healthcare use and charges for the children after considering their socio-demographic characteristics and their health characteristics. And so here’s the comparison of the P-CAT to the medical home. And I apologize. I meant to bring copies. I only have one copy showing how these cross map and the sample questions. I can pass around the one page that I have and we can get copies to you. But you can see that the dimension on the primary care assessment tool correspond very nicely to the medical home dimensions in terms of assessable. And the longitudinal relationship is broken out in two components.
It’s how long the child has been going to that doctor and how often that they go to that particular doctor for their needs and then also how much that doctor treats the child like a person and not as a medical issue or a problem. Comprehensiveness of care is broken down into the services that the family perceives are available and those that the family says that they got. And then, of course, care coordination and cultural competence and family-centered care. So they map out very nicely.
Now, what we did was we use our Title XXI population here and at the time this study was done Title V, although they were moving very severe children into that program, they were as aggressive as they are today. So there were a fair amount of children that really had some complex special healthcare needs. We screen the claims and then counter data looking for diagnoses that might indicate a special need. Eight percent of the children in this program had a diagnosis that fit our criteria. Then we use the QUICC to look to see if the child was having consequences of their condition. And then the consequences were compensatory mechanisms, elevated service use, or functional limitation. And the child had to have at least one of those consequences to actually be in our study. And what we found was 65 percent of the children that had a diagnosis did, in fact, have at least one consequence for the condition.
And just a little bit about the sample. It was predominately White and non-Hispanic, but there were a good percentage of Black and Hispanic families included. And I'm sorry about the scale on this. I don’t know what happened. But this is how it broke out in terms of how many met one, two, or three components. And basically, 30 percent of the children had one of the consequences on the QUICC, 35 percent had at least two consequences, and about 34 percent had all three consequences based on the QUICC. So the children, by and large, had at least, you know, two or things that they were experiencing here.
Now, in terms of the scores and the primary care assessment tool, here I’ve broken out all the dimensions and show what the scores are and largely it ranges from zero to four. There’s actually one scale that can go slightly above four. And you can see that remarkably the scores in a lot of areas are fairly low. And that’s one of things that when people have seen this they’ve commented. And said, "Jeez, these are some of the worst," they call it satisfaction scores, "I’ve ever seen." Well, it’s not satisfaction. It really is the family’s experience with this. Access really, when you think about how highly people rate care, is fairly low. But the lowest scores were on the comprehensiveness areas and in care coordination. So that’s how it shook out just descriptively.
Audience Member: And those were for healthy kids?
Betsy: Yeah. Well, it’s for the Healthy Kids program but it was for children with chronic conditions in the Healthy Kids program.
Audience Member: Oh, okay.
Audience Member: Okay, okay.
Betsy: Okay? Yeah.
Audience Member: --to be comprehensive there’s a lot of stuff that’s coming from the on-site as important, correct?
Betsy: Oh, yeah.
Audience Member: Community Health Center not a small practice could actually delivery for--in terms of [inaudible] structured--
Betsy: Yeah. Although I do--in this program now I know Healthy Kids sounds--the name of the program is Healthy Kids. Let me go back and reiterate. We screened for diagnoses for children that have a chronic condition. And then if the children had chronic condition, we administered the QUICC to be sure they were having a consequence of the chronic condition. The children that then did the survey, the families that did the survey, are only those whose children had a chronic condition and were experience a consequence of a chronic condition, at least one.
Audience Member: Oh, at least one.
Betsy: At least one, although, two-thirds of the sample had two or more. So I just want to clarify that. So these are children with special healthcare needs. They may not have been severe enough to be in Title V, but they do have special healthcare needs and they do have consequences to their condition. Most of these children, about 80 percent of these children were seeing a physician in a private office. These were not public health department, community health centers, or whatever. They predominately were cared for in pediatrician offices.
John: Just curious, if you had separated this out by age, would you have seen a difference in the adolescents with relatively young kids?
Betsy: The program only covers ages 3 to 19. So it doesn’t cover anybody below three years of age. And I don’t know. We did not do that, although, on our statistical models age was not significant. But we didn’t break out descriptively that way.
