View the presentation (55:35)Watch an interview with David Olds.
Opinions or points of view expressed are those of the author(s)/presenter(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice.
Research for the Real World SeminarDavid Olds, Ph.D.,
DirectorPrevention Research Center for Family & Child Health; University of Colorado, Denver
January 20, 2010
Kristina Rose: Good morning everyone. Good morning and welcome to NIJ's “Research in the Real World” seminar series. I want to extend a special welcome to our visitors that came from outside the Department of Justice. We are honored to have here as our guest today, Dr. David Olds from the University of Colorado Health Sciences Center, and he is the second Stockholm Prize winner in criminology that we have had in the last couple of months. So we are honored to have you here with us.
Dr. Olds today is going to talk about the 19-year follow-up study to his famous research, as what has come to be known as the Nurse Home Visiting Program. And what's impressive about Dr. Olds is his commitment to the research process and to getting it right. In doing a little background reading on Dr. Olds, I was struck about the fact that he actually turned down research grant dollars because he did not feel that his program, his model, was ready to be replicated; he wanted to test it even further, and I was very impressed with that. As OJP progresses in its efforts to become more evidence based, we can look to the work of Dr. Olds to inspire us to work harder for more rigorous testing of the programs that we fund and determining what works, while maintaining — as Dr. Olds did — the fidelity to the model as a guiding principal.
I'd like to introduce Laurie Robinson, and I must to tell you that Laurie is going to need to leave a little bit early today. She's got some other very important meetings to attend to, but we are thrilled that she is here to introduce Dr. Olds. And it is fitting that she should do this as we all know of her strong leadership and support for evidence-based policies and programs and the research and for the science, especially. And before I ask her to come to the podium, I want to remind everyone here that this program will be audio taped, and if you have questions, to please wait for the microphone to get to you because that's going to be very important for the audio-taping quality. So without further delay, I'd like you to welcome Laurie Robinson, our Assistant Attorney General for the Office of Justice Programs.
Laurie Robinson: Good morning. I'm really delighted to be here this morning and to have the chance to introduce our speaker; we're delighted to have Dr. David Olds here today. I think Dr. Olds is well known to many of us here at OJP and to anyone who has studied the important work of early childhood interventions. Dr. Olds, as you know, is here to tell us about the long-term follow-up on the Nurse-Family Partnership program, and as I think all of you know, that's a program that provides nurses to visit expectant mothers and continue to visit those mothers until the child reaches age 2. Now I want to just tell you that I had the opportunity — when I first came to the Justice Department back in the early '90s at the beginning of the Clinton administration, Shay Bilchick, who many of you may remember, headed OJJDP back in the Clinton years.
Shay and I became aware of the Nurse-Family Partnership program back in 1993 — at the beginning of the Reno era — and brought information about that to Janet Reno's attention. She was so fascinated by it and followed this throughout her term — and I'll get to this in a minute to tell you a little bit more about how she engaged and how we engaged with that — but the Nurse-Family Partnership program currently provides nurse visitations to more than 20,000 mothers each year in communities in 20 different states around the country. You know we often talk about programs that work, but it may be only in one or two locations — this has a very broad reach. And the results that Dr. Olds is going to be talking about today come from the Elmira study site, but also from sites in Memphis and Denver. He began initially in Elmira though with his first randomized, controlled trial more than 20 years ago, so that's a very long period for follow-up — so again, often times we only have studies where we'll have one- or two-year follow-up, and this kind of long follow-up I think gives us important kinds of results. Among the outcomes from this study that he has reported on are the health behaviors, arrests and convictions among the mothers, and the early childhood health and behavior of the children of those mothers.
In today's presentation, Dr. Olds is going to focus on the outcomes for the children 19 years after their mothers were home-visited during pregnancies and during the first two years of their children's lives. Key outcomes that he'll tell us about are the arrests and convictions on these children who are now well into their teen years. So again I think we can agree that that's quite critical information.
