European Journal of Pediatrics

https://doi.org/10.1007/s00431-020-03826-x
REVIEW

Association of food hypersensitivity in children with the risk

of autism spectrum disorder: a meta-analysis
Hong Li1 & Haixia Liu2 & Xin Chen2 & Jian Zhang2,3 & Guanglei Tong1 & Yehuan Sun2,4

Received: 6 June 2020 /Revised: 28 September 2020 /Accepted: 1 October 2020

# Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
This meta-analysis was performed to clarify the association between food hypersensitivity in children and autism spectrum
disorder (ASD) in detail. Relevant studies published in 8 databases before March 2020 were retrieved and screened
according to established inclusion criteria. The odds ratio (OR) with the 95% confidence interval (CI) was pooled to
estimate the effect. Subgroup analyses were performed in terms of publication year, study design, location, sample size,
definition of food hypersensitivity, definition of ASD, and study quality score. Furthermore, we stratified studies by
participant sex and age to perform a more detailed analysis. This meta-analysis included 12 published articles with 434,809
subjects. A significant association was observed between food hypersensitivity and the risk of ASD (OR = 2.792, 95% CI:
2.081–3.746). The risk of ASD among girls and subjects younger than 12 with food hypersensitivity may be greater than
that among boys and those older than 12. The results of sensitivity analysis and publication bias analysis show that the
association is relatively stable.
Conclusion: Our results showed a positive association between food hypersensitivity and autism spectrum disorder, and
girls and subjects younger than 12 may be more sensitive to this association. The role of food hypersensitivity in the onset
of ASD deserves more attention.

What is Known:
• Food hypersensitivity is a term used to describe food allergies and food intolerance.
• ASD is a group of neurodevelopmental disorders that are characterized by deficits in social interaction, repetitive or stereotypic behavior, and
verbal communication disorder. What is New:
• In this work, we reviewed and analyzed the available data and studies and found a positive association between food hypersensitivity and ASD.
• Girls and children younger than 12 may be more sensitive to have ASD than boys and children older than 12.

Keywords Food hypersensitivity . Foodallergy . Foodintolerance . Autismspectrum disorder . Meta-analysis

Hong Li and Haixia Liu contributed equally to this work.

Communicated by Gregorio Paolo Milani
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s00431-020-03826-x) contains supplementary
material, which is available to authorized users.

* Yehuan Sun Jian Zhang

yhsun_ahmu_edu@yeah.net zjnetmail@126.com
 Eur J Pediatr

Hong Li
kfk20200308@163.com Guanglei Tong
tong70409@163.com
Haixia Liu
lhx@stu.ahmu.edu.cn
Extended author information available on the last page of the article
Xin Chen
xinchen08110620@163.com
Abbreviations 2.24% in 2014, 2.41% in 2015, and 2.76% in 2016 [8];
AHRQ Agency for Healthcare Research and Quality children aged 8 years: 1.85% in 2016 [3]). The prevalence
ABC Autism Behavior Checklist ofever-diagnosed ASD inthe USA has state-level
ADI Autism Diagnostic Interview differences, which varied from 1.54% in Texas to 4.88% in
ADI-R Autism Diagnostic Interview-Revised Florida in 2016 [9]. A study conducted in Spain revealed
ADOS Autism Diagnostic Observation Schedule an ASD prevalence of 1.23% in 2017 with geographical
ASD Autism spectrum disorder differences, and the ASD incidence increased from 0.07%
CARS Childhood Autism Rating Scale in 2009 to 0.23% in 2017 [10]. In addition, the economic
CBM China Biology Medicine disc/Sinomed burden of supporting a child with ASD was estimated to be
CNKI China National Knowledge Infrastructure 1.4 million dollars both in the USA and UK [11] because
CCMD Chinese diagnostic criteria for mental disorders
of no cure, and the economic burden of ASD in the USA,
CIs Confidence intervals
combining annual direct medical, direct nonmedical, and
DSM Diagnostic and Statistical Manual of Mental
productivity costs, was forecasted to range from $162–
Disorders
FEM Fixed effects model $367 billion (0.884–2.009% of GDP) in 2015 and $276–
HRs Hazard ratios $1011 billion (0.982–3.600% of GDP) in 2025 [12]. The
IL Interleukin cause of ASD is widely believed to be a combination of
ICD International Statistical Classification of genetics and environment [13, 14]. Currently, several
Diseases studies have shown that immunediseases and allergies are
NHIS National Health Interview Survey closely related to ASD [15–17], especially in terms of food
NOS Newcastle-Ottawa Scale [15]. Food hypersensitivity is a common medical condition
ORs Odds ratios of the immune system, which may be one of the factors
PRISMA Preferred Reporting Items for Systematic related to ASD [15]. Food hypersensitivity is a term used
Review and Meta-Analyses to describe food allergies and food intolerance [18].
PRs Prevalence ratios According to a study from Russia, a high frequency of
REM Random effects model hypersensitivity to cereal and dairy products was found in
RRs Relative risks children with ASD [19]. Based on data from the National
TNF Tumor necrosis factor
Health Interview Survey (NHIS) collected between 1997
Introduction and 2016 in the USA, it was reported that the weighted
prevalence of food allergies was higher in children with
Autism spectrum disorder (ASD) is a group of
ASD (11.25%) than in children without ASD (4.25%) [15].
neurodevelopmental disorders characterized by deficits in
A recent review also supported the relationship between
social interaction, repetitive or stereotypic behavior, and
food and ASD, which found that maternal diet during
verbal communication disorder [1]. Researchers have
gestation and the diet of children with ASD were
found evidence of sex differences in ASD; that is, ASD is
modifiable risk factors for the development or worsening
more common in boys than in girls [2, 3]. Most children
of symptoms of ASD [20]. Harumi Jyonouchi’s studies
with ASD go to the doctor for speech dysplasia or delay
revealed that food allergies may account for some
[4], which could be diagnosed as early as 3 years of age
gastrointestinal symptoms observed in children with ASDs
[5]. A systematic review demonstrated that the global
[21]. The pathogenesis is more likely that mast cells
incidence of ASD in 2010 was 7.6‰ and there was little
(primarily in the hypothalamus and participate in the
regional variation in the prevalence of ASD worldwide [6].
regulation of behavior and language [22]) activated by
A long-term autism monitoring study conducted in the
allergens [22, 23] release proinflammatory and neurotoxic
USA found that the prevalence of ASD was increasing
molecules [23], which disrupt the gut-blood-brain barriers
annually (children aged 4 years: 1.34% in 2010, 1.53% in
[24] (the effect of the gutmicrobiota-brain axis [25]) and
2012, and 1.70% in 2014 [7]; children aged 3 to 17 years:
Eur J Pediatr

