OBJECTIVE. Thimerosal, a mercury compound used as a preservative in vaccines administered during infancy, has been suspected to affect neuropsychological development. We compared the neuropsychological performance, 10 years after vaccination, of 2 groups of children exposed randomly to different amounts of thimerosal through immunization.
METHODS. Children who were enrolled in an efficacy trial of pertussis vaccines in 1992–1993 were contacted in 2003. Two groups of children were identified, according to thimerosal content in vaccines assigned randomly in the first year of life (cumulative ethylmercury intake of 62.5 or 137.5 μg), and were compared with respect to neuropsychological outcomes. Eleven standardized neuropsychological tests, for a total of 24 outcomes, were administered to children during school hours. Mean scores of neuropsychological tests in the domains of memory and learning, attention, executive functions, visuospatial functions, language, and motor skills were compared according to thimerosal exposure and gender. Standard regression coefficients obtained through multivariate linear regression analyses were used as a measure of effect.
RESULTS. Nearly 70% of the invited subjects participated in the neuropsychological assessment (N = 1403). Among the 24 neuropsychological outcomes that were evaluated, only 2 were significantly associated with thimerosal exposure. Girls with higher thimerosal intake had lower mean scores in the finger-tapping test with the dominant hand and in the Boston Naming Test.
CONCLUSIONS. Given the large number of statistical comparisons performed, the few associations found between thimerosal exposure and neuropsychological development might be attributable to chance. The associations found, although statistically significant, were based on small differences in mean test scores, and their clinical relevance remains to be determined.
- ethylmercury compounds
- developmental disabilities
- controlled trial
Concern has been raised regarding the potential effect of thimerosal, a mercury compound used as a preservative in vaccines, on neuropsychological development in children.1 Autism,2 motor tics,3,4 mental retardation, language disorders, and attention-deficit/hyperactivity disorder5 have been alleged as adverse events attributable to thimerosal contained in vaccines. In contrast, the results of large, well-designed studies do not support such associations between thimerosal exposure and neuropsychological development.6–12
Several international agencies recommended different safe limits of exposure to ethylmercury (a metabolite of thimerosal), according to body weight, on the basis of the available pharmacokinetic data on methylmercury, a similar compound.12,13 However, the blood half-life of ethylmercury from thimerosal contained in vaccines is shorter than that of methylmercury.14 Moreover, none of the studies conducted to date has been able to demonstrate what amount of thimerosal administered with vaccines is associated with adverse neurotoxic events.9,12,13 Nonetheless, since 1999, regulatory agencies in both the United States and Europe have recommended eliminating thimerosal from vaccines when possible, as a precautionary measure.15–17
Assessing the effects of thimerosal in vaccines long after immunization is complicated by a variety of factors that potentially affect the quality of data on vaccine exposure and measures of neuropsychological development. Controlling for social and educational factors and some health conditions may be difficult in observational studies in the general population, in the absence of high-quality records including information on each of these factors.
A randomized, controlled, clinical trial on the efficacy of acellular pertussis vaccines started in Italy in 199218 and included a diphtheria-tetanus-acellular pertussis (DTaP) vaccine manufactured by Chiron Biocine that contained thimerosal and a DTaP vaccine manufactured by SmithKline Beecham that contained 2-polyphenoxyethanol (a compound with no known neurotoxicity) as a preservative. Because infants enrolled in this trial were assigned randomly to receive different vaccines, they also were exposed randomly to different cumulative quantities of thimerosal in their first year of life. Moreover, children were monitored actively, with monthly contacts up to the age of 6 years, and did not receive other vaccines containing thimerosal during this period. In the present study, we compared neuropsychological outcomes, 10 years after vaccination, for children who had been exposed to different amounts of thimerosal.
