This article was downloaded by: [Stanford University Libraries]On: 18 October 2012, At: 11:39Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
Journal of Toxicology and Environmental Health, PartA: Current IssuesPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/uteh20
A Positive Association found between AutismPrevalence and Childhood Vaccination uptake acrossthe U.S. PopulationGayle DeLong aa Department of Economics and Finance, Baruch College/City University of New York, NewYork, New York, USA
Version of record first published: 26 May 2011.
To cite this article: Gayle DeLong (2011): A Positive Association found between Autism Prevalence and Childhood Vaccinationuptake across the U.S. Population, Journal of Toxicology and Environmental Health, Part A: Current Issues, 74:14, 903-916
To link to this article: http://dx.doi.org/10.1080/15287394.2011.573736
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form toanyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses shouldbe independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly inconnection with or arising out of the use of this material.
Journal of Toxicology and Environmental Health, Part A, 74:903–916, 2011Copyright © Taylor & Francis Group, LLCISSN: 1528-7394 print / 1087-2620 onlineDOI: 10.1080/15287394.2011.573736
A POSITIVE ASSOCIATION FOUND BETWEEN AUTISM PREVALENCEAND CHILDHOOD VACCINATION UPTAKE ACROSS THE U.S. POPULATION
Gayle DeLong
Department of Economics and Finance, Baruch College/City University of New York, New York,New York, USA
The reason for the rapid rise of autism in the United States that began in the 1990s isa mystery. Although individuals probably have a genetic predisposition to develop autism,researchers suspect that one or more environmental triggers are also needed. One of thosetriggers might be the battery of vaccinations that young children receive. Using regressionanalysis and controlling for family income and ethnicity, the relationship between the propor-tion of children who received the recommended vaccines by age 2 years and the prevalence ofautism (AUT) or speech or language impairment (SLI) in each U.S. state from 2001 and 2007was determined. A positive and statistically significant relationship was found: The higher theproportion of children receiving recommended vaccinations, the higher was the prevalenceof AUT or SLI. A 1% increase in vaccination was associated with an additional 680 childrenhaving AUT or SLI. Neither parental behavior nor access to care affected the results, sincevaccination proportions were not significantly related (statistically) to any other disability orto the number of pediatricians in a U.S. state. The results suggest that although mercury hasbeen removed from many vaccines, other culprits may link vaccines to autism. Further studyinto the relationship between vaccines and autism is warranted.
Autism is an urgent and growing publichealth problem in the United States. The ill-ness impairs speech, language, social abilities,and behavior. In 1990, autism was considereda rare disease (Tebben 1990), but less than 2decades later autism affected an estimated 1 in91 U.S. children (Kogan et al. 2009). Althoughscientists generally agree that a genetic predis-position for autism exists (Rutter 2000), genesalone do not change quickly enough to createthe current epidemic. The recent explosion inthe prevalence of autism suggests the existenceof one or more environmental triggers (Blaxill
Received 29 October 2010; accepted 1 March 2011.This study includes data from the U.S. National Centers for Health Statistics (NCHS). Any analyses, interpretations, or conclusions
reached are the author’s own and not those of the NCHS, which is responsible only for the initial data. The author is grateful to JoséGarrofe Dorea, Anthony Mawson, Jonathan Rose, Paul Turner, and David Yermack as well as seminar participants at Baruch Collegeand two anonymous reviewers for thoughtful comments. Francis Donnelly provided invaluable assistance in creating the geographicinformation system graphic. The author has two children with pervasive development disorder, not otherwise specified. She has filed apetition in the U.S. Court of Federal Claims under the National Vaccine Injury Compensation Program for one of her children. The authoris on the board of directors and research committee of Sensible Action for Ending Mercury-Induced Neurological Disorders (SafeMinds).
Address correspondence to Gayle DeLong, PhD, Associate Professor, Department of Economics and Finance, Baruch College/CityUniversity of New York, One Bernard Baruch Way, Box B10-225, New York, NY 10010, USA. E-mail: [email protected]
2004). Could one of those triggers be thebattery of vaccinations given to young children?
Chronic, negative reactions to vaccinationshave been recognized in both humans andanimals. In the late 19th Century, Burnett(1884/1960) described long-term negativeeffects such as eczema, diarrhea, and fatiguein some individuals who received a series ofsmallpox vaccinations. By the 1990s, veterinar-ians began to notice that some animals devel-oped chronic ailments such as autoimmunedisorders and seizures after being vaccinated(Smith 1995; Dodds 2001). In the early
903
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
904 G. DELONG
2000s, vaccines were shown to be linked toautoimmune disorders and, possibly, autism inhumans (Shoenfeld and Aron-Maor 2000).
There are several reasons why vaccinesmay trigger autism. Certain vaccines containthimerosal, a preservative that is almost halfmercury (Hg) by weight, which was shown tobe associated with adverse effects includingautism. Nataf et al. (2006) found that childrenwith autism have higher levels of precopro-porphyrin, a biomarker for Hg toxicity, thanneurotypical children. This finding was con-firmed by Geier and Geier (2007) and Geieret al. (2009) in the United States, Austin andShandley (2008) in Australia, and Youn et al.(2010) in Korea. Thimerosal-containing hep-atitis B shots were associated with delayedacquisitions of vital reflexes in baby macaques(Hewitson et al. 2010). Although thimerosalwas removed from many vaccines from 2000,it is still present in almost all influenza shotsas well as eight other U.S. vaccines givento children (Centers for Disease Control andPrevention 2010). In addition, the CDC beganin the early 2000s to encourage the inoculationof pregnant women and children aged 6 to 23months against influenza (Centers for DiseaseControl and Prevention 2001; 2002). Giventhe increased use of influenza shots contain-ing thimerosal, children’s exposure to Hg viavaccines was likely increased in utero but notdecreased after fetuses were born, even thoughthimerosal was removed from other vaccines.
There are other possible links between vac-cines and autism besides Hg. Vaccines alsocontain the neurotoxin aluminum (Al) as wellas live viruses. The Al in vaccines has beenassociated with disorders in the central ner-vous system (Authier et al. 2001) as well aswith autism (Blaylock 2008). Combining Hgand Al magnifies the toxicity of each (Haley2005). Both metals also are known to suppressthe immune system (Havarinasab et al. 2005);thus, a susceptible person may not be able tomount an effective immunological response tothe live viruses found in certain vaccines suchas the measles–mumps–rubella shot. Measles-containing vaccines stimulate the production ofcytokines that inflame and damage the brain,
possibly contributing to autism (Ashwood et al.2004; Vargas et al. 2005; Singh 2009).