stimulate microglia [26], thus contributing to brain focal analyze; and (4) duplicate studies. The most recent and
inflammation and ASD [22–25]. However, the findings of complete data were included if studies were the same or
previous studies on whether individuals with ASD have progressive. Two authors (H Liu and H Li) independently
abnormal immunoglobulin (Ig) levels are slightly different, performed the eligibility assessment, and discrepancies
including increased IgG4 [27] and the absence of elevated were resolved through discussion (k, number of studies
IgE [28]. Growing studies show that early intervention [31, 32] = 12, Kappa coefficient [33] =
plays an important role in the prognosis of ASD [29, 30]. 0.856).
Given the high prevalence and high economic burden of
ASD, food, as an easily implemented and accepted Determination of exposure and outcome
intervention, caught our attention. Our meta-analysis aimed
to clarify the association between food hypersensitivity in Food hypersensitivity was clearly defined based on
children and ASD in detail. parental reports or clinical diagnosis or abnormal immune
indicators. ASD was defined clearly in articles by clinical
diagnosis or more detailed diagnostic criteria/abnormal
assessment scale scores. The studies that did not clearly
Methods distinguish between allergicdiseasesand
foodhypersensitivity, and the studies that did not confirm
This meta-analysis was performed with reference to the that this population was ASD (such as children with a low
Preferred Reporting Items for Systematic Review and score on one scale may be diagnosed with ASD in the
MetaAnalyses (PRISMA) guidelines. future) were not considered.
Literature search
Data extraction
Before 8 March 2020, relevant available articles were
identified by literature searching from the following The following information was extracted according to a
databases: (1) PubMed; (2) Cochrane Library; (3) predesigned extraction form by two independent authors:
ScienceDirect; (4) Web of Science; (5) China National first author’s name, publication year, location where the
Knowledge Infrastructure (CNKI); (6) Sinomed (CBM); study was carried out, sample size, study design, definition
(7) Wanfang Data; and (8) Chinese VIP Database. The of food hypersensitivity, definition of ASD, and study
systematic retrieval strategy was as follows: (Food quality score. The extracted data included relative risks
Hypersensitivity OR Food (RRs), oddsratios(ORs), prevalence ratios (PRs) or hazard
Hypersensitivities OR Food Allergy OR Food Allergies OR ratios (HRs) and 95% confidence intervals (CIs), of which
Food intolerance OR Food Anaphylaxis) and (Autism OR indexes adjusted for the most confounders were preferred.
Autism Spectrum Disorder OR Autism Spectrum Disorders If the above effect indexes were not available, the raw
OR Asperger Syndrome OR Autistic Disorder OR ASD). unadjusted data (such as the number of exposed and
Additionally, we reviewed the references from the relevant unexposed persons in the case group and control group or
articles to identify additional eligible studies. the number of patients and nonpatients in the exposed
group and unexposed group) in the original studies could
Study selection be extracted. If the data on the total population was
missing, the subgroup results in the article could be
The following inclusion criteria were applied: (1) a study combined to represent the total results. If the data on effect
that focused on the association between food size in one article were all missing, the article would not be
hypersensitivity and autism spectrum disorder, in which considered.
food hypersensitivity and autism spectrum disorder have
clear definitions; (2) an observational study published as an
original study; and (3) studies that provided odds ratios Study quality assessment
(ORs) and 95% confidence intervals (CIs) or data to
calculate them. The exclusion criteria were as follows: (1) According to the Agency for Healthcare Research and
studies for which the full text was unavailable or only the Quality (AHRQ), the Newcastle-Ottawa Scale (NOS) and
abstract was available; (2) reviews, case reports, 11-item Prevalence Study Quality checklist were
comments, or letters to the editor; (3) studies for which recommended to assess the quality of the included case-
data on effect size were unavailable or not adequate to control studies, cohort studies, and cross-sectional studies
 Eur J Pediatr

