Dales L, Hammer SJ, Smith NJ. Time Trends in Autism and in MMR Immunization Coverage in California. JAMA. 2001;285(9):1183–1185. doi:10.1001/jama.285.9.1183
Author Affiliations: Immunization Branch, California Department of Health Services, Berkeley.
Context Considerable concern has been generated in the lay and medical communities
by a theory that increased measles-mumps-rubella (MMR) immunization among
young children may be the cause of an apparent marked increase in autism occurrence.
Objective To determine if a correlation exists in secular trends of MMR immunization
coverage among young children and autism occurrence.
Design, Setting, and Participants Retrospective analyses of MMR immunization coverage rates among children
born in 1980-1994 who were enrolled in California kindergartens (survey samples
of 600-1900 children each year) and whose school immunization records were
reviewed to retrospectively determine the age at which they first received
MMR immunization; and of autism caseloads among children born in these years
who were diagnosed with autism and were enrolled in the California Department
of Developmental Services regional service center system.
Main Outcome Measures Measles-mumps-rubella immunization coverage rates as of ages 17 months
and 24 months and numbers of Department of Developmental Services system enrollees
diagnosed with autism, grouped by year of birth.
Results Essentially no correlation was observed between the secular trend of
early childhood MMR immunization rates in California and the secular trend
in numbers of children with autism enrolled in California's regional service
center system. For the 1980-1994 birth cohorts, a marked, sustained increase
in autism case numbers was noted, from 44 cases per 100 000 live births
in the 1980 cohort to 208 cases per 100 000 live births in the 1994 cohort
(a 373% relative increase), but changes in early childhood MMR immunization
coverage over the same time period were much smaller and of shorter duration.
Immunization coverage by the age of 24 months increased from 72% to 82%, a
relative increase of only 14%, over the same time period.
Conclusions These data do not suggest an association between MMR immunization among
young children and an increase in autism occurrence.
In the mid to late 1990s, a medical investigative group in Great Britain
postulated that measles immunization and/or immunization with measles, mumps,
and rubella vaccines given combined at the same time or in succession over
a relatively short time period might be a cause of the developmental disorder
autism.1 Strong interest in and debate over
this hypothesis has occurred. In this connection, some cite a 1999 report
from the California Department of Developmental Services2
showing an increase within the past 2 decades in its regional system caseload
of children with autism, wondering if increasingly widespread combined measles-mumps-rubella
(MMR) immunization of young children might have been responsible for this
increase.3 Also, multiple news media stories
and Internet Web site presentations on MMR vaccine and autism have cited these
California autism data.4,5
Autism, an incompletely understood developmental disorder or group of
related disorders, varies in clinical presentation, so diagnosis is not always
straightforward. Further, temporal associations between receipt of MMR vaccine
and autism onset can occur by chance alone, since MMR vaccine is typically
given at 12 to 15 months of age, and the median age at which parents of children
with autism first notice troubling signs in their children's development is
approximately 18 to 19 months of age.6
The aforementioned California Department of Developmental Services report
stresses that its patient caseload data cannot be used as a true measure of
changes over time in autism incidence because other factors can affect trends
in system case numbers.2 Further, "ecological
correlations," that is, observations of parallel trends over time or across
geographic locations for 2 events, generally do not constitute strong evidence
for a causal association between the 2 events.7
Other events besides a possible increase in early childhood immunization with
MMR vaccine have occurred over the time period of the increase in autism cases.
Nonetheless, given concerns expressed about this matter, we have examined
California data on early childhood MMR immunization levels over time along
with the data on autism caseloads during the same period. The combined MMR
vaccine was licensed for use in the United States in 1971, and over the remainder
of the 1970s, almost completely replaced use of the individual measles, mumps,
and rubella vaccines for routine immunization of young children. Since 1979,
the MMR preparation used in the United States has not changed in composition.
Thus, for children born in California from 1980 onward, the secular trend
in the proportion of children who received MMR immunization at an early age
is the logical parameter to compare with the apparent trend of increasing
Since the early 1970s, the California Department of Developmental Services
has funded a system of regional centers for persons with developmental disabilities,
including autism. These regional centers, now 21 in number and covering all
of the state's 58 counties, function as service hubs, coordinating, linking,
and funding services for eligible persons. The 1999 report on trends in the
system's autism caseload compiled demographic and patient developmental evaluation
record data electronically submitted to the department by the 21 regional
centers.2 Because the increase in this system's
autism caseload occurred within the past 2 decades, we examined the data for
children born between 1980 and 1994, the latter being the latest patient birth
year for which a relatively complete system autism caseload count is currently
available. The California Department of Developmental Services supplied caseload
numbers for the diagnostic category "autistic disorder" (International Classification of Diseases, 9th Edition, code 299.00),
which excludes other pervasive developmental disorders.
The early childhood MMR immunization level estimates for California
children by year of birth come from the California Department of Health Services
annual surveys of statewide random samples of public and private school kindergarten
pupils' immunization records. The number of children included in the surveys'
random samples selected each year increased from approximately 600 to 800
children born each year in 1980 through 1984 to between 1000 and 1900 annually
for children born in 1985 through 1994, but otherwise the sampling method
remained virtually the same throughout the whole time period. School immunization
records of the kindergarten children sampled, who were 5 to 6 years of age
at the time of the surveys, were assessed retrospectively to determine at
what earlier ages they received their immunizations. Per standard national
recommendations extant over this time period, children should have received
their initial MMR immunization at 12 to 15 months of age. Two immunization
coverage measures are presented here: the proportions of children in each
year's survey sample who had received MMR immunization by the age of 17 months,
and the proportions that had received such immunization by the age of 24 months.