Dr. Olds is a professor of pediatrics, psychiatry, preventive medicine and nursing at the University of Colorado, Denver, where he directs the Prevention Research Center for Family & Child Health. He's committed much of his professional life to the development and testing of this partnership program. And as Kris mentioned, he's won the 2008 Stockholm Prize in criminology, which is a very, very high honor. His work has been funded by — among others — NIH, the Robert Wood Johnson Foundation and our own OJJDP.
Under the Clinton administration, the Nurse-Family Partnership program — as I mentioned — had the strong support of Janet Reno, who in 1996 helped the program expand to six additional sites and helped make the program part of the activities that Weed & Seed sites helped to support — that's something that Shay and Steve Rickman, who many of you may remember headed Weed & Seed during the Clinton years, really fostered. But lest you think that Dr. Olds has known nothing but success, I wanted to point out that like many applicants for federal grants, when he was a young researcher he was turned down for his first federal grant application. This was back when HHS was HEW — back in the real olden days — and when he applied for his first federal grant to the Office of Child Development at HEW, he was turned down. But he revised and resubmitted that application, which then got funded, and I guess we can say the rest is history.
Now when I taught at Penn, I used to tell my students that tenacity — which some might say is stubbornness — but tenacity is one of the things that leads to success. And I think success is one thing that certainly would characterize our speaker this morning, so please join me in welcoming Dr. David Olds.
David Olds: Thanks so much for inviting me, Laurie and Thom and others. It's really an honor to be here to have a chance to talk with you this morning. I can say with absolute certainty that the strongest and most coherent support that we see across the various departments and professions within the federal government, and with states and local governments, comes from people involved in the justice system. And I think that part of that is really — it's a reflection of people in justice simply wanting to get the job done.
I got involved in this, and I guess Laurie, your characterization of me as being tenacious I think is probably, there are probably a lot of other adjectives that my kids might give, but tenacious is one of them. I actually started this work, got involved in this kind of work, when I finished up undergraduate school in Baltimore and went to work at the Union Square Day Care Center on West Lombard Street.
I was a product of the '60s. I hoped that if I could just help poor preschoolers get off to a good start that they would have a better chance of succeeding in elementary school and later on in society. But it was unfortunately too clear to me that for many of the children in my day care center, it was already too late. Some of the children had been abused, were being abused; some of the children had been exposed to toxicants like alcohol and their behavior was already compromised. I worked at that inner-city day care center, and one of the things that struck me was that the children who were doing better than others in our classroom had parents who would show up for parent group meetings that I held during nap time when the children were sleeping. The children who I suspected of being abused — I never saw their parents. So I knew that it was important for us to begin working with parents to help them provide more competent care of their children at much earlier ages.
But I also realized that what was going on in the neighborhood matters. If you walked outside of the basement steps of this day care center on West Lombard Street, the neighborhood conditions were terrible. There were no sources; there were no supermarkets; the rates of unemployment were off the charts; the housing stock was terrible. I took the children to a play-area park nearby where there were needles and alcoholics — needles in the ground, alcoholics roaming around. The rates of crime were very high; I had my car stoned on the way to work. It made me realize that context matters, and I'm going to come back to this issue as we … as I give my presentation this morning. I went to graduate school and eventually began collaborating with a team — actually there have been hundreds of people who have worked on this initiative over the last 30 years. Here are some of our key research team members who have worked on the studies that I am going to be talking about this morning.
The experience in inner-city Baltimore taught me that if we were going to make a difference in the lives of vulnerable children and families, we would have to start early. And perhaps most importantly, the program would have to make sense to parents. I think one of the key elements of this program is that it activates parents' instinctual drives to protect their children. I think that parents become engaged in the program because it makes sense to them, especially given — as you'll see in a moment — given the population that we've targeted for this service. Nurses, as Laurie mentioned, begin working with families as early as possible during pregnancy and follow the child through the second year of life. These are critical years in the child's development where neurologic development is proceeding very rapidly and the likelihood of compromised neurologic development as a result of exposure to tobacco and alcohol and illegal substances and so forth are very high. So the nurses have been provided with detailed visit-by-visit guidelines that help structure their work with families to reduce the child's exposures to early experiences that can compromise a developing fetal brain and capacity for later behavior regulation.
