Ninth Annual Maternal and Child Health Epidemiology Conference / December 10-12, 2003

What Does the New Data on Children with Special Health Care Needs Tell Us?

STEVEN BLUMBERG:  My role here today is to give you a flavor of the National Survey of Children With Special Healthcare Needs, to just highlight some of the State data from it and to talk a little bit about whether or not you can create a general indicator that shows how well a State, and perhaps the word isn't functions, but how well children with special healthcare needs are doing in a particular State in general.  Michael gave you a little bit of background.  Let me just highlight some of the key things that you need to know about the National Survey in order to best understand it.  Obviously there's lots of detail out there.  At the end of this there's a website that I'll give you that you can go out and find more information than you could possibly want about the Survey.  I'm going to be followed later today by Christi Bethell who will also then give you a website where you can find all you will ever want to know, well maybe not all you'll ever want to know, a lot of what you'll want to know about the data.  The National Survey of Children With Special Healthcare Needs was conducted in 2001.  It started October 2000 and lasted until April 2002.  As Michael said, we are grateful to the Maternal and Child Health Bureau of HRSA for funding this survey.  Its purpose was to produce National and State-based estimates of the prevalence and impact of special healthcare needs among children under 18 years of age. 

To do this, we conducted a telephone survey, random digit dial telephone survey in all 50 States, plus the District of Columbia.  The samples were independent.  That means that, the samples were drawn independently within each State so that we would be able to create State estimates from this data set.  In terms of screening, the basics of how this survey was conducted, once we called a household or called the telephone number and identified that in fact it was a household, we asked if there were any children living in the household.  If so, we collected some demographic information about those children and then screened those children for special healthcare needs.  The screening tool was the CSHCN screener that was developed by the Child and Adolescent Health Measurement Initiative.  Ultimately we screened 196,888 households with children, screening just over 373,000 children for special healthcare needs.  If we found a child with a special healthcare need in the household or if there were more than one, we randomly selected one child with a special healthcare need within that household and interviewed the parent or guardian who was most knowledgeable about that child's health.  So ultimately, we completed just under 39,000 interviews concerning children with special needs.  The goal was to get approximately equal size samples from every State and it works out to be approximately 750 children with special needs in each State.  The response rate was 61 percent. 

And here's the original first look at the prevalence of children with special healthcare needs.  If any of you were in the session yesterday afternoon you've seen some of these numbers.  For all 50 States and the District of Columbia, that's 12.8 percent of children have special healthcare needs.  Here you can see the distribution, the prevalence of rates are highest in the Northeast, along the Mason Dixon line, down in Louisiana.  They are lowest out in the West and Southwest, as well as Alaska and Hawaii.  Now you're going to see a lot of pages that look like this.  It's early, it's the last day of the conference, I decided to show you pictures instead of text.  A couple of things that I just want to make you aware of, the number that you'll see at the bottom, where it says, "All 50 States and D.C.," that's a National estimate, it's not an average of the 50 States.  So, larger States, obviously, are going to be contributing more to that number.  In all cases, I've tried to split the States up into sort of six groupings and assigned colors to them.  There's nothing special about those groupings.  They're not particular standard deviations or top 10 percent, bottom 10 percent. 

I was looking for nice round whole numbers to make it easier to interpret.  I want to just highlight 15 key indicators that come from the survey.  In selecting these 15, there really wasn't anything special, other than trying to identify a group of things that you can determine from the survey.  Obviously, there are plenty more indicators that one could chose.  These 15 that I have selected simply are, perhaps, highlights of the survey.  Eventually there will be a chart book that is released by the Maternal and Child Health Bureau providing National and State data.  These indicators, as well, are in that chart book.  So, two indicators on child health, three on health insurance coverage, five on access to care, one on family centered care, and four indicators on the impact of the child's special healthcare needs on the family.  Handouts are in the back of the room.  Hopefully you guys picked them up because I'm not going to have time to spend much time on any one slide here. 

So we'll move fairly quickly.  Obviously, I'm happy to take questions later on if you have any.  So, starting out with child health, the percent of children with special needs whose conditions affect their activities usually always or a great deal.  We asked two questions; first we asked about how often the child's special health need effects his or her ability to do things that other children do.  And here we've included anybody who said, "usually" or "always."  But we also recognize that there are certain conditions where the condition may not affect their activities on a regular basis, but when they're experiencing a flare-up in that condition, a particularly severe episode, it can be quite disabling at that time and so we've also included them here.  We look at school absences due to illness.  For most children the average is about three school absences due to illness in a given school year.  This shows you the percent of children with special healthcare needs with 11 or more days of school absences.  Turning to health insurance, we can address the percentage of children with special needs without insurance at some point in the past year, which is 11.6 percent nationally. 

We can look at what percent were without insurance at the time of the survey.  This number, obviously, would be smaller than ever in the past year and it's down at about five percent.  We also asked a number of questions about how adequate the insurance was.  Now when we talk about that, what we asked were things such as, "Does the insurance cover all of the healthcare needs that the child has?  Are the costs of what's not covered reasonable?  Does the insurance plan allow the child to see the providers that he or she needs?"  And, as you can see, among those who are insured, one-third reported at least on one of those characteristics that the insurance was not adequate.  Turning to access to care, the survey includes 12, 15 services needs from general preventive medicine to specialty care, to dental care, to communication aids and prescription medicines.  This is any unmet needs.  The way that the questions were set up we asked, "Did your child need this particular type of service at any point in the past 12 months?  If so, did the child get all of the services that he or she needed?"  When I talk about any unmet need, it's that for at least one service it was reported as needed and the child didn't get all that he or she needed in the past year.  Percent with any unmet need for family support services.  Here we're talking about things such as care coordination, Respite care, genetic counseling. 