Exposure to Thimerosal
In 1992–1993, 15 601 healthy, 2-month-old infants were enrolled in the Italian Trial on Pertussis Vaccines.18–20 In this trial, infants were selected from the general population in 4 of Italy's 20 regions (Fig 1) and were assigned randomly to receive, under double-blind conditions, 3 doses of 1 of 4 vaccines, 2 of which were DTaP vaccines from 2 different manufacturers. One DTaP vaccine contained 50 μg of thimerosal (or 25 μg of ethylmercury) per dose, and the other was thimerosal-free (2-phenoxyethanol was used as preservative). The 3 doses of DTaP vaccine were administered at 2, 4, and 6 months of age. To comply with Italy's vaccination schedule, all children also received 3 doses of hepatitis B virus vaccine (child formulation), each of which contained 25 μg of thimerosal (or 12.5 μg of ethylmercury), at 2, 4, and 12 months of age, and a fourth dose of diphtheria-tetanus vaccine, which contained 50 μg of thimerosal (or 25 μg of ethylmercury), at 11 months of age. Table 1 reports the vaccination schedule and the cumulative amounts of ethylmercury, the mercury metabolite of thimerosal, administered to infants in the trial according to age. These children received no other doses of thimerosal-containing vaccines until they were 6 years of age, when they received a booster dose of either diphtheria-tetanus vaccine or DTaP vaccine, in accordance with Italy's vaccination schedule. Therefore, in the first 12 months of life, the cumulative intake of ethylmercury, the mercury metabolite of thimerosal, was 137.5 μg for the children who were assigned randomly to receive the DTaP vaccine that contained thimerosal (“higher intake group”) and 62.5 μg for those who received the thimerosal-free DTaP vaccine (“lower intake group”).
Enrollment and Neuropsychological Assessment
The study population for the present study included children living in the Veneto Region who were vaccinated with 3 doses of 1 of the 2 DTaP vaccines during the trial. A random sample from all 3399 subjects originally included in the trial in the region was extracted progressively until the desired sample size was met. Parents were sent a letter of invitation; once they provided written informed consent, they were interviewed by trained nurses over the telephone to update the information on the children's medical history and sociodemographic characteristics. An appointment was scheduled for administration of neuropsychological tests to the children. No economic incentive was provided to participants. The tests were administered during a single 2-hour session during regular school hours, by 1 of 10 trained psychologists (who were blinded to the vaccine received by each child).
A total of 11 neuropsychological tests, which had been selected on the basis of previous studies,3,21 were administered to the enrolled children. The tests were used to explore areas of cognitive function allegedly affected by thimerosal exposure, were age-appropriate, and were standardized for the Italian population. The following domains were assessed: (1) memory and learning, with the California Verbal Learning Test-Children Version22 and the Wechsler Intelligence Scale for Children-Revised (WISC-R) digit span test23; (2) attention, with the continuous performance test (Conners’ version), a computer-assisted test that evaluates sustained attention, inhibition, and impulsivity24; (3) executive functions, with the WISC-R coding test23 and the phonemic verbal fluency test, which assesses the ability to access a lexicon through phonemic cues and to set up adequate verbal search strategies25; (4) visuospatial functions, with the WISC-R block design test23; (5) language, with the WISC-R vocabulary and similarities tests,23 the Boston Naming Test,26 and the semantic verbal fluency test, which assesses the ability to access a lexicon according to semantic categories25; and (6) motor skills, with the finger-tapping test,27 which provides an assessment of fine motor speed and coordination, and the writing praxia test,28 which measures the writing speed of the child.
Autism was considered as a separate outcome of the study and was assessed through review of the medical history with the parents and the medical charts of family pediatricians, according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria.29 The presence of any motor or phonic tics was screened by psychologists through direct observation of the children during the neuropsychological assessments. The psychologists were trained to recognize tic behaviors, and 2 binary variables were created to indicate the presence of motor and phonic tics.
Children were 10 to 12 years of age when the neuropsychological evaluations were conducted, between the end of 2003 and early 2005. The study was approved by the review boards of the Istituto Superiore di Sanità (Italy's National Health Institute) and the Veneto Region.
