Early markers of autism spectrum disorders in infants and toddlers prospectively identified in the Social Attention and Communication Study

Abstract

The Social Attention and Communication Study involved the successful implementation of developmental surveillance of the early markers of autism spectrum disorders in a community-based setting. The objective in the current study was to determine the most discriminating and predictive markers of autism spectrum disorders used in the Social Attention and Communication Study at 12, 18 and 24 months of age, so that these could be used to identify children with autism spectrum disorders with greater accuracy. The percentage of ‘yes/no’ responses for each behavioural marker was compared between children with autistic disorder (n = 39), autism spectrum disorder (n = 50) and developmental and/or language delay (n = 20) from 12 to 24 months, with a logistic regression also conducted at 24 months. Across all ages, the recurring key markers of both autistic disorder and autism spectrum disorder were deficits in eye contact and pointing, and from 18 months, deficits in showing became an important marker. In combination, these behaviours, along with pretend play, were found to be the best group of predictors for a best estimate diagnostic classification of autistic disorder/autism spectrum disorder at 24 months. It is argued that the identified markers should be monitored repeatedly during the second year of life by community health-care professionals.

Keywords

autism spectrum disorders early markers red flags prospective infants toddlers surveillance screening

As it is increasingly apparent that early intervention is critical in promoting better developmental outcomes for children with autism spectrum disorders (ASDs) and their families (Dawson, 2008; Rogers and Vismara, 2008), early identification is paramount. Despite our growing knowledge of the early signs of ASDs (Barbaro and Dissanayake, 2009), there is currently no recommended universal tool for early detection, as the majority of tools available to date lack sufficient specificity and sensitivity. Therefore, research needs to focus on pinpointing behaviours that will successfully identify children across the autism spectrum, encompassing those children with milder presentations, to increase sensitivity of the tools used in early identification.

Early markers of ASDs

Retrospective videotape and parental report studies (e.g. Adrien et al., 1993; Baranek, 1999; Clifford and Dissanayake, 2008; Werner et al., 2000; Young et al., 2003) have provided invaluable insight into the early signs of ASDs apparent between 12 and 24 months, which have consistently indicated deficits in social attention and communication (see review by Barbaro and Dissanayake, 2009). Prospective studies of ‘high-risk’ infant siblings of children with an ASD (ASD-sibs) have also been useful in charting the early phenotype (Rogers, 2009). The Autism Observation Scale for Infants (AOSI; Bryson et al., 2008) was successful in prospectively identifying infants by 12 months who were later diagnosed with an ASD. The behavioural markers included abnormal eye contact, visual tracking, disengagement of visual attention, orienting to name, imitation, social smiling, reactivity, social interest and sensory-orienting behaviours (Zwaigenbaum et al., 2005). Mitchell et al. (2006) found that ASD-sibs, who received a diagnosis of an ASD at 24 months, understood and produced fewer phrases and gestures by 12 months (giving, pointing, showing, shaking/nodding head, holding up arms) and had little knowledge of the appropriate use of real and toy objects. By 18 months, these toddlers showed delays in their use of gestures and in the understanding and production of phrases and single words. Furthermore, ‘response to joint attention’ skills, which has been found retrospectively to be an important marker, was also found by Sullivan et al. (2007) to be an important prospective marker for the screening of ASDs in the latter part of the second year of life.

In their study of ASD-sibs from 14 to 24 months, Landa et al. (2007) found that a subset of children later diagnosed with an ASD did not display social and communication deficits until 24 months. These toddlers were initially more or less indistinguishable from those without an ASD on the social and communication variables examined at 14 months of age. However, by 24 months, both the ‘early’ (identified at 14 months) and ‘later’ (identified after 14 months) diagnosis groups showed persistent impairments in triadic gaze, response to joint attention and initiation of joint attention (Landa et al., 2007). These data suggest that, in many cases, ASD has a progressive phase involving developmental arrest, slowing or even regression in social and/or language behaviours, which is consistent with previous findings (e.g. Interactive Autism Network, 2008; Lord et al., 2004; Werner and Dawson, 2005). Landa et al. therefore propose a continuum of impairment, where children reach the threshold for diagnosis at different times in their early years. Thus, ASDs in toddlers who are asymptomatic in the early part of their second year of life are likely to remain undetected if screening only occurs during this period.

Prospective, community-based studies have also provided evidence of early behavioural markers. The earliest signs identified were through the Early Screening of Autistic Traits (ESAT) Questionnaire at 14/15 months of age (Swinkels et al., 2006), with the most predictive items of ASD constituting a lack of bringing/showing objects, smiling and reaction when spoken to. The most sensitive items were a lack of eye contact and interest in people and the presence of stereotypical movements. Of these, stereotypical movements was the least predictive item (Dietz et al., 2006). The key markers found at 18 months using the Checklist for Autism in Toddlers (CHAT) were protodeclarative pointing, gaze monitoring and pretend play (Baron-Cohen et al., 1992, 1996). Unfortunately, the use of both the ESAT and CHAT with community-based samples has resulted in poor sensitivity, consequently, missing many of the children later diagnosed with an ASD (Baird et al., 2000; Dietz et al., 2006).

Using the Infant–Toddler Checklist (ITC), Wetherby et al. (2007) compared children with developmental delay and ASD aged 18–26 months and found five core social and communication deficits in the ASD group (gaze shifts, following of gaze/points, rate of communicating, acts for joint attention and inventory of conventional gestures). However, when later used with a community-based sample, the ITC was unable to distinguish between children with ASDs from those with developmental or language delays despite having excellent sensitivity between 9 and 24 months (93%; Wetherby et al., 2008).

The modified-CHAT (M-CHAT; Robins et al., 2001), designed for use with 24-month-olds, identified six key social relatedness and communication items that best discriminated between children with and without an ASD, including protodeclarative pointing, response to name, interest in peers, bringing things to ‘show’, following a point and imitation. The M-CHAT was also later used in a community-based sample of 16- to 30-month-olds but resulted in a low positive predictive value (PPV) when used alone (11%; Kleinman et al., 2008), consequently identifying many children without an ASD.

