Parent-reported patterns of loss and gain in communication in 1- to 2-year-old children are not unique to autism spectrum disorder

Abstract

We compared loss and gain in communication from 1 to 2 years in children later diagnosed with autism spectrum disorder (n = 41), language impairment (n = 110) and in children with typical language development at 7 years (n = 831). Participants were selected from a prospective population cohort study of child language (the Early Language in Victoria Study). Parent-completed communication tools were used. As a group, children with autism spectrum disorder demonstrated slower median skill gain, with an increasing gap between trajectories compared to children with typical development and language impairment. A proportion from all groups lost skills in at least one domain (autism spectrum disorder (41%), language impairment (30%), typical development (26%)), with more children with autism spectrum disorder losing skills in more than one domain (autism spectrum disorder (47%), language impairment (15%, p = 0.0003), typical development (16%, p < 0.001)). Loss was most common for all groups in the domain of ‘emotion and eye gaze’ but with a higher proportion for children with autism spectrum disorder (27%; language impairment (12%, p = 0.03), typical development (14%, p = 0.03)). A higher proportion of children with autism spectrum disorder also lost skills in gesture (p = 0.01), sounds (p = 0.009) and understanding (p = 0.004) compared to children with typical development but not with language impairment. These findings add to our understanding of early communication development and highlight that loss is not unique to autism spectrum disorder.

Keywords

autism spectrum disorder communication language regression skill loss trajectory

Introduction

Autism spectrum disorder (ASD) is one of the most common neurodevelopmental disorders and is known to be heterogeneous. In addition to phenotypic heterogeneity, children with ASD demonstrate heterogeneous patterns of onset and trajectories of development (Kim et al., 2015; Landa et al., 2013; Ozonoff et al., 2010; Shumway et al., 2011). There has been a substantial focus on early social communication development in ASD because a deeper understanding of the loss and gain in skills in the first few years of life has the potential to build knowledge of how ASD unfolds. It may also assist identification, help delineate potential subgroups of children with ASD and shed light on the neurobiological mechanisms that underpin the disorder. However, one of the challenges in studying early social communication development in ASD is that children are not typically diagnosed before 3 years of age (Bent et al., 2015).

Retrospective studies have traditionally described two distinct patterns of onset in ASD: early onset ASD and regressive ASD. Early onset ASD is the gradual unfolding of symptoms over the course of the first few years of life. Regressive onset is a pattern of normal or near normal development followed by an abrupt or gradual loss of skills (Stefanatos, 2008), primarily involving the loss of words and/or social communication skills (Lord et al., 2004). Approaches to defining and measuring regression are varied in the literature. The vast majority of retrospective studies have used the Autism Diagnostic Interview–Revised (ADI-R; Lord et al., 1994) to measure regression. This tool defines loss of skill as being ‘communicative use of at least five different words (other than “mama” and “dada”) on a daily basis for at least 3 months’ and requires that there has been a loss of the skill for at least 3 months (quoted from a test booklet of an assessment tool; Lord et al., 1994). A small number of retrospective studies have used the Regression Validation Interview (Luyster et al., 2005; Thurm et al., 2014). This is a detailed semi-structured parent interview that obtains more comprehensive information on skill attainment and loss. Prospective studies have generally taken a repeated measures approach by examining change in social communication skills and language across frequent time intervals (Brian et al., 2014; Landa et al., 2013; Ozonoff et al., 2010).

A systematic review of 85 studies (n = 29,035) of which 98% were retrospective, and the remainder high-risk sibling studies, included data on regression in ASD (Barger et al., 2013). The prevalence of regression in ASD was reported to be approximately 32% (95% confidence interval (CI) 30–35). The average age at which regression occurred was 20 months (95% CI 1.7–1.9; Barger et al., 2013). In children with typical development (TD), language impairment (LI) or developmental delay, regression has been reported as less common, ranging from 0% to 24% (Baird et al., 2008; Landa et al., 2013; Lord et al., 2004; Pickles et al., 2009; Thurm et al., 2014).

Retrospective studies of children with ASD carry a number of methodological limitations that substantially increase risk of bias and limit detailed assessment of development (Ozonoff et al., 2008, 2011). By contrast, high-risk sibling studies have been developed to prospectively monitor development on the pathway to ASD, recruiting siblings of children with ASD during pregnancy or soon after birth (Chawarska et al., 2014; Messinger et al., 2013; Ozonoff et al., 2010; see Jones et al., 2014 for a review). The findings from high-risk studies have challenged the way early social communication development in ASD has traditionally been understood, suggesting that dichotomous categories of regression do not adequately capture the many ways ASD can emerge (Kalb et al., 2010; Landa et al., 2013; Landa and Garrett-Mayer, 2006; Ozonoff et al., 2010; Shumway et al., 2011; Siperstein and Volkmar, 2004). Some studies, for example, have proposed a range of onset patterns including early onset (early delays, no loss), delay and regression (some early delays before loss), plateau (no early delays, no loss) and regression (no delays before a clear loss of skills; Kalb et al., 2010; Ozonoff et al., 2008; Shumway et al., 2011). Whereas two high-risk sibling studies grouped children by different developmental trajectory types or classes such as accelerated development, normative development, delay in some areas of development (e.g. fine/gross motor and language) and widespread delay with declining trajectories (Brian et al., 2014; Landa et al., 2012). These studies have highlighted the importance of placing loss of different types of skills within the context of trajectories of development.

