J Autism Dev Disord

DOI 10.1007/s10803-013-1785-z

ORIGINAL PAPER

Early Signs of Autism in Toddlers: A Follow-Up Study

in the Danish National Birth Cohort
Sanne Lemcke • Svend Juul • Erik T. Parner •

Marlene B. Lauritsen • Poul Thorsen

Ó Springer Science+Business Media New York 2013

Abstract To identify possible early signs of autism Keywords Autism spectrum disorder
Intellectual
spectrum disorder (ASD) within the Danish National Birth disability
Early signs
Parents’ interview
Follow-up
Cohort, we studied prospectively collected interviews from study
Birth cohort
76,441 mothers about their children’s development and
behaviour at 6 and 18 months. In Danish national regis-
tries, 720 children with ASD and 231 children with intel- Introduction
lectual disability (ID) were identified. At 6 months,
associations between early signs and ASD or ID were The early signs of autism spectrum disorder (ASD) have
found only in few areas. At 18 months social, language, been studied intensively, and distinctions in early devel-
and motor skills were delayed, and suspicion of vision and opment between children with ASD and normal develop-
hearing problems were increased for both groups. Signs ment have been identified in various domains. Studies have
distinguishing ASD from ID were unclear, and the positive confirmed that many young children with ASD in their first
predictive values regarding ASD were below 10 % for years of life show less interest in social interaction and
individual predictors and aggregated risk scores. have impairments in early language skills, play, and motor
function compared to children with normal development
(Yirmiya and Charman 2010). Intellectual disabilities (ID)
are common in children with ASD, with prevalence rates in
different studies between 30 and 68 % (Yeargin-Allsopp
S. Lemcke (&)
M. B. Lauritsen et al. 2003; Chakrabarti and Fombonne 2005). A number of
Regional Centre for Child and Adolescent Psychiatry, Aarhus studies have looked for signs that could distinguish
University Hospital, Harald Selmers Vej 66, 8240 Risskov,
between the early symptoms of ASD and ID in young
Denmark
e-mail: [email protected] children; in the second year of life some studies have found
differences in social development between these groups
S. Juul (Osterling et al. 2002; Werner et al. 2005; Ozonoff et al.
Department of Epidemiology, School of Public Health,
2008).
University of Aarhus, Aarhus, Denmark
Positive effects of prolonged intensive intervention
E. T. Parner programmes for children aged 3–5 years with ASD have
Department of Biostatistics, School of Public Health, been reported (Landa 2008; Dawson 2008). Until now,
University of Aarhus, Aarhus, Denmark
intervention studies of children younger than 2 years are
M. B. Lauritsen sparse, but yield hope that intervention at an early stage can
Research Unit for Child and Adolescent Psychiatry, Aalborg alter the course of development for children with ASD, and
University Hospital, Aalborg, Denmark that early identification may be beneficial (Warren et al.
2011). Parents are often the first to observe deviance in
P. Thorsen
Department of Obstetrics and Gynaecology, Lillebaelt Hospital, their children’s development, and their observations could
Kolding, Denmark provide important information for early detection of

123
 J Autism Dev Disord

children with ASD. The mean age of diagnosis for children Danish Civil Registration System (Pedersen et al. 2006)
with ASD is 4–6 years (Parner et al. 2008; Wiggins et al. was used to identify dead and emigrated children. In total,
2006); although up to 30 % of the parents in retrospective the study cohort comprised 76,441 children (Fig. 1).
interviews state that they have observed deviations in Each of the postpartum interviews in DNBC comprised
development before 12 months of age, and 60–80 % before more than 200 questions, including questions about general
24 months of age (De Giacomo and Fombonne 1998; child development. The questionnaires were not developed
Chawarska et al. 2007). Studies using different kinds of specifically for detecting ASD. For this study, we used all
methodology have confirmed parents’ experience of devi- data on development of motor, cognitive, language, and
ations in development before the age of 2 years (Yirmiya social skills, including vision and hearing abilities, from
and Charman 2010; Landa 2008). the interviews at 6 and 18 months after birth. If the inter-
In the Danish National Birth Cohort (DNBC), mothers views were completed more than 2 weeks after the planned
were interviewed about their children’s development and age, the mothers were asked to answer the questions as if
behaviour during the first 18 months of the child’s life. the child was the required 6 or 18 months. If a mother was
Outcomes of ASD, childhood autism, and intellectual dis- uncertain about an answer or had chosen not to answer a
ability among children in the birth cohort could be gathered question, we coded the answer as missing, and excluded
from public health registries and matched to the mother’s 6 the child from the particular analysis. Many of the ques-
and 18 month interviews. This allowed us, in a large tions in the DNBC combine several areas of development
population-based birth cohort, to analyse correlations in one question, e.g., the question ‘‘Can he/she take off
between early development and later diagnoses of ASD socks and shoes when you ask him/her to?’’ which requires
and the impact of intellectual disability on the early signs a certain level of development in skills like motor function,
of ASD. language, cognition, and social ability [the exact wording is
available on DNBC’s website (DNBC 2012)]. In this study
the questions were classified in the following develop-
Methods mental domains: social and language, cognitive, motor
function, and faculties. If a question required skills in
The Danish National Birth Cohort multiple areas, it was classified according to the domain of
importance for the ASD diagnosis.
The Danish National Birth Cohort (DNBC) was established
between 1996 and 2002 with the aim to collect data to be
used to study the short- and long-term effect on the off- Danish National Birth Cohort
101,042 pregnant women gave consent to participate
spring of exposures during prenatal life and early child-
hood (Olsen et al. 2001). Danish-speaking pregnant women
Live born singletons
were invited to take part in the DNBC by their general Mother participated in the week 12 interview in pregnancy
practitioners at their first antenatal care visit, and a total of 86,809 children