Problems with current screening tools

In an attempt to increase sensitivity, the ITC and M-CHAT identified many children without ASDs, albeit with other general developmental and language problems. Furthermore, the low sensitivity reported in the large-scale screening studies (using CHAT and ESAT) may be due to these tools being administered at a single age, leading to many missed opportunities for identifying ‘at-risk’ children. It is also likely that the low sensitivities are a result of the risk markers being heavily based on those behaviours related to ‘classic’ autism, rather than the broader spectrum. Many of the retrospective videotape and parental report studies, which identified the early markers later incorporated into these tools, focussed on children with autistic disorder (AD). Consequently, many children with milder or atypical presentation of symptoms were not identified by these tools (Baird et al., 2000; Dietz et al., 2006). The latest study using the ESAT in a clinical population is testament to this (Oosterling et al., 2009), as it was only able to detect children with ASD with low IQ and not those with higher IQ and perhaps ‘milder’ or atypical symptom presentation.

Another problem, with current screening tools is the use of ‘lack of’ terminology, such as ‘lack of pointing’ (Landa, 2008). Many toddlers with ASDs do, in fact, exhibit these behaviours but use them less often or inconsistently and may therefore pass the screening despite demonstrating atypical behaviours. The focus should, therefore, be on whether a particular behaviour is typical or atypical, rather than present or absent. Allison et al. (2008) is currently addressing this issue with a revision of the CHAT, the Quantitative-CHAT (Q-CHAT), which utilises a 5-point scale on a 25-item parent report measure; we currently await the results of their screening study on 20,000 18- to 30-month-old children.

The current study

Barbaro and colleagues (Barbaro and Dissanayake, 2010; Barbaro et al., 2011) utilised the Maternal and Child Health (MCH) service in Victoria, Australia, to conduct a large-scale, prospective, longitudinal study with a community-based sample. In the Social Attention and Communication Study (SACS), they recently demonstrated that routine and repeated monitoring of social attention communication behaviours by MCH nurses can accurately identify children across the autism spectrum from 12 to 24 months. The current study utilises the SACS data to identify which of the early behavioural markers monitored at 12 , 18 and 24 months were the most predictive of a best estimate ASD diagnosis at 24 months, so that these may be used in community-based centres to identify children across the autism spectrum. Few researchers have attempted to distinguish between symptoms indicative of both AD and ‘broader’ ASDs versus developmental and/or language delays in infants and toddlers, and no prospective community-based study has done so to date. It is important to identify behavioural markers across the autism spectrum, as children with a higher functioning form of autism or Asperger’s disorder are traditionally diagnosed much later (Mandell et al., 2005).

Method

Participants

The cohort in the SACS comprised 20,770 children monitored by their MCH nurse for signs of an ASD from 8 to 24 months of age. A detailed description of this cohort can be found in Barbaro and Dissanayake (2010). Children deemed to be ‘at risk’ for an ASD by the MCH nurses, based on the SACS behavioural items (detailed in the ‘Procedure’ section), were referred to the SACS team at the child development unit (CDU) for a thorough developmental and behavioural assessment. Children were only referred from 12 months onwards and were assessed at 6-month intervals until 24 months of age.

A total of 110 ‘at-risk’ children were assessed at the CDU and are the focus of this study. A cross-sectional design was chosen for the current study to maximise sample numbers due to the reduced sample of children followed prospectively; the 12-month sample comprises all the children assessed at 12 months (n = 10), the 18-month sample consists of those assessed only once at 18 months or at both 18 and 24 months (n = 37),¹ and the 24-month sample comprises only those children assessed once at 24 months (n = 62). Although the majority of children assessed at 12 and 18 months were also followed up and assessed at 18 and 24 months of age, respectively, these data were excluded at each subsequent age due to the cross-sectional approach taken in this study.²

Children’s best estimate diagnostic status was determined at 24 months using a combination of Module 1 (preverbal) of the Autism Diagnostic Observation Schedule (ADOS; Lord et al., 1999, 2000) and the Autism Diagnostic Interview–Revised (ADI-R; Lord et al., 1994), as well as clinical judgment by both authors. The first author (JB) was trained to research reliability on both instruments, and both authors have extensive prior knowledge and experience in the assessment of young children with ASDs. When there was a lack of agreement between ADOS and ADI-R classifications of AD, ASD or non-spectrum, a best estimate diagnosis was given based on the clinicians’ judgments utilising all the information gained from formal testing, observations and developmental history. Where necessary, videotapes were reviewed and tests were rechecked to help with the diagnostic decision.

Diagnosis of ASDs at 24 months has been found to be both accurate and stable over time using the ADI-R and the ADOS together, and in combination with clinical judgment, with reports between 85% and 100% stability of a best estimate classification very common (Chawarska et al., 2007; Chawarska et al., 2009; Le Couteur et al., 2008; Lord, 1995; Paul et al., 2008; Stone et al., 1999; Turner et al., 2006).³ Although there is greater shift between diagnoses within the spectrum (i.e. AD vs. ASD), it was of interest to determine how children in each classification (as determined at 2 years) present with regards to their social and communication deficits across the second year of life. Children were therefore classified as AD, ASD (children meeting criteria for the ‘broader’ autism spectrum, but not meeting criteria for AD), DD/LD (children showing developmental and/or language delays, but not meeting criteria for AD or ASD), or TD (typically developing).

Children’s cognitive and language development was assessed at each age using the Mullen Scales of Early Learning (MSEL; Mullen, 1995), with performance on these scales determining whether children possessed any developmental or language delays. Age equivalent scores were combined from the MSEL’s Receptive and Expressive Language subscales and the Visual Reception and Fine Motor subscales, to form verbal and non-verbal mental ages, respectively. Table 1 presents the sample characteristics, grouped according to their best estimate diagnostic status.⁴

Table 1.