Children with autism appear to have vul-nerabilities that their neurotypical peers do notpossess. Autistic children tend to exhibit higherlevels of oxidative stress and poorer methyla-tion, the process by which the body detoxifiesitself (James et al. 2004). This difficulty in detox-ifying could be associated with metals fromvaccines being sequestered in the brain andcausing neurological damage (Kern et al. 2007).Vaccines may also increase the oxidative stressof children with preexisting mitochondrial dys-functions to such an extent that the childrendevelop autism (Poling et al. 2006). In gen-eral, susceptibility to developing a neurologicaldisability after exposure to an environmentalinsult such as a vaccine depends on factors suchas a child’s age at time of exposure, amountof exposure, genetic predisposition, and stress(Kern and Jones 2006).
Compounding these biological issues is thefact that the number of vaccinations recom-mended for U.S. children by age 2 years hasmore than tripled, from 8 vaccinations in 1983to 27 in 2010 (Centers for Disease Control andPrevention 1983; 2010). Although individualvaccines are tested for safety and efficacy, nostudy has ever examined the safety of the entirevaccination schedule recommended for U.S.children by the CDC. Neither the short-termnor chronic interactions among all the vaccinesin a child’s recommended schedule have everbeen tested.
Examining the relationship between theproportion of children who receive vaccina-tions and the prevalence of autism may provideinsights into whether autism is an adverse reac-tion to vaccinations. If an association betweenreceiving vaccinations and developing autism isfound to exist across geography and throughtime, further investigation into the hypothesisis warranted.
METHODS
In this study, the relationship betweenthe proportion of U.S. children who receiveda series of vaccinations recommended by
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
ASSOCIATION BETWEEN AUTISM AND VACCINATION 905
the U.S. Centers for Disease Control andPrevention (CDC) by age 2 years and the preva-lence of autism in each U.S. state over time wasexamined.
MeasuresPrevalence of autism To determine autism
prevalence by U.S. state, the number of 8-year-old students classified with either (1) autismor (2) speech or language impairments (speechdisorders) was divided by the total numberof 8-year-olds in the state. The number ofchildren with disabilities came from the U.S.Department of Education, Office of SpecialEducation Programs (2007) and the total num-ber of students came from the U.S. Departmentof Education, National Center for EducationStatistics. Although the diagnosis of autism isusually made when a child is 3 or 4 yearsold, some children are not diagnosed until theyare older. Children who receive a diagnosisof autism usually do so by the time they are8 years old. The category of speech or lan-guage impairments was included with autism,because these impairments are closely linkedto autism (Conti-Ramsden et al. 2006; De Fosseet al. 2004; Herbert et al. 2007).
Exposure to a recommended vaccinationseries Since 1994, the CDC has commis-sioned an annual survey to estimate vaccina-tion coverage in the United States for preschoolchildren. Surveyors at the National OpinionResearch Center (NORC) at the University ofChicago randomly call homes to find house-holds with children aged 19 to 35 months.When such a household is found, the inter-viewer asks which vaccinations the child hasreceived. If the parent or guardian agrees,NORC follows up the telephone interviewwith a written survey to the vaccinationprovider. The survey reaches approximately30,000 households with children of the appro-priate age. The proportion of children in eachstate that receives the various vaccines recom-mended by the CDC is reported. Starting in1995, the CDC reports the percentage of youngchildren who have received the 4:3:1:3:3series of shots, which consists of at least four
doses of the diphtheria, tetanus, and pertussis(or diphtheria, tetanus, and acellular pertussis)vaccine, three doses of poliovirus vaccine, onedose of any measles-containing vaccine, threedoses of the Hib vaccine, and three doses ofhepatitis B vaccine. The results of this surveyas well as the follow-up verification from thevaccination provider are available to the pub-lic (Centers for Disease Control and Prevention,National Center for Health Statistics 2007).Since the possible magnification effect of thetoxins in vaccines is of interest as well asthe possible interactions between toxins andlive viruses, the proportion of children whoreceived the entire 4:3:1:3:3 series of vaccina-tions by the time they were 19 to 35 monthsold was examined in this study.
Children who are vaccinated at age 2 yearsmay not develop autism until they are older.To determine the prevalence of autism for aspecific cohort of children, the vaccinationdata from when the children were 2 years oldis compared with autism prevalence when theyare 8 years old. The relevant vaccination datafor children who were 8 years old in 2001 arethose from 1995, when the children were 2years old. For children who turned 8 years oldin 2002, the relevant vaccination data are from1996, and so on. The earliest available data—vaccination data from 1995—were matchedwith autism prevalence up to 2007. Table 1shows the vaccination and autism/speechdisorder data by state for the various years inthis study.
Figure 1 shows the relationship betweenthe prevalence of 8-year-old children withautism or speech disorders by state in 2005and vaccination proportions 6 years earlier. Thedarker the shading of the state, the higher isthe proportion of children who received the4:3:1:3:3 series of vaccinations by the timethey were 2 years old; the larger the cir-cle, the greater is the prevalence of autismor speech disorders. Groupings of data weredetermined using the Natural Breaks (Jenks)method in the Arc geographic information sys-tem (ArcGIS) software package (EnvironmentalSystem Research Institute Inc. 2009). The mappresents an ambiguous picture: Some states,
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
TABL
E1.
Vacc
inat
ion
Rate
(199
5–20
01)a
ndAU
T/SL
IPre
vale
nce
(200
1–20
07)b
yU
.S.S
tate
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
Vax
rate
AUT/
SLI
U.S
.sta
te19
9520
0119
9620
0219
9720
0319
9820
0419
9920
0520
0020
0620
0120
07
Ala
bam
a45
.84.
665
.24.
476
.54.
474
.24.
574
.14.
776
.14.
679
.14.
5A
lask
a54
.35.
867
.25.
168
.85.
574
.16.
074
.54.
870
.65.
071
.24.
8A
rizon
a51
.23.
861
.84.
162
.04.
169
.04.
867
.35.
267
.25.
468
.16.
2A
rkan
sas
53.6
4.8
61.4
4.8
74.8
5.0
65.9
5.3
70.4
5.8
67.1
6.3
69.1
5.9
Cal
iforn
ia57
.74.
565
.84.
667
.14.
669
.84.
670
.54.
672
.34.
772
.64.
