 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* National Immunization Program, Centers for Disease Control and Prevention (CDC), Atlanta, Georgia
Battelle Memorial Institute, Centers for Public Health Research and Evaluation, Atlanta, Georgia
National Center on Birth Defects and Developmental Disabilities, CDC, Atlanta, Georgia
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
Methods. A case-control study was conducted in metropolitan Atlanta. Case children (N = 624) were identified from multiple sources and matched to control children (N = 1824) on age, gender, and school. Vaccination data were abstracted from immunization forms required for school entry. Records of children who were born in Georgia were linked to Georgia birth certificates for information on maternal and birth factors. Conditional logistic regression was used to estimate odds ratios (ORs).
Results. The overall distribution of ages at MMR vaccination among children with autism was similar to that of matched control children; most case (70.5%) and control children (67.5%) were vaccinated between 12 and 17 months of age. Similar proportions of case and control children had been vaccinated before 18 or before 24 months. No significant associations for either of these age cutoffs were found for specific case subgroups, including those with evidence of developmental regression. More case (93.4%) than control children (90.6%) were vaccinated before 36 months (OR: 1.49; 95% confidence interval: 1.04–2.14 in the total sample; OR: 1.23; 95% confidence interval: 0.64–2.36 in the birth certificate sample). This association was strongest in the 3- to 5-year age group.
Conclusions. Similar proportions of case and control children were vaccinated by the recommended age or shortly after (ie, before 18 months) and before the age by which atypical development is usually recognized in children with autism (ie, 24 months). Vaccination before 36 months was more common among case children than control children, especially among children 3 to 5 years of age, likely reflecting immunization requirements for enrollment in early intervention programs.
Key Words: autism • autism spectrum disorders • MMR vaccine • immunizations • epidemiology
Abbreviations: MMR, measles-mumps-rubella • IOM, Institute of Medicine • MADDSP, Metropolitan Atlanta Developmental Disabilities Program • DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th Edition • ASD, autism spectrum disorder • MR, mental retardation • OR, odds ratio • CI, confidence interval
Autism is a serious, life-long developmental disorder characterized by marked impairments in social interactions; communication skills; and repetitive, restrictive, or stereotyped behaviors, interests, and activities.1 Recent studies have suggested that the prevalence of autism is higher (30–60 per 10 000 children)2–5 than in studies conducted 15 to 20 years ago (4–5 per 10 000).6–10 The apparent increase in prevalence, coupled with reports of increasing numbers of children with autism being served by schools and service agencies,11–14 has prompted concerns that environmental exposures might be causing autism. Vaccines, particularly the measles-mumps-rubella (MMR) vaccine, are among the exposures for which there has been a great deal of speculation of a possible association with autism.
Wakefield et al15 were the first to propose that MMR vaccine might be causally linked to autism. They published a report describing 12 pediatric patients with inflammatory bowel conditions and regressive developmental disorders, mostly autism. In 8 of the 12 cases, the children’s parents or pediatricians suggested that MMR vaccine might have contributed to the onset of behavioral problems. The same investigators subsequently proposed a new syndrome consisting of certain gastrointestinal conditions associated with behavioral regression16 and reported identifying laboratory evidence of measles virus genome in the peripheral white blood cells and bowel biopsy specimens of a few such patients. The investigators, however, did not distinguish whether the virus was wild measles virus or vaccine strain virus.17,18 Several epidemiologic studies have not found an association between MMR vaccination and autism.19–24 The Institute of Medicine (IOM)25 reviewed the MMR-autism hypothesis and rejected a causal association at the population level but encouraged additional studies to evaluate more fully the possibility that there are subgroups of children who might be at increased risk of autism from MMR vaccination.
To examine further a possible relationship between MMR vaccine and autism, including in different subgroups of children, we conducted a large case-control study in metropolitan Atlanta in which we compared the MMR vaccination histories of a population-based sample of children with autism and school-matched control children who did not have autism.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
For the current study, case children were derived from the 987 children who were originally identified in the 1996 MADDSP prevalence year. During the period from 1999 through 2001, we were able to locate school records with the required immunization documents for 660 case children. We were not able to find school immunization records for the remaining children because the children had moved out of state, transferred to a school in a county that was not under MADDSP’s jurisdiction, transferred to a private school that was not accessible by MADDSP, or were being home schooled. When a child moved or transferred, the child’s permanent school record, including immunization form, was transferred to the child’s new school. We were not able to quantify how many children were lost for each of these reasons because of incomplete record keeping at the schools. An additional 17 case children were excluded from the study because we could not identify matched control children for them.