Sample characteristics (M, SD and 95% CIs) at the 12-, 18- and 24-month assessments (N = 109)

	Group
	AD (n = 39)	ASD (n = 50)	DD/LD (n = 20)
Age in months
12-month assessment	n = 3	n = 6	n = 1
Chronological	13.7 (1.2)	12.7 (0.5)	15.0 (–)
Non-verbal	12.0 (4.3) ± 4.9	11.5 (2.2) ± 1.8	16.5 (–)
verbal	9.2 (2.9) ± 3.3	9.0 (2.5) ± 2.0	8.0 (–)
18-month assessment	n = 13	n = 17	n = 7
Chronological	19.4 (1.0)	19.2 (1.3)	20.1 (1.6)
Non-verbal	17.4 (2.3) ± 1.3	17.5 (2.7) ± 1.3	17.4 (2.0) ± 1.5
Verbal	8.9 (2.0) ± 1.1	11.9^a (1.6) ± 0.8	13.6^a,b (2.1) ± 1.6
24-month assessment	n = 23	n = 27	n = 12
Chronological	25.8 (1.8)	26.2 (2.6)	26.8 (3.1)
Non-verbal	18.9 (3.4) ± 1.4	21.5^a (3.0) ± 1.1	21.5^c (3.8) ± 2.2
Verbal	11.1 (3.2) ± 1.3	15.8^a (4.4) ± 1.7	16.9^a (4.0) ± 2.3
Overall ADOS score	17.25	10.00	3.25
ADI-R overall SI score	15.34	10.74	3.35
ADI-R overall communication score	9.55	7.40	6.19
ADI-R overall RRB score	3.41	3.03	1.94
Gender (male and female)	28 and 11	41 and 9	14 and 6

M: Mean; SD: standard deviation; CI: confidence interval; AD: autistic disorder; ASD: autism spectrum disorder; DD/LD: developmental and/or language delay; SI: social interaction; RRB: repetitive, restricted and stereotyped behaviours; ADI-R: Autism Diagnostic Interview–Revised; ADOS: Autism Diagnostic Observation Schedule.

The one typically developing child, referred and assessed at 18 and 24 months, has been excluded.

Significantly different from AD, p < .01

Significantly different from ASD, p < .05

Significantly different from AD, p < .05

The socioeconomic status of the 17 Local Government Areas (LGAs) the sample resided in was mostly high, with the mean Socio-Economic Indexes for Areas (SEIFA) score (M = 1066) being slightly higher than the mean SEIFA score of the whole of metropolitan Melbourne (M = 1033). In addition to the referred sample from the SACS, an additional sample of TD children was also observed at their MCH consultation at 12 months (n = 13), 18 months (n = 12) and 24 months (n = 11). The SACS behavioural items (see below) were completed on each child to determine reliability of the nurses’ monitoring of the SACS items (see Barbaro and Dissanayake, 2010).⁵

Procedure

SACS behavioural items

The MCH nurses were trained by the SACS team to identify infants and toddlers ‘at risk’ of an ASD by monitoring social attention and communication behaviours during their routine consultations. They were trained to identify when a behaviour was atypical, such that it was either reduced or poorly developed, as opposed to simply present/absent. These behaviours were listed on a SACS ‘items’ sheet, and nurses were instructed on how each specific item was to be monitored at each age (see Appendix 1 for these items). Performance on several ‘KEY’ items was used to refer infants and toddlers ‘at risk’ for an ASD to the SACS team.⁶ For the purposes of this article, children’s performance on all items, not just KEY items, was of interest.

The same SACS ‘items’ sheets utilised by the nurses for referral were also completed by the first author to score the same behaviours during children’s assessments at the CDU. All sessions were videotaped and used to assist in the coding of these sheets where necessary. The data used in the current study are based on these observations of the children undertaken during their assessments at the CDU, as they were deemed to be more accurate and reflective of the child’s ability, as each child was seen for a longer period of time (2 h vs. 15–30 min at the MCH centre) and by an expert in ASD assessments.

Interrater reliability

To determine reliability of the first author’s scoring of the ‘items’ sheets, a second rater, blind to diagnostic status, was employed to recode 15% of the ‘items’ sheets for each of the 12-, 18- and 24-month samples (43 items in total). Percentage agreement was .96 or higher for the 12-, 18- and 24-month ‘items’ sheets and .80 or higher for each individual item, with the exception of two items.⁷

Results

Frequencies of each behaviour for each group were compared using the Fisher’s exact probability test, which handles small expected cell frequencies (Howell, 2010). Fisher’s tests were only conducted between the AD and DD/LD groups, and the ASD and DD/LD groups, to limit the number of comparisons and control for Type I error. At each age, correlations between the dependent variables were mostly low to moderate; however, some were high. Therefore, to further control for Type I error, a nominal p value of .01 was adopted for the 18- and 24-month data; however, due to the small sample sizes at 12 months, a p value of .05 was maintained. Effect size was evaluated using the phi coefficient.

12-month items

As 9 of the 10 12-month-olds assessed were later classified as being on the autism spectrum, it was not possible to ascertain which signs were specific to AD and ASD as opposed to DD/LD. Therefore, the AD and ASD groups were combined for the analyses due to low n and compared to a sample of 13 TD 12-month-olds (mean age = 12 months; 7 males and 6 females).

As seen in Figure 1, a failure to engage in pointing, waving, imitation, eye contact and response to name significantly distinguished children with AD/ASD from the TD children at 12 months (all p < .05), with moderate to high effect sizes (phi coefficient range = .6–.8). Although fewer children in the AD/ASD group engaged in follows point and social smiles relative to the TD group, these differences were not statistically significant (both p > .05). While eye contact, response to name and social smiles were recorded as absent for all three children with AD, these behaviours were largely present among the children with ASD.

Figure 1.

Behaviours that differentiated the AD/ASD group from the TD group at 12 months. Percentage of ‘No’ responses.

