AMCHP 2005 ANNUAL CONFERENCE
DELIVERING RESULTS, IMPROVING PREGNANCY & BIRTH
February 19-23, 2005
MARIE McCORMICK: I'm going to talk a little bit more broadly about the influences of early childhood and the importance of early childhood on outcomes even for those who have been born with perinatal risks. The job of the public health community does not end with the end of pregnancy, as you well know. And let's talk a little bit about what we know about the interventions that are useful for different types of children in early childhood.
First, let's talk about what we know about the pathways for these environmental effects on neuro development. In particular, but it can be ‑‑ I have this for neuro development because that's what I've been focused on. But it certainly also includes nutrition and other kinds of outcomes.
First, there are what are called parental processes and these are probably what we know best. These are the things like the parental and physical and emotional health, their ability to sustain their children, their ability to provide the kinds of experiences that foster development, parental sensitivity to children and conversely harshness or mode of discipline are well established in developmental psychology, as things that affect children's outcomes.
And just as we know that poverty increases the risk of adverse outcomes, poverty can also increase the risk of later outcomes independent of perinatal outcomes. For example we know low maternal education may lead to low earnings, anti depressive syndromes low earnings and the absence of financial support leads to financial strain and without social support the mother, in particular, but both sets of parents are impaired in their ability to parent their children and achieve appropriate outcomes.
But we're learning that there are other kinds of processes that affect children's outcomes as well. For example, there are neighborhood processes, and this is an emerging and burgeoning literature in terms of thinking about children's outcomes.
Neighborhood processes include resources for parents. These are the schools, the clinics, the advice, the neighbors, relationships within the neighborhood that support their role in parenting. And the setting up of norms or collective efficacy, that is, the group feels they can raise their children and achieve their outcomes, not just individual sets of parents.
And finally, as Mrs. Wenck indicated, there are also child care processes that are very important.
And what I'd like to do now is spend some time on what you heard about which was the Infant Health and Development Program, which is an example of child healthcare process for those at risk from adverse perinatal outcomes. This was a multi‑site randomized control trial of early childhood intervention for low birth weight premature infants. The general structure is given here. Phase one, which was zero to three years of age, was the intervention phase. I'll talk about that in a moment. Phases two through four were follow‑up phases. Phase four is the 18‑year follow‑up which we're just getting into press.
Eligibility for this trial was the infant had to be less than 2500 grams at birth and less than 37 weeks. And birth in one of the eight selected study sites. Now these sites weren't selected because they were passionately interested in early intervention. They were in fact selected to be large birthing centers that could provide the volume of infants for recruitment in the time required. The infants were recruited in two different birth weight strata, those less than or equal to 2,000 grams and then the group 2001 to 2500. This proved to be an enormously important decision that we did not understand at the beginning, but because we did this prospectively by the rules of randomized trialism we can analyze them separately. In order to get this number, the number of children we needed, we screened 4,000, almost 5,000 children, who met the birth weight and gestational age criteria, and excluded almost over 3,000 by protocol exclusions which I'll show for a moment.
We actually had very few refusals for those who were eligible to be enrolled, and more importantly, we had very few refusals after they learned their study assignment. The reason is that as you will see, we were providing two years of free day‑care, and we were very afraid that when people found out that they were not getting the free day‑care they would drop out.
Therefore, we were left with 985 infants in the study. The reason for protocol exclusion was largely and almost half the cases they lived too far from the child intervention center. It was determined that you could only take a one‑year‑old on a bus for 45 minutes max. And so those were the ones who were most frequently excluded.
We recruited only English‑speaking mothers or mothers at least who could function in English because most of the materials and the assessments were done in English. We rapidly learned that many of them were bringing care for in non-English homes by non-English speaking grandparents. Medical exclusions were very rare. These were kids who were very sick on the ventilator for over three months. And we had very few maternal exclusions for maternal mental illness or drug abuse.
The infants were randomized, one‑third to the intervention group and two‑thirds to the follow‑up group. And the randomization within birth rate strata and into the intervention group was basically determined by the size of the classroom that we thought we could handle in terms of the early intervention. So about at each site the target was about 140 children, 45 in the intervention site.
The intervention group got routine follow‑up services, and I'll describe those in a moment. This wasn't a garden variety primary care. We were biasing ourselves against the outcomes we were expecting to see. In addition, the intervention got home visits center‑based education and parent support groups.