And we believed, I believed, that it was not enough to design a program that made sense, that had promise, but that we needed to test it. We needed to make sure that what was being offered was really going to produce the kind of results that all of us in this room are committed to; so that's a good part of what we're talking about. The program focuses on a segment of the population that is vulnerable — low-income pregnant women bearing their first child. As a result of focusing on that segment of the population, the program serves large portions of teens and women who are unmarried. Forty percent of the births to women in the U.S. are to women having their first child. That experience of going through your first pregnancy creates a sense of psychological vulnerability, and there is physical vulnerability that increases women's receptivity to offers of help, especially from nurses. And we think that this particular targeting of the program is really crucial —for, among other reasons, if the nurses are successful in helping women plan future pregnancies, there is greater likelihood that the program will have even greater impact by having an impact on the care of subsequent children as you'll see in just a moment.
The nurses in this program have three major goals. The first is to help women improve the outcomes of pregnancy by helping them improve their prenatal health — helping them cut down on the use alcohol, drugs, identification of early emerging obstetric complications so that those problems can be treated more promptly and reliably before they start to compromise the developing fetal brain. The nurses also focus a lot of attention on helping parents provide more competent care for the child so that the child's health and development will be promoted — we're especially concerned about reducing the rates of child abuse and neglect. The third major goal is to help parents become more economically self sufficient. And the key to this is for nurses to work with young parents at this critical phase in their development where they can start to imagine what kind of life they want for themselves and their children. And it's out of that envisioning process and that reflection of what life might be like that nurses guide young parents in decisions about staying in school, finding work and figuring out when they are going to have the next baby that they have. Because pregnancy timing, as you'll see in a moment, is a critical factor in helping women and fathers care well for the first child, but also to stay in school, find work. And it's a sentinel outcome we think in accounting for the way this program works; so we'll talk about it in a moment.
This program really is designed to reduce the risks — from the criminal justice perspective — for early onset and violent criminality. We know that there is … that there are certain genetic factors that increase susceptibility to behavioral disinhibitation. But we also know that there are prenatal exposures to substances and pregnancy complications that can also contribute to behavioral disinhibition. We know that child abuse and neglect are significant risk factors for behavioral dysregulation. We know also — the evidence suggests at least — that these early conditions, like exposure to tobacco and exposure to maltreatment, interact with genetic susceptibilities to increase the likelihood of early behavioral dysregulation. We know that compromised language development increases the risk for behavioral dysregulation and that children who have been exposed to violence early on are more likely to have negative attributions for neutral behavior on the part of people with whom they interact, and that creates a propensity for them to behave violently and defensively when they come into encounters with other individuals. So it's these earliest risks that the program is designed to affect. And it's less likely, I think, to affect the kinds of risks for adolescent, onset antisocial behavior that is mostly attributable to puberty itself; so we are going to talk more about this in just a moment.
The program, as Laurie indicated, has been tested in three separate, randomized control trials over the last 30 years. When we first published — first with low-income whites in Elmira, New York, many people when we first published the results of the Elmira trial said, “You know, gee, you got a program that works; you need make it more widely available.” We took the position that way ought not to do that because we didn't know, based on the Elmira trial, whether the findings would replicate with minorities living in major urban areas. So we … rather than offering the program up for public investment, chose to replicate the study in Memphis, Tennessee, with a large sample of African-Americans.
In the mean time, a lot of people took the results of the Elmira trial and said, you know, “This Olds guy, he's uptight about this research, but we know that home visiting works, so we are going to make home visiting available to prevent child abuse, reduce infant mortality and so forth.” And at first, frankly, I didn't stand in front of that — those efforts — because I didn't know whether the positive effects that we had seen in Elmira were attributable to our particular version of home visiting or whether they could be applied to all types of home visiting programs. But in the early '90s, we began looking at the scientifically controlled studies of other home-visiting programs and concluded that the results were really disappointing. So I had to ask the other home-visiting programs to stop using our evidence to promote their expansion, the expansion of these other programs models. But I came away from this experience wondering whether the small effects, negligible affects, that we were seeing with other home-visiting program models was due to the fact that the models were employing non-nurses or whether it was due to the fact that the programs themselves were insufficiently developed.