[34]. For NOS, studies that scored 0–3 points, 4–6 points, Results
and 7–9 points were classified as low, moderate, and high
quality in turn, respectively. For the 11-item checklist, the Study selection and characteristics of studies
quality of the studies was assessed as follows: low quality
= 0–3; moderate quality = 4–7; and high quality = 8–11. A total of 1254 records were initially retrieved from 8
databases. After study selection, our meta-analysis finally
includedonly12 published articles[14, 15, 38–47].The
Statistical analysis flowchart of the literature search is shown in Fig. 1. This
meta-analysis included 434,809 subjects, including
The strength of the association between food participants from 4 cross-sectional studies, 7 case-control
hypersensitivity and autism spectrum disorder was studies, and only 1 cohort study. Among these studies, 6
evaluated by calculating the pooled effect estimates and studies were performed in the USA, 3 in China, 1 in Italy, 1
their 95% CIs. To explore the sources of heterogeneity, we in Norway, and 1 in Iran. Study quality assessment
performed subgroup analysis in terms of publication year, indicated that all included studies were of moderate or high
location, sample size, study design, definition of food quality, and all included casecontrol studies were of high
hypersensitivity, definition of ASD, and study quality quality. The characteristics of the eligible studies are
score. Furthermore, eligible studies were stratified by sex summarized in Table 1.
and age. Heterogeneity analysis among studies was
performed using Cochran’s Q chi-square test and I2 analysis Quantitative synthesis
[31]. I2≥ 50% indicated the existence of significant
heterogeneity, and the random effects model (REM) was In general, we extracted raw unadjusted data from 7 studies
used as the pooling method; otherwise, the fixed effects [14, 38, 43, 44–47] and adjusted ORs from 5 studies [15,
model (FEM) was used. Sensitivity analysis was performed 39, 40–42] to pool the result of this meta-analysis
to assess the stability of the results [35]. Funnel plot (Bresnahan et al.’s study [41] was the only cohort study we
asymmetry, Begg’s tests, and Egger’s tests were used to included, which reported an OR value; thus we pooled
assess publication bias qualitatively and quantitatively [36, using OR finally). The pooled OR from the random effects
37]. All statistical analyses were performed using Stata model was 2.792 (95% CI: 2.081– 3.746; I2 = 75.2%),
(version 14.0; Stata Corp, College Station, TX) software. which indicated a statistically significant association
Differences were deemed statistically significant when P < between food hypersensitivity and autism spectrum
0.05. disorder. The main result is shown in Fig. 2.
Eur J Pediatr

Fig. 1 Flow diagram of the study search and selection process Table 1 Characteristics of studies included in the meta-analysis
First author’s Publication Location Sample Study design Definition of exposure Definition of case Study quality
name year size score

Total serum IgE

Renzoni E [38] 1995 Italy 86 Case-control DSM-III-R and ABC 8
≥ 200 kU/L
Gurney JG [39] 2006 USA 85,272 Cross- Parental report Clinical diagnosis 6
sectional
Geng X [45] 2010 China 102 Case-control sIgG(+) CCMD-3 and CARS 7
Garg N [40] 2014 USA 16,610 Cross- Clinical diagnosis Clinical diagnosis 7
sectional
Li S [46] 2014 China 60 Case-control Fasting blood test ICD-10 (2 diagnosis) and 7
≥ 50 U/mL ABC
Bresnahan M 2015 Norway 45,126 Cohort Parental report DSM-IV-TR and ICD-10 6
[41]
Lyall K [42] 2015 USA 951 Case-control Parental report ADOS and ADI-R 7
Zerbo O [43] 2015 USA 33,390 Case-control ICD-9 (2 diagnosis) ICD-9-CM (2 diagnosis) 8
 Eur J Pediatr