Behaviours that were not as problematic for the AD/ASD group at 12 months were conversational babble, speaks 1–3 words, cuddles and attending to sounds (see Figure 2), with no significant differences between the AD/ASD and TD groups (all p > .05, phi coefficient range = .09–.48). Although over 50% of the children with AD/ASD had deficits in understands simple instructions, 23% of the TD children also failed this item, resulting in a nonsignificant difference (p = .19; phi = .33).

Figure 2.

Behaviours that did not differentiate the AD/ASD group from the TD group at 12 months. Percentage of ‘Yes’ responses.

18-month items

As more children with DD/LD were referred at 18 months, it was possible to compare this group to the AD and ASD groups (separately). Data from a sample of 12 TD children were included in the figures for comparative purposes (mean age = 18.0 months; 6 males and 6 females) but were not included in the analyses to limit the number of comparisons.

It is apparent from Figure 3 that deficits in pointing, eye contact and showing (communicating socially with others by ‘showing’ them objects) clearly differentiated both the AD and ASD groups from the DD/LD group, who passed each of these items at 18 months (all p < .01). Effect sizes were moderate to high, with phi coefficients ranging from .6 to 1.0. Consistent with the findings from 12 months, all children with AD showed deficits in pointing and eye contact; furthermore, they all showed deficits in showing at 18 months. Although the vast majority of the children with ASD did not engage in pointing, over one-third of this group did engage in eye contact and showing.

Figure 3.

Behaviours that differentiated the ASD and/or AD groups from the DD/LD group at 18 months. Percentage of ‘No’ responses.

Behaviours that significantly differentiated the AD (but not ASD) group from the DD/LD group were social smile, response to name, follows point, uses 5–10 words and understands words (all p < .01; phi coefficient range = .6–.8). The vast majority of the children with ASD did engage in social smile, nearly 50% engaged in response to name and over one-third engaged in follows point. However, although there were no significant differences between the ASD and DD/LD groups on uses 5–10 words and understand words, a high percentage of children in the ASD group did not show these behaviours (see Figure 3).

There were no significant differences between the AD and ASD groups and the DD/LD group in imitation, pretend play, points to facial features, obeys simple instructions, waving, cuddles, affection/comfort and loss of skills (all p > .01). Very few children with AD or ASD showed deficits in cuddles and affection/comfort, with none of the DD/LD (or TD) groups showing deficits in these areas (see Figure 4).

Figure 4.

Behaviours that did not differentiate the AD and ASD groups from the DD/LD group at 18 months. Percentage of ‘No’ responses.

24-month items

In keeping with the 18-month analyses, percentage responses for each behaviour were analysed between the DD/LD group and the AD and ASD groups. The data from a sample of 11 TD children, included in the figures, were not analysed (mean age = 24.0 months; 8 males and 3 females). It is apparent from Figure 5 that deficits in pointing, eye contact, showing, pretend play and waving clearly differentiated both the AD and ASD groups from the DD/LD group (all p < .01; phi coefficient range = .4–.9). Consistent with the 12- and 18-month data, all the children with AD showed deficits in pointing; however, nearly 50% of the ASD group engaged in pointing and showing at 24 months. It is also of note that over 50% of the children in both the AD and ASD group engaged in waving, and a large percentage of these groups engaged in pretend play (30% and 48%, respectively).

Figure 5.

Behaviours that differentiated the AD and ASD groups from the DD/LD group at 24 months. Percentage of ‘No’ responses.

Behaviours that differentiated the AD (but not ASD) group and the DD/LD group were: follows simple commands, follows point, social smile, response to name and loss of skills (all p ≤ .001; phi coefficient range = .6–.7). Approximately one-half to three-quarters of the ASD group engaged in follows simple commands, follows point, social smile and response to name, compared to approximately 20% of the AD group. It is worth noting that 33% of the children in the DD/LD group, and 59% of the children in the ASD group, also displayed loss of skills (see Figure 6).

Figure 6.

Behaviours that differentiated the AD group (only) from DD/LD group at 24 months. Percentage of ‘No’ responses.

There were no significant differences between both the AD and ASD groups and the DD/LD group in uses 20–50 words, 2-word utterances, parallel play, imitation and affection/comfort (all p > .01; see Figure 7). This is due to a large percentage of the DD/LD group failing the language items and over 50% of the children with AD/ASD engaging in imitation and parallel play. Once again, affection/comfort did not differentiate the children with and without AD/ASD.

Figure 7.

Behaviours that did not differentiate the AD and ASD groups from the DD/LD group at 24 months. Percentage of ‘No’ responses.

Group predictors of an autism spectrum (AD/ASD) classification at 24 months

As there were sufficient participant numbers at 24 months, a logistic regression analysis was conducted to determine which group of behavioural items could best predict the probability of a diagnostic classification on the autism spectrum versus a classification of DD/LD. The AD and ASD groups were therefore combined. Logistic regression was utilised as it does not assume a normal distribution or equal variance between groups, and the results are independent of sample size (Howell, 2010).

The entire referred cross-sectional sample at 24 months was used in this analysis (N = 99; ASD: n = 42, AD: n = 37 and DD/LD: n = 20). Associations between the predictors were evaluated using cross-tabulations and phi coefficients. Predictors with a significant bivariate association of .3 or higher with the dependent variable (diagnostic classification) were included simultaneously in the model predicting group membership. The following predictors were included: pointing, pretend play, showing, eye contact, follows simple commands, follows points and social smile. Analyses were run to detect multicollinearity, which was not deemed to be a serious problem.

The results are shown in Table 2. The final model was significant (χ² (7, N = 99) = 72.32, p < .001) and allowed prediction of 92.9% of the cases: 80% of the DD/LD group and 96.2% of the AD/ASD group. Significant predictors of a diagnostic classification of AD/ASD at 24 months of age included pretend play, pointing, eye contact and showing. The odds ratio was highest for pretend play (75.5), followed by pointing (37.3), eye contact (28.5) and showing (19.9).

