J Autism Dev Disord

DOI 10.1007/s10803-013-1771-5

ORIGINAL PAPER

Feeding Problems and Nutrient Intake in Children with Autism

Spectrum Disorders: A Meta-analysis and Comprehensive Review
of the Literature
William G. Sharp • Rashelle C. Berry • Courtney McCracken •

Nadrat N. Nuhu • Elizabeth Marvel • Celine A. Saulnier •

Ami Klin • Warren Jones • David L. Jaquess

Ó Springer Science+Business Media New York 2013

Abstract We conducted a comprehensive review and Introduction

meta-analysis of research regarding feeding problems and
nutrient status among children with autism spectrum dis- Autism spectrum disorders (ASD) represent a range of
orders (ASD). The systematic search yielded 17 prospec- complex developmental disabilities involving severe
tive studies involving a comparison group. Using rigorous impairments in social interaction and communication
meta-analysis techniques, we calculated the standardized accompanied by behavioral inflexibility, repetitive behav-
mean difference (SMD) with standard error and corre- iors, and/or restricted interests (APA 2000). In addition to
sponding odds ratio (OR) with 95 % confidence intervals the core diagnostic features, children with ASD often
(CI). Results indicated children with ASD experienced present with comorbid ear infections (Konstantareas and
significantly more feeding problems versus peers, with an Homatidis 1987), increased use of antibiotics (Niehus and
overall SMD of 0.89 (0.08) and a corresponding OR of Lord 2006). constipation (Ibrahim et al. 2009), possible
5.11, 95 % CI 3.74–6.97. Nutrient analyses indicated sig- gastroenterological disturbances (Horvath et al. 1999), and
nificantly lower intake of calcium (SMD: -0.65 [0.29]; an array of challenging behaviors, including self-injury,
OR: 0.31, 95 % CI 0.11–0.85) and protein (SMD: -0.58 severe tantrums, feeding problems, aggression, toileting,
[0.25]; OR: 0.35, 95 % CI: 0.14–0.56) in ASD. Future and sleep disturbances (Whiteley 2004; Herzinger and
research must address critical questions regarding the Campbell 2007; Seiverling et al. 2010). Of these concerns,
cause, long-term impact, and remediation of atypical feeding arguably involves the most essential of human
feeding in this population. activities, necessary to assure appropriate development and
sustain life. Chronic feeding problems place children at risk
Keywords Diet
Food selectivity
Mealtime problems
for a number of detrimental medical and developmental
Nutrition
Picky eating
Pediatric feeding disorders outcomes, including malnutrition, growth retardation,
invasive medical procedures (e.g., placement of a feeding
tube), developmental delays, psychological and social
deficits, and poor academic achievement (Kerwin 1999;
Sharp et al. 2010). Researchers, however, have only
recently begun to systematically investigate eating and
nutrient intake patterns associated with ASD, and many
W. G. Sharp
C. McCracken
C. A. Saulnier
A. Klin
questions remain regarding prevalence, consequences, and
W. Jones
D. L. Jaquess
remediation of feeding problems in this population.
Department of Pediatrics, Emory University School of Medicine,
Atlanta, GA, USA Lack of research on this topic is remarkable given the
historical link between feeding and ASD. Leo Kanner’s
W. G. Sharp (&)
R. C. Berry
N. N. Nuhu
E. Marvel
initial description of the condition cited atypical eating
C. A. Saulnier
A. Klin
W. Jones
D. L. Jaquess
patterns as prominent in his sample and past diagnostic
The Marcus Autism Center, 1920 Briarcliff Road,
Atlanta, GA 30329, USA systems included feeding difficulties as a defining charac-
e-mail: [email protected] teristic (Kanner 1943; Ritvo and Freeman 1978). Further,

123
 J Autism Dev Disord

the social and behavioral demands of feeding situations tap In a more recent review, Cermak et al. (2010) identified
into all three areas of difficulty displayed by children with studies investigating food selectivity and nutrient adequacy
ASD. Communication, behavioral flexibility, and social in ASD. The authors identified 817 participants in 16
engagement each play important roles in promoting intake, studies with two foci: food selectivity (nine studies) or
increasing dietary diversity, and assuring the saliency of nutritional status related to dietary intake (four studies);
social reinforcement during meals. Related theories reflect three studies spanned both areas of inquiry. Findings sug-
this connection, with different authors positing different gested food selectivity was a significant problem in ASD;
etiologies, including idiosyncratic focus on detail, behav- however, Cermak et al. cited the lack of a comparison
ioral rigidity, sensory impairments, social skills deficits, group, present in only 6 of the 12 studies, as a key limi-
and/or communication deficits (Cumine et al. 2000; Ahearn tation to drawing definitive conclusions. Findings regard-
et al. 2001). Finally, research regarding feeding problems ing the nutritional status of children with ASD were
in ASD and related dietary vulnerabilities has important equivocal. Four studies involving comparison groups
implications for a growing interest regarding the use of reported conflicting results, with the nutrient intake of
dietary manipulation (e.g., gluten and/or casein free, GFCF children with ASD described as below, above, or at the
diet) in this population, as well as the possible role of same level as typically developing peers. Three remaining
dietary insufficiencies in the pathology of the condition, studies comparing the nutrient intake of children with ASD
such as vitamin D (Cannell 2008). to recommended dietary standards also reported both
Much of what is known regarding feeding patterns in nutrient deficits (e.g., vitamin D) and excesses (e.g., pro-
ASD is based on anecdotal and case reports describing tein); however, no consistent pattern emerged, and lack of
children with ASD as presenting with unusual eating pat- comparison groups precluded conclusions as to whether a
terns, rituals regarding food preparation/presentation, food deviation from recommended levels was unique to ASD.
refusal and/or displaying strong emotional responses to The works of Ledford and Gast (2006) and Cermak et al.
new foods (Cornish 1998; Ahearn et al. 2001). Food (2010) provide an important foundation for understanding
selectivity (by type, texture, and/or presentation) is the feeding concerns and nutritional status of children with
feeding problem most often associated with ASD, typically ASD, offering provisional evidence that feeding problems
involving strong preferences for carbohydrates, snacks, may be endemic in the ASD population. Recent growth in
and/or processed foods while rejecting fruits and vegeta- research into feeding in ASD, combined with the avail-
bles (Ahearn et al. 2001; Schreck et al. 2004; Williams ability of quantitative procedures for synthesizing outcome
et al. 2005). Many past reports, however, documenting this data, present the opportunity for a more detailed analysis of
trend involved children seeking intervention for severe the extant literature. The current review sought to (a) sur-
food selectivity, often in the form of behavioral interven- vey the medical, habilitative, and psychological literature
tion aimed at expanding dietary variety (e.g., Sharp et al. in order to identify studies using empirical methods to
2010), and a more general picture regarding the eating investigate the feeding behaviors and/or nutritional status
patterns and nutritional status of all children with ASD has of children with ASD and (b) summarize the evidence on
yet to emerge. the basis of both descriptive and meta-analytic procedures.
Ledford and Gast (2006) conducted the first literature To address limitations noted in previous reviews, we
review of feeding problems in ASD, identifying seven focused exclusively on prospective research involving a
studies (381 total children) published between 1994 and comparison group to quantify the magnitude of feeding
2004. All studies reported significant feeding difficulties, problems and/or nutrient deficiencies associated with ASD
primarily in the form of food selectivity by type and/or and used this information to develop ASD-specific rec-
texture, with estimates ranging from 46 to 89 % of children ommendations to guide future clinical and research activ-
with ASD with atypical feeding habits. While providing ities in this area.
evidence of widespread feeding problems, large variability
in prevalence estimates reflected wide methodological
variability among the studies. Less than half of the studies Method
included a comparison group, and the primary method of
data collection involved chart audits or study specific Study Identification and Eligibility Criteria
questionnaires. In addition, few studies presented infor-
mation regarding participants’ definitive diagnostic status Following the guidelines outlined by the Preferred
(i.e., autistic disorder, PDD-NOS, Asperger syndrome), Reporting Items for Systematic Reviews and Meta-Anal-
with 85 % having no specific ASD diagnosis and no yses (PRISMA) statement, we searched MedLine, PsychI-
standardized assessment of disability which limits gener- NFO, and PubMed databases (January 1980 and August
alizability of the findings. 2011), reviewed reference lists, and conducted ancestral