Hu, Q [47] 2017 China 100 Case-control IgG ≥ 50 U/mL DSM-V, CARS and ABC 7
Xu G [15] 2018 USA 199,520 Cross- Parental report Clinical diagnosis 5
sectional
Malek A [14] 2019 Iran 227 Case-control Mother’s report ADI and DSM-IV-TR 7
Tan Y [44] 2019 USA 53,365 Cross- Parental report Clinical diagnosis 7
sectional
DSM, Diagnostic and Statistical Manual of Mental Disorders; ABC, Autism Behavior Checklist; CCMD, Chinese diagnostic criteria for mental
disorders; CARS, Childhood Autism Rating Scale; ICD, International Statistical Classification of Diseases; ADOS, Autism Diagnostic
Observation Schedule; ADI (ADI-R), Autism Diagnostic Interview (Autism Diagnostic Interview-Revised)
Then, subgroup analysis was performed based on the than that of moderatequality studies (OR = 2.651; 95% CI:
characteristics of 12 studies. When grouping by publication 2.072–3.391; I2 = 57.9%). The detailed results of the
year, the pooled OR of the studies published before 2017 subgroup analysis are summarized in Table 2.
(OR = 2.685, I2 = 78.3%) was lower than those of studies We only selected two studies (1 study [15] adjusted for
published after 2017 (OR = 2.985, I 2 = 74.2%) and the the most confounders and 1 study with raw adjusted data
heterogeneity was higher; these differences were both [43]) with data about sex group and 6 studies (3 studies
statistically significant. Regarding study design, the [15, 41, 42] adjusted for the most confounders and 3
reported result of the only cohort study [41] (OR = 1.800, studies with raw adjusted data [43, 45, 46]) with data about
95% CI: 1.006–3.220) was lower than that of cross- age group. A significant association was found between
sectional studies (OR = 2.803; 95% CI: 2.151–3.651; I2 = food hypersensitivity and ASD in both girls and boys, and
62.3%) and may be much lower than that of case-control the pooled OR was 2.470 (95% CI: 1.691–3.608; I2 = 0.0%)
studies (OR = 3.459; 95% CI: 1.719– 6.962; I2 = 82.1%). in girls, which was higher than the pooled OR in boys (OR
Compared with the studies conducted in the other = 1.894; 95% CI: 1.314– 2.731; I2 = 68.1%). We found a
locations, statistical significance was found in the studies significant association between food hypersensitivity and
conducted in the USA (OR = 2.467; 95% CI: 1.923– ASD whether or not subjects were younger than 12. The
3.163; I2 = 67.6%) and in China (OR = 11.746; 95% CI: pooled OR was 2.494 (95% CI: 1.864–3.336; I2 = 48.7%)
5.869–23.510; I2 = 4.4%). The pooled OR of studies with in those younger than 12, which may be higher than the
more than 30,000 participants was 2.408 (95% CI: 1.833– pooled OR in those older than 12 (OR = 1.914; 95% CI:
3.163; I2 = 75.5%), which may be lower than that of studies 1.023– 3.581; I2 = 71.4%). The main results are
with fewer than 30,000 subjects (OR = 3.872; 95% CI: summarized in Table 3.
1.779– 8.429; I2 = 76.8%). Then, we reanalyzed the
remaining 7 studies without studies of Malek et al. [14],
Renzoni et al. [38], Geng et al. [45], Li et al. [46], and Hu
[47] (which sample size is small) and found that the
combined results (OR = 2.389, 95% CI: 1.898–3.008) are
similar to the combined results of the total 12 studies (OR
= 2.792, 95% CI: 2.081–3.746), which indicated that the
sample size of including studies did not influence the
correlation. Considering the definition of exposure, a
significant relationship was found for food hypersensitivity
reported by both diagnosis and assessment (OR = 3.861;
95% CI: 1.479–10.081; I2 = 85.0%) and parents 2.611
(95% CI: 2.109–3.232; I2 = 48.9%). Considering the case
definition, a statistically significant correlation was found;
the pooled OR was 3.070 (95% CI: 1.724–5.468; I2 =
79.4%) for studies that reported detailed ASD diagnostic
criteria, which may be higher than those studies that
reported no detailed ASD diagnostic criteria (OR = 2.803;
95% CI: 2.151– 3.651; I2 = 62.3%). When grouping by
study quality, the combined OR of high-quality studies
(OR = 3.459; 95% CI: 1.719–6.962; I2 = 82.1%) was higher
Eur J Pediatr

Fig. 2 Forest plot of the ORs with corresponding 95% CIs from 12 studies on food hypersensitivity with the risk of ASD
Sensitivity analysis and publication bias analysis hypothalamus together with peptide neurotensin and
release interleukin (IL) and tumor necrosis factor (TNF),
The results of sensitivity analysis conducted by omitting which in turn stimulate microglia proliferation and
one study at a time were similar, without great fluctuation, activation, leading to disruption of neuronal connectivity
suggesting that the pooled OR was relatively stable. The and aberrant immune response, thus contributing to focal
results of the roughly symmetrical funnel plot inflammation and ASD [22–25, 51]. However, it is unclear
(Supplementary Fig. S3) and Begg’s test (P = 0.304) and whether there are abnormal immunoglobulin levels in
Egger’s test (P = 0.327) showed no publication bias. patients with ASD, where a previous study reported
increased IgG4 [27] and another study reported an absence
of elevated IgE [28]. What’s more, a recent review found
Discussion that early impairments of visual behavior including
oculomotor abilities, visual attention, and visual-motor
Our meta-analysis concluded that food hypersensitivity integration may be involved in the onset of ASD [52].
was associated with an increased risk of ASD. Regardless Although there is currently no cure for ASD, many
of whether the study was published after 2017, the type of potential treatment approaches have been found to reduce
study design, the sample size, how food hypersensitivity symptoms and improve behavior, such as luteolin
was defined, how ASD was defined, and the quality of the decreasing serum IL-6 and TNF levels and improving
study, the results demonstrated a correlation between food behavior [53], the combination of the flavonoids luteolin
hypersensitivity and ASD, similar to the results of previous and quercetin reducing ASD symptoms [54], the natural
reviews [48–50]. The detailed relevant mechanism is most flavonoids luteolin and tetramethoxyluteolin inhibiting
likely that mast cells activated by food hypersensitivity, inflammatory processes and having neuroprotective actions
release inflammatory and vasoactive mediators that [55], methoxyluteolin inhibiting stimulation of mast cells a
increase blood-brain barrier permeability, and mast cells greater extent than luteolin or Torin1, and the novel
trigger corticotropin-releasing hormone secretion from the flavonoid methoxyluteolin inhibiting mast cell stimulation
 Eur J Pediatr

entirely [56]. Moreover, a previous review suggested that a locations. In addition, studies conducted in China were all
gluten-free and case-control studies with few subjects, which may
Table 2 The pooled odds ratios of subgroup analysis for 12 studies between food hypersensitivity with ASD