Percent needing specialty care who had difficulty getting a referral.  And here we first had to find out that the child needed specialty care, then also that the child's insurance requires a referral, and then we could ask, "What percentage of children needed specialty care, but had difficulty getting a referral?"  And as you can see, it's about one out of five nationally.  Percent of children with special healthcare needs without a usual source of care.  Our definition of usual source of care does not include the emergency room, and you can see that about nine percent of children with special needs are without a usual source of care.  Personal doctor or nurse, 11 percent nationally of these children do not have a single person they can identify as a personal doctor or nurse.  Family centered care.  This was a group of five questions that asked people to rate how often certain types of family centered care was provided.  How often does the provider treat you like a partner in your child's care?  Provide enough information to you to make decisions about your child's care?  Information such as that. We combined all of those.  If for any of those five items the parent said, "You know, that rarely happens or only sometimes happens," we considered them to be without family centered care.  And you can see that one-third of children with special needs were without family centered care.  We then turned to impact on the family.  Impact on the family first and foremost can be financial.  So we asked about medical expenses in the past year that were not covered by insurance, essentially out-of-pocket expenses.  And this shows that 11 percent paid $1,000 or more.  Of course, for some families $1,000 or more may not cause financial problems. 

For others $200 can cause financial problems.  So we asked specifically about whether the condition caused financial problems for the family.  In addition to finances, having a child with a special healthcare need can be draining on parents due to the time involved in providing or coordination care, so we asked about the number of hours that are spent per week providing or coordinating care.  And we asked whether the child's condition effected the employment of family members, and you can see that nearly one-third, or 30 percent, said that, "Yes," indeed, the condition had effected the employment of family members.  Can we put all of this together into a composite indicator?  Well, of course, we can.  But I do want to make sure that you recognize some of the limitations that occur any time you try to create a composite indicator.  First, is a composite indicator useful?  Does it make sense to try to rank States against each other on an overall measure, or should we really keep our focus on each of these components when we talk about States? 

If we decide that a composite indicator is useful, which indicator should be used as components of the composite?  For our purposes, I'm going to use those 15 indicators that I just gave you.  Recognize if you drop one of them, if you add something else to them, you could change these State rankings.  Should some indicators be more important than others?  It certainly could be argued that some of those indicators are particularly crucial, some may not be as important if you want to try to compare States.  I have not tried to make that determination.  For this purpose, I've kept them all the same.  Are relatively small differences between States on particular indicators meaningful? 

So, if you've got something that's .04 difference, do we care?  Well recognize, if we for instance averaged the ranks, here you can see that those two are one rank apart, but so are States C and D one rank apart.  So if you just average ranks, then small differences between the States can become magnified.  Are extreme values meaningful when comparing States?  If you've got one State that really is far beyond the others, should that come into the composite?  If you just average ranks, it won't.  Again, 50 to 51 is just one jump.  If, on the other hand, you average percents, it will.  But, there's a caution here.  If you were just averaged the percents for the States across all these 15 indicators, one of the complications there is that implicitly you're starting to say that some of the indicators are more important than others.  Here's an example, if you were in a class and you're taking two tests and on one of those tests there was a wide distribution, some people got A's some people got F's.  And on the second test everybody got a B.  When they average those two together and give you final scores, final grades in that class, guess which test made more of a difference in your final grade, or in your difference from the other students?  It's going to be that one where there's a wide distribution.  One way to avoid that problem is to use standard scores.  Think back to your statistics classes, it's Z scores. 

Here, rather than taking the average of the percents, you take each percent subtracted from the mean, then divided by the standard deviation and that gives you a standard score.  For this particular measure, how far was this State from the mean of the 50 States?  That's what I've done in terms of creating a composite indicator.  This maximizes the impact of extreme percentage scores on a particular indicator.  It minimizes the impact of small differences and each indicator now has an equivalent impact on the composite indicator.  Those of you who are familiar with the kids count data books, this is what they do.  And ultimately, here's the rank after averaging the standard scores.  The best scores, those States where children with special healthcare needs are doing the best are right in the Northeast there, Connecticut, Rhode Island, Massachusetts, also Michigan and Iowa.  The worst, Montana, Nevada, Florida, Mississippi and Washington, D.C.  What I've drawn here is the States across the board so you can look up particularly how your State compares, because you may see that, in fact, your State really doesn't differ much from the States that are right around it.  I'm running out of time, or I'm out of time, so let me just conclude with a couple of thoughts.  Once again, recognize that the choice of indicators that are used in the composite are particularly critical for any composite.  I'm not claiming these 15 are the most appropriate indicators.  I also want to make you aware that if you start asking yourself, "Well, why do States differ?"  One thing that I found was that the composite indicator is highly correlated with the percentage of children in each State who live in households with income below 200 percent of the Federal Poverty Line.  The correlation coefficient is .71. 

In other words, 50 percent of the variation between the States is due to or associated with the variation in income levels of households with children in that State.  I would argue that a composite indicator unrelated to income may be desirable, but I haven't yet figured out a way to create one.  That's something I'm in the process of doing.  That's it.  There's the website. That'll give you all the information that you want to know about SLAITS.