So in the third trial that we conducted in Denver, we enrolled a large sample of Hispanics and systematically compared the relative impact of the program when delivered by nurses versus paraprofessional visitors who shared many of the social characteristics of the families that they served. The basic program models for nurse-visited and paraprofessional-visited families were the same with the exception of the background in the visitors. So that would allow us to sort out the role that visitor background plays in accounting for differential effects. And we are going to talk more about what we've learned from this trial in just a moment.
Now what you'll see on this slide is the set of findings that we have the greatest confidence in because they replicate across trials with different populations, living in different contexts and different points in our country's history. Based on the results of these trials, we are confident that the nurses can improve women's prenatal health — especially the reduction in tobacco use and hypertension disorders in pregnancy. We have seen consistent reductions in childhood injuries. Injuries are the leading cause of death among children aged 1 and actually now through middle adulthood for people in our society. We've seen fewer subsequent pregnancies and greater intervals between the births of first and second children, and, as I mentioned earlier, we think that's an important ingredient in accounting for the long-term effects of this program. We have seen increases in father involvement, including greater stability in partner relationship over time. We have seen increases in maternal employment and reductions in welfare and food stamp use in our first two trials conducted prior to welfare reform. In our third trial we see ,now … impacts — and I'll show you this in a moment, but we see, in fact, impacts on mothers' earnings derived from administrative datasets over the first eight years of the child's life in our Denver trial. And we see improvements in children's school readiness and language functioning, but it's concentrated in those children born to mothers who are more psychologically vulnerable, who have limited intellectual functioning themselves and higher rates of depression and anxiety — it's those mothers who are least capable, we think, of managing the care of their child while living in concentrated social disadvantage. And I'm going to show you more about that in just a moment, as well.
Now one of the things that's been important from our perspective is that the impact of the program, when we look at a variety of different indicators, is greater where there is greater social disadvantage. This slide comes from our first report of the Elmira trial looking at the rates of state-verified reports of child abuse and neglect and shows how the treatment-control differences vary depending on socio-demographic risks. The first set of bars shows the overall treatment-control difference in reports of child abuse and neglect. While there's about a 50 percent reduction overall, it's not as statistically significant effect. When we look at — as the socio-demographic risk accumulates, the treatment-control difference grows because the rates of maltreatment in the control group are higher. Now this is only a trend so it's important to keep that in mind. What we also found in the Elmira trial is that even among those mothers who were low income, unmarried and teens, the treatment-control difference was further concentrated among mothers who had limited sense of control over their life circumstances measured at registration. So here we had early indications that the benefits of the program are greater where mothers are personally compromised in their ability to manage their lives and there is greater socio-demographic risk. And so this is a general finding that you'll see repeated as I go through this presentation this morning.
In the Elmira study, by the time the children were 15, there was significant treatment-control differences in mothers' own convictions and arrests and incarcerations. These differences that you see here — 60 percent difference in arrest, a 70 percent difference in convictions— were based on mothers' self report. These treatment-control differences were even greater when we used the New York State Division of Criminal Justice system's data to determine these program effects. And these treatment-control differences are even greater among those mothers who were low income and unmarried at registration. And for the children, by the time the children are 15 we see statistically significant reductions — about a 50 percent treatment-control difference in the rates of state-verified reports of child abuse and neglect, and about a 60 percent overall treatment-control difference in arrests by a child aged 15, and reductions in their adjudications as persons in need of supervision.
Now these treatment-control differences that you see displayed here are even greater among those children born to mothers at greater socio-demographic risk. And there were … At this age the differences are statistically significant and there were no gender differences — we're going to talk about in just a moment — in arrests and convictions for the children. But I also want to point out that we conducted and my colleague John Eckenrode lead the study, showing that this 48-percent, treatment-control difference in state-verified reports of child abuse and neglect was attenuated in households where there were moderate to higher levels of domestic violence. Now that's important because there were no program effects on rates of domestic violence. And our goal in this program of the Surgeon General is to understand not only for whom the program is working, but for whom it's not working. Because if we can understand for whom it's not working, we can go back to the drawing board to try to figure out how to make the program stronger at addressing those subpopulations or contexts that we are not affecting. This finding was reported in the
Journal of the American Medical Association in about 2000.