6C
olor
ado
51.4
3.1
66.3
3.5
64.4
3.6
67.9
3.6
69.6
3.9
71.6
4.0
71.5
4.1
Con
nect
icut
63.9
3.6
80.6
3.9
76.0
3.7
81.5
3.9
82.3
3.9
81.6
4.2
78.4
4.1
Del
awar
e54
.73.
373
.53.
268
.33.
368
.03.
669
.03.
166
.22.
168
.9N
/A
Dist
ricto
fCou
mbi
a49
.83.
266
.82.
362
.42.
763
.52.
670
.92.
770
.03.
774
.93.
3Fl
orid
a53
.46.
072
.16.
067
.35.
675
.45.
677
.95.
771
.76.
473
.06.
4G
eorg
ia61
.75.
276
.95.
572
.95.
775
.75.
977
.95.
977
.75.
978
.55.
7H
awai
i66
.22.
074
.41.
873
.21.
773
.01.
479
.21.
372
.81.
470
.81.
0Id
aho
40.7
4.0
51.9
3.9
63.9
3.8
66.2
4.2
65.0
4.2
70.7
4.6
70.2
4.4
Illin
ois
57.2
5.6
64.9
5.6
67.4
5.6
73.7
5.5
72.0
5.5
71.2
5.7
72.7
5.6
Indi
ana
41.8
8.8
56.7
8.9
62.7
9.1
68.8
9.1
65.3
9.2
72.0
9.8
71.1
10.0
Iow
a47
.71.
670
.32.
571
.02.
178
.12.
278
.92.
282
.52.
178
.62.
0Ka
nsas
35.7
5.1
58.2
5.1
72.1
5.2
72.1
5.0
70.7
5.1
71.3
4.9
72.8
4.9
Kent
ucky
59.6
6.0
70.4
6.0
70.2
5.8
75.5
6.3
84.4
6.6
77.0
7.1
75.9
7.1
Loui
siana
61.8
5.1
73.0
5.1
70.6
5.2
72.4
5.4
72.3
5.1
71.8
5.4
64.1
5.6
Mai
ne46
.66.
668
.46.
978
.46.
878
.37.
276
.87.
776
.07.
575
.17.
2M
aryl
and
59.2
4.5
64.9
4.7
73.7
4.8
72.3
4.6
72.7
4.5
75.4
4.7
73.4
4.6
Mas
sach
uset
ts70
.72.
279
.92.
681
.13.
079
.63.
381
.43.
681
.43.
876
.64.
1M
ichi
gan
46.7
5.1
66.2
5.5
69.3
5.5
73.9
5.7
70.9
5.9
73.7
6.2
70.0
6.2
Min
neso
ta41
.24.
063
.24.
363
.74.
473
.14.
578
.54.
882
.44.
876
.35.
1M
ississ
ippi
38.1
6.2
68.1
6.4
70.8
6.6
79.5
6.2
79.0
6.2
75.9
6.7
80.2
6.8
Miss
ouri
50.5
6.2
68.3
7.1
67.6
7.5
74.8
8.1
68.9
8.5
76.8
9.0
75.5
8.7
Mon
tana
44.7
5.7
59.2
5.1
64.3
5.7
75.2
6.0
76.4
6.2
71.1
N/A
77.9
5.9
Neb
rask
a49
.27.
365
.37.
066
.47.
271
.17.
779
.87.
775
.57.
678
.97.
9N
evad
e55
.94.
061
.94.
065
.54.
270
.43.
968
.54.
369
.15.
068
.15.
1N
ewH
amps
hire
72.7
3.5
77.3
3.2
76.8
3.3
75.8
3.4
78.4
3.7
78.9
3.6
77.6
3.9
New
Jers
ey60
.57.
168
.57.
370
.57.
176
.77.
275
.37.
271
.27.
673
.17.
7N
ewM
exic
o43
.85.
166
.25.
166
.15.
066
.04.
866
.64.
964
.55.
363
.25.
5
906
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
New
York
62.6
3.6
72.6
3.8
68.1
4.1
79.9
4.2
78.2
4.4
72.3
4.6
77.1
5.0
Nor
thC
arol
ina
62.5
5.1
69.8
5.0
76.5
5.1
76.7
5.5
77.1
5.5
82.8
5.8
80.4
5.9
Nor
thD
akot
a65
.76.
571
.96.
873
.26.
873
.07.
376
.36.
580
.36.
978
.76.
7O
hio
47.7
4.3
69.5
4.4
63.9
4.3
71.0
4.4
73.0
4.5
68.9
4.7
71.2
4.6
Okl
ahom
a42
.35.
659
.06.
165
.05.
970
.35.
970
.46.
568
.37.
070
.06.
7O
rego
n56
.66.
162
.46.
162
.06.
167
.46.
463
.86.
674
.76.
868
.56.
9Pe
nnsy
lvan
ia62
.24.
470
.14.
770
.94.
775
.74.
980
.85.
077
.85.
378
.85.
4Rh
ode
Isla
nd58
.56.
177
.06.
372
.06.
676
.96.
583
.26.
280
.56.
881
.76.
9So
uth
Car
olin
a72
.27.
183
.07.
474
.47.
383
.87.
278
.07.
378
.57.
078
.77.
3So
uth
Dak
ota
27.5
6.5
62.0
7.1
67.0
7.5
66.1
7.7
76.9
7.6
73.6
7.9
76.5
8.7
Tenn
esse
e57
.85.
372
.35.
471
.35.
473
.55.
370
.05.
576
.85.
979
.76.
0Te
xas
52.4
4.0
62.5
4.1
65.3
4.2
63.7
4.2
64.8
4.2
63.5
4.3
69.7
4.3
Uta
h43
.74.
753
.44.
853
.74.
964
.85.
465
.85.
268
.25.
766
.16.
6Ve
rmon
t55
.83.
476
.83.
078
.02.
380
.72.
985
.22.
577
.02.
880
.3N
/A
Virg
inia
52.8
4.2
68.3
4.5
66.6
4.6
75.1
4.6
74.9
4.7
70.7
4.9
74.9
4.7
Was
hing
ton
57.0
4.1
68.5
4.0
66.6
4.0
68.8
4.1
67.1
4.2
72.5
4.3
71.2
4.4
WVi
rgin
ia28
.99.
950
.59.
773
.410
.175
.610
.277
.810
.771
.911
.078
.111
.3W
iscon
sin53
.14.
267
.14.
470
.94.
770
.64.
978
.65.
174
.25.
779
.56.