Table 2.

Logistic regression analysis of behavioural items for a diagnostic classification of AD/ASD at 24 months of age (N = 99)

Behaviour	B	SE	Wald χ² test	df		OR	95% CI for OR
							Lower	Upper
Pretend play^a	4.32	1.64	6.95	1	.008	75.48	3.03	1878.44
Pointing^a	3.62	1.66	4.77	1	.029	37.26	1.45	959.19
Eye contact^a	3.35	1.51	4.92	1	.027	28.54	1.48	552.09
Social communication^a	2.99	1.44	4.31	1	.038	19.93	1.18	335.67
Follows point	3.00	2.06	2.12	1	.145	20.04	0.36	1130.52
Follows simple commands	2.79	2.85	0.96	1	.327	16.26	0.06	4296.88
Social smile	1.09	1.51	0.52	1	.470	0.34	0.02	6.50

AD: autistic disorder; ASD: autism spectrum disorder; SE: standard error; df: degrees of freedom; OR: odds ratio; CI: confidence interval.

Key group predictors of a diagnostic classification of AD/ASD at 24 months.

Discussion

The aim in this study was to identify the most discriminating and predictive behavioural markers of AD and ASD at 12, 18 and 24 months of age utilised in the SACS, so that these items may be used to identify children across the autism spectrum from 12 to 24 months. Each of the key individual markers and group predictors of both AD and ASD, identified in this study, are flagged in Appendix 1. It was found that the children in the AD group had pervasive deficits in the majority of the social attention and communication items monitored at each age, with the children in the ASD group presenting with less pervasive and severe deficits in these areas. Furthermore, although the DD/LD group displayed impairments in the language variables, they showed a very similar pattern to the TD children on most other variables.

Markers of the autism spectrum at 12 months

Deficits in pointing, waving, imitation, eye contact and response to name are important markers for the identification of both AD and ASD at 12 months, which is consistent with retrospective and prospective studies (Adrien et al., 1993; Nadig et al., 2007; Osterling et al., 2002; Watson et al., 2007; Werner et al., 2000; Zwaigenbaum et al., 2005). While response to name was recorded as absent for all children with AD, it was recorded as present for the majority of children with ASD, resulting in 46% of the combined AD/ASD group passing this item at 12 months. This finding is consistent with Nadig et al. (2007), who found that deficits in response to name is very specific at 12 months of age (.89) but not very sensitive (.50). Determining risk status for an ASD on this behaviour alone at 12 months will therefore miss half the children later diagnosed with an ASD. Furthermore, the majority of children later diagnosed with ‘broader’ ASD at 24 months did engage in eye contact, response to name, follows points and social smile at 12 months, despite many of the children with AD failing these items. Therefore, although these behaviours are important to investigate at 12 months to identify children at risk for AD, the presence of these behaviours does not rule out the possibility of ‘broader’ ASD. Thus, as found by Landa et al. (2007), a subset of children later diagnosed with an ASD will not present with deficits in some social attention and communication behaviours around their first birthday.

Markers of the autism spectrum at 18 months

Consistent with the findings at 12 months, deficits in pointing and eye contact continue to be key markers at 18 months, with both the AD and ASD groups showing deficits in these behaviours relative to the DD/LD group. Showing becomes a very important marker for the identification of AD/ASD in 18-month-olds, as none of the children in the DD/LD and TD groups showed deficits in this area, compared to all of the AD group and 65% of the ASD group. These behaviours have consistently been found to be important markers for the identification of ASDs between 12 to 24 months of age (e.g. Adrien et al., 1993; Baron-Cohen et al., 1996; Landa et al., 2007; Robins et al., 2001; Werner and Dawson, 2005; Young et al., 2003).

Despite both the AD and ASD groups differing significantly from the DD/LD group in their eye contact and showing skills, once again, many of the children with ASD did engage in these behaviours at 18 months. Other behavioural markers that will identify children with classic autism (AD) but not ‘broader’ ASD at 18 months included social smile, response to name, follows point, uses 5–10 words and understands words. As the majority of children classified with ‘broader’ ASD did engage in these behaviours at 18 months, screening tools utilising these behaviours as risk markers at this age may not identify these children.

Imitation and waving did not continue to be key markers at 18 months, which may be a result of the AD and ASD groups being compared to a DD/LD group at 18 months, rather than a TD group.Nearly 60% of the DD/LD group showed deficits in imitation and waving at 18 months and thus did not differ significantly from the AD and ASD groups. Similarly, the finding that pretend play was not a key marker for AD/ASD at 18 months was due to the DD/LD group also showing deficits in this area (57%). Thus, deficits in pretend play at 18 months may be more indicative of general developmental/language delays, rather than being specific to the autism spectrum. This finding contrasts with the results from the CHAT (Baron-Cohen et al., 1996), which identified pretend play as a key marker for AD/ASD in 18-month-olds.

While all of the children with AD in the current study displayed deficits in pretend play at 18 months, over a third of the children with ASD engaged in pretend play. This finding, combined with the previous finding that the majority of children with ‘broader’ ASD did engage in ‘follows point’, may explain the lack of sensitivity of the CHAT for ‘broader’ ASD (Baird et al., 2000), as it uses both of these behaviours as key items at 18 months.

Markers of the autism spectrum at 24 months

Once again, consistent with the findings from the 12- and 18-month data, deficits in pointing, eye contact and showing continue to be key markers for the identification of children with AD/ASD at 24 months, as both the AD and ASD groups showed deficits in these behaviours relative to the DD/LD group. Furthermore, the vast majority of the ASD group at 24 months (86%) showed deficits in eye contact, in comparison to only 17% of the ASD group at 12 months and 65% at 18 months. Therefore, it seems that children with ‘broader’ ASD may show some regression in this area across the second year of life, which is consistent with the pattern displayed in Landa et al.’s (2007) ‘later’ diagnosis group.