Immunization coverage by the age of 17 months is presented because this age
just precedes the median age when parents of children with autism first note
signs of developmental problems. Immunization coverage through the age of
24 months is presented to detect a potential secular trend on increasing vaccine
coverage that might have been missed when the 17-month age checkpoint was
The upper 2 curves in Figure 1
show the percentages of children who received MMR immunization by the age
of 17 months and by the age of 24 months (including children immunized by
17 months of age), by birth year, for each annual California kindergarten
retrospective immunization survey sample cohort, for birth years 1980 through
1994. As can be seen from the 2 upper curves, for the 1980 through 1987 birth
year cohorts, there was almost no change in MMR coverage. A modest increase
occurred with the 1988 birth cohort, followed again by a near plateau with
little change in coverage over the next 6 years, through the 1994 birth cohort.
The relative increase in MMR coverage by age 24 months from birth year 1980
to 1994 was 14%, from 72% to 82%.
The lower curve in Figure 1
shows a major trend of increase in autistic disorder cases beginning with
the 1985 birth year cohort (and possibly even earlier), well before the MMR
coverage increase that occurred with the 1988 birth cohort. Further, this
trend of autistic disorder caseload increase continues at least through the
1994 birth cohort. Thus, the trend of increasing autism caseload numbers persisted
long after the MMR immunization coverage curve leveled off again with the
1988 birth cohort.
The trend of increasing numbers of autistic disorder cases, from 176
cases for birth year 1980 to 1182 cases for birth year 1994 (a relative increase
of 572%), is many times greater in relative magnitude than the modest increase
observed for MMR coverage. If the autistic disorder case data are computed
as rates, using annual numbers of births in California as denominators, the
increase noted for the lower curve in Figure
1 is only slightly less pronounced, increasing from 44 per 100 000
births for the 1980 birth cohort to 208 per 100 000 births for the 1994
birth cohort, representing a 373% relative increase.
The lack of correspondence between the trends in MMR coverage and numbers
of autistic disorder cases does not support the hypothesis that increasingly
widespread MMR immunization of young children is associated with the marked
secular trend of increasing numbers of autism cases enrolled in the California
Department of Developmental Services regional service center system. Difficulties
in interpreting the increasing numbers of autism cases are highlighted in
the California Department of Developmental Services 1999 report, which repeatedly
stressed that its data cannot be interpreted as measuring trends in the actual
incidence of autism.2 It is unknown how many
children with autism have not enrolled in the system and, as the system expanded
and matured over time, the proportions of California children enrolling and
the distribution of ages at enrollment likely changed over time as a result.
Also, the proportions of children enrolling in the system who were born outside
California may have changed over this time period. In addition, diagnostic
methods and categorizations of persons with autism have changed over time.
Recommendations have been made for a detailed investigation into the apparent
secular increase in autism cases, carefully evaluating the extent to which
apparent increases in autism are real as opposed to artifact, as well as evaluating
factors that may have contributed to any true increases including immunizations.2,8 Such investigations are in progress
(Robert Byrd, MD, University of California Davis Medical Center, Department
of Pediatrics, written communication, November 1, 2000).
However, the issue here is, assuming that the secular trend noted in
the report represents a true increase in autistic disorder incidence, what
is the evidence that MMR immunization may be associated with this trend? That
is, does an ecological correlation exist between MMR immunization coverage
rates of young children and the system's autism caseload? If MMR immunization
of young children was associated with a secular trend in autism incidence,
one would expect a curve for the numbers of autism cases among the 1980-1994
birth year cohorts that is very similar to the shapes of the curves for temporal
trends in MMR coverage. That is, the numbers of cases should be rather stable
for the 1980-1987 cohorts, followed by a small increase and then more or less
a plateau at a new, modestly higher level for the 1988-1994 birth cohorts.
As can be seen from the lower curve in Figure
1, however, this is not the case.
The data presented herein have some limitations. It would have been
useful to examine individual immunization and autism records on the same children;
however, these could not be linked. Further, the childhood immunization coverage
data used in this study do not provide precise quantification of the percentages
of children who received the combined MMR vaccine product vs separate injections
of the measles, mumps, and/or rubella components. However, the historical
information on MMR vaccine usage in the United States, as well as information
in the immunization coverage data set itself, strongly suggest that separate
injections of measles, mumps and/or rubella vaccines to young children were
rare during the time period under consideration, particularly for the 1984-1994
birth cohorts, in which the trend of increase in the system's autistic disorder
caseload is most pronounced.
Thus, examination of available data from California has been unable
to demonstrate a correlation between secular trends in early childhood MMR
immunization coverage and autism caseload levels. A similar lack of correlation
between the trends in early childhood MMR immunization rates and autism occurrence
has been noted in Great Britain6 and Sweden.9 In conclusion, to date, published observations based
on empirical evidence do not suggest that increased MMR immunization among
young children is associated with secular increases in cases of autism.