We attempted to match 3 control children to each case child and were successful for 97% of the case children; the remaining case children had fewer matched control children. Control children were selected from regular education programs and were matched to case children based on age in 1996 (within 1 year), gender, and school of attendance at the time of abstraction. However, when a case child was attending a psychoeducational school, a special school for children with behavioral and developmental difficulties, control children were selected from the school in the child’s residential area that the child would have attended had the child not had a disability. In addition, when a case child was in the last elementary grade level before middle school and was older than other children in his or her grade level, control children were selected from the middle school that the case child normally would have attended.
We excluded case and control children from the study when they were missing a vaccination form (15 case children and 14 control children). We also excluded children with incomplete vaccination forms when the form did not list at least 1 diphtheria-tetanus-pertussis vaccine by 2 years of age or at least 1 MMR vaccination at any age (4 case children and 1 control child). Children with a religious or medical exemption (1 case and 1 control) were not excluded from the study. After all exclusions, 624 case and 1824 control children remained in the study.
Classification of Autism Subgroups
The MADDSP data files were reviewed by a developmental pediatrician (M.Y.A.) to identify subgroups of children with potentially different susceptibilities to development of autism from an environmental exposure, such as MMR vaccine. These groups, which were not mutually exclusive, included 1) children without any indication of developmental delay before 12 months of age (ie, before the recommended age of the first MMR vaccination) or a preexisting condition, 2) children with any indication of loss of age-appropriate developmental skills (regression) or appropriate skills that failed to progress (plateau), and 3) children with and children without coexisting mental retardation (MR). Children without any indication of developmental delay at <1 year of age were children who did not lack any speech at appropriate ages, including cooing and babbling, and were socially responsive in the first year of life (eg, cuddling, appropriate eye contact, responding to parents voices). Children without a preexisting condition included children who did not have a major birth defect, a co-occurring developmental disability, or a major perinatal or postnatal insult (eg, infection, injury) that could have contributed to developmental delays. Children without a preexisting condition and without evidence of delay before 1 year of age were grouped into a single category. MR was defined as an IQ of 70 or less on the most recent psychometric test. We also attempted to examine information on family history of autism spectrum conditions or other developmental disabilities, but this information was incomplete in the records and not useful for analysis.
Vaccination History
Trained abstractors collected vaccination histories for both case and control children from the standardized state immunization forms that are required for all children who attend school and early intervention programs in Georgia. The forms are placed in each student’s permanent school file that is kept at the school where the child is enrolled. During the period in which children in our study would have enrolled in school, Georgia law required at least 1 dose of measles, mumps, and rubella vaccines, usually administered at 15 months of age as the combined MMR vaccine. Vaccination was also required for enrollment in preschool special education programs for 3- to 5-year-old children with disabilities.
Other Data Collection
For children with autism, additional developmental disability–related information was obtained from MADDSP data files. This included information on the presence of other developmental disabilities, epilepsy, and IQ level (for categorization of MR). In addition, we identified major birth defects among the case children by matching with Centers for Disease Control and Prevention’s Metropolitan Atlanta Congenital Defects Program, a population-based surveillance program of major birth defects that covers the same geographic area.27
For all case and control children, we obtained demographic information, including date of birth, gender, race, and birth state, from the birth certificate or registration form that is kept in each child’s permanent school record. We matched 355 (56%) case and 1020 (56%) control children to Georgia state birth certificate records, which allowed us to obtain additional information, such as each child’s birth weight and gestational age and the mother’s parity, age, race, and education.