Now, about two weeks ago we reported in
Archives of Pediatrics and Adolescent Medicine that the impact on the percentage of children on both arrests and convictions was statistically significant, as you'll see in a moment, here we see there's about a 40 percent reduction in the rates of arrest by the time the children are 14 — these are all based on self-report by the way. And there's about a 60 percent reduction in what children were ever convicted by age 19. But these differences were limited to — especially in the second half of adolescence — to girls. What we're looking at here is a hazard analysis for the … whether or not children are arrested — girls are arrested, by age, and you can see in the control group, the dark line, the rates of arrest go up dramatically after age 14 but remain relatively low among the nurse-visited girls. The pattern for boys is in the right direction but it's not a statistically significant difference for boys, and this, frankly, is disappointing to us because boys account for such a large portion of the arrests and especially violent criminality in the U.S.
Now if we look at the number of arrests over time, we see the same kind of overall reduction in the counts of arrests by the time the children are 18 — about a 60 percent overall reduction and about a 70 or so percent, a 65 percent reduction in the number of lifetime convictions by age 19. And if we look at the age crime curve overall, it falls into the pattern that one might have predicted. That as the control group goes up quite precipitously in mid-adolescence and then starts to decline, and that age crime curve is moderated by the intervention. Now, if we look at the difference though, this difference is really accounted for by big differences among girls as this slide shows. This may not show up very well — this is the first time I've presented this particular finding. What this shows is that you can start to see that the age crime curve is really moderated for girls but not so for boys, and especially in the second half of adolescence, the treatment-control difference in counts of arrests is essentially zero for the boys. Now we can talk about why this is, maybe in the discussion session — frankly, we don't know. We've been trying to get to the bottom of this; we have some thoughts, and we are conducting some analyses to try to help us sort this out. We are pleased of course with the overall reduction in arrests; that's good, but we want to know what accounts for the differences between the large effects for females and the small negligible ... (Break in audio.) … especially in the second half of adolescence.
And when we look at other aspects of youth behavior at age 19, overall … when we use the word high-risk youth … when having sex they are more likely to have used condoms, but for nurse-visited boys they are more likely to have sexual partners. But there is a significant decrease — a large decrease in the likelihood that nurse-visited girls, especially those born to mothers who were low income and unmarried at registration, will become pregnant by age 19. And there is a large reduction in their use of Medicaid by the time they turn 19, as well. In the article we say that there is a certain kind of coherence in impact on crime and on life course among girls but not so among boys; something we don't yet, again, fully understand.
Now this slide … I'm showing you this slide because even though it's not a statistically significant effect, this is what we would have predicted at age 19, and it is a difference that is large in effect-size terms. This is the use report of felony assault in the year preceding the 19-year interview. And you can see that overall, nurse-visited boys and girls are less likely to have reported felony assault. Now it's this kind of violent criminality that the program, we think, is really positioned to have an impact on because it has an impact on prenatal health, early maltreatment — the kind of risks for early onset conduct disorder. The difference is large; it's in the expected direction; it doesn't achieve conventional statistical significance. And we are going to be looking at these kinds of patterns of antisocial behavior in the subsequent trials that we've conducted in Memphis, and hopefully, if we get the funds to do it, a follow-up of our Denver sample, as well. Now we're conducting now a 27-year follow-up of our Elmira study that will allow us to sample, that will allow us to look at mental disorders, including anti-social personality disorder, depression, post-traumatic stress disorder, rates of violent criminality, substance abuse, second-generation maltreatment — although it's going to be difficult for us to get a good handle on that unfortunately — and persistent unemployment.
Because the effects of the Elmira study were greater for mothers where they were unmarried and of low socio-economic status, we concentrated the recruitment in our Memphis trial on mothers who were low income and unmarried. And we recruited virtually the entire population of low–income, unmarried mothers in the Memphis community for this study. The neighborhood conditions in which the Memphis trial was conducted are among the worst in the country. The average neighborhood disadvantage in Memphis is about two-and-a-half standard deviations above the national average in terms of neighborhood disadvantage — concentrated disadvantage — so we are going to talk a little bit more about the implications of this.