Different elements of development have been reported to be lost by different studies. Specifically, loss of eye contact and gaze to faces has been reported to be the most common social communication skill lost by a number of retrospective studies (e.g. Ozonoff et al., 2005; Thurm et al., 2014). Eye gaze, shared positive affect, social interest, smiling and initiating interactive games have been reported to show declining trajectories in prospective high-risk sibling studies (Landa et al., 2007, 2013; Ozonoff et al., 2010) with word loss less common (Landa et al., 2013).

There has been debate about whether children in high-risk sibling studies are representative of children without a family history of ASD. For example, high-risk siblings may have different phenotypes to children without a family history of ASD (Mitchell et al., 2006; Pandey et al., 2008; Sebat et al., 2007) and may be more at risk of other developmental difficulties, including LI and learning difficulties (Landa et al., 2013; Szatmari et al., 2000).

To our knowledge no longitudinal population-based studies have been published that have investigated loss and gain in communication skills prior to the child’s ASD diagnosis. Nor have any studies used large samples of children without ASD as comparison groups. This study uses data from a prospective population-based study investigating language development collected at 1 and 2 years to describe early communication trajectories.

The aim of this study was to describe loss and gain of communication skills from 1 to 2 years in children later diagnosed with ASD, LI and in children with typical language development. Specifically, we wanted to investigate the proportion of children in each group who lost skills, the type of skills that were lost and the spread of loss across communication domains.

Methods

Study design

Participants were drawn from the Early Language in Victoria Study (ELVS), a prospective, longitudinal, population-based study in Victoria, Australia (Reilly et al., 2006, 2007). Children (n = 1910) were recruited from 6 of 31 geographical areas in Melbourne, 8–10 months of age and followed up almost annually. The Socio-Economic Index for Areas (SEIFA) from the 2001 census data was used (based on income, educational attainment and unemployment from the Census of Population and Housing) to stratify 31 local government areas (LGAs) into three levels (high, medium and low). Two LGAs from each level were then approached for recruiting participants. Participants were recruited via Maternal Child Health nurses, at universal hearing screening appointments, and through public notices. Exclusion criteria included cerebral palsy, known developmental delays or other significant intellectual or physical disability. Participants were also excluded if parents did not speak or read English to a satisfactory level to complete the questionnaires. Demographic information was collected from the primary caregiver when the child was 8 months of age. Parent questionnaires were completed annually, providing information about communication and behaviour development, family factors and the child’s environment. Direct assessments (Clinical Evaluation of Language Fundamentals (CELF)–fourth edition (Semel et al., 2003), and/or Preschool–second edition (Wiig et al., 2006) and the Wechsler Abbreviated Scale of Intelligence (WASI; block design and matrix reasoning subtests; Wechsler, 1999)) were completed by trained psychology graduates or speech pathologists in the child’s home or school, depending on the wave of the study and caregiver preference. The CELF was administered at 4, 5 and 7 years and the WASI at 7 years. Further details on the ELVS study design and sampling methods can be found in Reilly et al. (2006, 2007).

Ethical considerations

Ethical approval was obtained from the Human Research Ethics Committee at the Royal Children’s Hospital, Melbourne (#23018) and La Trobe University, Human Ethics Committee (#03–32).

Outcome measures

Two parent-completed questionnaires, the Communication Symbolic Behaviours Scales–Infant Toddler Checklist (CSBS-ITC; Wetherby and Prizant, 2002) and the MacArthur-Bates Communicative Development Inventories (CDI; Fenson, 1993) were used to assess early communication at 1 and 2 years. The CDI (Words and Gestures form and Words and Sentences form) is a widely used parent-completed measure of early communication. Only the expressive vocabulary data were used here. The CDI has been validated in the general population and in children with developmental difficulties and has frequently been applied in ASD studies (e.g. Bopp and Mirenda, 2011; Charman et al., 2003; Luyster et al., 2007; Mitchell et al., 2006). The CSBS-ITC is a parent-completed questionnaire on early communication, social and symbolic development. It has been validated as a broadband screener for ASD (Wetherby et al., 2008). There are seven clusters: emotion and eye gaze (0–8 possible points), communication (0–8), gestures (0–10), sounds (0–8), words (0–6), understanding (0–6) and object use (0–11). The cluster scores are collapsed into three composite scores: social (emotion, eye gaze, communication, gesture; 0–26), speech (words and sounds; 0–14) and symbolic (understanding and use of objects; 0–17). A total score (0–57) can also be calculated. We used raw scores from the CSBS-ITC and CDI in this study because we were specifically interested in examining loss of skills.

