These analyses revealed an absence of consistent differences between the scores of children in the amalgam and composite treatment groups on a battery of neuropsychological tests that assessed a wide range of domains, including intelligence, achievement, language, memory, learning, visual–spatial skills, verbal fluency, fine motor function, problem solving, attention, and executive function. The findings were similar when the dichotomous variable treatment-group assignment was replaced by two continuously distributed indices of exposure, one that combined the amount and duration of amalgam a child received and one that was a biomarker, urinary mercury concentration. Furthermore, no evidence was found to support the hypothesis that a subset of children in the amalgam group suffered substantial harm. The number of significant differences observed was similar to that which might have been expected to occur by chance.
Although neuropsychological deficits associated with amalgam exposure have been reported in several studies of dental professionals (Bittner et al. 1998
; Echeverria et al. 1995
; Ngim et al. 1992
) and others exposed occupationally to mercury (Rohling and Demakis 2006
), our findings are similar to those involving mercury exposure in cohorts drawn from the general population of adults and children (Brownawell et al. 2005
). In a cross-sectional study of 550 30- to 49-year-old healthy employed adults, scores on tests of verbal memory, nonverbal memory, attention, psychomotor speed, and fine motor coordination were not significantly associated with any of several exposure indices considered (number of visible amalgam surfaces, number of visible occlusal amalgam surfaces, urinary mercury concentration) (Factor-Litvak et al. 2003
). The mean urinary mercury concentration in that cohort of adults, 1.7 μg/g creatinine, was higher than the mean concentration of 0.9 μg/g creatinine among the children in the amalgam treatment group in our trial 5 years after placement of their first amalgam restorations. In a study of 1,663 Vietnam-era veterans, the total number of tooth surfaces with amalgam fillings was unrelated to clinical neurological signs (e.g., tremor, coordination, station, gait, strength, sensation, muscle stretch reflexes, or indices of peripheral neuropathy), although it was associated with vibrotacile sensation in non-diabetic participants (Kingman et al. 2005
). In a study of 384 German 6-year-olds, 24-hr urinary excretion of mercury, which averaged 0.16 μg, was not significantly related to scores on a variety of tests, including the Vocabulary and Block Design subtests of the WISC and five tests of the computerized Neurobehavioral Evaluation System 2 (pattern comparison, pattern memory, tapping, simple reaction time, continuous performance test) (Walkowiak et al. 1998
). In this cohort, some indices of visual contrast sensitivity did decline with increasing urinary mercury excretion, however (Altman et al. 1998).
Over the course of the follow-up interval, the scores of children in both treatment groups tended to change in the direction of improved performance, even on tests for which scores are standardized for age. Several factors might have contributed to improved performance over time. First, this could represent a type of sampling bias, reflecting the characteristics of families who are motivated to enroll in such a trial and to participate for its full duration. Second, all tests except the WISC-III and WIAT were administered yearly, so the general improvement in scores might reflect the familiarity that children developed with the test materials and expectations. Particularly large improvements tended to be on performance-based tests, such as the WRAVMA pegboard and the Processing Speed composite of the WISC-III, one component of which is Symbol Search, a timed task that involves matching symbols and digits. A substantial improvement was also noted on the WRAML Learning Index, which reflects the rapidity with which a child learns new material, such as sound–symbol pairs, a word list, and the locations of hidden designs. Repeated administration of these tasks, even at yearly intervals, might be expected to result in an increased rate of acquisition of the material.
As noted, the dental treatment needs of the children enrolled in the trial were substantial and exceeded those typical of the general population of U.S. children. For example, among 6- to 11-year-old children who participated in the National Health and Nutrition Examination Survey (NHANES) 1999–2002, the prevalence of dental caries in primary teeth was 22%, and the mean numbers of decayed or filled primary teeth and surfaces were 1.7 and 3.7, respectively (Beltran-Aguilar et al. 2005
). The prevalence of dental caries in permanent teeth was 20% (mean number of decayed, missing, and filled teeth and surfaces in permanent teeth were 0.1 and 0.4, respectively) (Beltran-Aguilar et al. 2005
). Therefore, children assigned to the amalgam group in the NECAT are likely to have experienced greater exposures to mercury vapor from amalgams than do most children in the United States. Given our failure to detect significant differences between the amalgam and composite groups in neuropsychological function, these results provide reassurance that the use of dental amalgam to repair caries is not producing substantial neuropsychological morbidity in the general population of children in the United States.
The conclusions must be tempered, first, by a recognition that the follow-up interval of 4–5 years might have been too short to allow for the expression of such deficits. Second, the critical window of children’s greatest vulnerability to elemental mercury might already have passed by the time the children were enrolled in the trial (≥ 6 years of age). Given the heightened sensitivity of the fetus to methylmercury, prenatal exposure to mercury vapor, which is known to cross the placenta, warrants increased attention. In the NHANES 1999–2000 survey, among women of child-bearing age, an increase of 10 dental surfaces restored with amalgam was associated with an estimated increase of 1.8 μg/L in urinary mercury concentration (Dye et al. 2005
). Mercury level in amniotic fluid is weakly associated with number of amalgam fillings (Luglie et al. 2005
). A recent case–control study did not, however, find an increased risk of delivering a low-birth-weight infant among women who had up to 11 amalgam restorations placed during pregnancy (Hujoel et al. 2005
). The results of studies of the reproductive outcomes of women with dental workplace exposures have been mixed (Dahl et al. 1999
; Elghany et al. 1997
; Ericson and Kallen 1989
). In some of these studies, distinguishing the potential impact of mercury exposure from the impacts of other workplace exposures, such as to disinfectants containing ethanol and benzene, is difficult. Third, the prevalence of children with enhanced sensitivity to elemental mercury might be too low among the children enrolled in the NECAT for us to have been able to detect their effects on the distribution of responses. Fourth, children with preexisting neuropsychological or behavioral disorders were not eligible for enrollment. Our findings therefore do not provide any information about the possibility that amalgam-related exposure to mercury vapor might exacerbate such disorders. Nevertheless, our results indicate that even among children with substantial dental needs, an increased risk of neuropsychological deficits could not be detected among children whose dental restorations contained elemental mercury.