123
J Autism Dev Disord

and online searches in English language journals for eli- procedures, feeding/nutrient assessment measures, and
gible studies. The search parameters included combinations summary of findings. Characteristics in each of these cate-
of key words regarding the target population (autism, gories were coded using a standardized checklist system. For
autistic, autism spectrum disorders, pervasive develop- feeding behaviors, we categorized item(s) and/or assess-
mental disorder, PDD-NOS, Asperger syndrome), meal- ment measure(s) and their content based in three categories:
time-related variables (diet, dietary intake, eating, feeding, food selectivity (e.g., by type, texture, or presentation),
food selectivity, nutrition, mealtime behaviors, pediatric food refusal (e.g., refusing food by crying, pushing away
feeding disorder), and evaluation methodology (assess- food, leaving the table)/poor oral intake (concerns regarding
ment, mealtime observation, food frequency). total calories or nutrients consumed), and/or behavioral
We focused on prospective studies utilizing a comparison rigidity during meals (e.g., difficulty eating across environ-
group to present quantitative information about feeding ments, insists on rituals at table). If food selectivity was
behaviors and/or nutrient intake in a pediatric population reported, we documented whether the pattern of food intake
(birth to 18 years of age) with ASD and sought to capture a was analyzed (e.g., preference or rejection of certain types).
wide range of children regardless of the presence of feeding For dietary information, data collection focused on the fol-
related difficulties. As a result, we excluded recent program lowing key dietary indicators: vitamin A, C, D, & E, zinc,
evaluations (Laud et al. 2009; Sharp et al. 2011) single- calcium, iron, fiber, fat, protein, carbohydrates, and total
subject designed studies (see Sharp et al. 2010 for a sum- energy (kcal). When available, we also recorded nutritional
mary), and chart reviews of children with ASD evaluated due risk based on the cut point method (Barr et al. 2002), a dif-
to atypical feeding patterns (Williams et al. 2005) in order to ferent approach to assessing dietary status that involves
avoid a known sampling bias. This procedure also excluded calculating an individual’s typical intake of each nutrient,
studies focusing on the impact of dietary manipulation (e.g., identifying the total number of nutrients falling within
GFCF diet) on nutrition or behavioral functioning (e.g., established standards (e.g., estimated average requirement),
Elder et al. 2006). To be included in the review, studies also and determining the proportion of children in each group
needed to meet the following criteria: meeting or not meeting recommended levels. The research
team involved a registered dietician, who was responsible for
1. Evaluated feeding through a standardized, replicable
calculating nutritional risk, as well as selecting and inter-
manner, such as dietary intake (e.g., 3-day food diary),
preting specific dietary indicators. To determine growth
feeding questionnaires [e.g., Children’s Eating Behav-
status, we also recorded anthropometric data (i.e., height,
ior Inventory-Revised (CEBI-R); Archer et al. 1991],
weight, body-mass index) when presented.
Brief Autism Mealtime Behavior Inventory (BAMBI;
Multiple researchers independently coded all studies. The
Lukens and Linscheid 2008), study specific question-
mean inter-rater agreement for categorical data was 97 %
naires involving set questions, and/or mealtime obser-
(range 87–100 %) with a corresponding Kappa of 0.94
vation with a detailed protocol.
(range 0.79–1). The overall intra-class correlation for inter-
2. Included a dependent variable(s) focused on feeding
val and continuous data was 0.93 (range 0.54–1). Coder
behavior (i.e., chronic food selectivity, food refusal/
agreement exceeded the 80 % standard widely adopted and
poor oral intake, and/or behavioral rigidity during
recommended during quantitative synthesis of research
meals), nutritional status, or dietary variety. Data
(Campbell 2003). Due to the wide range of assessment
obtained through these measures was presented in the
methods and item content related to feeding behaviors in
study, either descriptively (e.g., frequencies, percent-
ASD, two members of the research team with expertise in
ages, means) or statistically (e.g., p values, t scores).
autism and pediatric feeding disorders conducted a third
3. Focused on active chronic feeding concerns (i.e., not
level review of all extracted data to determine inclusion
studies or items pertaining to historical concerns alone
status and classify item/scale content based on the criteria
such as feeding during infancy, difficulty transitioning to
outlined above. The mean inter-rater agreement for items to
solids).
analyze was 91 % (range 66–100 % across studies) with a
corresponding Kappa of 0.80 (range 0.33–1).
Variables Coded, Data Extraction, and Reliability
Statistical Analysis
Data were extracted from articles using a three-phase
system. First, all articles identified through the literature To calculate the effect size (ES), we used means (standard
were screened for eligibility criteria. We then extracted deviations) or frequencies (percentages) and, if necessary,
descriptive information, collecting information regarding we estimated the ES from test statistics (e.g., Chi Square,
study descriptors, participant demographic variables, t tests). When summary statistics were not presented, we
composition of the comparison group(s), diagnostic attempted to contact the primary author via email before