Group allocation

ASD

From 4 years of age, all participating parents (n = 1623 at 4 years) were asked via the annual questionnaire whether their child had ever received a diagnosis of ASD. If yes, parents were telephoned by a clinical psychologist or speech pathologist (M.P., P.E.) experienced in the diagnosis of ASD during which information was obtained about how the diagnosis was made, the name and professions of the individuals making the diagnosis, child age at diagnosis and type of diagnosis (e.g. Asperger’s syndrome, autistic disorder) as allocated using classification systems of the time, like the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association (APA), 2000). Information about the child’s behaviour and co-morbidities was also obtained. The majority (93%) of children were given a clinical diagnosis by a multidisciplinary team that involved at least two professionals (typically a paediatrician and psychologist and/or speech pathologist). The other children were given a diagnosis by a paediatrician only. When the children were aged 7 years parents were asked to complete the Social Communication Questionnaire (SCQ), a widely used screening questionnaire derived from the ADI-R (Rutter et al., 2003). Mean scores on the SCQ were 14.53 (standard deviation (SD): 4.66; range 5–26; Veness et al., 2014). By 7 years, 44 children had been diagnosed with ASD with a mean age of reported diagnosis of 4 years 8 months (range 2–7 years). Further details on the children with ASD in this sample can be found in Veness et al. (2012, 2014).

LI and typical language development

Children were allocated to these categories following review of their performance on the CELF (Preschool–second edition and/or fourth edition) and the WASI. None of the children in the LI and TD groups were reported to have ASD. Figure 1 shows the tools used for this study, the ages they were administered and their role in group allocation and measuring outcomes of interest. Allocation to groups was completed when children were 7 years.

Figure 1.

Tools and ages at which they were used in this study to measure outcomes and to allocate groups.

Exclusion criteria

A total of 743 children were excluded because (1) there were incomplete or missing data at both time points (1 and 2 years), (2) they were not assessed or did not complete the CELF at 7 years and at one other time point and (3) they did not complete a non-verbal IQ measure at 7 years. They were also excluded if they did not fit into the pre-specified groups (ASD, LI and TD), for example, if the child was delayed on both the CELF-4 and the WASI and was not diagnosed as ASD. The total sample in this study consisted of 982 children (ASD: n = 41; LI: n = 110; and TD: n = 831).

Statistical procedures/data analysis

All analyses were conducted using Stata version 13.1. The ELVS sample is skewed towards families who are more advantaged and thus SEIFA quintiles, based on the Australian Bureau of Statistics reference, were used for adjusted data analysis where appropriate.

Trajectories

On inspection prior to analysis, it was clear that the data distribution was asymmetric for the children with ASD for the CSBS-ITC and CDI scores at each wave. Hence, non-parametric analyses were conducted and medians (interquartile ranges) were reported instead of means (SDs). Raw scores on the CSBS-ITC (social, speech and symbolic composites and total score) were plotted for each individual with ASD at 1 and 2 years. Classification into trajectory types was done as follows: acceleration: higher gradient than the mean gradient for TD children, improving: higher score at 2 than 1 year but no acceleration, plateau: same score at 1 and 2 years and loss: lower score at 2 than 1 year. Median CDI scores (number of words produced) for each group were plotted at 1 and 2 years. Comparison was made between median scores on CSBS-ITC and CDI for children with ASD, TD and LI at 1 and 2 years using the Wilcoxon rank sum test, corrected for multiple comparisons using Bonferroni correction with p < 0.002 required to reach statistical significance. Box plots were graphed to show the differences in medians (interquartile ranges) and the rate and direction of change on the CSBS-ITC composites and total scores from 1 to 2 years. Questionnaires at both 1 and 2 years were all completed within 2 months either side of the child’s birthday.

Definition of loss of skills

The period between 1 and 2 years is typically a period of great growth in communication skills (Fenson and Dale, 1994; Goldfield and Reznick, 1990) and it would be considered unusual for a child to lose skills previously obtained over this 1-year period, either by not using a previously acquired skill or by changing from being able to use a skill ‘always’ to only using the skill ‘sometimes’. Hence, children were classified as having lost skills if they achieved a lower raw score at 2 than 1 year on the CSBS-ITC and/or if they had fewer words on the CDI at 2 years compared to 1 year. The definition of ‘loss of skills’ for the CDI was developed because to our knowledge there is no published, standardised definition of how ‘loss of skills’ should be defined. Our definition is conservative because between 1 and 2 years of age children often have rapid word acquisition, and gain of only a single word or no word gain could also be considered atypical. In the absence of agreement, we decided a conservative approach was ideal.