The follow‑up services consisted of frequent, every three to six‑month visits, with frequent health and developmental surveillance, meaning annually they got detailed developmental assessments, and referral for any health or Social Services that were available in the community. This was considered the norm for follow‑up for very high risk premature infants and all of these children got all these kinds of services. The home visits started just about as a child would have expected to be born. They lasted every week until 12 months of age, corrected for duration of gestation. They used a curriculum called Partners for Learning, which was developed at the University of North Carolina , Frank Porter Graham Center , and adapted for premature infants. And they were also given some support in terms of parent problem solving and of course the social support of an individual coming into the home. At that time 12 months of age, they were enrolled in child development centers that were available five days a week, eight hours a day. I would also note we had transportation to these centers to assure that the children got there. We continued to use the partners for learning curriculum and we instituted parent support groups.
With one exception, we can't disentangle the effect of home visiting from the center‑based education, and we could probably say that the parent support groups weren't very effective because they weren't very well attended. As you can well see, this is the Cadillac variety of early childhood intervention. The reason we went to this kind of strong intervention package was we did not want the criticism that if we saw no effect in these children's it wasn't because we didn't mount a strong enough intervention.
What were the outcomes? First outcome of interest was child cognitive development or IQ. As you can see, at age 3, which was the end of phase 1, the trial, there were whopping big differences in IQ between the intervention and follow‑up group. But by age five these had diminished and by age eight were not seen.
However, when we looked at our group 2001 to 2500, we again saw the big differences at age three, and we saw about a four point and significant difference at age five and age eight. You can well imagine we were not seeing much in the way of differences in the group under 2,000 grams and we can come back to that in a moment. Again, looking at cognitive outcomes in terms of school achievement. This was on the Woodcock Johnson. The RD is reading. MA is math. For total, we saw no differences between the two groups; but, again, in the heavier birth weight group, the H group, we were seeing emerging differences at age eight in achievement but no differences in the lower birth weight group or the L group.
I would simply say these results parallel very closely the results that we're seeing in socially disadvantaged healthy term children in a number of studies. That is, whopping big IQ differences at the end of the intervention; diminution of the size of the IQ differences later on and the emergence of achievement differences in school. I will show you why the emergement of these achievement differences even if they were smaller than you would like to see remains very important.
We also looked at child behavior, which was at that time an emerging issue in terms of early intervention. Everybody did cognitive testing but very few people began to look at behavior. And at age three we did see a small but significant decrease in the number of behavior problems reported by the mothers. But, again, no differences at age five or age eight in either group in terms of reported behavior problems.
Now, we were taking low birth weight premature infants and we were putting them in daycare at one year of age, and we were extremely nervous about what that was going to do. And particularly in terms of health status. And indeed when we compare the intervention follow‑up groups at age three, the intervention groups did report more morbidity in terms of episodes of illness in both groups. However, when we looked at measures of more serious illness, the serious morbidity index here included hospitalizations, ER visits and conditions lasting longer than a month. There was no difference between the two groups. We asked, and activities of daily living scale, adapted for infants. The Stein Functional Status Scale and again there was no difference between the two groups. There was no difference on the effect of health status on growth, as you know if children are very sick they don't grow very well. Again there was no difference. And the parents didn't rate them as sicker ‑‑ more sickly in the intervention group than in the follow‑up only group. So we thought we bought the cognitive and behavioral gains at the expense of a few more episodes of non-serious relatively brief illness.
We also looked at secondary effects. Secondary effects we defined here were basically the effects of providing two years of free daycare. In other words, could we see either a negative economic impact as you might expect if there was an increased use of health services or more positive impact in terms of maternal outcomes. The only significant difference we saw in terms of healthcare use was the mean number of doctor's visits. If you can read that, it was probably two more visits over the course of three years that the intervention group had over the follow‑up group. We thought that was a tolerable expense.
In terms of maternal outcomes, we did see a difference, and that yellow line, I hope you can see the yellow line at the bottom, there was an increase in terms of the intervention group of women entering the work force earlier. But this was in quite stark contrast of earlier studies of early childhood intervention where there were stark sharp differences between the intervention group and the control group with much more work force participation by the mothers. And we were now in an era where mothers had to work regardless.