John: Sometimes I wonder whether in the adolescent age group because I don’t know. But do you think that maybe in that age group because the area is where you do the worst in coordinated [inaudible] healthcare.
Audience Member: Did you use the P-CAT as a telephone survey--
Betsy: Yes.
Audience Member: --or is this self-administered questionnaire?
Betsy: Telephone. Yeah. I would not have, yeah, sent it to anybody. That’s for sure.
Lynda: Betsy, can you give you an example--I hate to--of kids who would meet the criteria but not need the healthy--the Title V? I mean, the Title V program presumably--obviously it’s a program focused on providing care in children with special healthcare needs--.
Betsy: So who would have a diagnosis in it? Yeah.
Lynda: --So I guess I'm wondering, you know, and it is voluntary, right?
Betsy: Yes.
Lynda: So the kids that are in this Healthy Kids program but they don’t qualify for Title V or is that they don’t want to go into the Title V program? Or what’s the--
Betsy: Well, if they're found they have to go.
Lynda: If they're found they have to go?
Betsy: Yeah. If they're found they have to go. That’s absolutely true. There’s a variety of diagnoses both physical and mental health. Believe it or not, there is a significant cohort of kids either have schizophrenia, serious effective disorders, malignancies, cystic fibroses. Some portion of these children, according to the CRGs, actually were classified as catastrophic.
Lynda: And yet, they didn’t meet the criteria? I guess I'm kind of--
Betsy: Well, that’s why Children Medical Services actually over the last couple of years have become much more aggressive in the identification. And that’s why they wanted us to work with them to be sure that their identification processes are working. They said that they have huge variability throughout the state where some--the eligibility is determined at the district level. And even though they're state criteria and so they said that some children, some districts are very strict. Other districts are too liberal. Some kids in here certainly have asthma severe enough to have a consequence to it. There’s some of those. Some of the kids in here have ADD, ADHD, severe enough to have a consequence. So there’s some of those. It’s a mixture of children. And in keeping with the non-categorical approach, we intentionally left the children clustered together and characterized them by the number of consequences to their conditions. So you're right. I mean, now if we were to repeat this now we’d probably have a lot fewer children. We did interviews with about 1,300 children to do these analyses. I would suspect that perhaps half of these children might have been moved into Title V today. They're being much more aggressive.
Lynda: And just one more question. Have you done something like this for the kids who are Title V?
Betsy: Yes, we have. Yeah. And we don’t see as striking results with that group. And I'm not sure if it’s a sample size issue or if it’s simply because the severity of their illness is so condensed down that we’re not seeing some variability. But we’ve just repeated an analysis for Title V.
Lynda: I mean, is there a difference in the P-CAT’s scores for, I mean, going back to--this is a program that’s focused and organized around children with special healthcare needs. You would hope that the P-CAT’s for score for their medical home would be--
Betsy: They're somewhat higher--
Lynda: --higher.
Betsy: --not a whole lot higher.
Lynda: Not a whole lot higher.
Betsy: Yeah. Yeah.
Ardis: That's not surprising. We know we have a challenge delivering...
Betsy: So, just to talk now about how this related to outpatient use and charges. What we found was, of course, socio-demographic and health characteristics were extremely important. Outpatient charges were about $250.00 per member per month for these children. So you can see that they were certainly more expensive than your average child. But certainly not as high as what you’d see in a Title V program. But what we found was even after we considered health characteristics in the medical home, the charges per month were lower for Black children relative to White children. We all know that there’s a huge body of literature that says that minority families tend to receive less intensive treatment than non-minority families. And unfortunately, we are seeing this in a cadre or group of children here with special healthcare needs. And then as you’d expect, the more consequences the child was having to their conditions, the higher of the use in charges. So that’s not unexpected. But what happened with the medical home? Now, again, remember this is even after considering the child’s socio-demographic characteristics and their healthcare consequences, the access to care--the higher the access to care there was a 16 percent decrease in use and a 38 percent decrease in a child’s charges. The continuity of the relationship with the provider was very important in terms of decreasing use and decreasing charges. Comprehensiveness of services provided also resulted in a 27 percent decrease in charges. So there seem to be a lot of positive findings in terms of encouraging certain types of care for the children.