CSBS-ITC

Differences in scores were calculated for cluster, composite and total scores. Average magnitude of loss (and range) was calculated for each group. The number of children per group who lost skills in one or more clusters was calculated and graphed. Odds ratios (with CIs and p values) were used to compare loss of skills in the cluster, composite and total scores across the different groups (ASD vs LI; ASD vs TD). Odds ratios were adjusted for gender and socio-economic status.

CDI

The difference between scores at 1 and 2 years was calculated using the total expressive vocabulary score on the Words and Gestures and Words and Sentences forms. We used the full vocabulary from the latter form to provide a more conservative estimate. This ensured that substitution of early words with more complex words was not counted as a loss.

Results

There was a higher proportion of males in the ASD group compared to the TD and LI groups (p < 0.01), see Table 1 for descriptive characteristics. No significant difference was found in the ASD group compared to the TD and LI groups on any other feature.

Table 1.

Sample characteristics for children with ASD, language impairment and typical development.

	ASD (n = 41)	LI (n = 110)	TD (n = 831)	p
Male (%)	83	55	45	ASD > TD (<0.001) ASD > LI (=0.002)
Number of children at home (mean (SD; range))	1.7 (0.7; 1–3)	1.9 (0.9; 1–5)	1.7 (0.8; 1–5)	LI > TD (=0.0007)
English as main language spoken at home (%)	97.6	94.6	97.5	–
Indigenous (%)	0	0	0.12	–
Married or de facto (%)	97.6	91.8	96.9	TD > LI (=0.008)
SEIFA score (mean (SD; range))	1029.8 (65.5; 855.5−1104.1)	1012.4 (68.1; 834.4−1107.6)	1046.7 (51.8; 834.4−1147.0)	TD > LI (<0.001)
Maternal age at birth (mean (SD; range))	32.9 (5.4; 21−46)	32.2 (4.8; 19−43)	32.4 (4.2; 19−44)	–
Primary caregiver has completed high school (grade 12; %)	77.5	63.3	82.2	TD > LI (<0.001)

ASD: autism spectrum disorder; LI: language impairment; TD: typical development; SD: standard deviation; SEIFA: Socio-Economic Index for Areas.

Socio-economic status was measured using the SEIFA Disadvantage Index. SEIFA has a mean of 1000 and a SD of 100 with higher scores indicating greater advantage. Only the p values that reached significance (p < 0.01) for the group comparisons are reported.

Trajectories

Individual trajectories of raw scores from the CSBS-ITC were plotted for children with ASD (Figure 2). Substantial variation between children with ASD is shown, with four main trajectory types seen on both the CSBS-ITC and the CDI: worsening, plateau, improving and accelerating (Figure 3). The majority of children fell into the ‘improved’ trajectory type for both tools. The median composite and total scores at 1 year for the children in all three groups who lost skills were within the average range (see Figure 2, for example).

Figure 2.

Individual trajectories of communication development (social, speech, symbolic composites and total raw scores, CSBS-ITC) from 1 to 2 years for children with ASD (n = 41).

Figure 3.

Trajectory types for children with ASD (loss, plateau, improve and accelerate) for the CSBS-ITC and the CDI.

Median raw scores for all three groups increased over time for both the CDI and CSBS-ITC (Figures 4 and 5 and Table 2). At 1 year, no significant difference was seen in scores on the CDI between the children with TD, LI and ASD groups. However, there was a widening gap between the TD and ASD/LI groups over the next 12 months, which reached significance (p < 0.0001) at 2 years for the difference between the ASD and the TD group but not between the children with ASD and LI.

Figure 4.

Trajectory of median number of words produced for children with ASD, LI and TD at 1 and 2 years.

Figure 5.

Median and interquartile range for raw scores on the CSBS-ITC (total score and composite scores) from 1 to 2 years for children with ASD, LI and TD.

Table 2.

CDI median scores for each group (ASD, LI and TD) at 1 and 2 years.

	Median (interquartile range)
	Aged 1 year	Aged 2 years
ASD	2 (0−6)	99.5 (34−206)*
LI	1 (0−4)	114 (41−227)
TD	4 (0−8)	282 (175−404)*

CDI: Communicative Development Inventories; ASD: autism spectrum disorder; LI: language impairment; TD: typical development.

Significant difference between these two medians (p < 0.0001).

On the CSBS-ITC, there was a statistically significant difference between the median scores for children with ASD and children with TD at 1 year in the social composite (p < 0.002) and the total score (p < 0.002). At 2 years, a statistically significant difference between medians was found for all the composites and total score for the ASD group compared to the LI and TD groups (p < 0.002), except in the speech composite where there was not a statistically significant difference between the LI and ASD children.