101,042 women consented to take part, corresponding to

approximately 50 % of the women invited (Nohr et al. Excluded
2,998 children
2006). The data were collected via telephone interviews
during weeks 12 and 30 of pregnancy, and when the child
Potential participants
was 6 and 18 months of age. The participating mothers 83,811 children

were contacted for each interview, even if they had not

participated in the previous interview. Missing
6 and 18 months interviews
There were a considerable number of drop-outs during 7,370 children
the DNBC project. Ninety-two percent of the recruited
women completed the first interview in week 12 of preg-
Study cohort
nancy, 70 % took part in the interview at 6 months after 76,441 children

birth, and 66 % in the interview at 18 months (DNBC

2012). The present study population was limited to children Missing Missing
6 months interview 18 months interview
whose mothers had participated in the DNBC interview in 10,659 children 13,720 children
week 12 of pregnancy and at least one of the interviews at
6 or 18 months. In addition, the inclusion criteria were
Completed 6 months interview Completed 18 months interview
singleton pregnancy, alive and living in Denmark 65,782 children 62,721 children

throughout the follow-up period, and living with the

mother at the time of interview. Information from the Fig. 1 Flowchart for the study population

123
J Autism Dev Disord

Outcome unreliable results, predictor analyses were only performed

when at least five children with ASD, CA, or ID no ASD
In Denmark, children suspected of having ASD are usually had the problem in question. Some of the children in the
referred to a public child psychiatric hospital, where cohort are siblings (n = 14.996) and to account for the
evaluation and treatment is free of charge. Since 1995, all dependence of these data, we adjusted the analyses using a
contacts with psychiatric hospitals have been registered in robust variance estimator that allowed clustering of
the Danish Psychiatric Central Research Register (Mors data.(Kirkwood and Sterne 2003).
et al. 2011), which now is included in the Danish National The distribution of time to onset of walking and sitting
Patient Register (Lynge et al. 2011). The ICD-10 classifi- alone was described with one minus Kaplan–Meier curves,
cation has been used in Denmark since 1994. and groups were compared using the Breslow-Gehan test
By linking the DNBC with the Danish Psychiatric (Kalbfleisch and Prentice 2002). We analysed time to sit
Central Research Register via the personal identification alone until 12 months of age, and time to walk alone until
number from The Danish Civil Registration System 20 months of age.
(Pedersen et al. 2006), we identified the children in the An aggregate risk marker, sum score, was created as the
cohort with any main or subsidiary ASD diagnosis (ICD-10 total number of impairments listed in Tables 1 and 2 at 6
code F84.0, F84.1, F84.5, F84.8, F84.9). We also per- and 18 months, respectively. In addition, in the 18-month
formed a separate group of children with childhood autism scores were included parents suspicion of reduced hearing,
(CA) code F84.0. The quality of the diagnosis of CA in the vision, and squinting plus sitting later than 12 months, and
Danish Psychiatric Central Research Register has been walking later than 20 months of age. The scores were
validated, and 94 % of the CA diagnoses were verified as mainly grouped based on the number of impairments, but
true positive (Lauritsen et al. 2010). also numbers of participants in each category were taken
In Denmark, many children observed for ID are exam- into account. The scores were used as a categorical variable
ined in paediatric or child psychiatric wards, and the in Cox regression analysis with the group of children
diagnoses are reported to the Danish National Patient without any impairments as the reference.
Register. By linking the DNBC with this register, it was The frequencies of the background variables were cal-
possible to identify children diagnosed with ID (ICD-10 culated and distributions between groups (ASD, ID no
code F70-F79, corresponding to an IQ \70), but without a ASD, and study cohort) were tested for differences. The
diagnosis of ASD (ID no ASD). Chi squared test was used for categorical variables and the
Wilcoxon rank sum test for continuous variables. We used
Statistical Analyses the term ‘‘significance’’ for findings that were statistically
significant at the 5 % level without any formal correction
The study was performed as a longitudinal cohort study for multiple testing. Stata Statistical Software, version 11.0
within the DNBC. To adjust for differences in follow-up was used for the analyses.
time Cox proportional hazards regression model was used The study was approved by the Danish National Board
to analyse the associations between predictors and the of Health and registered with the Danish Data Protection
subsequent diagnosis (Kirkwood and Sterne 2003; Kim Agency. On inclusion in the DNBC, the mothers gave
2012). The effect measure was hazard ratio (HR) with age consent to the use of data for research purposes in the
of the child as the underlying time scale. For rare diseases future, and the steering committee for the DNBC gave
such as ASD, the HR is approximately equal to relative permission to use data from the cohort for the project.
risk. Follow-up time was determined from birth until first
ASD diagnosis or first ID but no ASD diagnosis or end of
follow-up on 20 August 2010. By definition, children in the Results
ID no ASD group had not been diagnosed as having ASD
during the follow-up period. For each disorder, the group By the end of follow-up, the mean age of the children in the
of children diagnosed was compared to the total study cohort was 9.8 years, with an age span from 7.2 to
cohort including the diagnostic groups. The analyses were 12.5 years. In all, 720 (0.9 %) children in the study cohort
performed both for the entire group of ASD and for the were diagnosed as having ASD, of these 268 were diag-
subgroup CA specifically. Because of overlap between nosed with CA, and 617 of the ASD mothers had taken part
these two groups, no comparisons were made between the in the 6 month interview and 591 in the 18 month inter-
ASD and CA groups. An assumption for Cox regression view. Another 231 children were diagnosed with ID no
model is that the ratios of the hazards for the comparing ASD. The average age of the children at the time of the
variables remain constant over time. It was assessed for third interview was 6 months in all groups, as planned, but
each variable using log-minus-log survival plots. To avoid at the fourth interview, the average age for all groups was

123
 J Autism Dev Disord

Table 1 Characteristics of the study population divided into diagnostic groups (n = 76,441)
ASD CA ID no ASD Study cohort

