consumption of fish contaminated by upstream gold-mining activities (Grandjean et al. 1999). Among children, the mean hair Hg concentration ranged from a geometric mean (range) of 11.9 (35.1) ppm for the lowest exposed of the three communities on the Rio Tapajos to 25.4 (82.9) ppm for the highest; 80% of the children in these villages exceeded 10 ppm. Most of the children reportedly ate fish for two meals per day. In the hair samples available for 63% of the children's mothers, the mean Hg concentration was 11.6 ppm. The battery of neurobehavioral tests administered to the children focused on motor function, attention, visual-spatial function, and short-term memory (finger tapping, Santa Ana dexterity test, WISC-III digit span forward, and Stanford-Binet copying (including recall), and bead memory subtests. In three villages, the tests were administered in Portuguese, although in the fourth (Village D), administration required the services of a Mundurucu interpreter. (The finger tapping, and Santa Ana dexterity tests could not be administered to children in that village.) Combining all four villages, children's hair Hg concentrations were significantly associated with their scores on finger tapping (both preferred and other hand; both p < 0.001), Santa Ana dexterity test (preferred hand, p = 0.005; other hand, p = 0.05), WISC-III digit span (p = 0.001), Stanford-Binet copying (p < 0.001) and recall (p < 0.001), and Stanford-Binet bead memory (p < 0.001). Adjustment for community generally reduced the magnitude of the associations, sometimes dramatically (e.g., from p < 0.001 to p = 0.99 for finger tapping preferred hand). Hair Hg concentrations and village of residence were so highly confounded, however, that adjustment for village might be inappropriate.
In the French Guiana cohort assembled by Cordier and Garel (1999) 206 children 5 to 12 years old were administered a battery of neuropsychological tests that included finger tapping, three subtests from the Stanford-Binet (block designs, copying designs, and bead memory), and two subtests from the McCarthy scales (numerical memory and leg coordination). Median maternal-hair concentration was 6.6 ppm (range, 2.6-17.8 ppm). With adjustment for potential confounders, increased Hg concentrations were associated with copying-design score especially in boys. The findings are complicated, however; when only the children living in the region that has higher exposures were considered and the analyses were stratified by sex, increased Hg concentrations were associated with poorer leg coordination in boys and poorer block-design scores in girls.