Abstraction of Other Information
Sociodemographic variables and health information at birth and in infancy were derived from the original trial database for all children included in the present study. Medical conditions identified during the telephone interviews and relevant therapies were verified with the family pediatricians through review of local medical charts. For each child, the following information was collected: gender, birth weight, gestational age, mother's age at birth, Rh immunoglobulins during pregnancy, hepatitis B immunization at birth, type of delivery, breastfeeding duration, family composition, parents’ education, presence of diseases of the central nervous system or other chronic diseases, and current prescription of antihistamines or antiepileptic drugs.
Sample Size and Data Analysis
Comparison between the 2 groups of children (higher and lower intake groups) was performed for a total of 24 neuropsychological outcomes. With the assumption that children in the lower intake group had mean scores for neuropsychological tests equal to reference values, a sample size of 700 children in each group (for a total of 1400 children) would be sufficient to yield a statistical power of 90% for identifying differences of at least ±5%, at a 5% significance level with a 2-tailed test.
We assessed differences between groups by using Student's t test for continuous variables. The difference between the mean neuropsychological test scores, with 95% confidence intervals (CIs), was used as a measure of effect size. The Mann-Whitney U test was used to validate results in cases of violation of normality and homoscedasticity assumptions. The χ2 test or Fisher's exact probability test were used for categorical variables. We also calculated risk ratios (RRs), with 95% CIs, to assess the association between thimerosal exposure and the presence of tics at the univariate level.
We performed separate, ordinary, least-squares, multivariate, linear regression analyses, using the raw score of each test as the dependent variable, to assess the association between thimerosal intake and the results of neuropsychological tests, with adjustment for the following potential confounders: age at test administration, gender, birth weight, gestational age, type of delivery, breastfeeding duration, family composition, mother's age at birth, parents’ education, presence of diseases of the central nervous system or other chronic diseases, current prescription of antihistamines or antiepileptic drugs, Rh immunoglobulins during pregnancy, and hepatitis B immunization at birth. Standardized regression coefficients (SRCs) were used as a measure of effect size, with the SRC representing the quantitative difference in the outcome scores, expressed in SD units, for the lower versus higher thimerosal intake group. We also assessed the effect of thimerosal exposure in separate analyses according to gender. In secondary analyses, we performed logistic regression analyses using as the dependent variable the presence of motor and phonic tics. The adjusted RR was used as a measure of effect size. During statistical analyses, the 2 study groups were masked with respect to the amount of thimerosal exposure. We did not correct for multiple comparisons, to maintain the highest possible sensitivity during statistical analyses.
A total of 1979 families, of the total regional cohort of 3399, were sent an invitation letter (Fig 1). We were unable to contact 160 families (8.1%) by telephone; of the remaining 1819 families, 114 (6.3%) declined participation, and 1 child (0.05%) had died. Telephone interviews were conducted for the remaining 1704 families (87.9% of the eligible population). We detected, through the telephone interviews with parents and reviews of medical charts, 1 case of autism among the 856 children in the lower thimerosal intake group and no cases among the 848 children in the higher thimerosal intake group.
A total of 301 children from the 1704 interviewed families did not undergo the neuropsychological evaluation. Therefore, a total of 1403 children (70.9%) were included in the analysis, 697 in the higher intake group and 706 in the lower intake group.
The proportions of families whose child did not undergo the neuropsychological tests did not differ significantly between the 2 groups. Parents failed to have their children evaluated for the following reasons: lack of time to participate, fear that the child would feel uncomfortable during the evaluation, or failure to keep the appointment for the tests (higher intake group: n = 150; lower intake group: n = 146); in some cases, the child had a medical condition (lower intake group: epilepsy, n = 1; autism, n = 1; leukemia, n = 1; retinitis pigmentosa, n =1; higher intake group: head injury, n = 1).
The characteristics of the 1403 children who underwent neuropsychological assessments are listed in Table 2. The 2 groups were similar in terms of sociodemographic characteristics, clinical characteristics, and parents’ educational level, whereas birth weight was slightly lower for the higher intake group.