Number 720 268 231 76,441

Gendera
Girls (%) 143 (19.9) 42 (15.7) 82 (35.5) 37,331 (48.8)
Boys (%) 577 (80.1) 226 (84.3) 149 (64.5) 39,110 (51.2)
Mean age at diagnosis, years 6.7 (0.6/12.1/2.23) 5.8 (1.9/11.6/2.06) 5.9 (0.5/11.4/1.83) NA
(min/max/SD)
Mean age at end of follow-up, years 10.0 (7.2/12.4/1.32) 9.9 (7.2/12.3/1.32) 10.0 (7.4/12.4/1.26) 9.8 (7.2/12.5/1.35)
(min/max/SD)b
Firstborn (%)b,c 379 (52.6) 135 (50.4) 107 (46.3) 35,760 (46.8)
Mother’s level of educationc,d,#
High (%) 356 (49.4) 140 (52.2) 101 (43.7) 39,129 (51.2)
Middle (%) 260 (36.1) 87 (32.5) 73 (31.6) 27,777 (36.3)
Low (%) 69 (9.6) 29 (10.8) 44 (19.1) 6,332 (8.3)
Missing (%) 35 (4.9) 12 (4.5) 13 (5.6) 3,203 (4.2)
Participated in the 6-month interview (%) 617 (85.7) 230 (85.8) 198 (85.7) 65,782 (86.1)
Mean age at the 6-month interview, months 6.4 (5.3/10.1/0.82) 6.4 (5.3/9.9/0.80) 6.4 (5.4/9.8/0.76) 6.4 (5.0/26.3/0.81)
(min/max/SD)
Participated in the 18-month interview (%) 591 (82.1) 216 (80.1) 193 (83.5) 62,721 (82.1)
Mean age at the 18-month interview, months 19.2 (17.5/24.1/1.16) 19.1 (17.5/24.0/1.00) 19.2 (17.7/23.4/1.12) 19.2 (17.2/34.5/1.14)
(min/max/SD)
NA not applicable, SD standard deviation
Statistical significant differences: p \ 0.05
a
Between all groups
b
Between ASD and cohort
c
Between ASD and ID no ASD
d
Between ID no ASD and cohort
#
Mother’s level of education divided into 3 groups: high: 4 years of education beyond high school or in management; middle: skilled workers or
with middle range training; low: unskilled or unemployed

19 months instead of the planned 18 months. The propor- With regard to fine and gross motor skills at the time of
tion of firstborns was not evenly distributed in the cohort the 6-month interview, significant differences were
and the diagnostic groups (p \ 0.005); the same was found observed for the ASD and the ID no ASD group in only
for gender distribution, mothers level of education, and one question ‘‘cannot sit up straight when put on lap’’,
mean age at follow-up (Table 1). resulting in a significantly increased prevalence of ASD
At 6 months of age, few predictors in the area of social (HR 1.8 [1.4; 2.2]) and ID no ASD (HR 3.5 [2.5; 4.8])
and language development were found to be significantly diagnosed children (Table 2).
more prevalent in the ASD and ID no ASD groups compared A significantly increased risk for all social and language
to the cohort. For instance, with regard to the question as to development predictors was found at 18 months in all three
whether the child did not look in the direction of voices (ASD groups; highest for the CA (HR between 2.4 and 13.5) and
HR 3.2 [1.6; 6.4]), the result was based on only eight children ID no ASD groups (HR between 3.2 and 11.6). The only
in the ASD group and six in the ID no ASD, making the question where the risk was higher for the CA group (HR
results uncertain (Table 2). The question as to whether child 13.5 [9.4; 19.4]) than the ID no ASD group (HR 10.7 [6.6;
does not like to be swung around like a airplane was most 17.4]) was for the predictor ‘‘cannot fetch things or take
prevalent for the CA group (HR 2.2 [1.3; 3.7]). them to others if asked to’’, but the difference between the
In the ASD group and especially in the ID no ASD two appeared not to be significant based on the confidence
group, an increased proportion of children did not throw interval (Table 3).
toys to the floor, grab things, or put toys in their mouths At 18 months, the strongest predictor within the cogni-
(ASD HR 1.7–2.6) at the age of 6 months; however, these tive area for all three groups was not attempting to draw
findings were only applicable for less than 11 % of the (HR between 3.0 and 8.0). Furthermore, questions com-
children with ASD (Table 2). bining cognitive and motor functions ‘‘cannot drink from a

123
J Autism Dev Disord

Table 2 Impairments at 6 months as predictors for ASD

Developmental fields ASD CA ID no ASD Study cohort
n (%) HR (95 % CI) n (%) HR (95 % CI) n (%) HR (95 % CI) n (%)