Subgroups No. of studies OR (95% CI) Z P Heterogeneity

I2 (%) P

All studies 12 2.792 (2.081–3.746) 6.84 0.000 75.2 0.000

Publication year
< 2017 8 2.685 (1.619–4.451) 3.83 0.000 78.3 0.000
≥ 2017 4 2.985 (2.062–4.320) 5.80 0.000 74.2 0.009
Study design
Cross-sectional study 4 2.803 (2.151–3.651) 7.64 0.000 62.3 0.047
Case-control study 7 3.459 (1.719–6.962) 3.48 0.001 82.1 0.000
Cohort study 1 1.800 (1.006–3.220) 1.98 0.048 NA NA
Location
USA 6 2.467 (1.923–3.163) 7.11 0.000 67.6 0.009
China 3 11.746 (5.869–23.510) 6.96 0.000 4.4 0.
351
Other 3 1.633 (0.864–3.084) 1.51 0.131 42.9 0.174
Sample size
≤ 30,000 7 3.872 (1.779–8.429) 3.41 0.001 76.8 0.000
> 30,000 5 2.408 (1.833–3.163) 6.31 0.000 75.5 0.003
Definition of exposure
Parental report 6 2.611 (2.109–3.232) 8.81 0.000 48.9 0.082
Diagnosis/assessment 6 3.861 (1.479–10.081) 2.76 0.006 85.0 0.000
Definition of case
Nondetailed diagnostic criteria 4 2.803 (2.151–3.651) 7.64 0.000 62.3 0.047
Detailed diagnostic criteria 8 3.070 (1.724–5.468) 3.81 0.000 79.4 0.000
Study quality
High 7 3.459 (1.719–6.962) 3.48 0.001 82.1 0.000
Moderate 5 2.651 (2.072–3.391) 7.75 0.000 57.9 0.050

casein-free diet could be implemented when someone was exaggerate the relationship. In terms of sex and age, girls
diagnosed with an allergy or intolerance to nutritional and subjects younger than 12 with food hypersensitivity
gluten or casein [57]. Additionally, the use of folic acid and may be more likely to have ASD than boys and those older
multivitamin supplements among women before and than 12. This may be related to the high prevalence of
during pregnancy has been reported to reduce the risk of immune diseases and low prevalence of ASD in girls [2,
ASD in offspring [58]. 59]. Differences in age may be because most children with
While significant results were found in the studies ASD were diagnosed early at aged 3 [5], the capacity of
conducted in the USA and China, no significant visuospatial attention increases slightly with age and the
relationship was found in the studies conducted in the other reaction time to initiate an eye movement decreases until

Table 3 The pooled odds ratios of included studies stratified by different ages and sex
No. of studies OR (95% CI) Z P
Heterogeneity

I2 (%) P

Sex 2
Boys 2 1.894 (1.314–2.731) 3.42 0.001 68.1 0.077
Girls 2 2.470 (1.691–3.608) 4.68 0.000 0.0 0.777
Age 6