Differences in proportions of children who lost communication skills across groups

Only two children in the study sample had a smaller vocabulary at 2 than 1 year on the CDI. One child diagnosed with ASD used five words at 1 year (grandma, baa, grr, mummy and daddy) and only 2 words at 2 years (mummy and daddy). The other child in the LI group used 3 words at 1 year (mummy, daddy and grandma) and only 1 word at 2 years (here).

A proportion of children in all three groups had lower scores on the CSBS-ITC at 2 than 1 year. Percentages are reported for CSBS-ITC cluster scores within each composite on the CSBS-ITC. The ASD group had a higher percentage of children who lost skills compared to the LI and TD groups in all clusters except ‘use of objects’ (Table 3). In total, 41% of children with ASD lost skills in at least one cluster compared to LI (30%) and TD (26%). The cluster with the greatest loss for all groups was ‘emotion and use of eye gaze’, with the ASD group having around double the percentage of children who lost skills compared to the other two groups (27% ASD, 12% LI and 14% TD). Few females in the ASD group demonstrated loss of skills, while the number who lost skills in the LI and TD groups were more evenly split by gender (Table 3).

Table 3.

Proportion of children with ASD, LI and TD with lower scores at 2 years compared to 1 year on the CSBS-ITC.

	ASD		LI		TD
	n = 41		n = 110		n = 831
	Males n = 34 (%)	Females n = 7 (%)	Males n = 62 (%)	Females n = 48 (%)	Males n = 374 (%)	Females n = 457 (%)
Emotion eye gaze	29	14	10	15	13	16
Communication	12	29	6	10	9	13
Gesture	18	0	3	6	4	5
Social composite	15	0	6	4	4	8
Sounds	18	14	8	13	3	5
Words	3	0	0	2	0.5	0
Speech composite	9	0	5	4	1	0.4
Understanding	6	0	3	4	0	0.7
Object use	0	0	2	2	0	0.2
Symbolic composite	3	0	3	0	0	0.2
Total score	6	0	0	0	0.3	0.7

ASD: autism spectrum disorder; LI: language impairment; TD: typical development; CSBS-ITC: Communication Symbolic Behaviours Scales–Infant Toddler Checklist.

Odds ratios were calculated to investigate whether children with ASD had greater odds of losing skills than the other two groups (Table 4). In the cluster of ‘emotion and use of eye gaze’, the odds of losing skills for the children with ASD were 2.3 to 3 times higher than those for children with TD (p < 0.05) and LI (p < 0.05) respectively. Loss of skills was also significantly higher for children with ASD than TD in the clusters of gesture (p < 0.05), sounds (p < 0.01) and understanding (p < 0.01). The proportion of children who lost skills was significantly higher in children with ASD compared to TD in the speech (p < 0.001) and symbolic (p < 0.05) composites as well as the total score (p < 0.01). Odds ratios were not statistically different for other comparisons that were possible between ASD and LI, and odds ratios could not be calculated for the total score in children with LI compared to ASD as no child in the LI group demonstrated a loss of skills.

Table 4.

Adjusted odds ratios for loss of skills in children with ASD compared to LI and TD.

	ASD versus LI		ASD versus TD
	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p
Emotion eye gaze	3 (1.1−7.9)*	0.03	2.3 (1.1−4.8)*	0.029
Communication	2.5 (0.7−8.5)	0.14	1.6 (0.6−3.9)	0.345
Gesture	3.1 (0.8−11.6)	0.09	3.5 (1.32−9.27)*	0.012
Social composite	1.9 (0.5−6.9)	0.35	2.5 (0.9−6.9)	0.08
Sounds	1.4 (0.3−6.7)	0.67	5.2 (2.0−13.3)**	0.001
Words	2.4 (0.1−43.7)	0.56	7.5 (0.64−87.8)	0.109
Speech composite	1.4 (0.3−6.7)	0.66	7.5 (1.6−34.1)**	0.009
Understanding	1.4 (0.2−9.5)	0.71	19.3 (2.6−145.4)**	0.004
Object use	–		–
Symbolic composite	1.4 (0.2−9.5)	0.90	31.1 (1.5−650.6)*	0.027
Total score	–	–	13.6 (2.0−92.2)**	0.007

ASD: autism spectrum disorder; LI: language impairment; TD: typical development; CI: confidence interval.

Odds ratios were adjusted for gender and neighbourhood disadvantage (using Socio-Economic Index for Areas (SEIFA) quintiles as previously described). Only adjusted figures are presented in the table. There was little difference between adjusted and unadjusted figures in all odds ratios except for the cluster of gesture for the ASD versus LI comparison, which no longer met significance once adjusted.

p < 0.05; **p < 0.01.

Range of loss of communication skills

The children in the ASD group were comparable to the other groups for magnitude of loss (mean and range) on the CSBS-ITC, with amount of loss ranging from −6 to −1 in the ASD group, −6 to −1 in the LI group and −7 to −1 in the TD group.

