STATS ARTICLES 2010
What's in those veggies?
Rebecca Goldin Ph.D. & Jing Peng, June 4, 2010
Does pesticide and insecticide exposure lead to ADHD?
Does pesticide and insecticide exposure lead to ADHD? According to media accounts, it does. Reuters declared without reserve, “Pesticides (are) tied to ADHD in children in U.S. study”. The study referenced by Reuters examined the presence of certain metabolites of pesticides, and the prevalence of attention-deficit/hyperactivity disorder (ADHD) among a group of children aged 8 to 15. It appears in the June issue of the journal Pediatrics. Time Magazine cited a litany of chemically induced problems by way of an introduction to its article covering the same study. In contrast, HealthDay gave proper context to the study by citing experts other than the authors of the original study, placing the research in the context of other research on the same topics, and providing some reflection about the value of vegetables in our diet.
That said, the study does implicate pesticides in an increased risk for ADHD, but the consequences of pesticide exposure are far less widespread than the media would have you believe. A closer look at the data reveal some startling caveats to the results.
The study’s results
The design of the study was relatively simple. Exposure to “organophosphates” is assumed to be proportional to the existence of metabolites in the urine. The authors had data from the National Health and Nutritional Examination Survey (2000-2004) on the presence of these metabolites. The children were evaluated for ADHD using the Diagnostic and Statistical Manual of Mental Disorders, with the exception of one of the required criteria for an ADHD diagnosis (that symptoms would be present for at least seven months). All told, the authors had data for 1,139 kids. Of these, 119 (10.4 percent) met the modified criteria for ADHD.
The concentration of urinary concentrations of dialkyl phosphate metabolites (DAPs) of organophosphates was broken into two large classes of metabolites: dymethyl alkylphosphate (DMAP) and diethyl alkylphosphate (DEAP) metabolites.
The results were not as clear as were represented in media accounts. For a ten-fold increase of DEAP metabolites found in the urine, there was no associated increase in ADHD. The adjusted odds ratio was .94, with a confidence interval of 0.69 to 1.28. In other words, in the group of kids examined, those with higher levels of DEAP were less likely to have ADHD. However, since the confidence interval contains 1, this “benefit” is unlikely to be present in the whole population. On the other hand, for the DMAP metabolites, there is a statistically significant association between the two. For a ten-fold increased level of DMAP in the urine, the odds of a positive ADHD test was 55 percent higher (odds ratio 1.55, confidence interval 1.14-2.10). Note that because the confidence interval does not contain 1, the association is statistically significant. For the (larger) group of kids either diagnosed with ADHD by the DISC-IV test or on ADHD medication, the association was stronger, with an odds ratio of 1.72 (confidence interval 1.31 to 2.28). So – assuming causality -- if one class of metabolites shows no negative impact on ADHD, and the other shows a statistically significant negative impact, what is the total effect?
Taken as a whole, these metabolites are termed DAPs. A ten-fold increase in DAPs was associated with a 21 percent increase in odds of an ADHD diagnosis. However, the association is not statistically significant (confidence interval .97-1.51), suggesting that any association, if present, is weak. In contrast, there is a 33 percent increase in odds of either having an ADHD diagnosis or having taken ADHD meds (but not satisfying the diagnostic criteria for ADHD)– and the result is statistically significant (confidence interval 1.10 to 1.67).
Subtypes of ADHD
Here’s the catch, however, with the results. Not only do we find conflicting results when we look closer at the metabolites, but when we break ADHD down into subtypes we find rather consistent results. Of the 119 cases of ADHD used for this study, 21 were subtyped “hyperactive/impulsive” while the large majority of cases (the other 98 kids) were classified as either “inattentive” subtype or a “combined” subtype.
For each of the classes of metabolites (DMAP and DEAP) and the combined metabolite measurement (DAP), there is a statistically significant association with a diagnosis of hyperactive/impulsive subtype of ADHD. In contrast, for both the inattentive subtype and the combined subtype, there is no statistically significant association.
In other words, the results are far less worrisome for eating vegetables than implied by media accounts, which seemed to believe that pesticides are categorically associated with increased odds of ADHD. The association is only for a minority of ADHD-diagnosed kids (17.6 percent), and affects less than two percent of the total population examined.
What about causality?
The model used by the authors is logistic regression. The idea is that the authors assume that the presence of metabolites (or the pesticide exposure, which leads to the presence of metabolites) cause ADHD, and then they see how big an effect is. In other words, if pesticides lead to ADHD, then how much an increase is it causing?
Of course such as study cannot prove causality, only an association, but in this case there is a large body of research suggesting that there may be a causal relationship. But the astute reader must keep in mind that the causality cannot be proven in a study of this nature. At best, in an observational setting, causality has to be established with mechanistic explanations and a high odds ratio.
What can be said about kids with higher levels of exposure?
While an analysis on what happens with a ten-fold increase in exposure sound like it captures all the information, it actually tells us very little about how much risk this actually entails, since it doesn’t capture what exposure levels exist.
Therefore it is extremely important to compare those kids who have higher exposure (the half of the group with the highest exposure levels) with those with undetectable levels of exposure. This is where Reuters accurately reports that “researchers found those with high levels were almost twice as likely to develop ADHD as those with undetectable levels.” Time Magazine, however, turned to a fear-inducing assessment that we are seeing large increases with low levels of exposure, stating the same result thus: “The effect was seen even at the low end of exposure: kids who had any detectable, above-average level of the most common pesticide metabolite in their urine were twice as likely as those with undetectable levels to record symptoms of the learning disorder.”
Indeed, this is worrisome, though perhaps there is consolation in the fact that those kids on the bottom half of exposure did not have an increase compared to those with undetectable levels, a point not noted in Reuters or Time. Certainly, half of our kids at higher risk of ADHD is serious, as the lead author of the study noted. Yet the message to the public would be clearer if the whole story were told rather than a part: the Pediatrics study did not convincingly show an overall increase in ADHD, but rather a minority subtype of ADHD. Kids at the highest levels of exposure are at increased risk, but those at low levels are not.
A next step might be to see whether washing vegetables and fruit is sufficient to lower these metabolite levels to those that are not harmful, or whether more drastic actions need to be taken. Hats off to the Huffington Post, which noted, “A 2008 Emory University study found that in children who switched to organically grown fruits and vegetables, urine levels of pesticide compounds dropped to undetectable or close to undetectable levels.” We cannot be sure of a causal relationship, but the research is increasingly pointing toward the likelihood that reducing pesticide consumption reduces the risk for hyperactive/impulsive ADHD, present in about two percent of kids.