Mean scores and SDs for each neuropsychological test according to thimerosal exposure are reported in Table 3, for the entire population included in the study and according to gender. All mean scores were in the normal range according to reference standards,22–28 and the study groups were similar with respect to almost all neuropsychological outcomes. Children in the higher intake group performed significantly worse than children in the lower intake group only in the finger-tapping test with the dominant hand, although the difference in mean scores was modest (difference: 1.01; 95% CI: 0.30–1.73; P = .006, Student's t test). Motor tics were observed for 2.56% of patients in the lower thimerosal intake group and 2.91% of patients in the higher intake group (RR: 1.13; 95% CI: 0.60–2.13; P = .744, Fisher's exact probability test). Phonic tics were detected for 0.43% of patients in the lower intake group and 0.87% of patients in the higher intake group (RR: 2.04; 95% CI: 0.51–8.13; P = .338, Fisher's exact probability test).
Girls in the higher intake group performed significantly worse than did those in the lower intake group in the finger-tapping test with the dominant hand (difference in scores: 1.08; 95% CI: 0.11–2.05; P = .029, Student's t test) (Table 3). In multivariate analyses, we did not find any statistically significant associations for the overall group of children and for boys (Table 4). We found worse performance in the higher intake group for the Boston Naming Test (SRC: −0.16; P = .025) and the finger-tapping test with the dominant hand (SRC: −0.16; P = .029) for girls. Finally, we did not find any significant association in the logistic regression analyses for phonic and motor tics, in the entire population or according to gender.
One of the major problems in assessing the effects of exposure to thimerosal during infancy on neuropsychological development is identifying groups of children who have good vaccination records, who received different amounts of thimerosal, and who, many years later, are still comparable with respect to the occurrence of factors that may affect neuropsychological development. In this regard, our study population, as a subset of children who had participated in the Italian Trial on Pertussis Vaccines, had several advantages, particularly the following: (1) the children belonged to groups that had been assigned randomly to receive different thimerosal exposures through vaccines in the first year of life, (2) the children did not receive any other thimerosal-containing vaccines until they were 6 years of age, (3) vaccinations were administered under strict rules and were carefully monitored, (4) vaccination records were very accurate, being collected under rigorous clinical trial conditions, and (5) children were monitored actively for 6 years. Furthermore, Italy's population has low population mobility, which contributed to the high response rate 10 years after enrollment in the original study. Moreover, the school curriculum until 13 years of age is standardized throughout the country, and the children had similar learning experiences. Finally, the neuropsychological assessments and the statistical analyses were performed under blinded conditions with respect to exposure to thimerosal.
According to our results, greater thimerosal exposure through vaccines administered in the first year of life was significantly associated with lower scores for 2 neuropsychological outcomes, in the domains of motor function (finger-tapping test) and language (Boston Naming Test), among girls. However, the differences in the mean scores were very small and of doubtful clinical relevance. Neuropsychological deficits in the domains of language and, to a lesser extent, motor functions were also observed in a cohort of children exposed prenatally to methylmercury,21 although a direct comparison may be inappropriate because ethylmercury, the metabolite of thimerosal, has different toxicokinetic features.14,30 To date, studies on the effects of thimerosal contained in vaccines have not provided a clear pattern of associations, and many authors found even better neurodevelopmental outcomes for children exposed to thimerosal.4,7,9 An association between thimerosal contained in vaccines and language delays was found in a study from the Vaccine Safety Datalink3 and in a recent study in 4 US health maintenance organizations.9 In the latter study, however, the Boston Naming Test outcome was not affected by thimerosal exposure. We failed to observe any association with tics, in contrast to the findings of some observational studies.3,4,9
Because we did not correct for multiple comparisons in the statistical analyses, we expected to find 4 significant associations among the 78 univariate statistical tests, at the 5% statistical significance level (Table 3), and 4 among the 72 multivariate analyses (Table 4) as effects of chance. Because we found only 2 significant associations between thimerosal exposure and neuropsychological outcomes, the observed associations might actually reflect the effects of chance. In addition, the associations were based on small differences in scores, they were detected only for girls, and they were not consistent with results from other studies.