0W
yom
ing
18.9
6.9
48.0
6.8
69.2
7.6
75.4
8.1
81.5
8.2
78.2
8.7
74.3
10.0
Not
e.Va
xra
te=
prop
ortio
nof
2-ye
ar-o
lds
(%)r
ecei
ving
4:3:
1:3:
3se
ries
ofva
ccin
atio
ns6
year
sea
rlier
.AU
T/SL
I=pr
eval
ence
ofau
tism
orsp
eech
/la
ngua
geim
pairm
ent
(%)i
nye
arin
dica
ted.
N/A
=no
tava
ilabl
e.
907
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
908 G. DELONG
FIGURE 1. Vaccination (1999) and Autism or Speech Disorder (2005) by U.S. State (color figure available online).
such as Texas, have low vaccination rates andlow prevalence of autism, while other states,such as Indiana, have low vaccination ratesand a high prevalence of autism. Conversely,Wyoming has a high vaccination rate and highprevalence of autism, while Vermont has ahigh vaccination rate and low prevalence ofautism. Additionally, Figure 1 merely presentsa snapshot, based on prevalence data in 2005.More rigorous analysis is needed that includesseveral years of data, as well as variables tocontrol for influences other than the vaccineseries.
Controlling for family environment Familyincome and ethnicity may influence whether achild receives a diagnosis of autism. More afflu-ent parents may be more prone to seek a diag-nosis (McAdoo and DeMyer 1977). Ethnicitymay be a factor in terms of the postula-tion that a deficiency in vitamin D is asso-ciated with autism: Dark-skinned people areknown to require more vitamin D and there-fore might be more prone to develop autism(Cannell 2008). For these reasons, variables
that measure household income and ethnicityare included.
To measure income, the median incomefor a four-person family reported by the U.S.Census Bureau (2008) was used. The inflation-adjusted median income (using the year 2000as the base) for the year the child was born wasused. For example, the prevalence of autismor speech disorder for 8-year-olds in a par-ticular U.S. state in 2001 was matched withthe median income in that state from 1993.Ethnicity figures were derived directly from theCDC’s National Immunization Survey. The sur-vey reported whether the child was Hispanic,African American, white non-Hispanic, orother. For each state and year, the percent-age of each ethnic group in the survey wasdetermined.
Statistical AnalysisTo understand whether vaccination might
be linked to autism, data on the prevalenceof autism or speech disorders between 2001
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
ASSOCIATION BETWEEN AUTISM AND VACCINATION 909
and 2007 were matched with vaccinationrates between 1995 and 2001 for each U.S.state. Regression analysis determines how a 1%change in the vaccination rate influences thepercent change in the prevalence of autismor speech disorders (Lewis-Beck 1990). Thestatistical model used took into considerationthe unique characteristics of each state. Forexample, each state had a unique mixture ofpollution, which may have affected the preva-lence of autism (Palmer et al. 2006; 2009), yetsuch an effect was not included in this study.A fixed-effects, within-group panel regression(Hall and Cummins 2005) controlled for theseunique yet undefined characteristics by deriv-ing a different starting point (intercept) for eachU.S. state. The 51 different intercepts—one foreach U.S. state and the District of Columbia—reflected the base level of autism or speechdisorders occurring in that state that werenot explained by the other independent vari-ables (vaccination rates, income, or ethnicity).The model then produced a single relation-ship between the independent variables andthe prevalence of autism or speech disorders.Although each state started from a differentprevalence rate of autism or speech disorder,the relationships between the dependent vari-able and independent variables was consideredthe same across state; a 1% change in vacci-nation rates was associated with the same per-cent change in prevalence of autism or speechdisorder across states. Similarly, the model con-trolled for the year in which an observationtook place. If autism awareness increased in aparticular year, the prevalence might also rise(Liu et al. 2010). To control for reasons thatoccur in a particular year, time dummy vari-ables for the year of observation were included:If an observation of prevalence occurred inthe year 2002, then the variable 2002 took avalue of 1 and the other year variables tooka value of 0. Heteroskedastic-robust standarderrors were calculated and used in determin-ing p values (Hall and Cummins 2005). Thestatistical package TSP 4.5 was used for theanalysis.
The model that results from the 7 yearsof prevalence data (years 2001 to 2007)
with control variables as well as time dummyvariables is:
Autism = a + b1∗Vaccination + b2∗Log(Income) + b3∗Hispanic + b4∗AfricanAmerican + b5∗Other + b6∗2002+b7∗2003 + b8∗2004 + b9∗2005+b10∗2006 + b11∗2007
If a combination of independent variablesperfectly predicts another independent vari-able, the model is said to suffer from per-fect multicollinearity and some coefficientswill be undefined (Lewis-Beck 1990). In thismodel, if the percentage of the populationin a U.S. state that was Hispanic, AfricanAmerican, and other is known, then the per-centage of the population that was white,non-Hispanic might be determined with cer-tainty. Similarly, if the year of the prevalenceobservation was not between 2002 and 2007,then the year of the observation was 2001with certainty. Therefore, to obtain meaning-ful coefficients, the dummy variable for theyear 2001 as well as the percentage of white,non-Hispanic children were removed from themodel.
RESULTS
The results are reported in the first columnof Table 2. The association between receivingthe 4:3:1:3:3 series of vaccinations and theprevalence of autism or speech disorders is apositive and statistically significant 1.7%. Thiscoefficient represents the average change inthe prevalence of autism or speech disordersfor a 1% change in vaccination rates, hold-ing the other independent variables constant.This result holds both across geography andover time. The results suggest that if a givenU.S. state has a 1% higher vaccination ratethan another U.S. state, then the state with thehigher vaccination rate might have, on average,a 1.7% higher prevalence of autism or speechdisorders. Further, if a given U.S. state decreasesits vaccination coverage by 1% from one yearto the next, prevalence of autism or speechdisorders may, on average, fall by 1.7%. If 100%
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
TABL
E2.
Anal
ysis
ofLe
arni
ngD
isabi
litie
s,U
nite
dSt
ates
,200
1–20
07,F
ixed
Effe
cts
Mod
el
Autis
mor
spee
chor
lang
uage
impa
irmen
tEm
otio
nal
dist
urba
nce
Hea
ring
impa
irmen
tM
enta
lre
tard
atio
nO
rthop
edic
impa
irmen
tO
ther
heal
thim
pairm
ent
Spec
ificl
earn
ing
disa
bilit
yTr
aum
atic
brai
nin
jury
Visu
alim
pairm
ent
Prop
ortio
nof
child
ren
rece
ivin
g4:
3:1:
3:3
vacc
inat
ion
serie
s
0.01
66∗∗
∗(0
.00)
0.00
10(0
.31)
0.00
26(0
.70)
0.00
08(0
.69)
–0.0
008
(0.4
2)0.