Number of children who lost communication skills in one or more cluster

A proportion of children from all groups lost skills in at least one domain (Figure 6). A higher proportion of children with ASD lost skills in more than one area compared to the other two groups (ASD > TD: p < 0.001; ASD > LI: p < 0.01). For those children who lost skills in more than one cluster, they were more likely to lose in the ‘eye contact and emotion’ and ‘communication’ domains.

Figure 6.

Proportion of children who lost skills in one or more clusters for each group (ASD, LI and TD).

Discussion

Trajectories

Children in all three groups demonstrated an overall median gain in all areas of communication from 1 to 2 years. However, individual trajectories were variable and there were differences between the groups in the rate of gain of skills. Our findings are consistent with those of other prospective studies reporting slower developmental trajectories from 1 to 2 years for children with ASD relative to their peers on language (Landa et al., 2007, 2012; Mitchell et al., 2006; Ozonoff et al., 2010) and social communication (Bryson et al., 2007; Landa et al., 2007, 2012; Ozonoff et al., 2010; Rozga et al., 2011; Veness et al., 2014; Young et al., 2009; Zwaigenbaum et al., 2005). The finding that there was no significant difference between children with ASD and LI at 1 and 2 years on both the CDI and the speech cluster on the CSBS-ITC has implications for differential diagnosis for these two groups at this young age. The CSBS-ITC speech composite includes questions about both speech (e.g. number of sounds used) and language (e.g. number of words produced or word combinations used). Prospective studies have found mixed evidence on whether there are differences in early language and speech between high-risk siblings who develop ASD and those who do not (see Jones et al., 2014 for a review). Very few of these prospective studies have compared speech and language development in children with ASD to a defined group of children later diagnosed with LI. Two studies with a comparison group of children with LI have produced similar findings to this study (Landa and Garrett-Mayer, 2006; Paul et al., 2008). One prospective community-based study did not find a significant difference between the language-impaired/developmentally delayed children (n = 20) and those with ASD (n = 42) on an expressive language measure at 24 months (Barbaro and Dissanayake, 2012). However, there was a significant difference between the children with LI and the children with autistic disorder. A study focused on speech found children with ASD had a higher proportion of atypical vocalisations (e.g. squeal, growl), compared to age and language matched peers but there were no significant differences in the proportion of correct consonants produced (Schoen et al., 2011). It seems likely that if there are differences between the groups in speech sound development they are unlikely to be detected by a parent-rated tool such as the CSBS-ITC. Given the similarities in trajectories for language-impaired and ASD groups, it would seem using speech and language ability for differential diagnosis may not be useful but, instead, contributes to an understanding of their ability to communicate and participate, that may impact other behaviours.

Loss of skills

Loss of social communication skills

A minority of children lost skills, either completely or with a decrease in their frequency of use, in all three groups but with a greater risk of skill loss and more pervasive skill loss in children with ASD. Of those children with ASD who lost skills, the majority lost skills in the ‘use of emotion and eye gaze’ cluster. This is in keeping with findings from past retrospective (Ozonoff et al., 2005; Thurm et al., 2014) and prospective studies (Landa et al., 2007, 2013; Ozonoff et al., 2011) investigating skill loss in ASD. Children in our study did not regress significantly more than other groups in the ‘communication’ cluster. This finding was unexpected given communication is considered a key diagnostic feature of ASD. One possible explanation for this finding is that a child’s ‘vulnerability’ to losing skills may occur at different stages of development for different skills. In support of this hypothesis, one study found some skills typically problematic in ASD (e.g. response to joint attention, amount of requesting and use of gesture) had improving trajectories between 1 and 2 years whereas other skills (e.g. eye contact and overall quality of rapport) had declining trajectories (Lord, 2014).

In general, the children in this study lost more in social communication skills than in other skills such as use of words. However, loss of skills was not as common as has been reported in other prospective studies (e.g. Ozonoff et al., 2010). There are several possible explanations for this. Our study assessed children at only two time points so may not have captured some of the subtle losses and variability that may have occurred in communication. Because several questions have been combined into clusters and composites, children may experience a loss in some areas and gain in others and this may counterbalance their overall score and underestimate loss of skills. It may also be argued that social communication skills are more challenging for parents to assess compared to direct clinical observations of behaviour such has been used in some prospective studies (Landa et al., 2007; Ozonoff et al., 2010).

Loss of words

Very few children in this study lost words on the CDI. This was perhaps unsurprising given children may have few words to lose at 1 year of age. Two children (one each from the ASD and LI group) lost words. Both had a small number of words at 1 year of age (3 and 5 words). The ages children were assessed in this study were limited to the waves data collected in the broader study and consequently may have underestimated loss as children may lose and regain words within the 1- to 2-year period.