A few things I want to emphasize about the Memphis program, we recruited … we were able to recruit … this is one of the reasons we went to the Memphis community, we were able to recruit virtually all indigent mothers, pregnant women, for this study through the Regional Medical Center in Memphis because all indigent women register for prenatal care at a single prenatal clinic. The program itself was conducted through the Memphis Shelby County Health Department, so it was conducted through a local institution. The study was conducted in the middle of a nursing shortage so about half the nurses left employment in the middle of the trial, and there was significant nurse turnover among families being visited. And, as I mentioned before, the study was conducted where there was concentrated social disadvantage.
By the way, we knew that we would not see replication of our effects on rates of state-verified reports of child abuse and neglect because we had conducted pretests and pilot work in the Memphis community that showed that the rates for state-verified reports of child abuse and neglect in the Memphis community were far too low for us to be able to detect the impact of the program on that outcome. What we didn't know, so we did not hypothesize, that we would see effects on child maltreatment. But what we didn't know was that the reason for that is that the system for recording child maltreatment in Memphis, but in Tennessee in general, was broken — the data system was broken —we learned that subsequently. But what we did know though that the rates of injury revealed in the medical record were sufficiently high for us to detect an impact on the rates of injuries to children. So what we found was that there was an overall 23 percent statistically significant reduction in all types of injury encounters among children in the first two years of life and an 80-percent reduction in the number of days that children are hospitalized with injuries. And this treatment-controlled difference in the … in this 80-percent treatment-control difference in the number of days that children are hospitalized with injuries — was concentrated among mothers who had the least capability of managing the care of their children because of limited intellectual functioning, limited sense of control over their life circumstances, higher rates of depression and anxiety. We created an index of psychological resources standardized to a mean of 100 and a standard deviation of 10. And what this index, which is reflected in the X-axis, the horizontal axis in this slide, shows is that that 80-percent treatment-control difference that I reported just a second ago in number of days that children are hospitalized with injuries is concentrated in the lower half of the distribution — that is those mothers with the least capability of managing the care of their children and living in concentrated disadvantage.
Now I want to tell you a little bit about these injuries because I think it's revealing. You may have noticed earlier that in the postnatal phase of the study, there were about 740 or so families randomly assigned to the treatment and control conditions — and there was disproportionate assignment between the control group and the nurse-visited condition. This slide shows the three children who were hospitalized with an injury or an ingestion in the first two years of a child's life. Notice that these children were all 12 months of age or older; two of these three children were hospitalized with ingestions. This slide shows the corresponding diagnoses — the reasons for hospitalization in the control group. Forty-four percent of these children were hospitalized before 6 months of age. These were not mobile; they were not creating risks for themselves. And 58 percent of these children are hospitalized with serious trauma, including fractured skulls, broken long bones — the kinds of conditions that strongly suggest that they were abused and neglected.
Also, we reported in
Pediatrics a couple of years ago that the rates of infant and childhood death in the control group were about four-and-half times higher in the control group than in the nurse-visited condition. Now this is only a trend, so it's important to keep this in mind, partly — fortunately — because childhood mortality is an infrequently occurring event. But look at the nature of the deaths in the control group — the causes of death in the control group versus the cause of death — for the one death in the nurse-visited condition. There were 10 deaths in the control group. One of the deaths in the control group was due to causes that are not preventable, that is a child who died of multiple congenital anomalies. The nine other deaths were due to either prematurity, sudden infant death syndrome or injury, and two of those three deaths due to injury were due to firearms. The one child who died in the nurses-visited condition died because of chromosomal abnormality. Again, it's revealing, I think, the kinds of conditions that children are having to contend with, that families are having to contend with, and the potential for preventing really bad things happening to children early in life.