Social and language

Doesn’t look in the direction of 8 (1.3) 3.2 (1.6; 6.4)* 3 (1.3) NA 6 (3.0) 7.6 (3.4; 17.1)* 274 (0.4)
new sounds or voices
Doesn’t make noises to himself 11 (1.8) 1.3 (0.7; 2.4) 4 (1.7) NA 5 (2.5) 1.9 (0.8; 4.5) 873 (1.3)
Cannot show that he wants to 53 (8.7) 1.2 (0.9; 1.6) 19 (8.3) 1.1 (0.7; 1.8) 25 (12.8) 1.8 (1.2; 2.8)* 4,802 (7.3)
make contact by reaching out
or making babbling sounds
Doesn’t try to imitate voice 162 (27.4) 1.2 (1.0; 1.4) 60 (26.9) 1.2 (0.9; 1.6) 56 (29.2) 1.3 (1.0; 1.8) 15,371 (23.4)
sounds when spoken to
Doesn’t like to be swung 29 (4.8) 1.6 (1.1; 2.3)* 15 (6.6) 2.2 (1.3; 3.7)* 11 (5.8) 1.9 (1.0; 3.5)* 1,994 (3.1)
around like an airplane
Cognitive
Doesn’t put toys in his/her 8 (1.3) 2.6 (1.3; 5.2)* 5 (2.2) 4.4 (1.8; 10.7)* 9 (4.6) 9.5 (4.9; 18.5)* 323 (0.5)
mouth
Doesn’t throw toys onto the 65 (10.6) 1.7 (1.3; 2.2)* 24 (10.5) 1.7 (1.1; 2.6)* 50 (25.4) 4.9 (3.6; 6.8)* 4,193 (6.4)
floor
Doesn’t try to grab things that 22 (3.6) 2.4 (1.6; 3.7)* 11 (4.8) 3.2 (1.8; 6.0)* 30 (15.2) 11.7 (7.9; 17.2)* 1,005 (1.5)
are out of reach
Doesn’t show that there are 63 (10.3) 1.0 (0.7; 1.2) 20 (8.8) 0.8 (0.5; 1.3) 25 (12.7) 1.2 (0.8; 1.8) 6,878 (10.5)
things he/she doesn’t like
Motor function
Cannot hold head straight when 3 (0.5) NA 2 (0.9) NA 9 (4.6) 27.0 (13.8; 52.8)* 118 (0.2)
picked up
Cannot sit up straight when put 100 (16.3) 1.8 (1.4; 2.2)* 41 (17.9) 2.0 (1.4; 2.8)* 54 (27.8) 3.5 (2.5; 4.8)* 6,486 (9.9)
on lap
Cannot roll over from his/her 179 (29.3) 1.1 (0.9; 1.3) 73 (31.9) 1.2 (0.9; 1.6) 84 (43.5) 2.0 (1.5; 2.7)* 18,004 (27.4)
back to his/her stomach
Cannot crawl on stomach 295 (48.1) 1.1 (0.9; 1.3) 111 (48.3) 1.1 (0.8; 1.4) 124 (62.6) 1.9 (1.5; 2.6)* 30,535 (46.4)
NA not applicable, n number, HR hazard ratio, CI confidence interval
Statistical significant differences: * p \ 0.05

cup without help’’ and ‘‘cannot climb stairs without sup- suspected impairment had been confirmed by a physician
port’’ were more frequent in the ID no ASD and the CA (Table 4).
groups than in the ASD group (Table 3). An aggregate risk marker, a sum score, at 6 and
The mean age for sitting alone in the cohort was 18 months showed a significant and increased association
6.5 months (Fig. 2). On average, children with ASD and in both the ASD and ID no ASD groups if the child failed
ID no ASD were not able to sit alone until later [ASD: three developmental predictors (Table 5). The general
6.9 months (95 % CI 6.7; 7.0), ID no ASD: 8.9 months picture for the sum score was similar to that of the indi-
(95 % CI 8.3; 9.5)]. Similar associations were seen for the vidual predictors; the scores for the ID no ASD group were
age of independent walking (Fig. 3), the mean age for the highest, followed by the CA, and lowest for the ASD.
cohort being 12.6 months, for the ASD group 13.7 months Whether a comorbid diagnosis of ID in children with ASD
(95 % CI 13.5; 13.9), and for the ID no ASD group is applied probably differs among the child psychiatric
16.4 months (95 % CI 15.9; 16.8). Significant differences hospitals in Denmark; so we do not expect a complete reg-
were seen between the study cohort, and the ASD, and ID istration of ID in children with ASD in the registries (Lau-
no ASD groups for the variables age of independent sitting ritsen et al. 2010). To clarify the effect of ID on the estimates
and age of walking (p \ 0.0001 on comparing all groups for ASD, supplementary analyses were performed for the
and comparing each pair of groups). children with ASD and no registered comorbid ID (ASD, no
Suspected hearing and vision impairments, including registered ID: 611 children). When the children with known
squinting, were more frequent in children who were later ID were removed, the associations were weaker for all
diagnosed with ASD or ID no ASD, and in many cases the predictors but with the same trends (results not shown).

123
 J Autism Dev Disord

Table 3 Impairments at 18 months as predictors for ASD

Developmental fields ASD CA ID no ASD Study cohort
n (%) HR (95 % CI) n (%) HR (95 % CI) n (%) HR (95 % CI) n (%)

Social and language

Doesn’t use word or word- 49 (8.3) 3.3 (2.5; 4.4)* 27 (12.5) 5.2 (3.5; 7.8)* 46 (24.0) 11.6 (8.3; 16.0)* 1,686 (2.7)
like sounds to indicate what
he wants
Doesn’t use more than ten 433 (73.4) 2.0 (1.7; 2.4)* 179 (83.3) 3.6 (2.5; 5.1)* 164 (86.3) 4.5 (3.0; 6.9)* 36,344 (58.0)
words
Doesn’t make sentences of 371 (68.5) 1.7 (1.4; 2.0)* 146 (77.3) 2.6 (1.8; 3.7)* 117 (80.7) 3.2 (2.1; 4.8)* 34,578 (56.8)
two words
Cannot take off socks or shoes, 166 (28.2) 1.8 (1.5; 2.2)* 74 (34.6) 2.4 (1.8; 3.2)* 86 (45.5) 3.9 (2.9; 5.2)* 11,153 (18.0)
when he/she is asked to
Cannot fetch things or take 55 (10.3) 6.1 (4.6; 8.1)* 38 (20.4) 13.5 (9.4; 19.4)* 20 (16.8) 10.6 (6.6; 17.2)* 1,181 (1.9)
them to others if asked to
Cognitive
Cannot be occupied with the 119 (20.2) 1.2 (0.9; 1.4) 45 (20.9) 1.2 (0.9; 1.7) 45 (23.4) 1.4 (1.0; 1.9) 11,250 (18.1)
same thing for at least
15 min without adult
involvement
Doesn’t try to make marks 81 (13.9) 3.0 (2.4; 3.8)* 41 (19.4) 4.4 (3.1; 6.2)* 58 (30.2) 8.0 (5.9; 10.9)* 3,272 (5.3)
with colour pencils or
similar objects
Doesn’t automatically turn the 301 (53.7) 1.2 (1.0; 1.4) 128 (62.1) 1.6 (1.2; 2.2)* 114 (62.6) 1.7 (1.3; 2.3)* 29,656 (50.1)
picture right when looking
in a book
Motor function
Cannot climb stairs without 83 (14.2) 3.8 (3.0; 4.8)* 42 (19.7) 5.7 (4.0; 7.9)* 82 (43.2) 17.7 (13.3; 23.6)* 2,606 (4.2)
support
Cannot drink from a cup 37 (6.3) 5.3 (3.8; 7.4)* 26 (12.0) 10.7 (7.1; 16.2)* 49 (25.5) 27.3 (19.7; 37.8)* 799 (1.3)
without help
NA not applicable, n number, HR hazard ratio, CI confidence interval
Statistical significant differences: * p \ 0.05