Although children with both AD and ASD were significantly different from the DD/LD group in their pointing and showing skills at 24 months, nearly 50% of the children in the ASD group did display these behaviours. Therefore, the presence of these behaviours at 24 months does not rule out the possibility of an autism spectrum classification. However, an absence of these behaviours strongly suggests a child will be on the spectrum, given that none of the children in the DD/LD and TD groups showed deficits in these skills. Other key markers of AD/ASD at 24 months include pretend play and waving, although some children with both AD and ASD did engage in these behaviours at 24 months.

Similar to the 18-month results, there were behavioural markers that were useful for identifying children with AD but not ‘broader’ ASD, and these included deficits in follows simple commands, follows point, social smile and response to name, as well as loss of skills. Thus, tools designed for use at 24 months of age that utilise these behaviours as key items may not identify those children with ‘broader’ ASD.

Also consistent with the findings from the 18-month data, imitation was no longer a marker of AD/ASD at 24 months, as the number of children with AD presenting with deficits in this area dropped by almost half from 18 to 24 months. Not surprisingly, the DD/LD group displayed comparable deficits to the AD and ASD groups on the language variables uses 20–50 words and 2-word utterances. These language variables are therefore useful in indicating general language/developmental problems but are not specific to AD/ASD at 24 months.

The pattern of results from the current study not only indicate the utility of individual key behavioural markers that change with age but also speak to the importance of relying on a group of markers rather than single behavioural items. The group of behaviours that were able to predict a diagnostic classification on the autism spectrum (both AD and ASD combined) at 24 months included pretend play, pointing, eye contact and showing. Although a deficit in pretend play was not a useful marker of AD/ASD at 18 months, it became one of the most important predictors for a diagnostic classification of AD/ASD at 24 months. This is due to the percentage of children with DD/LD displaying deficits in this area declining from 57% to 8% between 18 and 24 months. Developmental surveillance of children at 24 months should include the investigation of deficits in each of these behaviours together to determine risk status for AD/ASD. See Table 3 for a summary of the key markers of AD and ASD across the second year of life.

Table 3.

Summary table outlining the key markers of AD/ASD and AD only (ASD and DD/LD groups did not differ) across the second year of life

12 months	18 months		24 months
AD/ASD	AD/ASD	AD only	AD/ASD	AD only
Pointing	Pointing		Pointing
		Social smile		Social smile
Eye contact	Eye contact		Eye contact
Response to name		Response to name		Response to name
	Showing		Showing
		Follows point		Follows point
Waving			Waving
imitation
			Pretend play
		Understands words		Follows simple commands
		Uses 5–10 words
				Loss of skills

AD: autistic disorder; ASD: autism spectrum disorder; DD/LD: developmental and/or language delays.

A developmental approach to the detection of ASDs

It is clear from the preceding sections that due to the variable symptomology in children who fall across the autism spectrum, as well as the variability across different ages, a developmental approach must be taken when developing a tool or surveillance programme for ASDs. A deficit in eye contact is a preverbal deficit that begins early in life (at least by 12 months), permeates the entire spectrum and persists across the second year of life. Joint attention behaviours, such as pointing and the related behaviour showing, are also preverbal behaviours that are neither emerging at the developmentally appropriate ages (approximately 12 and 18 months, respectively) nor developing during the second year of life for children with AD and ASD. Alternatively, children with DD/LD do not show overall deficits in these areas during the second year of life. These preverbal behaviours are therefore consistent and reliable markers of ASDs that endure despite age and diagnostic classifications in the second year.

It is interesting to note that the consistent deficits for children with both AD and ASD are initiating behaviours (i.e. the child must initiate contact with others to point, make eye contact and show things to them); however, the behaviours that can identify children with AD (but not ASD) at 18 and 24 months are responding behaviours (i.e. responding when someone smiles at you, calls your name, points to something or gives you a simple instruction/command). Therefore, it seems that the children with ‘broader’ ASD, although able to respond appropriately to others in at least some behaviours, are still not initiating social interaction that is expected from even developmentally delayed and language-delayed children.

The two key markers ‘pretend play’ and ‘waving’ differed between 18 and 24 months due to the changes seen in the DD/LD group, rather than any changes seen in the AD and ASD groups. As children with AD, ASD and DD/LD were equally as deficient in these behaviours at 18 months, they were not key markers of AD or ASD at this age; however, the significant improvements seen in the DD/LD group by 24 months led these behaviours to subsequently become key markers of an ASD at 24 months of age. Thus, developmental improvements in pretend play and use of gestures late in the second year of life by children with DD/LD is important to consider when conducting developmental surveillance for the key signs of AD and ASD.

Strengths of children on the autism spectrum

Studies on the early markers of AD/ASD have focussed on what behaviours are ‘lacking’ amongst these children. However, the findings from the current study demonstrate not only deficits but also relative strengths, especially among the children with ‘broader’ ASD. This is perhaps why the prospective tools reviewed have low sensitivity, as they may have missed children displaying milder or atypical symptoms by using ‘lack of’ terminology (Landa, 2008).

Raters in the current study, although using ‘yes/no’ criteria, assessed behaviours according to whether they were typical or atypical, and MCH nurses were similarly trained to do so. However, despite this, many children with ‘broader’ ASD still passed some of these items. Thus, future research on tool development should (a) move away from ‘lack of’ and towards ‘typical/atypical’ terminology, in order to identify children with more subtle or atypical presentation of symptoms and (b) repeatedly monitor these behaviours, so that children who pass assessments at one age can still be identified at later examinations. This approach will facilitate the identification of not only those children with ‘broader’ ASD but also those children who show ‘regression’.