Now during the first three years of elementary school, there were significant improvements in children's math and reading achievement test scores. We see this effect, but it was limited to those children born to those mothers in the lower half of the distribution in terms of their psychological resources, their abilities to manage the care of their children, living in concentrated social disadvantage. We see the same pattern of results, by the way, when we look at what's happening to children at their 12th birthday, when we do direct tests of the children's reading and math achievement. And at age 12 we are also seeing significant reductions in the number of days that children use tobacco, alcohol or marijuana at the child's 12th birthday. And we see this whether we are looking at any one of these substances alone, the count of substances that they are using, or the number of days that they are using any of these substances over the 30-day period prior to this intervention or to this interview. And this is important, of course, because we know that early substance use is a significant risk factor for both depression later on, but also, especially for African-American males — for violence. But we don't know yet if we're going to have that kind of an effect on violent criminality later on in life, but we are…
Another piece of the findings — and it's important I'm sharing with you the full range of the findings because it helps you put this intervention in larger context. This slide shows government costs based on reviews of state Tennessee administrative records with data on month-to-month expenditures for food stamps, AFDC, TANF and then Medicaid. And what we see here is that over the 12-year period following the birth of the first child, the government saved about $12,600 in these types of government expenditures for these types of welfare expenditures alone. Now these are discounted savings, all expressed in 2006 dollars. The cost of the program, also expressed in 2006 dollars, is about $11,300, so by this measure alone the cost of the program is recovered.
We are in the middle of conducting a 17-year follow-up of the Memphis study to determine whether we're having affects on violent criminality, mental disorders, substance abuse disorders and sexually transmitted infections. Our Denver trial shows this pattern of results where nurses are producing effects that are consistent with what we've seen in our previous trials and the paraprofessionals are producing effects that are roughly half of what nurses are producing. Here, for example, we see changes in urine cotinine — cotinine is a major nicotine metabolite. We see significant reductions in tobacco use for the nurse-visited mothers, not so for the paraprofessional group.
This slide shows the timing to the first subsequent birth, or the first subsequent pregnancy in the Denver sample — and we can see that there is a significant reduction in closely spaced subsequent births among nurse-visited families compared to those in the control group, with the paraprofessional group falling right in between. We see significant improvements in children's preschool language scores at 4; we see the same things at age 2. These effects are clinically significant for the paraprofessional-visited group but not statistically significant. And we see the same kind of pattern when we look at children's executive functioning. Those of you who follow the developmental crime literature know that executive functioning is particularly important in reflecting children's executive cognitive capacity and the capacity for behavior regulation and inhibition of prepotent responses to distracting stimuli. These are the kinds of cognitive factors that can affect not only children's success in school, but susceptibility to later behavioral dysregulation; so we are optimistic about what we see. This is driven largely, by the way, by children's greater capacity for attention.
Now the Washington State Institute for Public Policy has conducted economic analyses of this program and estimated, several years ago, that the return on investment on a per-family basis is about $17,000 per family, and that did not … it did not take into consideration any of the welfare benefits that I just described for you, did not take into consideration any of the earnings benefits for the mothers that I just described for you, because those findings are not yet published.
We took the position, as Laurie indicated earlier, we were invited in 1996 or so to set up the program in high-crime neighborhoods. And it was about that time that some of these findings were starting to come in and starting to convince us that the program was really having replicable impacts of public health importance, and it was worth putting out for public investment. But we were concerned that the program would get watered down and compromised in the process of being scaled up. And so we worked and eventually developed and created a nonprofit organization that is designed to manage the national replication of the program with fidelity to the model.
And I just want to take a moment and just tell you a little bit about how this is structured. The University of Colorado owns the copyright for this program, and it gives to the Nurse-Family Partnership national service office a royalty-free license to replicate the program throughout the United States as long as they do it with fidelity to the model. The university earns nothing from the national expansion of this program. Although there is, as you'll see in a moment, there is a contract for us to conduct research to improve the program itself, but nothing from the sale of any kind of copyright or anything that would lead to any financial return to either me personally or to the university. The organization is set up that it licenses — in order for a local organization to deliver the program, they have to sign a contract in which they agree to adhere to 18 elements of model fidelity. And the national organization is set up to ensure fidelity by spending a lot of time making sure that communities and organizations and states are well prepared to support the development of the program — you can't just throw seeds on concrete.