123
 J Autism Dev Disord

using alternative methods and, if unsuccessful, we used outcomes using the funnel plot (Egger et al. 1997), failsafe
exact p values to calculate the ES. If an exact p value was N (Becker 2005), and the trim and fill method (Duval and
not provided, we adopted a conservative approach to esti- Tweedie 2000).
mating a p value closest to the level provided (Lipsey and
Wilson 2001).
The primary goal of the meta-analysis was to determine Results
the overall difference in feeding behavior and/or nutritional
status between children with and without ASD. We, Characteristic of Studies and Participants
therefore, calculated an overall mean study ES when
multiple comparisons had been made (Rosenthal and Rubin The search yields 17 articles meeting inclusion criteria out
1986). In line with these criteria, we combined outcome of a pool of 678 possible studies (see Fig. 1). Sixty-five
variables (e.g., food selectivity, food refusal), resulting in a percent of the articles were published in journals special-
single ES. Likewise, when studies separately presented izing in ASD or related DD, with the Journal of Autism and
individual items, or individual subscales, along with total Developmental Disorders contributing nearly half of the
scale scores, only items or scales pertaining to these criteria studies in this area (see Table 1). Ten articles (88 %) were
were used in the present analysis. For nutrient data, we published since 2000; five since 2010. Compared with
calculated a separate ES estimates for each nutrient across previous reviews, 2 of the 7 studies (29 %) identified by
studies. For studies involving multiple comparison groups, Ledford and Gast (2006) and 6 of the 16 studies (36 %)
we pooled the comparison groups, producing an overall ES. summarized by Cermak et al. (2010) met inclusion criteria.
Separate ES estimates for each comparison group [i.e., Ten articles were unique to this review.
ASD vs. typically developing peers (TD); ASD vs. siblings Feeding assessment methods primarily involved esti-
(SIB); ASD vs. children with other developmental dis- mates of nutrient intake (e.g., 3-day food diary) and
abilities (DD)] were also calculated to identify possible questionnaires specific to the study involving single item
moderator variables using the between groups Q test, with analysis (Table 2). Standardized questionnaires were uti-
a significance level of p \ 0.05. We did not conduct lized in only three studies (18 %). While most studies
additional analysis of potential moderators (e.g., age, sex, broadly assessed feeding behaviors using single items or
diagnostic status) given the lack of descriptive data pre- scales (e.g., eats a narrow range of foods, doesn’t try new
sented in the articles (described below). foods), three studies (Bandini et al. 2010; Emond et al.
Data were entered and analyzed using Comprehensive
Meta-Analysis 2 (Borenstein et al. 2005). We converted all
ES estimates to standardized mean difference (SMD). For Studies identified from
feeding behaviors, a positive SMD (SMD [ 0) indicated search N = 678
more feeding-related concerns in children with ASD com-
pared with the comparison group. We coded nutritional data Excluded on the basis of
so that a negative SMD (SMD \ 0) indicated more nutri- abstract n = 618
tional deficits in children with ASD. The point estimates and
standard error were calculated using a random-effects model Studies with potentially
of meta-analysis (Hunter and Schmidt 2004). We evaluated eligible abstracts n = 60
SMD magnitude using conventional standards (0.2 = small;
0.5 = medium; 0.9 = large; Cohen 1988). To aid in clinical Excluded on the basis of
interpretation of outcomes, we also calculated the corre- full text because study did
sponding odds ratio (OR) with 95 % CIs, with values not assess feeding behaviors
or nutrient intake n = 31
reflecting the odds of a child with ASD having a feeding Studies presenting data on
difficulty compared to child without ASD. feeding behaviors or
nutrient intake N = 29
To assess heterogeneity within subgroups and between
studies, effect sizes and associated 95 % CIs were calcu- Excluded because study did
not include a control group
lated for each subgroup. We also used the Q test to for- or focused on children with
mally determine if heterogeneity was present. To assess the ASD seeking assessment
robustness of our results, we conducted a sensitivity anal- and treatment for feeding
problems n = 12
ysis, which involved repeatedly calculating the effect size Studies included in the
with one study omitted per iteration and comparing the systematic review N = 17
results with the overall study effect. We analyzed the threat
of possible publication bias to the validity of the obtained Fig. 1 Flow diagram of included and excluded studies

123
J Autism Dev Disord

Table 1 Summary of articles by journal and year of publication Growth Status

Journal title n %
Seven studies (41 %) involving 426 children presented
Journal of Autism and Developmental Disorders 7 41 information regarding anthropometric parameters [e.g.,
Autism 1 5.9 height, weight, or body mass index (BMI)] compared with
Journal of the American Dietetic Association 1 5.9 typically developing peers. Six studies compared mean
Journal of Developmental and Physical Disabilities 1 5.9 values between groups, finding no statistically significant
Journal of Learning Disabilities 1 5.9 differences in anthropometric parameters. One study ana-
The Journal of Pediatrics 1 5.9 lyzed the percentage of children in each group identified as
Pediatrics 1 5.9 overweight (BMI C85th %) or underweight (BMI \5th %)
Physical & Occupational Therapy in Pediatrics 1 5.9 and reported no difference in the number of children falling
Research in Autism Spectrum Disorders 1 5.9 into these classifications (Bandini et al. 2010).
Research in Developmental Disabilities 1 5.9
Topics in Clinical Nutrition 1 5.9 Dietary Variety
Total 17 100
Year published Ten studies presented detailed food group preferences, six
2010–present 5 29 of which supported past reports indicating children with
2000–2009 10 59 ASD consumed fewer vegetables (Lukens and Linscheid
1980–1989 2 12 2008; Johnson et al. 2008; Martins et al. 2008; Bandini
et al. 2010; Emond et al. 2010) and fruits (Lukens and
Linscheid 2008; Martins et al. 2008; Emond et al. 2010), as
2010; Zimmer et al. 2012) calculated a dietary variety well as demonstrated preference for crispy/crunchy snack
score based on responses to a food frequency question- foods (Schmitt et al. 2008). One study reported signifi-
naire, focusing on the number of foods identified as never cantly fewer accepted foods across all food groups in
consumed. Food selectivity represented the primary type of children with ASD (Schreck et al. 2004), while another
feeding problem assessed among studies, representing study reported lower variety of dairy but equivalent variety
54 % of the items or scales (e.g., eats a narrow variety of of other food groups (Shearer et al. 1982). Raiten and
foods, obsessive eating habits), 21 % of the items or scales Massaro (1986) found no significant difference in food
focused on food refusal or poor oral intake (e.g., disrup- groups consumed, and Herndon et al. (2009) reported
tions/tantrums during meals, throw or spits food), 17 % increased intake of fruit among children with ASD but
assessed behavioral rigidity during meals (e.g., eats only in equivalent intake of other food groups.
specific places, requires specific utensils), and 7 %
involved overlapping content. Overall Measure of ES for Feeding Behaviors
The pool of studies involved a total sample of 881 and Nutritional Intake
children with ASD. Data regarding feeding behaviors were
gathered from 832 (94 %) participants, while 263 children Tables 4 and 5 present ES estimates calculated using ran-
(30 %) from eight studies provided data on micronutrient dom effects models. The overall test for heterogeneity of
intake. Only 29 % of studies presented data regarding study effect sizes was statistically significant (Q = 29.4,
diagnostic status, resulting in 669 participants (76 %) with df = 14, p = 0.009) indicating that the random effects
a nonspecific ASD diagnosis. In terms of comparison model was appropriate. The presences of heterogeneity
groups, most studies (82 %) involved typically developing within subgroups further supported the use of the random
peers or children drawn from the general population, fol- effects model.
lowed by studies involving children with developmental or All studies reported greater levels of feeding concerns
learning disabilities (18 %) or siblings (18 %). Most associated with ASD, regardless of the type of comparison
studies (82 %) reported equivalence between ASD and group or method of assessment. SMD estimates across
comparison groups in terms of age. Two studies (12 %) did studies ranged from 0.48 to 1.56 (Fig. 2) and the overall
not statistically analyze possible age difference across SMD involving all comparison groups was large and sta-
groups, while one study reported that the ASD group was tistically significant (p \ 0.001). Analyses involving indi-
significantly older. In terms of gender, the ASD groups vidual comparison subtypes suggested medium to large
tended to involve a higher ratio of males to females com- differences in feeding problems, ranging from a SMD of
pared to the comparison groups; four studies (24 %) sta- 0.69 (0.19) when the comparison group involved children
tistically analyzed this variable, all reporting higher with DD to 0.97 (0.22) when siblings were compared. The
numbers of males to females in the ASD groups (Table 3). corresponding overall OR involving all comparison groups