Loss across multiple skill domains

A majority of children lost skills in one cluster; yet, many also made gains in other clusters over the same time period. This suggests loss of skills, when it does occur, is not typically pervasive across all areas of communication (Meilleur and Fombonne, 2009; Thurm et al., 2014). Of those children who lost skills in more than one area, the majority had ASD. Only one child (of n = 41) in this study represented the more ‘classic’ description of regression with substantial loss of skills across multiple clusters. Consistent with this finding there has been increasing consensus in the literature that dramatic, catastrophic loss of skills in ASD is quite rare (Lord, 2014; Ozonoff et al., 2010; Siperstein and Volkmar, 2004; Thurm et al., 2014).

Loss of skills in children without ASD

Unexpectedly, a proportion of non-ASD children in this study were reported by parents to have a loss of communication skills from 1 to 2 years. One prospective study also found skill loss in children without ASD (Brian et al., 2014). In this study, 7.5% of the total sample (n = 29/389) plateaued or had a lower raw score on a subsequent developmental assessment. Here, 2.8% of low-risk infants, 5.2% of high-risk infants without ASD and 21% of high-risk infants with ASD demonstrated a reduction in raw scores, thus, suggesting loss of skills is not unique to ASD. To date, all studies investigating loss in non-ASD children have been retrospective (Baird et al., 2008; Lord et al., 2004; Pickles et al., 2009; but see Landa et al., 2013 and Brian et al., 2014 for exception). Few high-risk studies have compared developmental trajectories of the non-ASD children in detail and to our knowledge no prospective study has included a sample of TD and LI children as large as this study (Landa et al., 2007, 2013; Mitchell et al., 2006; Ozonoff et al., 2010). It is possible the smaller samples of non-ASD children in high-risk samples may not have captured the broad range of developmental trajectories that may occur in TD children. No prospective studies have compared ASD to a defined group of language-impaired children. In addition, they have not included children with ASD who were being diagnosed as late as middle childhood. As a result, our study may include a higher proportion of children with milder phenotypes and later diagnoses.

Another important difference between this study and others is the tools used to determine loss of skills. The questions in the CSBS-ITC and CDI are quite specific and based on skills being used at the current time in multiple natural contexts. In contrast, the majority of retrospective studies (such as Baird et al., 2008; Pickles et al., 2009) have relied on general questions about regression from tools that were not specifically designed to study regression. One prospective study of regression that investigated development of specific social communication behaviours reported a mean decrease in some communicative behaviours over discrete time periods in children with TD. This included gaze to faces between 6 to 18 months and directed vocalisation from 24 to 36 months (Ozonoff et al., 2010). It is likely that a proportion of TD and LI children experience some natural wax and wane in the development of specific communication skills, and loss of skills may be a natural part of development for some children. This suggestion is supported by other studies of communication development in children, which have reported a lack of stability in language skills measured over time (Lord et al., 2012; Ukoumunne et al., 2012). Our knowledge of loss of skills in children who have LI or TD remains limited, as there are very few studies that have specifically investigated loss of communication skills at an individual level.

Study limitations include that the ASD diagnostic assessments were not completed within the study but in the community and therefore were not uniform. The study may be prone to selection bias since we only included those children that had completed specific waves of the study. However, a high proportion of children were retained in the ASD group (93%). Our classification and cut-off points for loss of skills, plateau, improving and accelerating were quite narrowly defined. We are not aware of any published, consistently used guidelines or definitions of loss and trajectory classifications that have been used in longitudinal studies. Given there is substantial instability in these early years and our cut-offs required just one point difference, it is possible some children were misclassified. We have made our definitions as clear as possible in the article, so the reader can interpret the findings along with their limitations. Measurement error and classification bias can occur in parent-reported tools such as the CDI and CSBS-ITC. This may have influenced the number of children who had a reduction in scores and may have impacted the study findings. The use of other groups (TD and LI) for comparison in this study, however, may mitigate the effect of classification bias to some extent. One would not expect between group differences for data collected prior to problems being identified, unless there are characteristics of parents of children with ASD that are in some way different to parents of non-ASD children that influence the way they score.

Clinical implications

This study demonstrated that children have variable pathways to ASD, TD or LI and these pathways may or may not include some loss of skills. The variability and overlap in trajectories between the groups highlights the challenge in making differential diagnoses at this young age (Lord et al., 2012; Veness et al., 2014). Consistent with the findings of other prospective studies, children in this study demonstrated more loss in social communication domains than in words. Change in social communication may be subtle, and it may be that tools have not yet been adequately developed to capture these changes over time. Children who experience dramatic loss of skills were rare in our sample, but these children warrant special investigation and attention. A deeper understanding of the rate of change, patterns of derailment and individual variation in trajectories will help inform the development of targeted, developmentally sensitive intervention, the most effective and efficient resource allocation as well as inform prognosis. This is highly relevant for families, clinicians and policymakers.

Footnotes

Funding

Amanda Brignell is supported by an Australian Postgraduate Award scholarship. Infrastructure support was provided by the Victorian Government’s Operational Infrastructure Support Program. We wish to thank the William Collie Trust Fund for their financial support. This funding organization was not involved in the development, design, analysis, or interpretation of the study.