Limitations

It is unfortunate that the sample size at 12 months was small and that there was no DD/LD comparison group at this age to determine which behaviours were more specific to AD/ASD as opposed to DD/LD. Thus, the results from the 12-month analyses should be interpreted with caution. Furthermore, as a cross-sectional method was taken in this study to maximise sample numbers at each age, it was not possible to examine change across time within individuals. In addition, while it is feasible that some of the children in this study will cross diagnostic boundaries as they age, it remains the case that children who move from a diagnosis of AD to ‘broader’ ASD nonetheless present with the ‘AD’ profile of symptoms from 12 to 24 months and can thus be identified on the basis of this profile. Moreover, the number of children, if any, who move off the spectrum, are expected to be minimal on the basis of previous findings (Chawarska et al., 2007; Chawarska et al., 2009; Le Couteur et al., 2008; Lord, 1995; Paul et al., 2008; Stone et al., 1999; Turner et al., 2006). However, only a follow-up study can confirm this outcome, which is currently underway with this sample at 48-months of age. Finally, only children identified ‘at risk’ in the SACS were included in this study; therefore, we do not know the true–false negative rate of the original large sample. However, as shown in Barbaro and Dissanayake (2010), the false negative rate would have been low based on the current prevalence rates of 1% (216 out of 20,770 children were identified ‘at risk’ for an ASD, with a false positive rate of 19%). Nevertheless, the issue of false negatives must be taken into consideration when interpreting the results in this article, as those few children missed may have milder symptoms of ASD that were not captured though the original SACS.

Conclusions/future directions

Developmental surveillance of the early markers of ASDs during routine consultations at MCH centres in the SACS was successful in identifying children with ASDs as well as some children with DD/LD (Barbaro and Dissanayake, 2010). These items were therefore identified as useful during Level 1 surveillance. Detailed analysis of each of these items in this article revealed the most discriminating individual markers at each age and a predictive group of markers at 24 months, which can be used during Level 2 surveillance to more accurately identify children with AD and ASD. Overall, the key markers comprised social and joint attention behaviours, which is consistent with previous findings from both retrospective and prospective studies.

The key markers of AD and ASD differed across the ages, and tools must therefore be tailored according to the age at which children are assessed. Future studies should aim to identify whether subgroups of children later diagnosed with AD/ASD emerge as showing a particular ‘pattern’ of deficits, with children in these subgroups potentially showing similar deficits at comparable ages. The results suggest that while an absence of key markers will identify the majority of children on the spectrum, the presence of any single behaviour should not be used to negate or ‘rule out’ the possibility of an autism spectrum diagnosis, as many children later diagnosed with ‘broader’ ASD will, in fact, pass some of the key items during their second year of life. Due to this variable symptom presentation across the second year in children with ASDs, screening tools administered at only one age will continue to have problems with sensitivity. It is thus vital that the markers identified in this article are monitored at regular intervals during the second year of life to increase the early identification of young children across the spectrum of autism.

Footnotes

Appendix 1 Funding

This research was supported by a Telstra Foundation Community Development Fund and the first author was supported by a Sir Robert Menzies Allied Health Scholarship.

Notes

References

Adrien

Lenoir

Matineau

. (1993) Blind ratings of early symptoms of autism based upon family home movies. Journal of the American Academy of Child and Adolescent Psychiatry 32: 617–626.

Allison

Baron-Cohen

Wheelwright

. (2008) The Q-CHAT (Quantitative CHecklist for Autism in Toddlers): a normally distributed quantitative measure of autistic traits at 18-24 months of age: preliminary report. Journal of Autism and Developmental Disorders 38: 1414–1425.

Baird

Charman

Baron-Cohen

. (2000) A screening instrument for autism at 18-months of age: a six-year follow-up study. Journal of the American Academy of Child and Adolescent Psychiatry 39: 694–702.

Baranek

(1999) Autism during infancy: a retrospective video analysis of sensory motor and social behaviors at 9-12 months of age. Journal of Autism and Developmental Disorders 29: 213–224.

Barbaro

Dissanayake

(2009) Autism spectrum disorders in infancy and toddlerhood: a review of the evidence on early signs, early identification tools, and early diagnosis. Journal of Developmental and Behavioral Pediatrics 30: 447–459.

Barbaro

Dissanayake

(2010) Prospective identification of autism spectrum disorders in infancy and toddlerhood using developmental surveillance: the Social Attention and Communication Study (SACS). Journal of Developmental and Behavioral Pediatrics 31: 376–385.

Barbaro

Dissanayake

(2012) Developmental profiles of infants and toddlers with autism spectrum disorders identified prospectively in a community-based setting. Journal of Autism and Developmental Disorders. Epub ahead of print 7 February 2012. DOI: 10.1007/s10803-10012-11441-z.

Barbaro

Ridgway

Dissanayake

(2011) Developmental surveillance of infants and toddlers by maternal and child health nurses in an Australian community-based setting: promoting the early identification of autism spectrum disorders. Journal of Pediatric Nursing 26: 334–347.

Baron-Cohen

Allen

Gillberg

(1992) Can autism be detected at 18-months? The needle, the haystack, and the CHAT. British Journal of Psychiatry 161: 839–843.

10.

Baron-Cohen

Cox

Baird

. (1996) Psychological markers in the detection of autism in infancy in a large population. British Journal of Psychiatry 168: 158–163.

11.

Bryson

Zwaigenbaum

McDermott

. (2008) The Autism Observation Scale for Infants: scale development and reliability data. Journal of Autism and Developmental Disorders 38: 731–738.

12.

Chawarska

Klin

Paul

. (2007) Autism spectrum disorder in the second year: stability and change in syndrome expression. Journal of Child Psychology and Psychiatry 48: 128–138.

13.

Chawarska

Klin

Paul

. (2009) A prospective study of toddlers with ASD: short-term diagnostic and cognitive outcomes. Journal of Child Psychology and Psychiatry 50: 1235–1245.

14.

Clifford

Dissanayake

(2008) The early development of joint attention in infants with autistic disorder using home video observations and parental interview. Journal of Autism and Developmental Disorders 38: 791–805.

15.

Dawson

. (2008) Early behavioral intervention, brain plasticity, and the prevention of autism spectrum disorder. Development and Psychopathology 20: 775–803.