123
 Table 2 Description of experimental characteristics and assessment methodology by study
Study

123
Bandini Collins Dominick Emond Herndon Johnson Lockner Luckens Martins Matson
et al. et al. et al. et al. et al. et al. et al. and Linsheid et al. et al.
(2010) (2003) (2007) (2010) (2009) (2008) (2008) (2008) (2008) (2009)

Outcomes presented
Feeding behaviors/Food selectivity X X X X X X X X X
Micronutrient analysis of dietary intake X X X X X
Setting
Community wide X X X X X X X
Diagnostic clinic/Early intervention X X X
Other
Feeding measure(s)*
Standardized questionnaires X X
Estimates of nutritional intake X X X X X X
Subtypes: Food diary o o o
24 h recall o o
Food frequency inventory o o o o
Study specific questionnaire X X X X X X X
ASD Diagnostic indicator
Parent report X
ASD rating scale X X
Clinical provider X X X
ADOS X X X
ADI-R X X X
Not specified X X
Cognitive functioning/IQ X Xa Xa
Anthropometric data*
Weight X X
Height X X
BMI X X X
J Autism Dev Disord
 Table 2 continued
Study
Nadon Provost Raiten Schmitt Schreck Shearer Zimmer N % of total
et al. et al. and Massaro et al. et al. et al. et al. studies
(2011) (2010) (1986) (2008) (2004) (1982) (2012) (17 total)
J Autism Dev Disord

Outcomes presented
Feeding behaviors/Food selectivity X X X X X X 15 88
Micronutrient analysis of dietary intake X X X 8 47
Setting
Community wide X X X 10 59
Diagnostic clinic/Early intervention X X 5 29
Other X X 2 12
Feeding measure(s)*
Standardized questionnaires X 3 18
Estimates of nutritional intake X X X X X 11 65
Subtypes: Food diary o o o 6 35
24 h recall 2 12
Food frequency inventory o o 6 35
Study specific questionnaire X X X X 11 65
ASD Diagnostic indicator
Parent report
ASD rating scale X 3 18
Clinical provider X X 5 29
ADOS X 4 24
ADI-R X 4 24
Not specified X X X 5 29
Cognitive functioning/IQ 3 18
Anthropometric data*
Weight X X 4 24
Height X X 4 24
BMI X X 5 29
* Subheadings may not add up to 100 % due to multiple measures used in a study
a
Data only presented for ASD group

123
 Table 3 Description of participants
Study

Bandini Collins Dominick Emond Herndon Johnson Lockner Luckens Martins Matson

123
et al. et al. et al. et al. et al. et al. et al. and Linsheid et al. et al.
(2010) (2003) (2007) (2010) (2009) (2008) (2008) (2008) (2008) (2009)

ASD group
Sample size 53 107 54 79 46 19 20 68 41 112
ASD diagnosis X X
Autistic disorder 45 72
PDD–NOS 1 40
Asperger syndrome
Age (months) X X X X X X X X X X
Mean 79.2 96 91.2 6, 15, 24, 38, 54* 55.9 39.2 72.8 85.2
SD 25.2 43.3 29.8 13.9 8.9 29.8 34.4
Range 36–216 33–96 24–48 36–60 36–132 36–132 36–192
Gender X X X X X
Male (%) 44 (83 %) 47 (87 %) 44 (96 %) 56 (82 %) 34 (83 %)
Female (%) 9 (17 %) 7 (13 %) 2 (4 %) 12 (18 %) 7 (17 %)
Comparison group
Sample size 53a 331 (DD: 262; SB: 69) 38a 12,901a 31a 15a 20a 40a 55a (TD: 41; SB: 14) 167a
Subtype**
TD X X X X X X X X
DD X X X
SB X X
Age (months) X X X X X X X X
Mean 80.4 DD: 95.8; SIB: 99.4 95.3 6, 15, 24, 38, 54* 59.9 36.4 72.8
SD 28.8 DD: 50.2; SIB: 45.1 33.1 16.5 9.46 29.8
Range DD: 24–216.9; SIB: 24–216 24–48 36–132 TD: 12–48; SIB: 24–132
Gender X X X Xc X
Male (%) 45 (78 %) 21 (71 %) 23 (74 %) 20 (50 %) TD: 23 (56 %) SIB: 7 (50 %)
Female (%) 13 (22 %) 11 (29 %) 8 (26 %) 20 (50 %) TD: 18 (44 %); SIB: 7 (50 %)
J Autism Dev Disord
 Table 3 continued
Study

Nadon Provost Raiten Schmitt Schreck Shearer Zimmer Total % of total

et al. et al. and Massaro et al. et al. et al. et al. sample/N studies
(2011) (2010) (1986) (2008) (2004) (1982) (2012) (17 total)
J Autism Dev Disord

ASD group
Sample size 48 24 40 20 138 12 22 881
ASD diagnosis X X X 5 29
Autistic disorder 10 12 22 5 29
PDD–NOS 3 3 18
Asperger syndrome 4 1 6
Age (months) X X X X X X X 17 100
Mean 94.8 51.2 127 99 96 98.4 14 82
SD 30 10.6 52 29 9.6 38.4 13 76
Range 45.6–154.8 36–70 84–120 53–152 11 65
Gender X X X X X X 11 65
Male (%) 44 (92 %) 18 (75 %) 28 (70 %) 20 (100 %) 121 (88 %) 20 (91 %)
Female (%) 4 (8 %) 6 (25 %) 12 (30 %) 14 (10 %) 2 (9 %)
Comparison group
Sample size 48a 24a 34b 18a 298a 12 20a 13,544
Subtype**
TD X X X X X X 14 82
DD 3 18
SB X 3 18
Age (months) X X X X X X X 15 88
Mean 92.4 51.2 105.6 108 100.8 97.2 13 76
SD 34.8 9.8 57.6 7.2 39.6 11 65
Range 37.2–153.6 36–72 84–120 60–144 8 47
Gender Xc X Xc X Xc X 11 65
Male (%) 20 (42 %) 18 (75 %) 19 (56 %) 18 (100 %) 158 (53 %) 10 (45 %)
Female (%) 28 (58 %) 6 (24 %) 15 (44 %) 140 (47 %) 12 (55 %)

TD typically developing, DD other developmental delay, SB siblings

* Longitudinal design
** Subheadings may not add up to 100 % due to multiple groups used in a study
a
Reported matched for age
b
Reported age difference with ASD group
c
Reported higher ratio of males to female in ASD group