We saw no differences in subsequent child bearing, which was good for us in this phase, because we were looking at the potential effect of the intervention on younger siblings, and we saw no further educational attainment. So where does the infant health and development program fit within the modalities of early intervention?
Early intervention as it's being delivered can be delivered either through home visiting or center‑based services. We combine them, too. The activities that can in early intervention can be educational or developmental, these were the primary activities that we were providing in the infant health and development program. However, as an early intervention they can be therapeutic services as well. Physical therapy, occupational therapy. And then there are the home visiting programs such as the Hawaii program and the Alaska program, which are really functioning, focused on parenting skills to prevent child abuse.
In terms of what we know about some of the early intervention programs, the target programs have been the socially disadvantaged infant beginning with HEADSTART. There have been infants with identified neurodisabilities and low birth way, premature, IHDP. One of the reasons, in fact, people were surprised that we got the results we did was we were biased against finding results for two reasons: One is because the target populations have been socially disadvantaged infants, people were concerned we weren't going to get effects because our population was much more heterogenous. We actually had some very advantaged families in our study. Particularly in my own site, which was Boston . We had in fact one of the wives of the Harvard house masters in our program. We thought it was terrific. And, in fact, the only site in which there was no intervention effect was Harvard, because of the relative affluence of our population.
I will also say we had the same effect size at every site controlling for socioeconomic status. But the effect size, even if it was 10 to 15 IQ points, got you into in the normal range in Philadelphia and got you into the retarded range in Miami which was a very disadvantaged site where the maternal IQs averaged 60.
The other reason, of course, is that we had low birth weight infants who were heterogenous, in terms of their neurologic development and their potential neurologic handicaps and indeed the children with Frank CP did not benefit from the early intervention program.
But our result, as I mentioned, are very consistent with the socially disadvantaged. And I also want to point out that they're also consistent with early intervention for disabled children. This is a table from Jack Shonkof's meta analysis of this data and as you can see the average overall was a half standard deviation in whatever developmental test they were using. That's eight IQ points for those who don't deal with that all the time. But it could range as high as almost a full standard deviation in programs that dealt with children with mixed disabilities.
So why did we go back at age 18? So we've got four IQ points. I mean that's significant, but there's a lot of people who would say they're not convinced about that. The reason we went back at age 18 and why the early emergence of achievement differences in school was important lay in the fact that there were two recent publications of studies. One actually just this year published their age 41 results. And that is the Perry preschool study. This was a randomized trial for disadvantaged children in Epsilani , Michigan of early intervention before they went to school with nothing else happening afterwards.
At age 27, as you can see here, the intervention group was likely to earn more money, to own their own home and never to have been on welfare as adults by substantial measures.
They were also less likely to have had risky behavior or adverse behavioral outcomes. As you can see, the program people averaged only 2.3 arrests as opposed to the non‑program people with 4.5. I think most of us would consider 2.3 not the best possible outcome, but this is again a very impoverished community and they were clearly having an effect. Again, this is 27 years. This is like 24 years after the intervention, without anything else in between.
I will also like to say when people talk about cost‑effectiveness of early intervention services they're largely quoting this study and it's largely jail time that makes it, decreased jail time that reduces the costs.
The second study that I'd like to bring to your attention is the Absinario project. It was a randomized trial for disadvantaged childhood of early childhood education from North Carolina and in fact is the direct antecedent of IHDP; Partners for Learning was the curriculum that was developed there. They have followed their cohort out to age 21. Their result at age eight parallel very closely what we saw at age eight in our 2,001 to 2500 gram birth weight cohort. And as you can see, the difference here in reading scores remains all the way out to age 21. And the difference in math achievement scores also remains the same, out to age 21. Again, without any subsequent intervention on the parties.
We all lose math scores as we go along. So what do we conclude? First, postnatal environment is a powerful predictor of child outcomes for all children except for perhaps ‑‑ I would say for all children regardless of their prenatal outcomes.
Evidence is increasing that early childhood interventions can improve outcomes for children for a variety of reasons, both physical and environmental.
And finally, intensive programs may have a lasting effect. What I will say is, however, we don't know how intensive intensive needs to be in order to get these effects and there's still a substantial amount that we need to know about modalities, the curriculum and comparisons of various approaches to various types of children. Thank you.