My attention was drawn by an observant reader to a flurry of American newspaper reports asserting that poverty was more damaging to children’s intelligence than crack cocaine. The newspaper accounts followed a pattern we have come to expect: the conclusions are trumpeted, the implications explored with excitement, and although the author’s name is mentioned there is no way of tracing the story to an actual publication.
The reported bare bones of the story are that the lead researcher, Dr Hallam Hurt, at the Albert Einstein Medical Center, Philadelphia, looked at full-term children born to crack abusing mothers, and found that this group was not significantly different from control children in behaviour and intelligence. Commendably, she followed them for 25 years. It seem that the “crack cocaine baby” concerns can be laid to rest. Hurt is clear that using crack is not a good thing, but then goes on to claim that it is poverty which really causes the major impact on these children’s lives. Considering that I have just published a paper on US scholastic achievement, in which we briefly considered the effects of the cocaine epidemic, I thought it worth while to explore these claims.
By way of background, a 1989 study in Philadelphia found that nearly one in six newborns at city hospitals had mothers who tested positive for cocaine. In a city which then had 1.58 million residents, assuming 13 births per thousand residents per year, I calculate there would have been 20,540 births, of which 3,420 were born to cocaine-using mothers. Gathering 10,000 crack babies for a study would take about three years.
Dr Hurt organized a study of 224 near-term or full-term babies born at Einstein between 1989 and 1992 - half with mothers who used cocaine during pregnancy and half who were not exposed to the drug in utero. All the babies came from low-income families, and nearly all were African Americans. The study enrolled only full-term babies so the possible effects of prematurity did not skew the results. The babies were then evaluated periodically, beginning at six months and then every six or 12 months on through young adulthood. Their mothers agreed to be tested for drug use throughout the study. Attrition cut the final longitudinal sample by half.
Because of the decision to not study premature babies, there may be a slight under-estimate of cocaine effects, since one of the consequences is premature birth and low birth weight. 224 is a respectable sample for a clinical study, though rather small for more detailed analysis of confounding effects. As we have so often lamented, humans are a nuisance to study. People who take one drug also tend to take other drugs, including alcohol, and are less respectful of their health generally, less able to get and keep employment, and less able or willing to save money. The confounding effects are the usual mess of multiple possible causes, many self imposed.
On the issue of whether crack cocaine mothers give birth to significantly damaged children, the published work suggests not. Frank et al. reviewed the literature (37 studies) up to 2000 and concluded: “After controlling for confounders, there was no consistent negative association between prenatal cocaine exposure and physical growth, developmental test scores, or receptive or expressive language. Less optimal motor scores have been found up to age 7 months but not thereafter, and may reflect heavy tobacco exposure. No independent cocaine effects have been shown on standardized parent and teacher reports of child behavior scored by accepted criteria. Experimental paradigms and novel statistical manipulations of standard instruments suggest an association between prenatal cocaine exposure and decreased attentiveness and emotional expressivity, as well as differences on neurophysiologic and attentional/affective findings.”
In that review, Frank quotes 5 papers by Hurt:
79. Hurt H, Brodsky NL, Betancourt L, et al. Cocaine-exposed children. J Dev Behav Pediatr.1995;16:29–35. [PubMed]
80. Hurt H, Brodsky NL, Betancourt L, et al. Play behavior in toddlers with in utero cocaine exposure. J Dev Behav Pediatr. 1996;17:373–379. [PubMed]
81. Hurt H, Malmud E, Betancourt L, et al. A prospective evaluation of early language development in children with in utero cocaine exposure and in control subjects. J Pediatr. 1997;130:310–312. [PubMed]
82. Hurt H, Malmud E, Betancourt L, et al. Children with in utero cocaine exposure do not differ from control subjects on intelligence testing. Arch Pediatr Adolesc Med. 1997;151:1237–1241. [PubMed]
83. Hurt H, Malmud E, Braitman LE, et al. Inner-city achievers: who are they? Arch Pediatr Adolesc Med. 1998;152:993–997. [PubMed]
I have looked one other paper in more detail, having come across it in separate searches (it looks pretty similar to no. 82). At age 6 there are no significant differences in Wechsler scores, which are in the low 80s for children and mothers in both groups. I have also contacted Dr Hurt to make sure I have the full list of relevant papers.
It seems that cocaine is no big deal for the developing brain. Brains are often spared dreadful insults, even short famines as in the Dutch hunger winter. Frankly, this feels like a very significant finding. Cocaine does things to the brain of the habitual user, so its inability to affect a vulnerable baby in the womb is very interesting. Is it a vivid example of the environment not have much effect on anything? Or is it just an exception? If a heavy dose of crack cocaine in-utero doesn’t touch you, what is the chance of a rap video doing so?