123
 J Autism Dev Disord

Table 4 Effect sizes, 95 % confidence limits and within-group tests for heterogeneity for studies included in the meta-analysis for feeding
behavior problems by comparison groups
ASD versus Number of Random effects model Within-groups
subgroup contributing
studies SMD (SE) OR 95 % confidence limits p value v2 test (Q) p value
LCL UCL

All groups 15 0.89 (0.08) 5.11 3.74 6.97 \0.001

TD 13 0.94 (0.11) 5.49 3.77 7.98 \0.001 29.9 0.003
SB 3 0.98 (0.22) 5.89 2.73 12.71 \0.001 0.45 0.798
DD 2 0.67 (0.19) 3.36 1.69 6.67 0.001 0.012 0.913
TD typically developing, DD other developmental delay, SB siblings

Table 5 Effect sizes, 95 % confidence limits and within-group tests for heterogeneity for studies included in the meta-analysis for nutritional
data
Nutrient Number of Random effects model
contributing
studies SMD (SE) OR 95 % confidence limits p value
LCL UCL

Calcium 8 -0.65 (0.29) 0.31 0.11 0.85 0.022

Carbohydrates 7 -0.02 (0.07) 0.97 0.76 1.24 0.810
Energy 6 0 (0.06) 0.99 0.80 1.25 0.995
Fiber 6 0.09 (0.12) 1.18 0.77 1.78 0.448
Iron 7 0.17 (0.20) 1.35 0.66 2.76 0.414
Protein 7 -0.58 (0.25) 0.35 0.14 0.86 0.021
Total fat 6 0.03 (0.06) 1.05 0.84 1.30 0.690
Vitamin A 6 -0.51 (0.35) 0.39 0.11 1.37 0.143
Vitamin C 7 -0.13 (0.19) 0.98 0.52 1.87 0.507
Vitamin D 6 -0.07 (0.19) 0.88 0.45 1.71 0.703
Vitamin E 5 0.05 (0.17) 1.10 0.61 1.98 0.742
Zinc 6 -0.03 (0.09) 0.95 0.69 1.31 0.758

was 5.11 (95 % CI 3.74–6.97), suggesting that the odds of each study. No study significantly altered the overall mean
having a feeding problem in children with ASD are 5 times ES estimates for feeding behaviors or nutrient intake.
the odds for children without ASD. Visual inspection of the funnel plots indicated no potential
Analyses involving nutritional data suggested children publication bias for outcome related to feeding behaviors
with ASD had significantly lower consumption of calcium or analyses involving calcium or protein. Furthermore, the
(p \ 0.05) and protein (p \ 0.05) compared to TD peers. failsafe N analysis indicated that there would need to be
No other significant differences in nutrient consumption 858 published studies with non-significant findings related
were detected between groups. Bandini et al. (2010) and to feeding behaviors to change the current effect size to
Zimmer et al. (2012) also assessed risk of inadequate non-significant. The failsafe N was 37 for the calcium
nutrient intake using the cut point method; both reporting intake outcome was 33 for the protein intake outcome. This
children with ASD were significantly more likely to have evidence lends credence to the robustness of our findings.
inadequacies compared to TD children (p \ 0.03). Given
the small sample of studies, we did not estimate an ES for
cut point data. Discussion

Sensitivity Analysis, Publication Bias and Reliability This meta-analysis shows a strong association between
of Results feeding difficulties and ASD, corroborating anecdotal
reports and descriptive studies documenting this trend.
Sensitivity analysis involved visual inspection of confi- While previous reviews summarized the literature, this
dence intervals for the overall effect size after removing systematic evaluation of the research base quantifies the

123
J Autism Dev Disord

Study Name OR (95 % CI) OR (95 % CI)

Bandini et al. (2010) 4.91 (2.40 - 10.1)

Collins et al. (2003) 4.08 (1.79 - 9.24)
Dominick et al. (2007) 16.8 (5.76 - 49.1)
Emond et al. (2010) 2.39 (1.79 - 3.19)
Johnson et al. (2008) 4.79 (1.29 - 17.8)
Lockner et al. (2008) 10.5 (1.31 - 84.3)
Luckens & Linsheid (2008) 8.45 (3.93- 18.2)
Martins et al. (2008) 5.84 (2.92 - 11.7)
Matson et al. (2009) 5.18 (2.55 - 10.5)
Nadon et al. (2011) 4.56 1.29 - 16.1)
Provost et al. (2010) 4.48 (0.71- 28.4)
Raiten & Massaro (1986) 3.07 (0.59- 15.8)
Schmitt et al. (2008) 4.28 (0.99- 18.4)
Schreck et al. (2004) 4.72 (3.24 - 6.89)
Zimmer et al. (2012) 10.7 (3.29 - 34.9)
Mean ES 5.11 (3.74 - 6.97)

Fig. 2 Forest plot of feeding problems with 95 % confidence intervals

magnitude of effect. By conventional standards, findings regarding the impact of aberrant feeding patterns on health
reflect a ‘‘large’’ difference in the presence of feeding and developmental in the ASD population.
problems between children with and without ASD, corre- When considering the impact of chronic feeding prob-
sponding with an estimated fivefold increase in the odds of lems, growth and nutrition represent key barometers of
having a feeding problem in this population. Higher rates health status. Findings from the current review, however,
of feeding problems were detected regardless of the make- indicate that feeding problems and subsequent nutritional
up of the comparison group or the assessment methodol- intake deficits do not necessarily translate into greater risk
ogy, providing convergent evidence that feeding problems for compromised growth. All seven studies analyzing
are more likely to occur in children with ASD. This sug- growth parameters reported no significant difference in
gests, at a minimum, assessment of feeding problems in height, weight, and/or BMI between children with and
ASD should be included as part of routine screenings in without ASD. This parallels nutrient data indicating com-
pediatric settings, which would necessitate enhanced parable intake of energy, carbohydrates, and fats when
awareness among caregivers and practitioners regarding compared to typically developing peers. This suggests,
this issue. Encouragingly, the pool of studies included in despite increased feeding problems, children with ASD
this review reflects a relative surge in case–control pro- apparently consume enough volume of food to meet gross
spective research of feeding problems in ASD, with more energy needs and relying exclusively on anthropometric
than a quarter of the studies published since 2010. Despite parameters to assess health status may in fact mask
greater empirical attention in this area, a closer examina- underlying nutritional deficits. It may also explain why
tion suggests a sizable gap between studies of feeding and feeding problems are often overlooked in relation to other
other areas of inquiry. For example, even after removing area of clinical concern in the ASD population, since
our conservative inclusion criteria, the largest source of failure to thrive or a declining growth velocity are the
articles in the current review, Journal of Autism and standard nutritional health indicators (WHO 2006) that
Developmental Disorders, published only nine studies on trigger clinical attention in pediatric settings (Ledford and
feeding problems in ASD between 1980 and 2011; this Gast 2006). Closer examination of nutrient intake, how-
represents 0.3 % of the 2,485 articles published in the ever, indicates significant specific deficits (lower intake of
journal over the same time period. Further, only two calcium and protein) and a higher number of nutritional
studies were published in pediatric journals despite the deficits overall among children with ASD. These patterns
frontline role pediatricians play in screening and identify- may well place this population at risk for long-term med-
ing health concerns among children with ASD. Given the ical complications not captured by broad anthropometrics
significant level of feeding concerns associated with ASD or energy intake. For example, lower levels of calcium,
and the biological and social significance of healthy eating, compounded by the increased need for this nutrient during
greater clinical and research scrutiny in this area are clearly childhood to promote growth of bones, may portend risk of
needed to improve assessment methods, increase access osteomalacia and osteoporosis. This assertion is consistent
to treatment, and develop more definitive conclusions with findings indicating decreased bone cortical thickness