16.

Dietz

Swinkels

van Daalen

. (2006) Screening for autistic spectrum disorder in children aged 14 to 15 months. II: population screening with the Early Screening of Autistic Traits Questionnaire (ESAT). Design and general findings. Journal of Autism and Developmental Disorders 36: 713–722.

17.

Howell

(2010) Statistical Methods for Psychology. Belmont, CA: Wadsworth.

18.

Interactive Autism Network (2008) IAN research report #6, regression. Available at: http://www.iancommunity.org (accessed 17 August 2008).

19.

Kleinman

Robins

Ventola

. (2008) The Modified Checklist for Autism in Toddlers: a follow-up study investigating the early detection of autism spectrum disorders. Journal of Autism and Developmental Disorders 38: 827–839.

20.

Landa

(2008) Diagnosis of autism spectrum disorders in the first 3 years of life. Nature Clinical Practice Neurology 4: 138–147.

21.

Landa

Holman

Garrett-Mayer

(2007) Social and communication development in toddlers with early and later diagnosis of autism spectrum disorders. Archives of General Psychiatry 64: 853–864.

22.

Le Couteur

Haden

Hammal

. (2008) Diagnosing autism spectrum disorder in pre-school children using two standardised assessment instruments: the ADI-R and the ADOS. Journal of Autism and Developmental Disorders 38: 362–372.

23.

Lord

(1995) Follow-up of two-year-olds referred for possible autism. Journal of Child Psychology and Psychiatry 36: 1365–1382.

24.

Lord

Risi

Lambrecht

. (2000) The Autism Diagnostic Observation Schedule-Generic: a standard measure of social and communication deficits associated with the spectrum of autism. Journal of Autism and Developmental Disorders 30: 205–223.

25.

Lord

Rutter

DiLavore

. (1999) Autism Diagnostic Observation Schedule–WPS Edition. Los Angeles, CA: Western Psychological Services.

26.

Lord

Shulman

DiLavore

(2004) Regression and word loss in autism spectrum disorder. Journal of Child Psychology and Psychiatry 45: 1–21.

27.

Mandell

Novak

Zubritsky

(2005) Factors associated with age of diagnosis among children with autism spectrum disorders. Pediatrics 116: 1480–1486.

28.

Mitchell

Brian

Zwaigenbaum

. (2006) Early language and communication development of infants later diagnosed with autism spectrum disorder. Journal of Developmental and Behavioral Pediatrics 27: S69–S78.

29.

Mullen

(1995) Mullen Scales of Early Learning (MSEL). Circle Pines, MN: AGS.

30.

Nadig

Ozonoff

Young

. (2007) A prospective study of response to name in infants at risk for autism. Archives of Paediatric and Adolescent Medicine 161: 378–383.

31.

Oosterling

Wensing

Swinkels

. (2009) Advancing early detection of autism spectrum disorder by applying an integrated two-stage screening approach. Journal of Child Psychology and Psychiatry 51: 250–258.

32.

Osterling

Dawson

Munson

(2002) Early recognition of one year old infants with autism spectrum disorder versus mental retardation: a study of first birthday party home videotapes. Development and Psychopathology 14: 239–251.

33.

Paul

Chawarska

Cicchetti

. (2008) Language outcomes of toddlers with autism spectrum disorders: a two year follow-up. Autism Research 1: 97–107.

34.

Robins

Fein

Barton

. (2001) The Modified Checklist for Autism in Toddlers: an initial study investigating the early detection of autism and pervasive developmental disorders. Journal of Autism and Developmental Disorders 31: 131–144.

35.

Rogers

. (2009) What are infant siblings teaching us about autism in infancy?. Autism Research 2: 125–137.

36.

Rogers

Vismara

(2008) Evidence-based comprehensive treatments for early autism. Journal of Clinical Child and Adolescent Psychology 37: 8–38.

37.

Stone

Lee

Ashford

. (1999) Can autism be diagnosed accurately in children under 3 years?. Journal of Child Psychology and Psychiatry 40: 219–226.

38.

Sullivan

Finelli

Marvin

. (2007) Response to joint attention in toddlers at risk for autism spectrum disorder: a prospective study. Journal of Autism and Developmental Disorders 37: 37–48.

39.

Swinkels

SHN

Dietz

van Daalen

. (2006) Screening for autistic spectrum in children aged 14 to 15 months. I: the development of the Early Screening of Autistic Traits Questionnaire (ESAT). Journal of Autism and Developmental Disorders 36: 723–732.

40.

Turner

Stone

Pozdol

. (2006) Follow-up of children with autism spectrum disorders from age 2 to age 9. Autism 10: 243–265.

41.

Watson

Baranek

Crais

. (2007) The First Year Inventory: retrospective parent responses to a questionnaire designed to identify one-year-olds at risk for autism. Journal of Autism and Developmental Disorders 37: 49–61.

42.

Werner

Dawson

(2005) Validation of the phenomenon of autistic disorder regression using home videotapes. Archives of General Psychiatry 62: 889–895.

43.

Werner

Dawson

Osterling

. (2000) Brief report: recognition of autism spectrum disorder before one year of age: a retrospective study based on home videotapes. Journal of Autism and Developmental Disorders 30: 157–162.

44.

Wetherby

Brosnan-Maddox

Peace

. (2008) Validation of the Infant-Toddler Checklist as a broadband screener for autism spectrum disorders from 9 to 24 months of age. Autism 12: 487–511.

45.

Wetherby

Watt

Morgan

. (2007) Social communication profiles of children with autism spectrum disorders late in the second year of life. Journal of Autism and Developmental Disorders 37: 960–975.

46.

Young

Brewer

Pattison

(2003) Parental identification of early behavioural abnormalities in children with autistic disorder. Autism 7: 125–143.

47.

Zwaigenbaum

Bryson

Rogers

. (2005) Behavioral manifestations of autism in the first year of life. International Journal of Developmental Neuroscience 23: 143–152.