0018
(0.3
2)0.
0064
∗(0
.09)
0.00
06(0
.17)
0.00
03∗
(0.1
0)
Log(
hous
ehol
dIn
com
e)–0
.002
9(0
.70)
0.00
06(0
.72)
–0.0
081
(0.3
1)0.
0008
(0.3
4)–0
.001
4(0
.46)
–0.0
011
(0.7
3)0.
0004
(0.9
5)–0
.001
4∗∗∗
(0.0
1)–0
.000
1(0
.71)
Hisp
anic
(%)
0.02
13(0
.13)
0.00
06(0
.83)
–0.0
269∗
(0.0
6)0.
0117
∗∗∗
(0.0
0)–0
.004
2(0
.20)
–0.0
029
(0.5
6)–0
.006
1(0
.72)
–0.0
008
(0.4
1)0.
0005
(0.2
9)Af
rican
Amer
ican
,not
Hisp
anic
(%)
0.02
16(0
.13)
0.01
08(0
.15)
–0.0
443
(0.1
2)0.
0053
(0.9
7)–0
.007
3(0
.19)
0.00
55(0
.29)
0.02
31(0
.23)
–0.0
001
(0.9
0)0.
0001
(0.7
9)
Oth
er,n
otH
ispan
ic(%
)–0
.020
8(0
.41)
–0.0
124∗
∗(0
.03)
–0.0
074
(0.6
4)0.
0024
(0.6
9)–0
.002
5(0
.60)
–0.0
078
(0.2
6)–0
.047
9∗∗∗
(0.0
1)–0
.000
9(–
0.77
)0.
0013
∗∗(0
.04)
Adju
sted
R2.9
636
.935
8.1
803
.953
6.7
935
.909
0.9
006
.928
0.5
312
n35
435
434
835
432
735
235
623
130
2
Not
e.St
anda
rder
rors
are
hete
rosk
edas
tic-r
obus
tusin
gW
hite
’sm
etho
d.Ti
me
dum
my
varia
bles
incl
uded
;pva
lues
are
inpa
rent
hese
s.Th
em
axim
umnu
mbe
rofo
bser
vatio
nsis
357.
Not
ever
yst
ate
repo
rtsea
chdi
sabi
lity
fore
very
year
soth
enu
mbe
rofo
bser
vatio
nsco
uld
bele
ssth
an35
7.Si
gnifi
canc
ein
dica
ted
as∗∗
∗(∗
∗ )(∗
)p<
.01
(.05)
(.10)
.
910
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
ASSOCIATION BETWEEN AUTISM AND VACCINATION 911
children received this series of vaccinations, theprevalence of autism or speech disorders wouldbe 1.7% higher than the prevalence withoutvaccination. With more than 4 × 106 babiesborn in the United States each year, this find-ing translates into an additional 680 children(= number of children [4 × 106] × coeffi-cient [0.017] × 1% [0.01]) exhibiting autism orspeech disorders for every 1% rise in childrenreceiving the 4:3:1:3:3 series of vaccinationsby age 2 years.
Robustness TestsThe association between receiving vacci-
nations and developing autism or a speechdisorder might be driven by parental behavioror access to medical care. A parent or guardianwho obtains timely vaccinations for a child mayalso be more prone to seek medical diagnosessuch as autism for the child. Similarly, parentswho live in areas with greater access to medicalcare such as cities may be better able to obtainboth vaccinations as well as autism diagnosesfor their children.
To test whether the association betweenautism and vaccination is spurious, two meth-ods were used. The first was to analyze diag-noses of other disabilities. If the autism resultsare driven by parental behavior or access tomedical care, one should also see prevalencerates of other disabilities—especially those thatrequire parental action—are positively associ-ated with receiving the vaccination series. Testswere conducted to examine the associationbetween the proportion of children receiv-ing the 4:3:1:3:3 series of vaccinations andthe prevalence of other disabilities. The samemodel that was used to analyze autism was alsoused to determine other disabilities. The sourceof the data for the number of children receivingservices from schools for particular disabili-ties was the U.S. Department of Education,Office of Special Education Programs (2007),the same as the source for the number of chil-dren receiving services for autism. The otherdisabilities were emotional disturbance, hear-ing impairment, mental retardation, orthopedicimpairment, other health impairment, specific
learning disability, traumatic brain injury, andvisual impairment.
In columns 2 through 9 of Table 2, the rela-tionships between the proportions of childrenreceiving the 4:3:1:3:3 series of vaccinationsand the prevalence of 8 other disabilities arereported. None of the relationships is signif-icant at the 5% level. The prevalence ratesof two classifications—specific learning dis-ability and vision impairment—are marginallypositively related to the proportion of chil-dren receiving the 4:3:1:3:3 series of vaccina-tions (p = .09 and .10, respectively). A specificlearning disability is defined as “a disorder inone or more of the basic psychological pro-cesses involved in understanding or in usinglanguage, spoken or written, that may mani-fest itself in an imperfect ability to listen, think,speak, read, write, spell, or to do mathematicalcalculations” (U.S. Department of Education,Office of Special Education Programs 2007).Since the disability relates to language orspeech impairments, certain school systemsmight classify children with specific learningdisabilities, while other schools classify chil-dren with similar impairments with speechor language impairments. The positive rela-tionship between visual impairment and vac-cination may be the result in autistic chil-dren of the influence on the retina of thepertussis toxin (found in the DTP vaccine),which produces visual impairments (Megson2000).
To further test whether access to med-ical care influenced the positive relation-ship between vaccination proportions and theprevalence of autism, the relationship betweenthe number of pediatricians in a state and theprevalence of autism was examined. If autismdiagnoses are driven by access to medical care,the greater the number of pediatricians perU.S. state, the higher should be the prevalenceof autism. However, the relationship betweenautism prevalence and number of pediatri-cians per 1000 children by state (Freed et al.2004), as measured by the correlation coef-ficient between the two sets of numbers, is–0.29. The result is not statistically significant,suggesting there is no significant relationship
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
912 G. DELONG
between the number of pediatricians per 1000children and the prevalence of autism.
The results from examining the number ofpediatricians by state and the analysis of otherdisabilities suggest that the association betweenthe proportion of 2-year-olds receiving the4:3:1:3:3 series of vaccinations and the preva-lence of autism is driven neither by parentalbehavior nor by access to medical care.