There is now really extensive training and ongoing technical assistant to nurses and supervisors to make sure that the program is delivered thoroughly. There are detailed visit-by-visit guidelines designed to support the nurses' work with families to achieve the goals and objectives that I described for you earlier. There is a Web-based information system into which sites enter data on every attempted or completed visit so that they can use that information for continuous quality improvement in comparison of their implementation performance according to national standards that were achieved in our randomized control trials. That information is used for continuous quality improvement. All the identifiers are stripped, but that information can be used for supervision of nurses, can be used for agencies to figure out how they stand in relationship to other sites around the country and so forth. We think that this is, in fact I know, that it's unique to have a large program have this level of ongoing accountability in its administration.
So as Laurie indicated before, the program is now up and running in 380 counties — actually in 28 states, serving over 20,000 families per day. And we are also now beginning some work internationally; we've been invited by other governments to set up the program outside of the U.S., and we've taken the position that we ought to not sell the program to them — that we don't really know whether this program is going to work in international context. And we tell those who have invited us to set up the program elsewhere that we don't know whether it's going to work, but we've developed a model that we think is a responsible replication, that involves for pretest and adaptation to the program to local context, and pretest and small-scale formative work in those societies. Our contracts with these other societies calls for them to conduct their own separate, randomized control trial of the program to know whether it really is producing the kinds of results they would like and we would like before they invest significant resources into the expansion of the program. And we're doing this work with the Netherlands, Germany, England, Australia, where we are serving under the [Prime Minister Kevin] Rudd administration; there's a significant effort to serve aboriginal and Torres Strait islander populations. And there's a large effort going on in England where we are — it's now up in 60 sites, and there is an 18-site, randomized control trial of the program being conducted under the auspices of the Department of Health.
Now here in the U.S., we want to continue to find out what the long-term effects of the program are, but we are also trying to — we think of this program as a work in progress. We are busy at the university conducting research to build the next generation of this program. We have studies — extramural funding to conduct work in the areas that you see on this slide, and we've been invited by, among other places, the British government to help design new innovations in primary health care that will follow not just the model that the nurses use in working with families, but the approach to intervention development in building an evidence base to guide practice. So that's part of what we're doing with the university. As you can see, we've tried to be systematic and thoughtful about how we go about doing this work.
There is an element of this work, though, that is not something that we can easily structure, and I want to talk a little bit about this in closing, and that is the nurses' passion to do this work. This kind of work will not be effective unless nurses have a sense of deep passion and commitment to it. This cannot be simply institutionalized. One African-American nurse in one of our Southern states was given instructions on how to find a new family. And the instructions went like this: “Go out Snyder Hill Road, and at that brown cow with the white spots standing by an elm tree, turn left. And go down that dirt road, and when you get to the end of that dirt road, you'll find your new family.” So this nurse did that, and when she got down to the end of the dirt road, she pulled up in front of a rusting-out trailer with a big confederate flag hanging on the outside. Being black, she thought for a moment about whether she was going to go in there, but she thought for half a second; she got out of the car, and she went inside this trailer. And when she got inside, she saw that the walls were painted with racial epithets. It turns out that this mother's boyfriend was a white supremacist, and when he came out and saw that she was black, he said, ”You know, I ain't having no ‘N' nurse come into my house.” And the mother said, “But I need her.” So the nurse stood her ground, and the boyfriend stormed out.
The nurse came back to the next visit; she heard the boyfriend rustling around in the bedroom area. And later she paid a visit, and he was in the living area, kind of listening in on them at the kitchen table. Not too long after that he was sitting at the kitchen table, and not too long after that he was participating in the visits. And sometime after that, the nurse pulled up out front of this trailer and looked up and saw that the confederate flag had been taken down. And when she went inside, she saw that the racial epithets had been painted over. People said to this nurse, “Where did you get the courage to go into that place over and over again?” And the nurse's response was, “I couldn't stand the thought of one more child being raised a racist.” And I think that it's that kind of passion that is so essential to the effectiveness of these kinds of services. And it's so important as we think about the expansion of these kinds of services that we recognize and protect that passion because it's that kind of passion that will ultimately produce the kinds of changes that we're all — everyone in this room is committed to. So thank you.