Some limitations should be considered in the interpretation of our results. The cumulative intake of thimerosal was relatively low, compared with that in other countries including the United States, where vaccination schedules included more thimerosal-containing vaccines in the first year of life. Moreover, there was no comparison group with no exposure to thimerosal, although our setting was appropriate to identify a dose-response effect in the absence of any evidence suggesting a threshold dose for observation of an effect. Our analysis included only healthy children who were selected during enrollment in the original trial, and some families might have declined to participate in the present study because their children had cognitive developmental problems. This might have reduced the prevalence of adverse neuropsychological conditions and might have made potential differences hard to detect. The eligibility criteria of the original trial also limited the participation of low birth weight children, and only 55 children with birth weights of <2500 g underwent the neuropsychological evaluation (data not shown). Moreover, only 1% of children in this study received hepatitis B virus vaccine at birth. Although no effect of birth weight according to thimerosal intake was detected through multivariate analyses, our study was not powered to detect an association of thimerosal exposure and neuropsychological development in low birth weight infants.
No study conducted to date has been able to provide conclusive evidence of an effect of thimerosal on neuropsychological development. Final judgments regarding this association must rely on the entire body of results from studies conducted in different settings and with different levels of validity and on the coherence of results. The lack of consistency among the results of our study and other available studies suggests that an association between thimerosal exposure through vaccination in infancy and neuropsychological deficits is unlikely or clinically negligible. Additional data from populations with wider ranges of exposure to thimerosal and additional neuropsychological assessments at older ages may help to clarify the issue of potential associations between thimerosal and neurodevelopmental outcomes.
The study was supported in part by the US Centers for Disease Control and Prevention, through contract 2002-N-00448 with the Istituto Superiore di Sanità.
We thank the Regional Health Authorities in the Veneto Region, all of the families that participated in the Italian Trial on Pertussis Vaccines, the staff members at the children's schools, and the pediatricians in the Veneto Region, who contributed to the logistics and performance of the neuropsychological evaluations. We also thank the study staff members, who were supported with the funds from the US Centers for Disease Control and Prevention, including consultants (Michela Cendron and Bernardo dalla Bernardina), psychologists (Katia Battistella, Alessia Ciccola, Francesca Gnoato, Laura Mattiuzzi, Sonia Mele, Andrea Melendugno, Francesca Offredi, Davide Paganini, Sara Prioni, Gabriella Scala, and Irene Spera), and study nurses (Gianna Ceccato, Liliana Cortese, Mara Girelli, Nadia Grandin, Ave Marchesin, Loretta Meneghin, and Emma Nami), without whom such a large study would not have been possible. Thanks also go to Francesca Romana Meduri at the Istituto Superiore di Sanità for the administrative management of the contract, to Mark Kanieff, who helped to revise the manuscript, and to Nancy Binkin, MD, for her friendly capable support. During the contract, Dr Tozzi moved from the Istituto Superiore di Sanità to Bambino Gesù Hospital, maintaining his role as principal investigator.
- Accepted May 13, 2008.
- Address correspondence to Alberto E. Tozzi, MD, Epidemiology Unit, Bambino Gesù Hospital, Piazza S. Onofrio 4, 00165 Rome, Italy. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
The issue of thimerosal neurotoxicity and infant vaccines remains controversial, although many studies support the lack of association between thimerosal and neuropsychological development.
What This Study Adds
The weak and inconsistent associations found in this study suggest that the association between thimerosal exposure through vaccination in infancy and neuropsychological deficits is unlikely or clinically negligible.