LimitationsThis study examined aggregate data, which
introduced at least four limitations. (1) Sincethe dependent variable was a percentage,the regression analysis showed association,not causation. If individual children had beenexamined, the dependent variable might havebeen 1 if the child developed autism or speechdisorder and 0 otherwise. The results of sucha regression could have been used to pre-dict health outcomes for children not in thestudy. However, such analysis is beyond thescope of this study. (2) The data in this studywere not exact. Learning disability classifica-tions were assigned by individual school dis-tricts, which may have implemented classifica-tions differently; vaccination proportions werebased on limited surveys, not entire popula-tions. (3) Aggregation bias assumes each indi-vidual in a given group acts according to theaverage of the group, but that is rarely thecase. In this study, children in each U.S. statewere divided into two groups, fully vaccinatedand not fully vaccinated. However, the varia-tion among the not fully vaccinated childrenwas not known. Even if a child missed onlyone shot in the series, that child was classi-fied as not fully vaccinated. A child who missedonly one shot was different from a child whowas completely unvaccinated, yet in this studyboth children were classified as not fully vac-cinated. (4) Confounding factors were also anissue. Factors such as prenatal exposure totoxins (Austin 2008) and toxin exposure fromsources other than vaccines (Palmer et al. 2006;2009) were not considered yet might influencewhether a child develops autism. Althoughthe study found that, on average, children
who were not fully vaccinated were less likelyto develop autism or speech disorders, anygiven child—especially a child who was almostfully vaccinated or was exposed to toxins inutero—may have developed the disabilities.As a result of “ecological fallacy”—applyingresults from the study of aggregate populationsto individuals—epidemiological studies such asthis one are better for creating hypothesesthan for establishing causation (Washio et al.2008).
DISCUSSION
The results of this study add support tothe hypothesized link between vaccines andautism, yet many studies conclude that alink between vaccines and autism cannot beestablished. How can this study be reconciledwith the studies that find no link? Recall thatthis study asked whether a series of vacci-nations could be associated with autism. Itused a constant definition of the prevalenceof autism and speech disorders—school chil-dren receiving services for autism or speechdisorders as a % of all school children—aswell as a constant age of 8 years old fordiagnosis. By using a special type of regres-sion analysis (fixed effects, within-group panelregression) along with dummy variables fortime, this study also controlled for confound-ing factors introduced since the study lookedat different U.S. states over time. Most studiesthat were not able to establish a link focusedon a single vaccination or vaccine ingredientand did not consider the interaction amongvaccinations. Questions about the methodol-ogy or databases used have also been raised.Madsen et al. (2002) investigated a possible linkbetween the measles–mumps–rubella (MMR)vaccine and autism. Data showed the preva-lence of autism among children who receivedthe MMR was the same as the prevalenceamong children who did not, and the studyconcluded that a link between the MMR vac-cine and autism was not established. However,many children in the study were too young tohave been diagnosed with autism even if theyhad received the MMR vaccination (Goldman
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
ASSOCIATION BETWEEN AUTISM AND VACCINATION 913
and Yazbak 2004). If Madsen et al. (2002)had examined only older children, the resultsmight have been different. A series of articlesappeared in 2003 and 2004 that used a Danishdatabase to show that the number of autismcases increased even though thimerosal wasremoved from vaccines. However, the defini-tion and catchment area of autism cases used inthe database expanded during the time of thestudies. Had the narrow definition of autismthat was used in the beginning of the studiesbeen maintained, the results might have beendifferent (Geier and Geier 2004). In anotherstudy, Verstraeten et al. (2003) found a linkbetween exposure to thimerosal and autism,but were not able to confirm the result with fur-ther study. However, the U.S. National Instituteof Environmental Health Sciences (2006) andthe Centers for Disease Control and Prevention(2008) acknowledged that the database usedin the study was inadequate for studying apossible link between thimerosal and autism.Concerns included the fact that the studyexamined different medical facilities over timebut provided no controls for differing defini-tions of autism across institution and over time.Had the studies controlled for these issues, theresults might have been different.
One study did examine the entire vac-cination schedule. Smith and Woods (2010)evaluated the long-term effects of the timingof vaccinations and found that children whowere vaccinated on-time had fewer neurolog-ical issues than children who were vaccinatedlate, which was defined as a child receiv-ing at least 1 vaccination more than 30 daysafter the recommended date. However, almostall the children in the study were exposedto vaccines, so the study did not addressthe question of whether exposure to vaccineswas associated with negative neurological out-comes. Moreover, by dividing children into2 groups—those who received vaccinationson time and those who did not—the studyaggregated what may be a disparate groupof late vaccinators. A child who received allbut 1 vaccination on time might be differentfrom a child who received no vaccinations,yet both were in the group of children who
did not receive timely vaccinations. Had theresearchers examined fully vaccinated versuscompletely unvaccinated children, the resultsmight have been different.
Future Directions for ResearchComparing the prevalence of autism
among children who are fully vaccinated andthose who are not vaccinated at all wouldbe enlightening. In their study “Children WhoReceived No Vaccines: Who Are They andWhere Do They Live?” Smith et al. (2004)used data from the U.S. National ImmunizationSurvey to determine the location of unvacci-nated children. A follow-up study could inves-tigate the prevalence of autism among unvac-cinated children. Other children who typicallyare not vaccinated could be surveyed. Thesegroups include the Amish and children servedby Homefirst, a health clinic near Chicago(Eisenstein, 2009), as well as some home-schooled children or younger siblings of chil-dren with autism whose parents decided notto vaccinate. Incremental analysis could alsodetermine the increase or decrease of theprevalence of autism or speech disorders as thenumber or type of vaccinations increased. Astudy of vaccinated versus unvaccinated chil-dren is useful and feasible.
CONCLUSIONS
Evidence presented in this paper suggestsa possible link between susceptible childrenreceiving a battery of vaccinations and devel-oping autism or speech disorders. AlthoughHg has been removed from many childhoodvaccines, other ingredients could link vaccinesto autism. Aluminum, which is found in atleast 20 U.S. childhood vaccines (Centers forDisease Control and Prevention, 2010), is notonly a neurotoxin, but also an immunosuppres-sant that may allow measles-containing vac-cines to create cytokines that damage the brain.Enhanced exposure to aluminum via vaccinesmay be associated with an increase in theprevalence of neurological disorders such as
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
914 G. DELONG
autism, especially if an aluminum-containingvaccine is administered along with a measles-containing vaccine. Reducing thimerosal andobserving an increase in autism exonerates nei-ther thimerosal nor vaccines from being poten-tial links to autism. Further research into therelationship between vaccines and autism iswarranted.