100 100
Study cohort Study cohort
ASD ASD
80 CA 80
Sit alone in percent

CA
Walk alone in percent

ID no ASD ID no ASD

60 60

40 40

20 20

0 0
0 2 4 6 8 10 12 6 8 10 12 14 16 18 20
Age in months Age in months

Fig. 2 Distribution of age in months for sitting independently, Fig. 3 Distribution of age in months for walking independently,
interview at 18 months. Information included until 12 months of age. interview at 18 months. Information included until 20 months of age.
ASD autism spectrum disorder, CA childhood autism, ID intellectual ASD autism spectrum disorder, CA childhood autism, ID intellectual
disability disability

123
J Autism Dev Disord

Table 4 The parents’ information on impairments in use of faculties as predictors for ASD at 18 months
ASD CA ID no ASD Study cohort

n (%) HR (95 % CI) n (%) HR (95 % CI) n (%) HR (95 % CI) n (%)

Parental suspicion of reduced hearing 122 (20.7) 2.4 (2.0; 2.9)* 59 (27.4) 3.5 (2.6; 4.7)* 59 (30.7) 4.1 (3.0; 5.5)* 6,203 (9.9)
A doctor has confirmed the reduced 44 (7.5) 2.3 (1.7; 3.2)* 24 (11.2) 3.6 (2.4; 5.5)* 22 (11.5) 3.7 (2.4; 5.8)* 2,140 (3.4)
hearing
Parental suspicion of reduced visiona 24 (4.7) 5.3 (3.5; 7.9)* 13 (6.8) 7.7 (4.4; 13.4)* 36 (20.6) 27.8 (19.2; 40.1)* 528 (1.0)
A doctor has confirmed the reduced 11 (2.2) 5.4 (3.0; 9.9)* 7 (3.7) 9.2 (4.3; 19.5)* 20 (11.5) 32.4 (20.3; 51.7)* 234 (0.4)
visiona
Parental suspicion of squintinga 61 (12.1) 2.1 (1.6; 2.7)* 33 (17.2) 3.1 (2.2; 4.5)* 38 (21.7) 4.2 (2.9; 6.0)* 3,467 (6.3)
A doctor has confirmed the squintinga 31 (6.1) 2.7 (1.9; 3.9)* 19 (9.9) 4.6 (2.9; 7.3)* 28 (16.0) 7.9 (5.3; 11.9)* 1,309 (2.4)

NA not applicable, n number, HR hazard ratio, CI confidence interval

Statistical significant differences: * p \ 0.05
a
Only in version 2 of interview at 18 months, n = 48,075

Table 5 Aggregate risk scores based on mothers’ report of atypical development

Total number of ASD CA ID no ASD Study cohort
questions
n (%) HR (95 % CI) n (%) HR (95 % CI) n (%) HR (95 % CI) n (%)

6 months
0–2 questions 478 (77.5) Ref. 177 (77.0) Ref. 113 (57.1) Ref. 55,226 (84.0)
3–4 questions 117 (19.0) 1.4 (1.1; 1.7)* 40 (17.4) 1.3 (0.9; 1.8) 62 (31.3) 3.1 (2.3; 4.3)* 9,631 (14.6)
5–13 questions 22 (3.6) 2.7 (1.7; 4.1)* 13 (5.7) 4.3 (2.5; 7.6)* 23 (11.6) 12.0 (7.6; 18.8)* 925 (1.4)
18 monthsa
0–2 questions 211 (35.7) Ref. 44 (20.4) Ref. 33 (17.1) Ref. 34,077 (54.3)
3–4 questions 223 (37.7) 1.5 (1.3; 1.9)* 81 (37.5) 2.7 (1.9; 3.9)* 48 (24.9) 2.1 (1.4; 3.3)* 23,568 (37.6)
5–6 questions 90 (15.2) 3.4 (2.7; 4.4)* 50 (23.2) 9.0 (6.0; 13.5)* 39 (20.2) 9.4 (5.9; 15.0)* 4,377 (7.0)
7–15 questions 67 (11.3) 16.8 (12.7; 22.1)* 41 (19.0) 47.3 (30.9; 72.4)* 73 (37.8) 116.3 (77.2; 175.3)* 699 (1.1)
n Number, HR hazard ratio, CI confidence interval, Ref. reference group
Statistical significant differences: * p \ 0.05
a
Including parental suspicion of reduced hearing, vision, and strabismus plus sitting up later than 8 months and walking later than 15 months