< 12 6 2.494 (1.864–3.336) 6.15 0.000 48.7 0.058

≥ 12 2 1.914 (1.023–3.581) 2.03 0.042 71.4 0.061
Eur J Pediatr

14– 15 years of age approximately [52]. Therefore, larger mainly reviewed the paper and Guanglei Tong assisted. Hong Li and
Haixia Liu had primary responsibility for the final content. All
and well-designed studies are needed to confirm the
authors read and approved the final manuscript.
differences in regard to sex and age.
This meta-analysis has several advantages. First, to our Funding This study was funded by National Natural Science
knowledge, this meta-analysis was the first to discuss the Foundation of China (grant number: 81872704).
correlation between food hypersensitivity and ASD.
Second, we performed detailed subgroup analyses, Compliance with ethical standards
including by publication year, study design, location,
sample size, definition of food hypersensitivity, definition Conflict of interest The authors declare that they have no conflict of
interest.
of ASD, and study quality score. Third, we extracted data
from different sex and age groups to explore sensitive
Ethical approval This article does not contain any studies with human
populations. Finally, the adjusted ORs were preferentially participants or animals performed by any of the authors.
extracted for pooling. In addition, sensitivity analysis and
publication bias analysis showed that our conclusion was Informed consent Informed consent was obtained from all individual
stable and convincing. However, our meta-analysis also has participants included in the study.
several limitations. First, the extracted data included
unadjusted and adjusted data, and the adjustment factors
were not the same in some studies. That may result in
References
underestimating or exaggerating risk. Second, the
heterogeneity of all studies (I 2 = 75.2, P = 0.000) was high, 1. Newschaffer CJ, Croen LA, Daniels J et al (2007) The
which may decrease reliability. The source of high epidemiology of autism spectrum disorders. Annu Rev Public
Health 28:235–258.
heterogeneity may be the differences in the study quality,
https://doi.org/10.1146/annurev.publhealth.28.021406.144007
study design, study location, and ASD or food 2. Wing L (1981) Sex ratios in early childhood autism and related
hypersensitivity definition of including studies. Finally, on conditions. Psychiatry Res 5(2):129–137.
account of the relatively small number of studies, our meta- https://doi.org/10.1016/ 0165-1781(81)90043-3
analysis included only one cohort study and we know that 3. Maenner M, Shaw K, Baio J et al (2020) Prevalence of autism
the results of cohort studies are more reliable than those of spectrum disorder among children aged 8 years - Autism and
Developmental Disabilities Monitoring Network, 11 Sites,
crosssectional studies. Therefore, more high-quality cohort United States, 2016. Morbidity and mortality weekly report.
studies are needed in the future. Surveill Summaries (Washington, DC : 2002) 69(4):1–12.
https://doi.org/ 10.15585/mmwr.ss6904a1
4. Chawarska K, Paul R, Klin A, Hannigen S, Dichtel LE,
Volkmar F (2007) Parental recognition of developmental
Conclusion problems in toddlers with autism spectrum disorders. J Autism
Dev Disord 37(1):62–72. https://doi.org/10.1007/s10803-006-
In summary, our meta-analysis shows that food 0330-8
5. Ospina MB, Krebs Seida J, Clark B et al (2008) Behavioural
hypersensitivity is associated with ASD. This correlation
and developmental interventions for autism spectrum disorder:
was not influenced by the publication year, design type, a clinical systematic review. PLoS One 3(11):e3755.
sample size, definition of food hypersensitivity, definition https://doi.org/10. 1371/journal.pone.0003755
of ASD, or the quality of included studies but may be 6. Baxter AJ, Brugha TS, Erskine HE, Scheurer RW, Vos T, Scott
affected by location where the study was carried out. The JG (2015) The epidemiology and global burden of autism
spectrum disorders. Psychol Med 45(3):601–613.
ASD risk among girls and subjects younger than 12 with https://doi.org/10.1017/ s003329171400172x
food hypersensitivity may be greater than that among boys 7. Christensen DL, Maenner MJ, Bilder D et al (2019) Prevalence
and those older than 12. Given the high prevalence of and and characteristics of autism spectrum disorder among children
lack of cure for ASD in the general population, the role of aged 4 years - Early Autism and Developmental Disabilities
Monitoring Network, Seven Sites, United States, 2010, 2012,
food hypersensitivity in the onset of ASD deserves more and 2014. Morbidity and mortality weekly report. Surveill
attention. Certainly, our finding requires more studies to be Summaries (Washington, DC : 2002) 68(2):1–19.
confirmed further. https://doi.org/10.15585/ mmwr.ss6802a1
8. Xu G, Strathearn L, Liu B, Bao W (2018) Prevalence of autism
Authors’ Contributions Hong Li, Haixia Liu, and Yehuan Sun spectrum disorder among US children and adolescents, 2014–
designed the research; Hong Li, Haixia Liu, and Xin Chen conducted 2016. JAMA 319(1):81–82.
the research; Hong Li, Haixia Liu, and Jian Zhang analyzed the data; https://doi.org/10.1001/jama.2017.17812
Hong Li and Haixia Liu, and Xin Chen wrote the paper. Yehuan Sun
 Eur J Pediatr

