Skip to Navigation
Skip to Content
Share this page


Welcome to PSR's Environmental Health Policy Institute, where we ask questions -- then we ask the experts to answer them. Join us as physicians, health professionals, and environmental health experts share their ideas, inspiration, and analysis about toxic chemicals and environmental health policy.


More Topics »

Perspectives on Estimating the Population Burden of Children’s Exposures to Environmental Chemicals

Posted on April 4, 2012

By David C. Bellinger, MD

This essay is in response to the question: How does the environment influence brain development? What are the exposures of greatest concern? What is the latest science and how can we translate that science into protective public health policy?

Conventional approaches to calculating the societal impact of children’s exposures to environmental chemicals are likely to yield underestimates.  It is typically assumed that the burden associated with a chemical depends solely on the number of children who, as a result of exposure, meet diagnostic criteria for a disease such as mild mental retardation (i.e., IQ

A focus on the magnitude of the effect size associated with an exposure, without concomitant consideration of its prevalence, also tends to result in underestimation of its population impact.  For example, it is often claimed that an exposure that causes a modest impact, such as a 3 point decline in IQ for a 10 μg/dL increase in blood lead level, is “clinically insignificant” because the Full-Scale IQs of most children who suffer a lead-associated decline will remain within normal limits.  But if an exposure is prevalent and causes a dose-dependent decrement in children’s IQ, the total number of IQ points lost in the population as a whole might be surprisingly large, and the reduction in the intellectual resources of a society substantial.

I suggest that a population-based approach that takes account of both the effect size associated with an exposure and the prevalence of the exposure is the most appropriate strategy for estimating its public health import.  I used such an approach to estimate and compare the contributions of environmental chemical exposures to children’s neurodevelopment to the contributions of a number of important risk factors (Bellinger, 2012).  The other risk factors included congenital heart disease (CHD, the most common structural birth defect), acute lymphocytic leukemia (ALL, the most common malignancy in children), preterm birth, neurodevelopmental disorders such as autism spectrum disorders (ASDs) and attention deficit hyperactivity disorder (ADHD), traumatic brain injuries, and socioeconomic and nutritional risks such as nonorganic failure to thrive and iron deficiency.  Full-Scale IQ (FSIQ) served as the basis for comparison insofar as this is the endpoint most commonly measured in studies of children’s neurodevelopment.  Based on published prevalence estimates, I calculated the numbers of cases of each condition among the 25.5 million children under 5 years of age in the United States.  The effect sizes (FSIQ losses) were derived from meta-analyses. The total FSIQ points lost as a result of a condition were estimated by multiplying the number of cases in this age group by its associated effect size.

Calculations were carried out for the chemicals for which the necessary data were available: lead, methylmercury, and organophosphate pesticides (OPs).  The distributions of biomarker levels in the U.S. children were obtained from the NHANES surveys.  Meta-analyses or pooled analyses provided the estimates of the effect sizes (dose-effect relationships).

The estimated total FSIQ losses associated with chemicals were surprisingly large.  The greatest loss (34,000,000 points) was associated with preterm birth, but the losses for lead (23,000,000) and OPs (17,000,000) were as large or larger than those estimated for many other risk factors (e.g., CHD:105,000; ALL: 136,000; brain tumors: 37,000; ASDs: 7,000,000; ADHD: 17,000,000; iron deficiency: 9,000,000; traumatic brain injuries: 6,000,000).  This is not because the effect sizes for the chemicals are so large but because the exposures that are associated with adverse effect are so prevalent.  For example, because no threshold has been identified for lead’s impact on FSIQ, all children contribute, at least a little, to the lead-associated FSIQ loss in the population.  This stands in contrast to CHD or ALL, for which only a relatively small number of children contribute to the total FSIQ loss.  This illustrates Rose’s contention, articulated in the context of radiation exposure, that, “ is the total dose falling on the whole population which determines the burden of health effects” (Rose, 1991). 

While the societal implications of a modest chemical-associated shift in the mean of the FSIQ distribution toward lower values have been noted previously, the focus has typically been on the resulting increase in the number of very low performers and the reduction in the number of high performers (e.g., Needleman et al., 1982).  The contribution to the societal burden of the losses of the much larger number of children in the middle of the FSIQ distribution has been largely overlooked, despite the fact that these children can contribute more to the total FSIQ loss than do children in the extreme tails.

Many caveats attend the method used to make the comparisons (see Bellinger, 2012), but the results suggest that in continuing to apply standards appropriate to evaluating the impact of chemical exposures on an individual child rather than on the population as a whole, we risk underestimating the public health burden associated with these exposures.


Bellinger DC. A Strategy for Comparing the Contributions of Environmental Chemicals and Other Risk Factors to Children’s Neurodevelopment. Environmental Health Perspectives, in press.

Lewington S, Clarke R, Qizilbash N, Peto R, Collins R; Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet 2002; 360(9349): 1903-13.

Needleman HL, Leviton A, Bellinger D. Lead-associated intellectual deficit. New England Journal of Medicine 1982; 579.

Rose G. Sick individual and sick populations. International Journal of Epidemiology 1985; 14; 32-38.

Rose G. Environmental health: problems and prospects. J R Coll Physicians London 1991; 25: 48-52.


Leave your comment

Enter this word: Change