Statistical Analyses
Determining exposure–disease associations requires knowledge of dates of exposure and onset of illness. Autism, however, usually does not have a well-demarcated date of onset. Other studies have tried to address the possible relationship to MMR vaccination by examining the temporal relationship between vaccination and onset of initial parental concern, date of first diagnosis of autism, or onset of regression (if present). We had incomplete information on these events, so we compared the distribution of ages at first MMR vaccination between case and control children. The assumption implicit in this exposure comparison is that if the MMR vaccine increases the risk of autism, which usually develops before 24 months of age, then children who are vaccinated at younger ages would have a higher risk of developing autism. The age at exposure was examined in a number of ways. First, we compared the overall distributions of age at vaccination. We then analyzed associations using 3 specific age cutoffs: 1) <18 months of age, as an indicator of "on-time" vaccination according to the recommended vaccination schedule for MMR vaccine28; 2) <24 months of age, the age by which atypical development has become apparent in most children with autism23,29–32; and 3) <36 months of age, the age by which autistic characteristics must have developed to meet DSM-IV criteria for autism.1
We used the 
2 statistic for categorical comparisons of the characteristics of case and control children. We compared the overall distributions of ages at first MMR vaccination using a likelihood ratio test in a conditional logistic regression model stratified by matched sets in which age at vaccination was included as a categorical variable with 5 age categories. We also used conditional logistic regression models to estimate the odds ratios (ORs) for the association between autism and age at MMR vaccination dichotomized according to the 3 prespecified age cutoffs (18 months, 24 months, and 36 months).
In the subgroup of children that we matched to birth certificate files, we were able to adjust for additional factors. Potential confounding variables were evaluated individually for their association with autism case status. Those with a P < .20 were included as covariables in conditional logistic regression models to estimate adjusted ORs.33 In analyses stratified by birth or maternal characteristics, we were not able to maintain the matched sets in the analysis. However, we did include the matching factors (age, gender, and school) as covariables in the regression models.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
|
Demographic Characteristics of Case Children and Matched Control Children
In the total sample, case and control children were matched appropriately on age and gender, with a preponderance of boys in both groups (Table 2). The racial distributions were also fairly similar, although a larger proportion of control (10%) than case (6%) children were classified as "other" race and both groups had an appreciable number for which race information was missing.
|
Comparisons of Ages at MMR Vaccination
The overall distributions of ages at first MMR vaccination were similar (P = .22) for case and control children (Fig 1). Most case (70.5%) and control (67.5%) children were vaccinated between 12 and 17 months of age.
|
|
Results for Subgroups of Case Children
When we performed analyses within the nonmutually exclusive clinical subgroups of case children, we found no associations with vaccination before 18 months or 24 months of age among cases without preexisting conditions before 1 year of age, case children with regression or plateau, and case children with and without MR (Table 4). Using a 36-month age cutoff, the ORs in all subgroups of case children were above 1.0, but only the OR among case children without MR had a CI that excluded 1.0. None of the adjusted results using the birth certificate sample was statistically significant. In the birth certificate sample, however, only 3 case children without mental retardation were vaccinated after 36 months of age, resulting in a highly unstable OR estimate for this subgroup.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
We compared the distribution of ages at vaccination between case and control children because we lacked an unvaccinated comparison group and we had incomplete information for determining date of onset of autism. Determining onset of autism, however, is difficult even under the best of circumstances. In most instances, brain abnormalities associated with autism probably occur prenatally,34–36 but parents might not become aware of their children’s problems until later in life, when communication delays and characteristic behaviors become apparent. Analyses of videotapes made of children before ASD diagnosis indicate that identifying onset of developmental problems is very difficult, especially retrospectively,37,38 and that children who were reported as losing skills often had abnormal behaviors before the time when the loss was first noted.32,39
The US vaccination schedule recommends that the first dose of MMR vaccine be administered between 12 and 15 months of age.28 Thus, our results for vaccination before 18 months of age evaluated possible increased risks of autism associated with vaccination by or shortly after the recommended age. Parental concerns about development or the first indications of atypical development usually occur before 24 months of age in children with autism,23,30–32,40 and developmental regression, if it occurs, usually is noted between 12 and 24 months of age.23,41–43 In Wakefield’s case series, 10 of the 12 children had ASD and 9 (90%) of the ASD cases had atypical behaviors noted by 21 months of age.15 Thus, we would expect that exposures that could be causally associated with autism would most likely occur before 24 months of age.
To meet DSM-IV criteria for autism, some manifestation of atypical development must be apparent before 36 months of age.1 Of the 41 case children who were vaccinated after 36 months of age in our study, 32 (78%) had documented delays in development before 36 months of age. Rather than representing causal relationships, associations with the 36-month cutoff would be more likely than associations with earlier age cutoffs to have been influenced by factors related to the evaluation, management, and treatment of the child. For example, case children might have been more likely than control children to have been vaccinated as a requirement for enrollment in early intervention or preschool special education programs. This possibility is supported by the finding that the difference between case and control children in the proportion vaccinated before 36 months of age was strongest in the 3- to 5-year-old age group. In 1991, the Individuals with Disabilities Education Act11 mandated the provision of special education programs for children with autism beginning at approximately 36 months of age. Thus, the case children who were 3 to 5 years of age in 1996 would have been most affected by the Individuals with Disabilities Education Act special education requirement and 98% of these children had been enrolled in preschool special education programs.