REFERENCES
Ashwood, P., Anthony, A., Torrente, F.,and Wakefield, A. J. 2004. Spontaneousmucosal lymphocyte cytokine profiles inchildren with regressive autism and gas-trointestinal symptons. J. Clin. Immunol. 24:664–73.
Austin, D. 2008. An epidemiological analysis ofthe ‘autism as mercury poisoning’ hypothe-sis. Int. J. Risk Safety Med. 20: 135–42.
Austin, D. W., and Shandley, K. 2008. An inves-tigation of porphyrinuria in Australian chil-dren with autism. J. Toxicol. Environ. HealthA 71: 1349–51.
Authier, F. J., Cherin, P., Creange, A., Bonnotte,B., Ferrer, X., Abdelmoumni, A., Ranoux, D.,Pelletier, J., Figarella-Branger, D., Granel, B.,Maisonobe, T., Coquet, M., Degos, J. D., andGherardi, R. K. 2001. Central nervous sys-tem disease in patients with macrophagicmyofasciitis. Brain 124: 974–83.
Blaxill, M. F. 2004. What’s going on? The ques-tion of time trends in autism. Public HealthRep. 119: 536–51.
Blaylock, R. L. 2008. A possible central mech-anism in autism spectrum disorders, Part 1.Altern. Ther. Health Med. 14: 46–53.
Burnett, J. C. 1960. Vaccinosis and its cure bythuja; With remarks on homæoprophylaxis.Hindhead, Surrey: Health Science Press.Originally printed 1884.
Cannell, J. J. 2008. Autism and vitamin D. Med.Hypoth. 70: 750–59.
Centers for Disease Control and Prevention.1983. Recommended schedule for activeimmunization of normal infants and children.Available at http://www.cdc.gov/vaccines/pubs/images/schedule1983s.jpg (accessed10 October 2010).
Centers for Disease Control and Prevention.2001. Prevention and control of influenza.Morbid. Mortal. Weekly Rep. 50 (RR04):1–46.
Centers for Disease Control and Prevention.2002. Prevention and control of influenza.Morbid. Mortal. Weekly Rep. 51 (RR03):1–31.
Centers for Disease Control and Prevention.2008. Report to Congress on vaccine safetydatalink, House Appropriations Commitee.Available at http://evidenceofharm.com/VaccineDataLinkReporttoCongressFinal.pdf(accessed August 26, 2008).
Centers for Disease Control and Prevention.2010. Recommended immunization sched-ules for persons aged 0 through 18 years—United States. Morbidity and MortalityWeekly Report 58: 1–4.
Centers for Disease Control and Prevention.2010. Vaccine excipient & media sum-mary, Part 2. Available at http://www.cdc.gov/vaccines/pubs/pinkbook/downloads/appendices/B/excipient-table-2.pdf(accessed October 10, 2010).
Centers for Disease Control and Prevention,National Center for Health Statistics. 2007.National immunization survey, Public usedata files. Available at: http://www.cdc.gov/nis/datafiles.htm (accessed November 10,2008).
Conti-Ramsden, G., Simkin, Z., and Botting, N.2006. The prevalence of autistic spectrumdisorders in adolescents with a history ofspecific language impairment (SLI). J. ChildPsychol. Psychiat. 47: 621–28.
De Fosse, L., Hodge, S. M., Makris, N.,Kennedy, D. N., Caviness, V. S., Jr., McGrath,L., Steele, S., Ziegler, D. A., Herbert, M.R., Frazier, J. A., Tager-Flusberg, H., andHarris, G. J. 2004. Language-associationcortex asymmetry in autism and specificlanguage impairment. Ann. Neurol. 56:757–66.
Dodds, W. J. 2001. Vaccination protocols fordogs predisposed to vaccine reactions. J. Am.Anim. Hosp. Assoc. 37: 211–14.
Eisenstein, M. 2009. Vaccine choice: Homefirstgives its families choice. Available at
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
ASSOCIATION BETWEEN AUTISM AND VACCINATION 915
http://homefirst.com/info-1/vaccine-choice.html (accessed June 12, 2009).
Environmental System Research Institute,Inc. 2009. Introduction to map design.Available at http://www.esri.com/industries/k-12/PDFs/intrcart.pdf (accessed January 20,2009).
Freed, G. L., Nahra, T. A., and Wheeler, J. R.2004. Relation of per capita income andgross domestic product to the supply anddistribution of pediatricians in the UnitedStates. J. Pediatr. 144: 723–28.
Geier, D., and Geier, M. R. 2004.Neurodevelopmental disorders followingthimerosal-containing childhood immuniza-tions: a follow-up analysis. Int. J. Toxicol. 23:369–76.
Geier, D. A., and Geier, M. R. 2007. A prospec-tive study of mercury toxicity biomarkersin autistic spectrum disorders. J. Toxicol.Environ. Health A 70: 1723–30.
Geier, D. A., Kern, J. K., and Geier, M. R.2009. A prospective blinded evaluation ofurinary porphyrins verses the clinical sever-ity of autism spectrum disorders. J. Toxicol.Environ. Health A 72: 1585–91.
Goldman, G. S., and Yazbak, F. E. 2004.An investigation of the association betweenMMR vaccination and autism in Denmark. J.Am. Physicians Surgeons 9: 70–75.
Haley, B. 2005. Mercury toxicity: Geneticsusceptibility and synergistic effects. Med.Veritas 2: 535–42.
Hall, B. H., and Cummins, C. 2005. TSP 5.0Reference Manual, 294–295, 312–314. PaloAlto, CA: TSP International.
Havarinasab, S., Haggqvist, B., Bjorn, E.,Pollard, K. M., and Hultman, P. 2005.Immunosuppressive and autoimmuneeffects of thimerosal in mice. Toxicol. Appl.Pharmacol. 204: 109–121.
Herbert, D., Tran, Y., Craig, A., Boord, P.,Middleton, J., and Siddall, P. 2007. Alteredbrain wave activity in persons with chronicspinal cord injury. Int. J. Neurosci. 117:1731–1746.
Hewitson, L., Houser, L. A., Stott, C., Sackett,G., Tomko, J. L., Atwood, D., Blue, L.,and White, E. R. 2010. Delayed acquisition
of neonatal reflexes in newborn primatesreceiving a thimerosal-containing hepatitisB vaccine: Influence of gestational age andbirth weight. J. Toxicol. Environ. Health A 73:1298–1313.