123
 J Autism Dev Disord

in a group of 75 boys with ASD when compared to peers, deficits (Lukens and Linscheid 2008). With this in mind,
highlighting the need to investigate the calcium intake and caregivers should employ utmost caution when deciding to
bone growth in children with ASD, as well as identify pursue this form of treatment. At a minimum, families
possible etiologies (Hediger et al. 2007). Together, the wishing to pursue a possible dietary intervention should do
available evidence suggests the need to look beyond gross so under the guidance of a healthcare professional (e.g.,
anthropometric parameters, such as incorporating idiosyn- registered dietitian) who can assess the impact of further
cratic analysis of nutritional intake as part of routine restrictions on a child’s nutritional status and work to
medical care in ASD. It will also be important to determine ensure the child’s nutritional needs are met during the
the long-term health burden associated with atypical pat- intervention. Similarly, untested interventions that may
terns of intake on a population level, particularly high secondarily affect nutritional status, such as chelation
consumption of snack and fats in ASD, which may portend therapy, could compound an already risky situation by
increased risk for diet-related diseases (e.g., obesity, car- further reducing the bioavailability of key nutrients, such as
diovascular disease) in adolescence or adulthood. calcium. Clearly, such potential iatrogenic effects should
Two candidates for explaining reduced nutrient intake in be carefully investigated prior to recommending any
ASD are food selectivity and/or elimination diets (e.g., the treatment. To assist caregivers with making an informed
GFCF diet). Only three studies in the current review, decision, pediatric practitioners must screen for preexisting
however, specifically investigated the relationship between feeding concerns, highlight the tenuous empirical support
restricted patterns of intake and nutritional status. Herndon for diet modification as treatments of ASD and review
et al. (2009) reported fewer servings of dairy and that this potential consequences (e.g., further nutritional deficits,
relationship remained after excluding children following a stigmatization, diversion of treatment resources; Mulloy
GFCF diet. Zimmer et al. (2012) excluded children on et al. 2010) and barriers (e.g., resources to purchase spe-
elimination diets and still reported that selective eaters with cialized foods, strategies for ensuring dietary compliance;
ASD had lower intake of calcium, vitamin B12, and vita- Elder 2008) associated with dietary interventions. Addi-
min D, compared to non-selective eaters with ASD and tional research will also be needed to more clearly eluci-
lower intake of protein, calcium, vitamin A, and vitamin D, date the impact of dietary manipulations on growth,
compared with typically developing peers. Finally, Bandini nutrition, and family resources.
et al. (2010) reported children with ASD experienced more The higher rate of feeding concerns in ASD also
nutrient inadequacies than typically developing children, a emphasizes a subsequent need to identify and disseminate
finding that persisted after excluding children on special empirically-supported treatments for feeding problems
diets. Together, there is evidence suggesting that nutri- associated with ASD. At this time, behavioral intervention
tional issues associated with ASD may be related to the represents the only empirically supported treatment for
patterns of food selectivity beyond what could be attributed pediatric feeding disorders Sharp et al. (2010) and there is
to parent-mediated dietary manipulations. Going forward, provisional evidence that these benefits apply to children
it will be important to control for the use of vitamin/min- with ASD. With this said, support for behavioral treatment
eral supplements, which may mask an even greater risk of to expand dietary variety has primarily been documented at
compromised dietary status among children with ASD. day-treatment or inpatient feeding programs (Laud et al.
Provisional evidence suggests higher use of supplements 2009; Sharp et al. 2011). Unfortunately, few inpatient and
among caregivers concerned about increased levels of food day-treatment programs exist, which curtails adequate
selectivity or food refusal (Yu et al. 1997), and parents of access to care. Given the need for feeding intervention in
children with ASD may be more likely to try dietary sup- this population, an important goal moving forward will be
plementation in general (Lockner et al. 2008). to develop additional treatment options, such as organizing
The combination of increased feeding problems and disciplines involved in providing care along clinical service
nutritional concerns raises important questions regarding lines and expanding training and educational opportunities
the use (and possible detrimental impact) of dietary for community providers regarding behavioral strategies
manipulations in the ASD population. Many of these diets for targeting food selectivity. It will also be important to
(e.g., the GFCF) eliminate dairy proteins, placing addi- determine whether intervention to address food selectivity
tional restrictions on a population vulnerable for lower in ASD can be adapted for delivery through less intensive
calcium intake, and provisional evidence suggests that this methods of service delivery, such as outpatient treatment,
may lead to greater deficits in bone development among group therapy, or caregiver training.
children with ASD (Hediger et al. 2007). Elimination diets This review also highlights important areas for future
also target starches and snack foods often identified as research to enhance understanding of feeding problems and
preferred foods among children with ASD, which may nutrient status in ASD (Table 6). More detailed diagnostic
increase the risk for weight loss and further nutritional characterization continues to be needed to better define