9. Xu G, Strathearn L, Liu B et al (2019) Prevalence and treatment 24. Angelidou A, Alysandratos K-D, Asadi S et al (2011) Brief
patterns of autism spectrum disorder in the United States, 2016. report: “Allergic symptoms” in children with autism spectrum
JAMA Pediatr 173(2):153–159. https://doi.org/10.1001/ disorders. More than meets the eye? J Autism Dev Disord
jamapediatrics.2018.4208 41(11):1579–1585. https://doi.org/10.1007/s10803-010-1171-z
10. Pérez-Crespo L, Prats-Uribe A, Tobias A et al (2019) Temporal 25. Petra AI, Panagiotidou S, Hatziagelaki E, Stewart JM, Conti P,
and geographical variability of prevalence and incidence of Theoharides TC (2015) Gut-microbiota-brain axis and its effect
autism spectrum disorder diagnoses in children in Catalonia, on neuropsychiatric disorders with suspected immune
Spain. Autism Res 12(11):1693–1705. dysregulation. Clin Ther 37(5):984–995.
https://doi.org/10.1002/aur.2172 https://doi.org/10.1016/j.clinthera. 2015.04.002
11. Buescher AV, Cidav Z, Knapp M, Mandell DS (2014) Costs of 26. Theoharides TC, Stewart JM, Panagiotidou S, Melamed I
autism spectrum disorders in the United Kingdom and the (2016) Mast cells, brain inflammation and autism. Eur J
United States. JAMA Pediatr 168(8):721–728. Pharmacol 778: 96–102.
https://doi.org/10.1001/ jamapediatrics.2014.210 https://doi.org/10.1016/j.ejphar.2015.03.086
12. Leigh J, Du J (2015) Brief Report: forecasting the economic 27. Enstrom A, Krakowiak P, Onore C et al (2009) Increased IgG4
burden of autism in 2015 and 2025 in the United States. J levels in children with autism disorder. Brain Behav Immun
Autism Dev Disord 45(12):4135–4139. 23(3):389–395. https://doi.org/10.1016/j.bbi.2008.12.005
https://doi.org/10.1007/s10803-0152521-7 28. Theoharides TC, Angelidou A, Alysandratos K-D et al (2012)
13. Liu Q, Chen M, Sun L et al (2019) Rational use of Mast cell activation and autism. Biochim Biophys Acta (BBA) -
mesenchymal stem cells in the treatment of autism spectrum Mol Basis Dis 1822(1):34–41.
disorders. World J Stem Cells 11(2):55–72. https://doi.org/10.1016/j.bbadis.2010. 12.017
https://doi.org/10.4252/wjsc.v11.i2.55 29. Camarata S (2014) Early identification and early intervention in
14. Malek A, Farhang S, Amiri S, Abdi S, Rezaih AR, Asadian M autism spectrum disorders: accurate and effective? Int J
(2019) Risk factors for autistic disorder: a case-control study. SpeechLanguage Pathol 16(1):1–10.
Iran J Pediatr 29(3). https://doi.org/10.5812/ijp.80935 https://doi.org/10.3109/17549507. 2013.858773
15. Xu G, Snetselaar LG, Jing J, Liu B, Strathearn L, Bao W (2018) 30. Tachibana Y, Miyazaki C, Ota E et al (2017) A systematic
Association of food allergy and other allergic conditions with review and meta-analysis of comprehensive interventions for
autism spectrum disorder in children. JAMA Netw Open 1(2): pre-school children with autism spectrum disorder (ASD).
e180279–e180e79. https://doi.org/10.1001/jamanetworkopen. PLoS One 12(12): e0186502.
2018.0279 https://doi.org/10.1371/journal.pone.0186502
16. Shibata A, Hitomi Y, Kambayashi Y et al (2013) 31. Higgins J, Thompson S, Deeks J, Altman D (2003) Measuring
Epidemiological study on the involvements of environmental inconsistency in meta-analyses. BMJ (Clin Res ed) 327(7414):
factors and allergy in child mental health using the Autism 557–560. https://doi.org/10.1136/bmj.327.7414.557
Screening Questionnaire. Res Autism Spectr Disord 7(1):132– 32. Huedo-Medina T, Sánchez-Meca J, Marín-Martínez F, Botella J
140. https://doi.org/10.1016/j.rasd. 2012.06.003 (2006) Assessing heterogeneity in meta-analysis: Q statistic or
17. Lin TY, Lin PY, Su TP et al (2014) Autistic spectrum disorder, I2 index? Psychol Methods 11(2):193–206.
attention deficit hyperactivity disorder, and allergy: is there a https://doi.org/10.1037/ 1082-989x.11.2.193
link? A nationwide study. Res Autism Spectr Disord 33. Tang W, Hu J, Zhang H, Wu P, He H (2015) Kappa coefficient:
8(10):1333–1338. https://doi.org/10.1016/j.rasd.2014.07.009 a popular measure of rater agreement. Shanghai Arch
18. Lomer MC (2015) Review article: The aetiology, diagnosis, Psychiatry 27(1):62–67. https://doi.org/10.11919/j.issn.1002-
mechanisms and clinical evidence for food intolerance. Aliment 0829.215010
Pharmacol Ther 41(3):262–275. 34. Rostom A, Dubé C, Cranney A et al (2004) Celiac disease.
https://doi.org/10.1111/apt.13041 Evidence Reports/Technology. Appendix D. Quality
19. Cнerevko NA, Skirnevskaya AV, Rosenstein MY, Novikov PS, Assessment Forms. Assessments, No. 104. Rockville (MD):
Muraveinik OA, Denisov AA (2018) Features of specific food Agency for Healthcare Research and Quality (US). Available
hypersensitivity to dairy and cereal products in children with from: https:// www.ncbi.nlm.nih.gov/books/NBK35156/)
autism spectrum disorder. Byulleten' Sibirskoi Meditsiny 35. Haidich AB (2010) Meta-analysis in medical research.
17(1):159–166. https://doi.org/10.20538/1682-0363-2018-1- Hippokratia 14(Suppl 1):29–37
159-166 36. Begg CB, Mazumdar M (1994) Operating characteristics of a
20. Peretti S, Mariano M, Mazzocchetti C et al (2019) Diet: the rank correlation test for publication bias. Biometrics
keystone of autism spectrum disorder? Nutr Neurosci 50(4):1088–1101
22(12):825–839.
37. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias
https://doi.org/10.1080/1028415x.2018.1464819 in meta-analysis detected by a simple, graphical test. BMJ
21. Harumi J (2009) Food allergy and autism spectrum disorders: is (Clinical research ed) 315(7109):629–634
there a link? Curr Allergy Asthma Rep 9(3):194–201.
38. Renzoni E, Beltrami V, Sestini P, Pompella A, Menchetti G,
https://doi. org/10.1007/s11882-009-0029-y Zappella M (1995) Brief report: Allergological evaluation of
22. Theoharides TC (2013) Is a subtype of autism an allergy of the children with autism. J Autism Dev Disord 25(3):327–333
brain? Clin Ther 35(5):584–591. https://doi.org/10.1016/j.
clinthera.2013.04.009
23. Theoharides T, Valent P, Akin C (2015) Mast cells,
mastocytosis, and related disorders. N Engl J Med 373(2):163–
172. https://doi. org/10.1056/NEJMra1409760
Eur J Pediatr