Audience Member: The definition of use is a claim?
Betsy: Yes.
Audience Member: State claim?
Betsy: Claims.
Unknown Speaker: So that would mean that they're not making telephone contact or other kinds of use of services.
Betsy: That’s true. We had to rely on claims data, but again, we were looking at it related to what was meaningful to health plans. And so this is related to paid claims what we saw.
Audience Member: I just want to make, to understand this. The controls were for kids who were not in a medical home?
Betsy: There’s no controls.
Audience Member: So--okay.
Betsy: It’s looking at the scores for this group of children.
Audience Member: Okay.
Betsy: Susan?
Susan: To a certain degree the--if access is high and the use was correlated with lower use, couldn’t that also be just a function of less complexity of need? So, you know, my needs are small, therefore, access to what I need is high. Therefore, utilization is--
Betsy: Well, again, within this group we controlled for the consequences of their condition--so one, two, or three. So even after considering that we got these results. So remember that in the statistical models are the consequences to the condition and the socio-demographic characteristics.
Ardis: That doesn’t capture the severity of the illness but may drive care.
Betsy: It maybe inadequately getting at the health status, that’s true. Jim.
Jim: --fascinating data--These are cross sectional, is that right?
Betsy: Yes.
Jim: So you're really talking about the....
Betsy: The survey was administered and we look at their use one year after this survey. Then we looked at care coordination and family-centered care. Now, care coordination did increase use rates and charges by quite a huge amount. Family-centered care decreased use rates. Now, we looked at the odds of ER use and there were only two things that were related to emergency room use. The more consequences to the condition, the higher the odds of ER use. But we found that the more culturally competent the care, the lower the odds of ER use. So that was--
Ardis: Can you go back one second?
Betsy: Sure.
Ardis: Can we go back to care coordination issue? Why do you think you're saying that you increased charges when care coordination was avaiable?
Betsy: Why do I think that?
Ardis: Yeah. That's intuitively not fitting with--
Betsy: --Well, actually, we spent a long time talking to the physicians about this. And one of the things that the physicians commented on is that sometimes initially when you provide care coordination, you actually might see an increase in service use. Because the children are getting services that they otherwise would not have gotten. And what they suggested was because we had these longitude claims data that we look another year and see perhaps has this changed or declined to cross time for this group of children.
Ardis: You're catching the kid a at 1.5, not at the beginning of getting care coordination.
Betsy: That’s correct. We’re catching them at one point in time.
Ardis: They may have had an inferior [inaudible].
Betsy: Sure. And we did control for the length of time that they were enrolled in the program for these findings, and we didn’t see--that was not significant.
Arids: I wonder if you're getting disease severity here, because who get care coordination are more complicated--
Betsy: That certainly could be. That certainly could be.
Unknown Speaker: --that could account for the charges here--
Faye: Could you explain a little bit about how you define the finding family-centered care what it actually was consisted of?
Betsy: Yes. There’s a series of items on the primary care assessment tool that ask about the nature of the family-centered care. To what degree the family’s opinion is considered, to what degree they’re considered in the decision-making. There’s a series of questions on the primary care assessment tool that do address the Family-centered care issue. And it’s scored, according to what the developers at Johns Hopkins recommended, and I have don’t have the tool with me. It actually is publicly available. It’s very easy to get a hold of.
Lynda: Betsy, did you--were you able, I don’t know if your sample size was large enough. Were you able to--you said most of the kids saw private physicians who had been contracts with the plans. But did you see any difference in terms of size of practice and whether or not there were, you know, group practice verses solo practice, for example. Or by health plan or where there any staff models or group model HMOs?
Betsy: Well, one of things that we’re doing right now is we’re actually doing an analysis to break this out by health plan characteristics, which is a huge challenge in and of itself because so many of the health plan characteristics are correllated with each other. And so it’s very difficult to try to break those out and do any of those kinds of analyses. So that part we’re doing.