Although we found many significant associations, the early signs of ASD. A pattern of associations evolves for
positive predictive values (the probability of a later diag- the majority of questions in the areas of social behaviour,
nosis if specific concerns or impairments were found) of all language, and motor development at 18 months for the
findings were small, in most cases below 5 %. For example, children who later were diagnosed as having ASD or ID no
the variable ‘‘cannot fetch things or take them to others if ASD, and only few associations at 6 months. For most
asked to’’, where the HR is 6.1 for the ASD group but only 55 predictors, hazard ratios were highest for the ID no ASD
children with ASD fulfilled the criteria corresponding to a group followed by the CA and the entire ASD group.
positive predictive value of 4.6 %. For the aggregate risk Our findings correlate with studies based on parents’ ret-
scores, the positive predictive values at 18 months for the rospective information, where two-thirds of the parents of
highest scores ‘‘7–15 questions’’ were 9.6 % for the ASD children with ASD stated that they noted atypical devel-
group and 10.4 % for the ID no ASD. opment in their children before the age of 24 months
(Chawarska et al. 2007; De Giacomo and Fombonne 1998).
In another birth cohort, the Avon Longitudinal Study of
Discussion Parents and Children (ALSPAC), signs of ASD were also
identified in prospective information from parents before
Findings the child was 2 years of age and in the same developmental
fields as we found (Bolton et al. 2012).
The study investigated whether prospectively collected In line with findings in prospective studies of high-risk
information from mothers about children’s early develop- groups, our study did not identify clear distinctions in
ment and behaviour in this large birth cohort could identify social skills at 6 months (Ozonoff et al. 2010; Bryson et al.

123
 J Autism Dev Disord

2007). At 18 months, the child’s inability to fetch things also found an increased proportion of children in whom
showed an increased frequency in children with ASD and physicians had confirmed the impairment.
ID no ASD, and especially in children with CA. Using ID has been found in 30–68 % of children with ASD in
different research designs, other studies have found that other studies (Yeargin-Allsopp et al. 2003; Chakrabarti and
deviance within this area is a predictor of ASD (Ozonoff Fombonne 2005; Fombonne 2009), and in studies of the
et al. 2010; Clifford et al. 2007; Bolton et al. 2012). Some early signs of autism, distinguishing between ASD and ID
of the more ASD-specific social symptoms in the second has been difficult (Osterling et al. 2002). As mentioned
year of life found by others, such as impaired eye contact previously, the completeness of data on concurrent ID
and the child not responding to its name or to pointing among children with ASD in the Danish registers is
(Yirmiya and Charman 2010), were not included in the unknown (Lauritsen et al. 2010). It was therefore not
DNBC interviews. possible in this study to explore further the impact of ID on
Studies of early language development have shown that possible associations in children with ASD; however, all
deviances in early communication are more often reported associations became weaker when the analyses were
in children with ASD than in normally developed children. repeated after exclusion of the children with known ID
Studies involving retrospective parent interviews have from the ASD group. In the ALSPAC birth cohort, the
stated that delayed language was the first deviation in same trend was found, and it could indicate that intellectual
development many parents recognised in their child, and in disability is part of the explanation for the delayed devel-
the ALSPAC birth cohort communication and language opment in the children (Bolton et al. 2012). Gender also
problems were identified by parents in children from has an impact on children’s development; we did not
the age of 6 months (De Giacomo and Fombonne 1998; investigate the effect of gender in this study but are plan-
Chawarska et al. 2007; Bolton et al. 2012). In a follow-up ning an upcoming study in the cohort on this issue.
study of high risk children and in retrospective parent
interviews on language development, differences were Strengths
found between children with ASD and normal children at
12 months, but not between the ASD group and children To our knowledge this is the largest study on early
with ID (Wetherby et al. 2007; Watson et al. 2007). The symptoms of autism based on prospectively collected
questions in the DNBC confirmed a similar frequency of information from a population-based birth cohort. Studies
language delay in children with ASD and ID no ASD at based on retrospective parental reports have indicated that
18 months but not at 6 months. in many cases parents are the first to be aware of deviations
In the present study, as in previous studies, many chil- in a child’s development. Data in the DNBC were collected
dren with ASD and ID had delayed gross motor skills, like prospectively before the mothers became aware that the
sitting and walking (Esposito and Venuti 2008; Ozonoff child had autism, thus preventing recall bias. Our study is
et al. 2008; Landa and Garrett-Mayer 2006). With respect based on the mothers’ impressions, and the results are
to fine motor skills at 18 months, being unable to drink consistent with the findings in parents’ retrospective
without help and not attempting to make marks with a interviews (Sivberg 2003; Young et al. 2003) and in the
pencil were more frequent both among ASD and ID no ALSPAC birth cohort (Bolton et al. 2012). The findings in
ASD children. In other studies, children with ASD also the study can be used in primary care settings to increase
developed their fine motor skills more slowly than other professionals’ awareness of parental concerns about their
children (Landa and Garrett-Mayer 2006; Lloyd et al. child’s developmental and behavioural problems.
2011; Bolton et al. 2012). We found that the only distinctly Another strength of the study is that we have shown in a
increased risk for the ASD group at 6 months was showing large group the way ID contributes to the developmental
no interest in grabbing at things or throwing toys to the delays seen in small children with ASD and as a consequence
floor, and not being able to sit on a lap with straight back, influences the early signs of ASD at both 6 and 18 months. A
all skills which involve motor functions. question raised in many other studies because of the high
In some of the children in DNBC who were diagnosed comorbidity of ID in children with ASD.
with ASD or ID no ASD, the parents had suspected impaired
vision or hearing. The same pattern was found in the Limitations
ALSPAC birth cohort, and studies using home movies and
retrospective parental interviews have also reported con- With respect to identifying general deviations in children’s
cerns in relation to vision and hearing abilities (Vostanis development, the validity of the questions in the DNBC has
et al. 1998; Baghdadli et al. 2003; Bolton et al. 2012). The not been investigated. Our study showed that at 18 months
child’s abnormal or absent response towards social contact of age, some of the developmental delays detected by the
may be the explanation for the concern of the parents, but we mothers of children with ASD could be determined using