- ↵Institute of Medicine. Immunization Safety Review: Thimerosal-Containing Vaccines and Neurodevelopmental Disorders. Washington, DC: National Academy Press; 2001
- ↵Verstraeten T, Davis RL, DeStefano F, et al. Safety of thimerosal-containing vaccines: a two-phased study of computerized health maintenance organization databases. Pediatrics.2003;112 (5):1039– 1048
- ↵Andrews N, Miller E, Grant A, Stowe J, Osborne V, Taylor B. Thimerosal exposure in infants and developmental disorders: a retrospective cohort study in the United Kingdom does not support a causal association. Pediatrics.2004;114 (3):584– 591
- ↵Geier D, Geier MR. Neurodevelopmental disorders following thimerosal-containing childhood immunizations: a follow-up analysis. Int J Toxicol.2004;23 (6):369– 376
- ↵Heron J, Golding J; ALSPAC Study Team. Thimerosal exposure in infants and developmental disorders: a prospective cohort study in the United Kingdom does not support a causal association. Pediatrics.2004;114 (3):577– 583
- Fombonne E, Zakarian R, Bennett A, Meng L, McLean-Heywood D. Pervasive developmental disorders in Montreal, Quebec, Canada: prevalence and links with immunizations. Pediatrics.2006;118 (1). Available at: www.pediatrics.org/cgi/content/full/118/1/e139
- McCormick M, Bayer R, Berg A. Report of the Institute of Medicine: Immunization Safety Review: Vaccines and Autism. Washington, DC: National Academy Press; 2004
- ↵Parker SK, Schwartz B, Todd J, Pickering LK. Thimerosal-containing vaccines and autistic spectrum disorders: a critical review of published original data. Pediatrics.2004;114 (3):793– 804
- ↵Ball LK, Ball R, Pratt RD. An assessment of thimerosal use in childhood vaccines. Pediatrics.2001;107 (5):1147– 1154
- ↵Pichichero ME, Gentile A, Giglio N, et al. Mercury levels in newborns and infants after receipt of thimerosal-containing vaccines. Pediatrics.2008;121 (2). Available at: www.pediatrics.org/cgi/content/full/121/2/e208
- ↵European Agency for the Evaluation of Medicinal Products. EMEA public statement on thiomersal containing medicinal products, July 8, 1999. Available at: www.emea.europa.eu/pdfs/human/press/pus/2096299EN.pdf. Accessed December 12, 2008
- ↵Salmaso S, Mastrantonio P, Tozzi AE, et al. Sustained efficacy during the first 6 years of life of 3-component acellular pertussis vaccines administered in infancy: the Italian experience. Pediatrics.2001;108 (5). Available at: www.pediatrics.org/cgi/content/full/108/5/e81
- ↵Grandjean P, Budtz-Jorgensen E, White RF, et al. Methylmercury exposure biomarkers as indicators of neurotoxicity in children aged 7 years. Am J Epidemiol.1999;150 (3):301– 305
- ↵Delis DC, Kramer JH, Kaplan E, Ober BA. The California Verbal Learning Test-Children Version. New York, NY: Psychological Corp; 1994
- ↵Wechsler D. WISC-R: Scala di Intelligenza Wechsler per Bambini Riveduta: Manuale [WISC-R: Wechsler Intelligence Scale for Children-Revised: Manual; in Italian]. Florence, Italy: Organizzazioni Speciali; 1998
- ↵Conners CK; Multi-Health Systems Staff. Conners’ Continuous Performance Test. Toronto, Canada: Multi-Health Systems; 2000
- ↵Bisiacchi PS, Cendron M, Gugliotta M, Tressoldi P, Vio C. Batteria di Valutazione Neuropsicologica per l'Età Evolutiva [Battery for Neuropsychological Evaluation in Childhood; in Italian]. Trent, Italy: Erickson; 2005
- ↵Kaplan E, Goodglas H, Weintraub S. Boston Naming Test. Philadelphia, PA: Lea & Febinger; 1983
- ↵Nussbaum NL, Bigler ED. Halstead-Reitan neuropsychological test batteries for children. In: Reynolds C, ed. Handbook of Clinical Child Neuropsychology. 2nd ed. New York, NY: Plenum Press; 1997
- ↵Tressoldi PE, Cornoldi C. Batteria per la Valutazione Della Scrittura e Della Competenza Ortografica Nella Scuola Dell'Obbligo [Battery for the Evaluation of Writing and Spelling Competence in Elementary and Junior High School; in Italian]. Florence, Italy: Organizzazioni Speciali; 1991
- ↵American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994
- Copyright © 2009 by the American Academy of Pediatrics