The provider part actually is an enormous source of frustration because we have tried for years over and over again to interview physicians about the characteristics of their practices. The response rates regardless of what we do are notoriously horrible. They hover in the 30 to 35 percent range. We even commissioned a famous artist in Florida to do an original oil painting of an angel holding two children and gave that as a raffle. This woman’s famous and she does these kinds of things. I personally didn’t like the painting, but be that as it may. And the response rates just were not very good. And so you're left with 30 to 35 percent of the physicians responding about whether they're in large group or what they are. And we do use Department of Health databases, but there’s limited information that’s on there. So, you know, I'm certainly open to ideas and suggestions about getting better responses from physicians. We put the survey on the Internet, told them they could do it on the Internet. And we did on--we offered them a phone survey, we offered to do it with their office manager. I mean, everything we tried and it’s very difficult.
Okay let me go through because there’s actually just a fair amount of things here. And again, I know this is hard to have just quick highlights. The odds in-patient use we saw the same thing here where there certainly were some dimensions associated with increased odds of in-patient use, although access to care, the higher the access to care the lower the odds and in-patient use. But in some of these other areas we did see increased odds. And some of it certainly maybe related to picking up severity issues. Okay. So that--
Audience Member: [Inaudible].
Betsy: Yes. Yes. But again, that may be inadequate, although in a lot of work that we've done we've seen a strong relationship between the number of consequences and elevated use in charges. I mean they do seem to function as a severity measure.
In terms of looking at the unmet need out-of-pocket spending and health status, this was done with a Title V population. And we statistically looked at new enrollees in the program. They were in Title V less than three months. And compared them to children that were in the program for 12 months or longer. This was not longitudinal. It was cross sectional for each. But we did control for whether they were new or established enrollee. And then we looked at unmet need, out-of-pocket spending and health status between the two groups after considering variables like socio-demographic characteristics and their health consequences.
And just descriptively you can see that 44 percent of new enrollees reported some kind of out-of-pocket expense in the amount of--for a total pool of over $1,000 each month, and of course, a huge standard deviation here. Twenty-three percent said they had unmet needs and then for the established enrollees the percent reporting out-of-pocket expenses dropped dramatically. It dropped to around $300 per month on an average. But again, with a very standard deviation, and about 16 percent reported unmet needs.
Bobby: Do you know where that unmet need came from?
Betsy: Yes. The survey is actually very detailed and what we do is we go through everything. We say physical therapy, speech therapy, occupational therapy, there’s a whole sequence that we go through. And we asked about whether they needed that service, if they needed it and got it, who paid for it or if they needed it and didn’t get it why not. And there’s a whole series of question that go for each specific need. We simply aggregated them for the purposes of these analyses. But we know each one.
In terms of the health status measures, this is using the child health questionnaire. And we looked at the physical summary score and psychosocial summary score just for this part of it, but we have the results for the whole thing. Then you can see that the third bar was the United States average and that both the new and established enrollees are below the U.S. average in their physical and psychosocial functioning. But you can see not a lot of difference between new and established enrollees and the results.
Lynda:. This is all--average of all kids verses new and established enrollees of children with special healthcare needs?
Betsy: These two children--this is a new enrollee group in Title V, this is the established enrollee group in Title V. So they’ve been in 12 months or longer. And this is results that the developers got for the United States average. They did a lot of survey work. We didn’t do this.
Lynda: And once the survey--what about the physical and psychosocial? What do they measure?
Betsy: We use the child health questionnaire and the child health questionnaire has various dimensions on it. There’s a physical summary score and psychosocial summary score that you can get from it, along with other things. There’s many other dimensions that are measured on the CHQ. We just pulled out two of these just as illustration. And basically, the children, as you would expect in Title V, are less healthy than the U.S. average is or lower scores than the U.S. average is. But you don’t see a whole lot of difference in the scores.
Lynda: Well, that’s what I'm puzzled. I guess that’s where I'm going with this is I'm puzzled that there isn’t a bigger difference.
Betsy: Between the new and established enrollees?
Lynda: No. Between those--that group as a whole and the U.S. average. I mean, that’s only what, less than a 10 percent difference? About 10 percent?
Betsy: Actually, it’s a significant difference.
Lynda: Is that significant? Okay.
Betsy: It is.
Lynda: So for a--for someone who--
Audience Member: What you're looking at is activities of daily living type of stuff?
Betsy: It’s looking at a whole range of things in the [inaudible]). Yeah.