123
J Autism Dev Disord

Table 6 Summary of key recommendations for clinical and research nature of feeding problems associated with ASD remain
activities for feeding and ASD elusive. The current study collapsed food selectively and
In clinical settings, healthcare providers are encouraged to: food refusal under the larger umbrella of feeding problems
1. Include assessment of feeding problems as part of routine due to the heterogeneity of item content, which reflects a
medical evaluations more global need to develop consensus regarding the def-
2. Screen for nutritional deficits/excesses in addition to inition of specific feeding concerns (e.g., food refusal vs.
measurement of gross anthropometric parameters food selectivity) in the pediatric feeding disorder literature.
3. Engage in caregiver education regarding tenuous empirical Increased diagnostic clarity would, in turn, aid in the
support for diet modification as treatment of ASD
development of standardized feeding measures.
4. Review potential consequences of pursuing an elimination diet
Finally, findings also raise important questions regarding
with consideration to the child’s unique feeding and nutritional
presentation how best to measure the nutritional status of children with
To enhance the literature moving forward, researchers should seek to: ASD. Idiosyncratic food choices among selective eaters will
1. Include detailed diagnostic characterization (e.g., ADOS, ADI) to likely result in different patterns of nutrient deficiencies
confirm ASD status based on the core foods that comprise an individual’s diet,
2. Further develop assessment methods to quantify feeding which may explain conflicting results among past reports.
problems and nutrient status We recommend that studies present data regarding overall
4. Identify and disseminate empirically-supported treatments for group analysis of nutrient intake, as well as an individual
feeding problems in ASD analysis regarding number of deficiencies using the cut point
5. Determine the long-term health burden associated with atypical method. Research would also benefit from increased stan-
patterns of intake on a population level (e.g., obesity,
cardiovascular disease), as well as the relationship with other
dardization in the measurement of nutritional intake (e.g.,
areas of functioning (e.g., quality of life, gastrointestinal issues) food diary, 24 h recall), consistent documentation of
anthropometric data, and long-term assessment regarding
the stability of dietary patterns over time. Finally,
samples of children with ASD. The dearth of studies in this researchers are also encouraged to extend the net of inquiry
review that provided a well characterized sample utilizing to include additional related outcomes, including quality of
diagnostic measures, like the Autism Diagnostic Observa- life, family functioning, relationship with gastrointestinal
tion Schedule (ADOS; Lord et al. 2000) and Autism issues, impact on developmental and cognitive status, and
Diagnostic Interview (ADI; Lord et al. 1994) that have etiological factors influencing dietary preference in ASD.
been standards of best practice in research for over a
decade, is striking and further emphasizes the limitations of
our knowledge of feeding profiles in ASD. Without stan-
Conclusion
dardized measures across samples, questions regarding the
relationship between ASD symptomatology and feeding
Our results confirm that children with ASD have more
behaviors remain unanswered. The relationship of atypical
feeding problems compared with peers. We also found a
feeding and intellectual status in children with ASD also
trend of lower intake of calcium and protein on a popula-
remains unclear given the lack detailed psychometric data,
tion level, and higher levels of nutritional inadequacies in
but represents an important focus for future research.
ASD, detected via idiosyncratic analyses using the cut
There is also a clear need to develop a frontline feeding
point method. Provisional data suggest food selectivity
screening tool to support research which can also be effi-
contributes to nutritional concerns related to ASD outside
ciently applied during medical appointments. Outcomes
of parent-mediated restrictions. Clinicians are encouraged
summarized in this review primarily involved study-
to increase screening for feeding concerns in children with
specific single-item measures, which limit conclusions
ASD and to use this information when counseling care-
regarding prevalence and topography that can be drawn
givers interested in pursuing an elimination diet.
across studies. Specifically, prevalence rates varied
depending on the content of the item or assessment Conflict of interest This was an unfunded study and no member of
method, with estimates as high as 95 % of a sample the research team has a conflict of interest.
describe as resisting trying new foods (Lockner et al.
2008). This could explain—at least in part—the high var-
iability in prevalence estimates cited by Ledford and Gast References
(2006). Without increased standardization in the measure-
ment of feeding concerns, the true prevalence of feeding Ahearn, W. H., Castine, T., Nault, K., & Green, G. (2001). An
assessment of food acceptance in children with autism or
concerns in ASD populations remains unknown at this pervasive developmental disorder—Not otherwise specified.
time. In addition, definitive conclusions regarding the exact Journal of Autism and Developmental Disorders, 31, 505–511.

123
 J Autism Dev Disord

American Psychiatric Association. (2000). Diagnostic and statistical Journal of Autism and Developmental Disorders, 38(5),
manual of mental disorders (4th ed, Text Revision ed.). 848–856.
Washington, DC: American Psychiatric Association. Herndon, A. C., DiGuiseppi, C., Johnson, S. L., Leiferman, J., &
Archer, L. A., Rosenbaum, P. L., & Streiner, D. L. (1991). The Reynolds, A. (2009). Does nutritional intake differ between
children’s eating behavior inventory: Reliability and validity children with autism spectrum disorders and children with
results. Journal of Pediatric Psychology, 16(5), 629–642. typical development? Journal of Autism and Developmental
Bandini, L. G., Anderson, S. E., Curtin, C., Cermak, S., Evans, E. W., Disorders, 39, 212–222.
Scampini, R., et al. (2010). Food selectivity in children with Herzinger, C. V., & Campbell, J. M. (2007). Comparing functional
autism spectrum disorders and typically developing children. assessment methodologies: A quantitative synthesis. Journal of
Journal of Pediatrics, 157(2), 259–264. Autism and Developmental Disorders, 37, 1430–1445.
Barr, S. I., Murphy, S. P., & Poos, M. I. (2002). Interpreting and using Horvath, K., Papdimitriou, J. C., Rabsztyn, A., Drachenberg, D., &
the dietary references intake in dietary assessment of individuals Tildon, J. T. (1999). Gastrointestinal abnormalities in children
and groups. Journal of the American Dietetic Association, with autistic disorder. Pediatrics, 135, 559–563.
102(6), 780–788. Hunter, J. E., & Schmidt, F. L. (2004). Methods of meta-analysis:
Becker, B. J. (2005). Failsafe N or file-drawer number. In H. Correcting error and bias in research findings (2nd ed.).
P. Rothstein, A. J. Sutton, & M. Borenstein (Eds.), Publication Thousand Oaks, CA: Sage.
bias in meta-analysis: Prevention, assessment and adjustments Ibrahim, S. H., Voigt, R. G., Katusic, S. K., Weaver, A. L., &
(pp. 111–125). Chichester: Wiley. Barbaresi, W. J. (2009). Incidence of gastrointestinal symptoms
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2005). in children with autism: A population-based study. Pediatrics,
Comprehensive meta-analysis (2nd ed.). Englewood, NJ: Biostat. 124(2), 680–686.
Campbell, J. M. (2003). Efficacy of behavioral interventions for Johnson, C. R., Handon, B. L., Mayer-Costa, M., & Sacco, K. (2008).
reducing problem behavior in persons with autism: A quantita- Eating habits and dietary status on young children with autism.
tive synthesis of single-case research. Research in Developmen- Journal of Autism and Developmental Disorders, 20, 437–448.
tal Disabilities, 8, 165–178. Kanner, L. (1943). Autistic disturbances of affective contact. The
Cannell, J. J. (2008). Autism and vitamin D. Medical Hypotheses, 70, Nervous Child, 2, 217–250.
750–759. Kerwin, M. E. (1999). Empirically supported treatments in pediatric
Cermak, S. A., Curtin, C., & Bandini, L. G. (2010). Food selectivity psychology: Severe feeding problems. Journal of Pediatric
and sensory sensitivity in children with autism spectrum Psychology, 24, 193–214.
disorders. Journal of the American Dietetic Association, Konstantareas, M. M., & Homatidis, S. (1987). Ear infections in
110(1), 259–264. autistic and normal children. Journal of Autism and Develop-
Cohen, J. (1988). Statistical power analysis for the behavioral mental Disorders, 17(4), 585–594.
sciences (2nd ed.). New Jersey: Lawrence Erlbaum Associates. Laud, R. B., Girolami, P. A., Boscoe, J. H., & Gulotta, C. S. (2009).
Collins, M. R., Kyle, R., Smith, S., Laverty, A., Roberts, S., & Eaton- Treatment outcomes for severe feeding problems in children with
Evans, J. (2003). Coping with the usual family diet: Eating autism spectrum disorder. Behavior Modification, 33, 520–536.
behavior and food choices of children with Down’s syndrome, Ledford, J. R., & Gast, D. L. (2006). Feeding problems in children
autistic spectrum disorders or cri du chat syndrome and with autism spectrum disorders: A review. Focus on Autism and
comparison groups of siblings. Journal of Learning Disabilities, Other Developmental Disabilities, 21, 153–166.
7(2), 137–155. Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis.
Cornish, E. (1998). A balanced approach towards healthy eating in Thousand Oaks, CA: Sage.
autism. Journal of Human Nutrition and Dietetics, 11, 501–509. Lockner, D. W., Crowe, T. K., & Skipper, B. J. (2008). Dietary intake
Cumine, V., Leach, J., & Stevenson, G. (2000). Autism in the early and parents’ perception of mealtime behaviors in preschool-age
years. New York: David Fulton. children with autism spectrum disorder and in typically devel-
Dominick, K. C., David, N. O., Lainhart, J., Tager-Flusberg, H., & oping children. Journal of the American Dietetic Association,
Folstein, S. (2007). Atypical behaviors in children with autism 108, 1360–1363.
and children with a history of language impairment. Research in Lord, C., Rsis, S., Lambrecht, L., Cook, E. H., Jr, Leventhal, B. L.,
Developmental Disabilities, 28, 145–162. DiLavore, P. C., et al. (2000). The autism diagnostic observation
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot- schedule-generic: A standard measure of social and communi-
based method of testing and adjusting for publication bias in cation deficits associated with the spectrum of autism. Journal of
meta-analysis. Biometrics, 56(2), 455–463. Autism and Developmental Disorders, 30, 205–223.
Egger, M., Davey Smith, G., Schneider, M., & Minder, C. (1997). Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism diagnostic
Bias in meta-analysis detected by a simple graphical test. BMJ interview—Revised: A revised version of a diagnostic interview
(Clinical Research Ed), 315(7109), 629–634. for caregivers of individuals with possible pervasive develop-
Elder, J. H. (2008). The gluten-free casein-free diet in autism: An mental disorders. Journal of Autism and Developmental Disor-
overview with clinical implications. Nutrition in Clinical ders, 24, 659–685.
Practice, 23(6), 583–588. Lukens, C. T., & Linscheid, T. R. (2008). Development and
Elder, J. H., Shankar, M., Shuster, J., Theriaque, D., Burns, S., & validation of an inventory to assess mealtime behavior problems
Sherrill, L. (2006). The gluten-fee, casein-free diet in autism: in children with autism. Journal of Autism and Developmental
Results of a preliminary double blind clinical trial. Journal of Disorders, 38, 342–352.
Autism and Developmental Disorders, 36(3), 413–420. Martins, Y., Young, R. L., & Robson, D. C. (2008). Feeding and
Emond, A., Emmett, P., Steer, C., & Golding, J. (2010). Feeding eating behaviors in children with autism and typically develop-
symptoms, dietary patterns, and growth in young children with ing children. Journal of Autism and Developmental Disorders,
autism spectrum disorders. Pediatrics, 126(2), 337–342. 38, 1878–1887.
Hediger, M. L., England, L. J., Molly, C. A., Yu, K. F., Manning- Matson, J. L., Fodstad, J. C., & Dempsey, T. (2009). The relationship
Courtney, P., & Mills, J. L. (2007). Reduced bone cortical of children’s feeding problems to core symptoms of autism and
thickness in boys with autism or autism spectrum disorder. PDD-NOS. Research in Developmental Disabilities, 3, 759–766.