39. Gurney JG, McPheeters ML, Davis MM (2006) Parental report 51. Theoharides TC, Tsilioni I, Patel AB, Doyle R (2016) Atopic
of health conditions and health care use among children with diseases and inflammation of the brain in the pathogenesis of
and without autism: National Survey of Children's Health. Arch autism spectrum disorders. Transl Psychiatry:6.
Pediatr Adolesc Med 160(8):825–830. https://doi.org/10.1001/ https://doi.org/10.1038/tp. 2016.77
archpedi.160.8.825 52. Apicella F, Costanzo V, Purpura G (2020) Are early visual
40. Garg N, Silverberg JI (2014) Association between childhood behavior impairments involved in the onset of autism spectrum
allergic disease, psychological comorbidity, and injury disorders? Insights for early diagnosis and intervention. Eur J
requiring medical attention. Ann Allergy Asthma Immunol Pediatr 179(2): 225–234. https://doi.org/10.1007/s00431-019-
112(6):525–532. 03562-x
https://doi.org/10.1016/j.anai.2014.03.006 53. Tsilioni I, Taliou A, Francis K, Theoharides T (2015) Children
41. Bresnahan M, Hornig M, Schultz AF et al (2015) Association of with autism spectrum disorders, who improved with a
maternal report of infant and toddler gastrointestinal symptoms luteolincontaining dietary formulation, show reduced serum
with autism evidence from a prospective birth cohort. JAMA levels of TNF and IL-6. Transl Psychiatry 5:e647.
Psychiatry 72(5):466–474. https://doi.org/10.1038/ tp.2015.142
https://doi.org/10.1001/jamapsychiatry.2014.3034
42. Lyall K, Van de Water J, Ashwood P, Hertz-Picciotto I (2015)
Asthma and allergies in children with autism spectrum
disorders: results from the CHARGE Study. Autism Res
Affiliations
8(5):567–574. https://doi.org/10.1002/aur.1471
54. Taliou A, Zintzaras E, Lykouras L, Francis K (2013) An open-
43. Zerbo O, Leong A, Barcellos L, Bernal P, Fireman B, Croen LA label pilot study of a formulation containing the anti-
(2015) Immune mediated conditions in autism spectrum inflammatory flavonoid luteolin and its effects on behavior in
disorders.
children with autism spectrum disorders. Clin Ther 35(5):592–
Brain Behav Immun 46:232–236. https://doi.org/10.1016/j.bbi. 602. https://doi.org/10. 1016/j.clinthera.2013.04.006
2015.02.001 55. Theoharides T, Tsilioni I (2018) Tetramethoxyluteolin for the
44. Tan Y, Thomas S, Lee BK (2019) Parent-reported prevalence of treatment of neurodegenerative diseases. Curr Top Med Chem
food allergies in children with autism spectrum disorder: 18(21):1872– 1882.
National health interview survey, 2011–2015. Autism Res https://doi.org/10.2174/1568026617666181119154247
12(5):802–805. https://doi.org/10.1002/aur.2106 56. Patel A, Theoharides T (2017) Methoxyluteolin inhibits
45. Geng X, Wu L, Song L (2010) Study on food intolerance in neuropeptide-stimulated proinflammatory mediator release via
children with autism and the curative effect of avoiding mTOR activation from human mast cells. J Pharmacol Exp
intolerance food. Chin J Appl Clin Pediatr 7:511–512 Ther
46. Li S, Wang X, Sun Z et al (2014) The food intolerance in 361(3):462–471. https://doi.org/10.1124/jpet.117.240564
patients with autism spectrum disorder. China J Health Psychol 57. Lange KW, Hauser J, Reissmann A (2015) Gluten-free and
22(2):181– 182 caseinfree diets in the therapy of autism. Curr Opinion Clin
47. Hu Q (2017) The study of food intolerlance specific IgG Nutrit Metabol Care 18(6):572–575.
antibodiy in children with autism spectrum disoeder [Master]. https://doi.org/10.1097/mco. 0000000000000228
Nanhua University 58. Levine S, Kodesh A, Viktorin A et al (2018) Association of
48. Lu M, Hu Y (2018) Review on the effects and mechanisms of maternal use of folic acid and multivitamin supplements in the
food allergy on autism spectrum disorders. Chin J Child Health periods before and during pregnancy with the risk of autism
Care 26(04):389–91+95 spectrum disorder in offspring. JAMA Psychiatry 75(2):176–
49. Guo J, Li E (2016) Research progress on food intolerance in 184. https://doi. org/10.1001/jamapsychiatry.2017.4050
children with autism. Chin J Clin Rational Drug Use 9(11):176– 59. Voskuhl R (2011) Sex differences in autoimmune diseases. Biol
177. https://doi.org/10.15887/j.cnki.13-1389/r.2016.11.122 50. Sex Differ 2(1):1. https://doi.org/10.1186/2042-6410-2-1
Wu Y, Hou X, Feng Y (2016) Research progress on children
with autism and food intolerance. Chin J Child Health Care
24(10): 1052–54+58 Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Hong Li1 & Haixia Liu2 & Xin Chen2 & Jian Zhang2,3 & Guanglei Tong1 & Yehuan Sun2,4
1 3
Department of Rehabilitation Medicine, Anhui Provincial Children’s Department of Neonatology, Anhui Provincial Children’s
Hospital/Children’s Hospital of Anhui Medical University, Hospital/
Hefei 230051, Anhui, China Children’s Hospital of Anhui Medical University,
2 4 Hefei 230051, Anhui, China
Department of Epidemiology and Health Statistics, School of Public
Health, Anhui Medical University, No. 81 Meishan Road, Center for Evidence-Based Practice, Anhui Medical University,
Hefei 230032, Anhui, China No.
81 Meishan Road, Hefei 230032, Anhui, China