Lynda: So that is significant. For somewhat, you know, who didn’t do so good in statistics I guess!
Betsy: Yeah.
Lynda: You're telling me that’s significant.
Betsy: Yeah, it is.
Lynda: Okay.
Betsy: It is.
Lynda: Never mind.
Audience Member: [Inaudible]
Betsy: Yes. It’s got a whole variety of things about their activity and what they could do.
Audience Member: [Inaudible].
Betsy: Yeah. Yes. But there’s no good one. Okay. Is there--did you have a question? No. Okay. Okay. And I know this is an awful lot information. We did a lot in the four years and it’s hard--and I was trying to get it across as quickly as possible. So I know it’s hard to go through all these details. This is, again, a Title V population. The first bar shows new enrollee, the second bar shows the established enrollee. Largely again a White non-Hispanic group, but you do a fairly significant percentage of black and Hispanics. And in the other little section over here we did do the CSHCN screener with this group and it shows how many of those in Title V didn’t meet any of the components, one component, two component, or three components. And, of course, you’ll very quickly noticed that about 22 percent to 20 percent of the kids, even though they were in Title V, parents that didn’t meet any of the components in the CSHCN screener. And again, we talked about what those issues may be.
Lynda: And again, were there any trends in terms of who those kids were? I mean, do you have sense of who were--
Betsy: Slightly lower educational level on the part of the respondent. Less than high school. But we didn’t see any race ethnicity effect. It was predominately an education effect. Now, what we found is that even after--
John: Age differences?
Betsy: No. No. In terms of the out-of-pocket spending, we did find a significant decrease in out-of-pocket spending after considering health and socio-demographic factors for established enrollees verses new enrollees. But it was--that for establishment was just .65 times that of new enrollees. We did find that out-of-pocket spending for Blacks and Hispanic families even after we considered everything else was lower than what it was for White and non-Hispanic families. And that was even after controlling for income. And, of course, out-of-pocket spending remained higher--the more consequences to the condition, the higher the out-of-pocket spending. Age, income and education was not significant and that was child age, family income and respond and education.
Now, we found that unmet needs also declined between the new and established enrollees, but that was only marginally significant. No other variables were significantly related to unmet needs. And here is where we get into these health measures.
And I want to bring up these health measures because I don’t know how it is in your state, but at least in Florida and also in Texas there’s a huge emphasis at the state level about wanting to prove that the children are healthier when they get in these programs. And basically what we consistently find no matter what we do is that missed school days, bed days, really it’s related to the health of the child. And we really don’t find anything else significantly related to that. Now, part of it I think is problems with our measures, and I think there’s a whole array of issues that we could discuss here. But essentially, what we found is no program effect. That it was really related to the number of components that they met on the screener in terms of how many bed days and missed school days we’re going to have. No other variables were significant at all.
So just quickly in terms of implementation, Florida’s Title V program is going to a capitated model, although that’s against I think what a lot of other people are doing but they’ve decided to do this. They do want to use the medical home concept extensively in their further work, as they switch to capitation and we’ll be helping them with that. And they are going to incorporate that as part of their ongoing quality measures. Out-of-pocket spending is an important component, but we really don’t see any other effects, at least in our ability right now. I think some of it may be measurement issue in terms of looking at unmet needs or looking at overall the child’s health status. So we will be working with incorporating the medical home concept into our ongoing assessment at a state level. I'm not sure what we will do with some of these other quality measures.
Some of the barriers that we encountered, it’s difficult to implement these things at a health plan level. They really don’t want to get involved in the cost of the survey, how expensive it is. Both for identify children and also for some of the detailed quality measures that are done. There is a lot of concern on the medical home results that I presented about the findings related to in-patient use because people wanted everything to show this nice, clean decrease in use for in-patient and ER, and it didn’t turn out that way. And so there’s a lot of concern about how people at the MCOs will perceive it. And how people at the state level will perceive that are doing budgeting. We did send the paper to the American Association of Health Plans for review and they are looking at it. But of course, the initial reaction was like "Oh, my goodness. We can’t do anything that’s going to increase in-patient use."