James, S. J., Cutler, P., Melnyk, S., Jernigan, S.,Janak, L., Gaylor, D. W., and Neubrander, J.A. 2004. Metabolic biomarkers of increasedoxidative stress and impaired methylationcapacity in children with autism. Am. J. Clin.Nutr. 80: 1611–17.
Kern, J. K., Grannemann, B. D., Trivedi, M. H.,and Adams, J. B. 2007. Sulfhydryl-reactivemetals in autism. J. Toxicol. Environ. HealthA 70: 715–21.
Kern, J. K., and Jones, A. M. 2006. Evidence oftoxicity, oxidative stress, and neuronal insultin autism. J. Toxicol. Environ. Health B 9:485–99.
Kogan, M. D., Blumberg, S. J., Schieve, L.A., Boyle, C. A., Perrin, J. M., Ghandour,R. M., Singh, G. K., Strickland, B. B.,Trevathan, E., and van Dyck, P. C. 2009.Prevalence of parent-reported diagnosisof autism spectrum disorder among chil-dren in the US, 2007. Pediatrics 124:1395–03.
Lewis-Beck, M. S. 1990. Applied regression: Anintroduction, Quantitative application in thesocial sciences, 49, 58–61. Newbury Park,CA: Sage.
Liu, K.-Y., King, M., and Bearman, P. S. 2010.Social influence and the autism epidemic.Am. J. Sociol. 115: 1387–34.
Madsen, K. M., Hviid, A., Vestergaard, M.,Schendel, D., Wohlfahrt, J., Thorsen, P.,Olsen, J., and Melbye, M. 2002. Apopulation-based study of measles, mumps,and rubella vaccination and autism. N. Engl.J. Med. 347: 1477–82.
McAdoo, W. G., and DeMyer, M. K. 1977.Research related to family factors in autism.J. Pediatr. Psychol. 2: 162–66.
Megson, M. N. 2000. Is autism a G-alpha pro-tein defect reversible with natural vitamin A?Med. Hypoth. 54: 979–83.
Nataf, R., Skorupka, C., Amet, L., Lam,A., Springbett, A., and Lathe, R. 2006.Porphyrinuria in childhood autistic disorder:
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012
916 G. DELONG
Implications for environmental toxicity.Toxicol. Appl. Pharmacol. 214: 99–108.
Palmer, R. F., Blanchard, S., Stein, Z., Mandell,D., and Miller, C. 2006. Environmentalmercury release, special education rates,and autism disorder: An ecological study ofTexas. Health Place 12: 203–9.
Palmer, R. F., Blanchard, S., and Wood, R.2009. Proximity to point sources of envi-ronmental mercury release as a predictor ofautism prevalence. Health Place 15: 18–24.
Poling, J. S., Frye, R. E., Shoffner, J., andZimmerman, A. W. 2006. Developmentalregression and mitochondrial dysfunction ina child with autism. J. Child Neurol. 21:170–72.
Rutter, M. 2000. Genetic studies of autism:from the 1970s into the millennium. J.Abnorm. Child Psychol. 28: 3–14.
Shoenfeld, Y., and Aron-Maor, A. 2000.Vaccination and autoimmunity-‘vaccinosis’:a dangerous liaison? J. Autoimmun. 14:1–10.
Singh, V. K. 2009. Phenotypic expression ofautoimmune autistic disorder (AAD): a majorsubset of autism. Ann. Clin. Psychiat. 21:148–61.
Smith, C. 1995. Are we vaccinating too much?J. Am. Vet. Med. Assoc. 207: 421–26.
Smith, M. J., and Woods, C. R. 2010. On-time vaccine receipt in the first year doesnot adversely affect neuropsychological out-comes. Pediatrics 125: 1134–41.
Smith, P. J., Chu, S. Y., and Barker, L. E.2004. Children who have received no vac-cines: who are they and where do they live?Pediatrics 114: 187–95.
Tebben, M. 1990. HHS awards $2.4 millionfor rare disease research. Public Health Rep.105: 212.
U.S. Census Bureau. 2008. Median incomefor 4-person family, by state, various years.Available at http://www.census.gov/hhes/www/income/4person.html (accessedNovember 12, 2008).
U.S. Department of Education, NationalCenter for Education Statistics. 2002.Private school universe survey (PSS)1999–2000 and 2001–2002. Available
at http://nces.ed.gov/surveys/pss/tables.asp(accessed July 15, 2007).
U.S. Department of Education, National Centerfor Education Statistics. 2008. Digest ofeducation statistics, various years. Availableat http://nces.ed.gov/programs/digest/d05/tables/dt05_034.asp (accessed November15, 2008).
U.S. Department of Education, Officeof Special Education Programs. 2007.Individuals with Disabilities Education Actdata, Part B, Child count, various years.Available at https://www.ideadata.org/PartBChildCount.asp (accessed November12, 2008).
U.S. Department of Education, Office ofSpecial Education Programs. 2007.OSEP IDEA, Part B, Data dictionary.Available at http://www.ideadata.org/docs/bdatadictionary.pdf (accessed June 7, 2007).
U.S. National Institute of Environmental HealthSciences. 2006. Thimerosal exposure in pedi-atric vaccines: Feasibility of studies using thevaccine safety datalink. Available at http://www.niehs.nih.gov/health/topics/conditions/autism/docs/thimerosalexposureinpediatricvaccines102606.pdf (accessed December30, 2008).
Vargas, D. L., Nascimbene, C., Krishnan, C.,Zimmerman, A. W., and Pardo, C. A. 2005.Neuroglial activation and neuroinflamma-tion in the brain of patients with autism. Ann.Neurol. 57: 67–81.
Verstraeten, T., Davis, R. L., DeStefano, F.,Lieu, T. A., Rhodes, P. H., Black, S. B.,Shinefield, H., and Chen, R. T. 2003. Safetyof thimerosal-containing vaccines: A two-phased study of computerized health main-tenance organization databases. Pediatrics112: 1039–1048.
Washio, M., Oura, A., and Mori, M. 2008.Ecological studies on influenza infec-tion and the effect of vaccination: Theiradvantages and limitations. Vaccine 26:6470–6472.
Youn, S. I., Jin, S. H., Kim, S. H., and Lim,S. 2010. Porphyrinuria in Korean childrenwith autism: correlation with oxidative stress.J. Toxicol. Environ. Health A 73: 701–710.
Dow
nloa
ded
by [
Stan
ford
Uni
vers
ity L
ibra
ries
] at
11:
39 1
8 O
ctob
er 2
012