In addition to being a large, population-based study, our study had a number of other strengths. We included a detailed review of case records by a panel of autism experts to confirm the case definition for autism according to DSM IV criteria. We were able to obtain additional clinical information that allowed us to evaluate associations within subgroups of case children according to developmental course (eg, regression) or presence of other coexisting conditions (eg, MR). We ascertained vaccination histories from standard immunization forms, eliminating possible recall bias. Information bias was further reduced by the fact that the clinical and behavioral data and the vaccination data came from independent record sources and the information on both exposure and outcome was recorded before the publicity about a possible association between MMR and autism. Furthermore, by linking with birth records, we were able to evaluate and control for potential confounding by demographic and birth characteristics.
Although the original group of 987 autism case children identified by MADDSP in 1996 probably was a fairly complete enumeration of cases in metropolitan Atlanta, we were able to locate vaccination records for only approximately two thirds of these children during 1999 through 2001. This is primarily because when a child moved or changed schools, the permanent school record was transferred to the child’s new school and we did not have access to records for children who were no longer attending a school in a metropolitan Atlanta public school district that participates in MADDSP. Thus, factors related to moving or changing schools might have influenced our results. However, we did not find any significant differences in demographic characteristics or cognitive level between case children who were included and those who were excluded from the study.
Among case and control children whose records we were able to match with Georgia birth certificate files, we performed a subanalysis to evaluate possible confounding by differences in birth and maternal characteristics. For the most part, the results were not greatly different from those in the total sample. The differences that were noted were predominantly of lower ORs in the birth certificate sample. These differences seemed to be primarily a result of restricting the analysis to children who were born in Georgia and could be matched to a state birth certificate and not to confounding by maternal or birth characteristics. Thus, the differences between the 2 samples could represent random fluctuation or a possible bias related to being born outside Georgia.
A number of other epidemiologic studies have failed to find an association between MMR vaccination and autism.19–24 A recent retrospective cohort study from Denmark is particularly persuasive.24 The study contained data on more than half a million Danish children, including nearly 100 000 who had not been vaccinated with MMR. Through linkages of various national registries and medical databases, the study found that the relative risk associated with MMR was 0.92 (0.68–1.24) for autistic disorder and 0.83 (0.65–1.07) for other ASDs. An Immunization Safety Review Committee of the IOM25 reviewed the epidemiologic and other evidence on MMR vaccine and risk for ASDs and concluded that the evidence favors rejection of a causal relationship at the population level. Other review panels have reached similar conclusions.44,45 The IOM committee, however, did recommend additional studies to evaluate potential high-risk subgroups of children.
The caveat by IOM relates primarily to "autistic enterocolitis," which has been proposed by Wakefield and colleagues16,18 to be a new clinical syndrome that is associated with MMR vaccination as supported by laboratory evidence of persistent measles virus infection in the intestines of affected children. The syndrome is characterized by developmental regression along with gastrointestinal disturbances. We were not able to evaluate the syndrome because we lacked information on gastrointestinal symptoms, but we did not find an association between age at vaccination, most notably by 18 months or 24 months, and autistic regression. We found a lower proportion of cases with regression, however, than has been reported in other studies.40,46 Our number of regression cases is likely to be an underestimation because we relied on abstracted information rather than interview with the parent and we may not have captured all of the behavioral information needed to determine whether the child had regression. Analyses by other investigators have found no support for a new variant of autism40,46 or for the association of the MMR vaccination with regressive autism23 or gastrointestinal disorders.47
In addition to regression, we evaluated other clinical subtypes of ASDs, including case children with and without MR, and case children who did not have congenital malformations or early evidence of developmental problems (and thus were at risk for onset of developmental disabilities at the recommended age for MMR vaccination). We generally did not find increased risks for any of these case subtypes associated with MMR vaccination at any age. The only exception was that case children without MR were more likely to have been vaccinated before 36 months of age than their matched control children.