The other issue that we had to fight was in terms of getting the CSHCN screener on the application, it’s very costly actually to change an application. And a lot of the costs have to do with the third party administrator who processes all these applications and enters it into a database. And it resulted in a lot of programming changes and cost for them. So there were actually a lot of barriers predominately related to money in terms of accomplishing a lot of things.
Audience Member: How long is it? How many questions?
Betsy: How long is--I'm sorry? Well, the CSHCN screener actually is five questions but it has three components. And the problem is that the state only allowed five lines. You may have five lines of this application and you may not have anymore. And so we ended up having to do was create a stem that said, you know, does your child have this condition--are these following condition that has lasted or likely to last, you know. So we had to create a stem, which I'm sure that the developers would have a fit and say, "Oh, jeez, this changes things." But the fact of the matter is it was five lines or you don’t get it at all. And so we had to make a change.
Audience Member: We didn’t do that with uninsured survey in our state in the same way. We coudn't get it on...
Betsy: Right. Right. So there are--there is a technical report that we developed on identification, which revising and repost in our web. The Medical Home paper has sent to a journal. There’s also a couple other articles that we are sending off that will appear in article format, but they’ll be issue briefs to go along with them. And then a lot of our products really have been in terms of state consultation. We worked very closely with Texas. We’re also now working with two other states where we’ll duplicate some analyses for them. And show them what we did, how we did it so that they can try to do that within our own states. So some of it also has been at the individual consultation. So I appreciate your attention. This was longer than I thought and I know it was a lot of detail to go through. But that was our project.
Lynda: Betsy, don’t feel bad. Everyone has gone over [time].
Betsy: Okay.
Lynda:. Does anyone have any other comments on this? Just as note, Betsy’s also going to be talking a little bit later about her work through the cooperative agreement where they're focusing on issues around cost and financing. So they're doing some very, very exciting things under that cooperative agreement. Faye?
Faye: Coming from Illinois, a state that feels the only identification need for Title XIX and Title XXI related to special healthcare needs is whether or not you're an SSI recipient. Do you have any advice or guidance about how to promote the fact that Florida and hopefully other states are doing some identification, and why another state might want to do it?
Betsy: You know, one of things that I think is really striking is some of the results we represented for the Healthy Kids Program that there are children in there that really probably would benefit from being in a more specialized network. And that’s why Florida’s taking the active step to move them out. And unfortunately, it’s too soon to say what will happen to these kids. And we’re are going to look at their care from the time they were in this Healthy Kids Program to the time that there--from the time that were moved then into Children’s Medical Services. And what the outcomes were for these kids. But frankly, it’s going to be hard to justify without that kind of information. Without outcome information, without looking at where they more or less expensive in one program versus another. And then you have all the turkey nature of the development changes in these kids as they age and the changes in their conditions. So I don’t have really, frankly, a good answers for you. I mean Florida just has always been extraordinarily liberal in how they’ve dealt with children with special healthcare needs. Even with a Republican governor and a Republican legislature. They continue to be very liberal when it comes to kids with special needs. Jim.
Jim: Betsy, I think that--I'm just sort of wondering how to frame your in-patient care stuff, which is interesting one. And--you’ve looked at using some of your risk adjusters as well to see whether that effect the findings--Because being cross sectional I guess I’d be less worried about the findings frankly. If I knew that you had an experimental study where you have--you've got care coordination, you’ve cared for kids with or without over time and you still found out, I’d be far more worried. But this doesn’t sound too surprising.
Betsy: Yeah. I'm not surprised either. And one of the things that actually I think is if you do have a physician that’s more sensitive and more in tune to the family, because you can see that it clustered also with family-centered care and some of the those more softer dimensions, if you will, that I just think that you have a physician that may be more in tune to needs and coordinating the care, and getting immediate care to the kids. I'm not, I'm personally not that particularly worried about it, but it really raised a lot of red flags to a lot of people.
Jim: We've look at children who--with chronic conditions and getting care from primary care physicians almost entirely. Children are getting predominately from sub-specialist and children are getting sort of from both. And we’re not sure what’s going on. In fact, we’re continually puzzled by the data. That middle group kids are getting care from both sub-specialists and primary care doctors on a regular basis are far more expensive--
Betsy: Yeah. We said the same thing actually.