123
J Autism Dev Disord

the questions in the DNBC. For some of the questions, group can have influenced their answers and thereby the
especially in the language field at 18 months, nearly half of results of the study.
the children in the entire study cohort did not reach the
expected developmental level, so the specificity of these
questions is quite poor. Furthermore, the questions asked in
the interviews did not include some signs identified in other Conclusions
studies as early signs of ASD, such as deviant eye contact,
reduced response to name, lack of response to pointing and A large part of the questions in the DNBC, especially at
imitation. Questions regarding these developmental skills 18 months, showed an increased risk in delayed develop-
may have identified more children at risk of ASD. ment within various developmental areas for children later
The findings of an increased risk of developmental delay diagnosed with ASD; however, as isolated predictors their
in a number of skills among children later diagnosed with value is limited. The same was observed for ID no ASD,
ASD do not currently translate into predictors useful for and it was not possible in this study to distinguish between
screening purposes. The findings were not very specific, and the two groups. Our study is based upon mothers’ infor-
for most findings the positive predictive value was below mation, and the results emphasise the importance of taking
5 %; the same pattern was seen for the aggregate risk score, parents’ worries about deviant development in their chil-
where the positive predictive value was around 10 %. For the dren seriously.
majority of predictors, the associations were stronger for ID
Acknowledgments This study was funded by grants from The
no ASD than for ASD, and the questions in the interviews Danish Medical Research Council, The Elsass Foundation, The Tryg
were not suitable to differentiate between ASD and ID. Foundation and The Beatrice Surovell Haskell Fund for Child Mental
However, both groups will probably benefit from being Health Research of Copenhagen. The Danish National Research
diagnosed and referred to intervention as early as possible; Foundation has established the Danish Epidemiology Science Centre
that initiated and created the Danish National Birth Cohort. The
thus from a primary health care point of view, it is of less cohort is furthermore a result of a major grant from this Foundation.
importance whether the problems are caused by ASD or ID. Additional support for the Danish National Birth Cohort is obtained
The participants in the DNBC deviate slightly from the from the Pharmacy Foundation, the Egmont Foundation, the March of
background population (Nohr et al. 2006), but we have no Dimes Birth Defects Foundation, the Augustinus Foundation, and the
Health Insurance Foundation.
reason to suspect that this affects the observed associations
between early signs and disease. At the end of the follow- Conflict of interest The authors declare that they have no conflict
up period, the youngest children in the DNBC cohort were of interest.
around 7 years of age. The average age of diagnosis for CA
in Denmark is 4–5 years and for ASD somewhat higher
(Parner et al. 2008), consequently some of the children may
not yet have been diagnosed. Further, it is possible that
screening and assessment of the cohort for ASD could have References
identified cases undetected by the public register-based
Baghdadli, A., Picot, M. C., Pascal, C., Pry, R., & Aussilloux, C.
system, but prevalence of ASD in the study population was (2003). Relationship between age of recognition of first distur-
9 per 1,000 children, corresponding well with prevalence bances and severity in young children with autism. European
studies in Denmark (Petersen et al. 2006; Parner et al. Child and Adolescent Psychiatry, 12, 122–127.
Bolton, P. F., Golding, J., Emond, A., & Steer, C. D. (2012). Autism
2008). We have no information about regression among the
spectrum disorder and autistic traits in the avon longitudinal
children in the DNBC, and it is therefore impossible to study of parents and children: Precursors and early signs.
comment on the role of regression in our estimates (Wig- Journal of the American Academy of Child and Adolescent
gins et al. 2009; Werner and Dawson 2005; Palomo et al. Psychiatry, 51, 249–260.
Bryson, S., Zwaigenbaum, L., Brian, J., Roberts, W., Szatmari, P.,
2006). However, we expect that it could minimise some of
Rombough, V., et al. (2007). A prospective case series of high-
the differences between the ASD group and the study risk infants who developed autism. Journal of Autism and
cohort if children with ASD and regression had had a Developmental Disorders, 37, 12–24.
normal development in their early life. The parents’ Chakrabarti, S., & Fombonne, E. (2005). Pervasive developmental
disorders in preschool children: Confirmation of high preva-
experiences with older siblings can be assumed to influence
lence. American Journal of Psychiatry, 162, 1133–1141.
their observations (De Giacomo and Fombonne 1998; Chawarska, K., Paul, R., Klin, A., Hannigen, S., Dichtel, L. E., &
Baghdadli et al. 2003). The proportion of firstborns was not Volkmar, F. (2007). Parental recognition of developmental
evenly distributed in the cohort and the diagnostic groups; problems in toddlers with autism spectrum disorders. Journal of
Autism and Developmental Disorders, 37, 62–72.
therefore experiences with older siblings can have influ-
Clifford, S., Young, R., & Williamson, P. (2007). Assessing the early
enced the differences among the groups in this study. Also, characteristics of autistic disorder using video analysis. Journal
the mothers’ lower level of education in the ID no ASD of Autism and Developmental Disorders, 37, 301–313.