Other concerns have been raised about vaccinations and autism, especially about thimerosal, the mercury-containing preservative that until recently had been included in multidose preparations of certain vaccines.25 In the present study, we were not able to evaluate the potential association between thimerosal exposure and autism. The routinely recommended infant vaccines that used to contain thimerosal were diphtheria-tetanus-pertussis, hepatitis B, and Haemophilus influenzae type b. Hepatitis B and Haemophilus influenzae type b vaccines, however, were not required for school attendance during the period of our study, and they were incompletely recorded in the school records. MMR vaccine has never contained thimerosal. Single-antigen measles vaccine has been hypothesized to be safer than MMR,48 but we had too few children who received measles vaccine alone to be able to evaluate this possibility. We performed an analysis in which we evaluated associations with any measles-containing vaccine; the results were similar to those for MMR vaccine (data not shown). We did not evaluate associations with the second dose of MMR vaccine because it is usually administered between 4 and 6 years of age, which is after the 36-month age limit for autism onset as defined by the DSM-IV.1
|
CONCLUSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
|
DISAPPOINTING RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
Simons SHP et al. Routine morphine infusion in preterm newborns who received ventilatory support. J Am Med Assoc. November 12, 2003
Submitted by Student
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Reprint requests to (F.D.) Centers for Disease Control and Prevention (E-61), 1600 Clifton Rd, NE, Atlanta, GA 30333. E-mail: fxd1@cdc.gov
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONDISAPPOINTING RESULTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C. P. Johnson, S. M. Myers, and and the Council on Children With Disabilities Identification and Evaluation of Children With Autism Spectrum Disorders Pediatrics, November 1, 2007; 120(5): 1183 - 1215. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. D'Souza, E. Fombonne, and B. J. Ward No Evidence of Persisting Measles Virus in Peripheral Blood Mononuclear Cells From Children With Autism Spectrum Disorder Pediatrics, October 1, 2006; 118(4): 1664 - 1675. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Castillejo, M. Bullo, A. Anguera, J. Escribano, and J. Salas-Salvado A Controlled, Randomized, Double-Blind Trial to Evaluate the Effect of a Supplement of Cocoa Husk That Is Rich in Dietary Fiber on Colonic Transit in Constipated Pediatric Patients Pediatrics, September 1, 2006; 118(3): e641 - e648. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Fombonne, R. Zakarian, A. Bennett, L. Meng, and D. McLean-Heywood Pervasive Developmental Disorders in Montreal, Quebec, Canada: Prevalence and Links With Immunizations Pediatrics, July 1, 2006; 118(1): e139 - e150. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S J. James, P. Cutler, S. Melnyk, S. Jernigan, L. Janak, D. W Gaylor, and J. A Neubrander Metabolic biomarkers of increased oxidative stress and impaired methylation capacity in children with autism Am. J. Clinical Nutrition, December 1, 2004; 80(6): 1611 - 1617. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. C. Meissner, P. M. Strebel, and W. A. Orenstein Measles Vaccines and the Potential for Worldwide Eradication of Measles Pediatrics, October 1, 2004; 114(4): 1065 - 1069. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. K. Parker, B. Schwartz, J. Todd, and L. K. Pickering Thimerosal-Containing Vaccines and Autistic Spectrum Disorder: A Critical Review of Published Original Data Pediatrics, September 1, 2004; 114(3): 793 - 804. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Trevathan Seizures and Epilepsy Among Children With Language Regression and Autistic Spectrum Disorders J Child Neurol, August 1, 2004; 19(1_suppl): S49 - S57. [Abstract] [PDF]
|
|
|
|
 |
|
 B. Bardenheier, H. Yusuf, B. Schwartz, D. Gust, L. Barker, and L. Rodewald Are Parental Vaccine Safety Concerns Associated With Receipt of Measles-Mumps-Rubella, Diphtheria and Tetanus Toxoids With Acellular Pertussis, or Hepatitis B Vaccines by Children? Arch Pediatr Adolesc Med, June 1, 2004; 158(6): 569 - 575. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Gottlieb Study finds no connection between MMR vaccine and autism BMJ, February 21, 2004; 328(7437): 421 - 421. [Full Text] [PDF]
|
|
|
|
|
|
E. Trevathan Seizures and Epilepsy Among Children With Language Regression and Autistic Spectrum Disorders J Child Neurol, January 1, 2004; 19(1_suppl): S49 - S57. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