Jim: --than the other two categories however we looked at them.
Betsy: Yeah. Yeah. We see the same thing. And we have done some things, running through the ACGs and it really changes things a whole overwhelming amount.
Lynda: Susan.
Susan: Well, I was going to say that I think one--I agree with what you were saying about it not being surprising. And I think one of the important things about these data are to frame it as based on some indicators that are associated with medical home. Not based on the intervention establishing that problem because you clearly haven’t sort of measured those things and then created--
Betsy: Oh, right. Right.
Susan: --new interventions and then measure, you know--
Betsy: Yeah. No. It’s definitely not experimental. That’s true.
Susan: But I think that, you know, the use of the word "medical home" may carry some communication around expectations as oppose to the tool that you use that it is highly correllated with some of the components...
Betsy: And we think the findings of overall are extremely positive. I mean there are a lot of very positive findings about reducing, you know, ER use, at least in one instance. Access to care does reduce the odds of in-patient stay, and there are a lot of components associated with reduced outpatient use and charges even after you consider the kid’s consequences to their condition. So I think all and all it’s extraordinary positive. But, you know, I guess when you deal with some groups right away they look at the one thing they don’t like.
Audience Member: Can you give us your web site?
Betsy: Sure. It’s www.ichp.edu.
Lynda: And did any of the--
Audience Member: I just want to say we've had similar experiences in our special kids special care program for complex--medical complex children who are in DSS foster care. The data is coming out soon. But the first year the expenses are higher because they're getting care. And your study is only one year out and I think looking over time and the satisfaction on the part of the family with care coordination is really an important measure I think.
Betsy: Now, we do, just as an aside, have a separate group of children that we have found longitudinally. And we have on the longitudinal follow up striking drops in their charges. So actually even though this was cross sectional, actually have some hope for our longitudinal group that we will see some of this play out in terms of dropping--to get what they initially will drop.
Lynda: You know, the other thing I think we have to not forget is that from health plans perspective, pediatrics is small piece, a very small piece of the pie. Then we’re talking about a group within that population. So, even though you’ve seen this increase in hospital charge, I mean, how much does it significantly change the overall premium for a health plan?
Betsy: Right.
Lynda: I don’t think it’s--you know, we’re talking big numbers.
Betsy: Right. Right.
Lynda: So--and then the other thing I want to go back to, Faye, your comment about, you know, Illinois and only focusing on SSI. That I do think, unfortunately, that’s going to be--Illinois is going to be in the majority and not the minority because again, seeing the potential impact of the change of the criteria and allowing states to either choose the MTHB definition, a categorical definition or a functional definition. My prediction is, and I hope I'm wrong, is that a lot of states are either going to either pick categorical and focus on SSI or go the way of Ohio and look at functional and pick, you know, three of their top diagnoses, which ends up being asthmas, diabetes, and ADHD. So, you know, obviously, I think our jobs are going to be really get states to appreciate the importance of using a broader definition. But it’s going to be hard.
Faye: Just look at SSI, there was something I’ve been trying to find out about five years that I cannot get information about out. Social Security Administration reports over 45,000 children on SSI in Illinois. The Department of Public Aid says every year in the block grant for the last five years, there are 14,000 who are on Medicaid and SSI.
Lynda: I mean, I don’t know. Tom, do you have any sense of where--
Tom: No. Those numbers should be the same.
Lynda: Match. Yeah.
Tom: SSA would have the--I would go with SSA’s numbers. Those are the right numbers for kids collecting a check.
Faye:: The question is where are--
Tom: Right.
Faye: --where are those 30, 000 kids?
Jim: They're probably on Medicaid but they're probably not labeled correctly in the Medicaid database. We actually work on those, with SSA four or five years ago to try to link Maryland Medicaid data with Maryland SSI data, SSA data. And, I tell you, the linkage is very, very difficult, first of all for it’s technical reasons. But the results were very inaccurate linkages. And in fact, the states don’t normally have a good SSI maker on their Medicaid identifiers. So, they may not even be accurately counting those 14 or 15 thousand kids. SSA is more accurate certainly.
###
End of Transcript