123
J Autism Dev Disord

Mulloy, A., Lang, R., O’Reilly, M., Sigofoos, J., Lancioni, G., & Sharp, W. G., Jaquess, D. L., Morton, J. S., & Herzinger, C. (2010).
Rispoli, M. (2010). Gluten-free and casein-free diets in the Pediatric feeding disorders: A quantitative synthesis of treatment
treatment of autism spectrum disorders: A systematic review. outcomes. Clinical Child and Family Psychology Review, 13,
Research in Autism Spectrum Disorders, 4(3), 328–339. 348–365.
Nadon, G., Feldman, D.E., Dunn, W., & Gisel, E. (2011). Mealtime Sharp, W. G., Jaquess, D. L., Morton, J. F., & Miles, A. G. (2011). A
problems in children with autism spectrum disorder and their retrospective chart review of dietary diversity and feeding
typically developing siblings: A comparison study. Autism, behavior of children with autism spectrum disorder before and
15(1), 98–113. after admission to a day treatment program. Focus on Autism and
Niehus, R., & Lord, C. (2006). Early medical history of children with Other Developmental Disabilities, 26, 37–48.
autism spectrum disorders. Journal of Developmental and Shearer, T. R., Larson, K., Neuschwander, J., & Gedney, B. (1982).
Behavioral Pediatrics, 27, S120–S127. Minerals in the hair and nutrient intake of autistic children.
Provost, B., Crowe, T. K., Osbourn, P. L., McClain, C., & Skipper, B. Journal of Developmental and Physical Disabilities, 12(1),
J. (2010). Mealtime behaviors of preschool children: comparison 25–34.
with autism spectrum disorder and children with typical Whiteley, P. (2004). Developmental behavioral and somatic factors in
development. Physical and Occupational Therapy in Pediatrics, pervasive developmental disorders: Preliminary analysis. Child:
30(3), 220–233. Care Health and Development, 30, 5–11.
Raiten, D. J., & Massaro, T. (1986). Perspectives on the nutritional WHO Multicentre Growth Reference Study Group. (2006). WHO
ecology of autistic children. Journal of Autism and Develop- Child Growth Standards: Length/height-for-age, weight-for-age,
mental Disorders, 16(2), 133–143. weight-for-length, weight-for-height and body mass index-for-
Ritvo, E. M., & Freeman, B. J. (1978). National society for autistic age: Methods and development. Geneva: World Health
children definition of the syndrome of autism. Journal of Autism Organization.
and Childhood Schizophrenia, 8, 162–170. Williams, K. E., Gibbons, B. G., & Schreck, K. A. (2005). Comparing
Rosenthal, R., & Rubin, D. (1986). Meta-analytic procedures for selective eaters with and without developmental disabilities.
combining studies with multiple effect sizes. Psychological Journal of Developmental and Physical Disabilities, 17,
Bulletin, 99(3), 400–406. 299–309.
Schmitt, L., Heiss, C. J., & Campdell, E. (2008). A comparison of Yu, S. M., Kogan, M. D., & Gergen, P. (1997). Vitamin-mineral
nutrient intake and eating behaviors of boys with and without supplement use among preschool children in the United States.
autism. Topics in Clinical Nutrition, 23(1), 23–31. Pediatrics, 100, e4.
Schreck, K. A., Williams, K., & Smith, A. F. (2004). A comparison of Zimmer, M. H., Hart, L. C., Manning-Courtney, P., Murray, D. S.,
eating behavior between children with and without autism. Bing, N. M., & Summer, S. (2012). Food variety as predictor of
Journal of Autism and Developmental Disorders, 34, 433–438. nutritional status among children with autism. Journal of Autism
Seiverling, L., Williams, K., & Sturmey, P. (2010). Assessment of and Developmental Disorders, 42(4), 549–556.
feeding problems in children with autism spectrum disorders.
Journal of Developmental and Physical Disabilities, 22,
401–413.

123