Tuesday, June 19, 2007

Early Childhood Blood Lead Levels and School Achievement

Has your child been tested for lead? No? A recent study by Duke researchers reported online at Environmental Health Perspectives might change your mind. And those of you who have might want to pay attention too. Seems even relatively small amounts of lead in the blood are related to lower math and reading achievement. Amounts smaller than what the Centers for Disease Control (CDC) considers “safe."

Researchers collected data on lead levels and school achievement at the end of 3rd grade for more than 8,600 children in several counties in North Carolina, a little more than half black, the others, white. That’s a large sample for sure. The North Carolina Department of Environment and Natural Resources maintains records of children’s blood lead levels. They estimate that about 25 percent of kids in the state have been tested for lead. So the Duke folks matched the kids tested for lead with school records from the North Carolina Education Research Data Center. They used more than a few items to link kids in the lead database with their scholastic records. To minimize any effects due to a language barrier, they limited their sample to White and Black students with no English deficiencies.

In case you’re wondering, lead in the blood is measured in terms of micrograms per deciliter (e.g. 2.00 ug/dL). The CDC currently lists 10 ug/dL as the “threshold” – anything lower than 10 being unsafe. A few decades ago the threshold was 60 ug/dL. The kids in the study? Averaged 4.50 ug/dL (black children, 5.19, white children, 3.70).

But back to the research. Did high lead levels correspond to lower reading and math scores? Yes. Higher blood levels were associated with lower scores. More to the point, poorer achievement was linked to levels at or below 5 ug/dl, below the CDC’s threshold of 10 ug/dL. The more the lead, the lower the scores. So yes, you might be thinking this was only a correlational study. There is no cause and effect here. Right. But then again it isn’t possible to do an experiment, you know, assigning some kids to have high lead levels, and others, none.

Okay, what about other things that might influence test scores?? The analyses did control for a variety of variables that might account for the link between the kids’ lead levels and achievement. Like their parent’s IQ and income. Or their home environment. Though some of these were not assessed so, well, well. Parental income and education stood in for their IQ, and although related, they’re not always identical. I know more than a few cognitively well-endowed types dishing out dinners and driving trucks. Of course, I also know some folks in cushy, well-paying desk jobs with various degrees and Blackberries who aren’t exactly clever.

Those control variables are not as worrisome as some that weren’t tested and might make a difference. For example, what about kids with existing learning disabilities? Or physical illnesses? Say one that might impact learning and thus test scores. Try as I might to find mention of these in the research analyses, they were absent. The kids in the study? Their parents had them tested for lead. Means they and/or the doctor had some reason to believe their children were at a disadvantage. So at least some of the kids were exhibiting some type of potentially troubling symptoms or behavior.

Brings me round to the next issue. The sample of kids. Not a random sample. Only the kids who got tested. So I assume the doctor asked the 5 questions on the lead checklist and decided to test the kid. Not sure what those 5 questions were or what the criteria were for deciding to test. And remember, only about 20 to 30 percent of kids in the state are tested. So something about those kids (or, okay, the doctors, or their parents!) made them different from those who were not tested.

I have to play devil’s advocate. Who knows about the other kids. The ones not tested for lead. Maybe some had little or no lead in their systems. But some might have had more. And maybe they’re functioning okay with higher blood levels. Unlikely, but possible. So it would be nice to have kids included in the study whose parents didn’t ask for lead tests – or whose doctors didn’t feel the need to do them. These kids would be a good control group. Of course to study them researchers would have to design a multi-year study in which all children in a sample were tested for lead. I’d assume they would have lower lead levels, but we don’t know for sure. On the basis of the Duke results we’d assume that these control kids would have even higher reading and math scores. But we just don’t know.

Parents breath a sigh of relief here. The apparent impact of lead on each child’s performance is quite small. A child with a 5 ug/dL had a decline in reading equal to about 15% of the interquartile range. The interquartile range? Technical speak. The middle fifty percent. Excluding the bottom 25% and the top 25%. In 2003 – 2004 this reading interquartile range spanned 12 points. The math, 10 points (with a 14% IQR decline). The mean reading and math scores are not provided but from the graph they look to around 255. So we’re talking a decline of about 1.8 points in reading and 1.2 in math. Doesn’t look like a large loss for each child but on a school level or a national level, it could make a profound difference. Other studies have also found small effect sizes for lead (see the Duke article for references). It must be noted as well that a host of other factors have a stronger relationship with cognitive functioning. For example, in this study, parental education and socio-economic status.

The troubling issue, of course, is the sheer number of children with high, potentially problematic lead levels. A recent national survey suggests over 2% of 1 -5 year olds have blood lead levels at or above the CDC threshold of 10 μg/dL. The National Center for Health Statistics reports children tested have a mean of 2.1 μg/dL. Who knows the mean of the kids not tested. High levels of lead have been found again and again to be associated with poverty as JLowe over at Impact Analysis points out.

Bottom Line: New evidence that relatively low levels of lead ( 3 to 5 ug/dL) might impair children’s cognitive functioning. Impairments appear to be relatively small. Evidence here would be even more convincing if study had included a control condition - the large portion of children who aren’t tested for lead.


Read the article for free. Introduction is a good read. The stastical modeling in the results section gets heavy. Check out the graphs by blood level on the final pages.

Miranda, Kim, Overstreet Galeano, Paul, Hull, & Morgan. 2007. The Relationship between Early Childhood Blood Lead Levels and Performance on End of Grade Tests. Environmental Health Perspectives Online, April.

1 comment:

Anonymous said...

I would love to know what the average kid's blood lead level is in this country. Makes you wonder. Does anyone have any data on this issue?