Now let us turn to poverty:
“Researchers Probe How Poverty Harms Children's Brains” Chronicle of Higher Education, 2008, Vol.54(25)
- Description: In 1989, Hallam Hurt, a neonatologist in Philadelphia, started recruiting poor inner-city women for a study of how cocaine use during pregnancy affects the developing fetus. Dr Hurt enrolled women who had used cocaine while pregnant and balanced them against a control group of equally poor women who had not taken any drugs. Years later, when the physician conducted follow-up intelligence tests, the results provided a shocking indictment of American society. The tests showed that children exposed to cocaine in the womb had a mean IQ of 79 at age 6, a full standard deviation below the average. Even more disturbing, though, were the results from the control group. The 6-year-olds who had never been exposed to drugs had virtually the same IQ as the children who had endured cocaine running through their veins while inside the womb. Dr. Hurt's data suggests that while cocaine no doubt does cause harm inutero, poverty itself presents an even graver threat to a child's intellectual development. "Growing up poor is bad for your brain" says Martha J. Farah, director of the Center for Cognitive Neuroscience at the University of Pennsylvania, who has collaborated with Dr. Hurt in recent years. What Ms Farah and other researchers are now trying to tease apart is exactly how poverty--which affects 12.8 million people under age 18--takes its toll on the brain. The preliminary results from her studies and others are pointing the way toward methods that might ameliorate some of the effects of poverty.”
What have the authors left out? Well, nearly all their subjects were African Americans. In repeated population surveys their intelligence scores are in the IQ 85-90 range. This sample is no different. Whether the causes are genetic or environmental or some blend of both is not discussed. Indeed, in the subsequent work two issues are underplayed or ignored.
Issue one, as just mentioned, is that African American intelligence and scholastic ability are well behind European American levels, by about 0.6 to 1.0 standard deviations depending on the measure (the Rindermann and Thompson Intelligence paper is available online, and we believe there was some significant narrowing of the ability gap up to 1989, which stalled until 2000, though we don’t know what caused it).
Issue two is that when you rank people by social class that is not a pure measure, imposed upon helpless citizens by heaven. It is likely to be caused, in part or to a very large degree, by ability and character differences. Hardworking, restrained, future-oriented persons tend to get better jobs, and try to pass on those values to their children, as well as passing on their genes. There seems to be a whole academic industry looking at the effects of SES without looking at intelligence or genetics. I believe that the proper policy should be to let all schools contend, and all hypotheses compete.
From their crack cocaine sample they pick 30 low socio-economic status children and contrast them with 30 not so low economic status children drawn from the community. Yes, that is right, 60 children in total. On this basis they contrast outcomes, attributing the differences to social class and, thereby, to poverty. They do this on 7 measures of cognition. 30 pairs, 7 measures. They do not report on the means and standard deviations of their tests. They say that they have given the Peabody Picture Vocabulary test, a measure of intelligence. I cannot find the results in the paper. I have asked if there is a technical appendix. The paper seems to be written back to front: findings first, some results next, and then the boring realities last, if mentioned at all.
They do their structured equation modelling on the basis of socio economic status, and do not appear to include intelligence as a variable in the equation. It is well nigh impossible to find their results. They may have done a better paper somewhere else, so I have asked for a complete list.
So, are the reporters and researchers on crack cocaine? I doubt it. They are on stronger stuff. They are on “Selective Attention”. To giver them their due, the researchers seem to have proved one important thing, which is that in-utero exposure to cocaine does not affect intelligence and behaviour. Their sample are virtually all African Americans who also take other drugs, often lead disorganised lives, and end up in poor and dangerous neighbourhoods, where other people take drugs and lead disorganised lives. Researchers and reporters are right to report their findings. Clinical samples tend to be small, and not always representative, but they are perfectly capable of finding valid results.
However, the researcher’s wish to show that all these unfortunate behaviours are caused by poverty requires a larger sample, other methodologies and other control groups of different ethnicities. Vietnamese and Koreans are also often poor when they arrive in the US, but few of them remain entirely so, and few take cocaine. You cannot simply assume that social class is an explanatory variable. Intelligence as a cause of behaviour must have a fair chance of participating in the equation modelling. This is where selective attention to causes is most troublesome. If all you see are social forces imposed on people from outside you will never see the way in which personal choices can lead to unfavourable outcomes. Selective attention prevents a full understanding of what may be happening.
Disclaimer: This post was written under the influence of coffee. If I get more and better papers out of this research group I will change my views.