123
 J Autism Dev Disord

Dawson, G. (2008). Early behavioral intervention, brain plasticity, Parner, E. T., Schendel, D. E., & Thorsen, P. (2008). Autism
and the prevention of autism spectrum disorder. Development prevalence trends over time in Denmark: Changes in prevalence
and Psychopathology, 20, 775–803. and age at diagnosis. Archives of Pediatrics Adolescent Medi-
De Giacomo, A., & Fombonne, E. (1998). Parental recognition of cine, 162, 1150–1156.
developmental abnormalities in autism. European Child and Pedersen, C. B., Gotzsche, H., Moller, J. O., & Mortensen, P. B.
Adolescent Psychiatry, 7, 131–136. (2006). The Danish Civil Registration System. A cohort of eight
DNBC. (2012). Danish National Birth Cohort. Retrieved from http:// million persons. Danish Medical Bulletin, 53, 441–449.
www.ssi.dk/English/RandD/Research%20areas/Epidemiology/ Petersen, D., Bilenberg, N., Hoerder, K., & Gillberg, C. (2006). The
DNBC/For%20researchers/Data%20available.aspx. Accessed population prevalence of child psychiatric disorders in Danish 8
28 Jan 2013. to 9 year old children. European Child and Adolescent Psychi-
Esposito, G., & Venuti, P. (2008). Analysis of toddlers’ gait after six atry, 15, 71–78.
months of independent walking to identify autism: A preliminary Sivberg, B. (2003). Parents’ detection of early signs in their children
study. Perceptual and Motor Skills, 106, 259–269. having an autistic spectrum disorder. Journal of Pediatric
Fombonne, E. (2009). Epidemiology of pervasive developmental Nursing, 18, 433–439.
disorders. [Review]. Pediatric Research, 65, 591–598. Vostanis, P., Smith, B., Corbett, J., Sungum-Paliwal, R., Edwards, A.,
Kalbfleisch, J. D., & Prentice, R. L. (2002). The statistical analysis of Gingell, K., et al. (1998). Parental concerns of early development
failure time data (2nd ed.). New York: Wiley. in children with autism and related disorders. Autism: Interna-
Kim, J. (2012). Survival analysis. Pediatrics in Review, 33, 172–174. tional journal of research, 2, 229–242.
Kirkwood, B., & Sterne, J. (2003). Essential medical statistics (2nd Warren, Z., McPheeters, M. L., Sathe, N., Foss-Feig, J. H., Glasser,
ed.). Oxford: Blackwell Science. A., & Veenstra-Vanderweele, J. (2011). A systematic review of
Landa, R., & Garrett-Mayer, E. (2006). Development in infants with early intensive intervention for autism spectrum disorders.
autism spectrum disorders: A prospective study. Journal of Child Pediatrics, 127, e1303–e1311.
Psychology and Psychiatry, 47, 629–638. Watson, L. R., Baranek, G. T., Crais, E. R., Steven, R. J., Dykstra, J.,
Landa, R. J. (2008). Diagnosis of autism spectrum disorders in the first & Perryman, T. (2007). The first year inventory: retrospective
3 years of life. Nature Clinical Practice Neurology, 4, 138–147. parent responses to a questionnaire designed to identify one-
Lauritsen, M. B., Jorgensen, M., Madsen, K. M., Lemcke, S., Toft, S., year-olds at risk for autism. Journal of Autism and Develop-
Grove, J., et al. (2010). Validity of childhood autism in the mental Disorders, 37, 49–61.
Danish Psychiatric Central Register: Findings from a cohort Werner, E., & Dawson, G. (2005). Validation of the phenomenon of
sample born 1990–1999. Journal of Autism and Developmental autistic regression using home videotapes. Archives of General
Disorders, 40, 139–148. Psychiatry, 62, 889–895.
Lloyd, M., Macdonald, M., & Lord, C. (2011). Motor skills of Werner, E., Dawson, G., Munson, J., & Osterling, J. (2005). Variation
toddlers with autism spectrum disorders. Autism. doi:10.1177/ in early developmental course in autism and its relation with
1362361311402230. behavioral outcome at 3–4 years of age. Journal of Autism and
Lynge, E., Sandegaard, J. L., & Rebolj, M. (2011). The Danish Developmental Disorders, 35, 337–350.
National Patient Register. Scandinavian Journal of Public Wetherby, A. M., Watt, N., Morgan, L., & Shumway, S. (2007).
Health, 39, 30–33. Social communication profiles of children with autism spectrum
Mors, O., Perto, G. P., & Mortensen, P. B. (2011). The Danish disorders late in the second year of life. Journal of Autism and
Psychiatric Central Research Register. Scandinavian Journal of Developmental Disorders, 37, 960–975.
Public Health, 39, 54–57. Wiggins, L. D., Baio, J., & Rice, C. (2006). Examination of the time
Nohr, E. A., Frydenberg, M., Henriksen, T. B., & Olsen, J. (2006). between first evaluation and first autism spectrum diagnosis in a
Does low participation in cohort studies induce bias? Epidemi- population-based sample. Journal of Developmental and Behav-
ology, 17, 413–418. ioral Pediatrics, 27, S79–S87.
Olsen, J., Melbye, M., Olsen, S. F., Sorensen, T. I., Aaby, P., Wiggins, L. D., Rice, C. E., & Baio, J. (2009). Developmental
Andersen, A. M., et al. (2001). The Danish National Birth regression in children with an autism spectrum disorder identi-
Cohort-its background, structure and aim. Scandinavian Journal fied by a population-based surveillance system. Autism, 13,
of Public Health, 29, 300–307. 357–374.
Osterling, J. A., Dawson, G., & Munson, J. A. (2002). Early Yeargin-Allsopp, M., Rice, C., Karapurkar, T., Doernberg, N., Boyle,
recognition of 1-year-old infants with autism spectrum disorder C., & Murphy, C. (2003). Prevalence of autism in a US
versus mental retardation. Development and Psychopathology, metropolitan area. The journal of the American Medical
14, 239–251. Association, 289, 49–55.
Ozonoff, S., Iosif, A. M., Baguio, F., Cook, I. C., Hill, M. M., Yirmiya, N., & Charman, T. (2010). The prodrome of autism: early
Hutman, T., et al. (2010). A prospective study of the emergence behavioral and biological signs, regression, peri- and post-natal
of early behavioral signs of autism. Journal of the American development and genetics. Journal of Child Psychology and
Academy of Child and Adolescent Psychiatry, 49, 256–266. Psychiatry, 51, 432–458.
Ozonoff, S., Young, G. S., Goldring, S., Greiss-Hess, L., Herrera, A. Young, R. L., Brewer, N., & Pattison, C. (2003). Parental identifi-
M., Steele, J., et al. (2008). Gross motor development, movement cation of early behavioural abnormalities in children with
abnormalities, and early identification of autism. Journal of autistic disorder. Autism, 7, 125–143.
Autism and Developmental Disorders, 38, 644–656.
Palomo, R., Belinchon, M., & Ozonoff, S. (2006). Autism and family
home movies: A comprehensive review. Journal of Develop-
mental and Behavioral Pediatrics, 27, S59–S68.

123