References

American Psychiatric Association (APA) (2000) Diagnostic Statistical Manual of Mental Disorders (DSM-IV-TR). 4th ed. Text revision. Washington, DC: APA.

Baird

Charman

Pickles

. (2008) Regression, developmental trajectory and associated problems in disorders in the autism spectrum: the SNAP study. Journal of Autism and Developmental Disorders 38: 1827–1836.

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 42: 1939–1948.

Barger

Campbell

McDonough

(2013) Prevalence and onset of regression within autism spectrum disorders: a meta-analytic review. Journal of Autism and Developmental Disorders 43: 817–828.

Bent

Dissanayake

Barbara

(2015) Mapping the diagnosis of autism spectrum disorders in children aged under 7 years in Australia, 2010–2012. Medical Journal of Australia 202: 317–320.

Bopp

Mirenda

(2011) Prelinguistic predictors of language development in children with autism spectrum disorders over four-five years. Journal of Child Language 38: 485–503.

Brian

Roncadin

Duku

. (2014) Emerging cognitive profiles in high-risk infants with and without autism spectrum disorder. Research in Autism Spectrum Disorders 8: 1557–1566.

Bryson

Zwaigenbaum

Brian

. (2007) A prospective case series of high-risk infants who developed autism. Journal of Autism and Developmental Disorders 37: 12–24.

Charman

Drew

Baird

. (2003) Measuring early language development in preschool children with autism spectrum disorder using the MacArthur Communicative Development Inventory (Infant Form). Journal of Child Language 30: 213–236.

10.

Chawarska

Shic

Macari

. (2014) 18-month predictors of later outcomes in younger siblings of children with autism spectrum disorder: a baby siblings research consortium study. Journal of the American Academy of Child and Adolescent Psychiatry 53: 1317–1327.

11.

Fenson

(1993) MacArthur Communicative Development Inventories: User’s Guide and Technical Manual. San Diego, CA: Singular Publishing Group.

12.

Fenson

Dale

(1994) Variability in early communicative development. Monographs of the Society for Research in Child Development 59: 1–173.

13.

Goldfield

Reznick

(1990) Early lexical acquisition: rate, content, and the vocabulary spurt. Journal of Child Language 17: 171–183.

14.

Jones

Gliga

Bedford

. (2014) Developmental pathways to autism: a review of prospective studies of infants at risk. Neuroscience and Biobehavioural Reviews 39: 1–33.

15.

Kalb

Law

Landa

. (2010) Onset patterns prior to 36 months in autism spectrum disorders. Journal of Autism and Developmental Disorders 40: 1389–1402.

16.

Kim

Macari

Koller

. (2015) Examining the phenotypic heterogeneity of early autism spectrum disorder: subtypes and short-term outcomes. Journal of Child Psychology and Psychiatry and Allied Disciplines. Epub ahead of print 12 August. DOI: 10.1111/jcpp.12448.

17.

Landa

Garrett-Mayer

(2006) Development in infants with autism spectrum disorders: a prospective study. Journal of Child Psychology and Psychiatry and Allied Disciplines 47: 629–638.

18.

Landa

Gross

Stuart

. (2012) Latent class analysis of early developmental trajectory in baby siblings of children with autism. Journal of Child Psychology and Psychiatry 53: 986–996.

19.

Landa

Gross

Stuart

. (2013) Developmental trajectories in children with and without autism spectrum disorders: the first 3 years. Child Development 84: 429–442.

20.

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.

21.

Lord

(2014) Early regression and worsening in autism spectrum disorders. In: Australasian society for autism research conference, Melbourne, VIC, Australia, 4–5 August.

22.

Lord

Luyster

Guthrie

. (2012) Patterns of developmental trajectories in toddlers with autism spectrum disorder. Journal of Consulting and Clinical Psychology 80: 477–489.

23.

Lord

Rutter

Le Couteur

(1994) The Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders 24: 659–685.

24.

Lord

Shulman

DiLavore

(2004) Regression and word loss in autistic spectrum disorders. Journal of Child Psychology and Psychiatry and Allied Disciplines 45: 936–955.

25.

Luyster

Qiu

Lopez

. (2007) Predicting outcomes of children referred for autism using the MacArthur-Bates Communicative Development Inventory. Journal of Speech Language and Hearing Research 50: 667–681.

26.

Luyster

Richler

Risi

. (2005) Early regression in social communication in autism spectrum disorders: ACPEA study. Developmental Neuropsychology 27: 311–336.

27.

Meilleur

Fombonne

(2009) Regression of language and non-language skills in pervasive developmental disorders. Journal of Intellectual Disability Research 53: 115–124.

28.

Messinger

Young

Ozonoff

. (2013) Beyond autism: a baby siblings research consortium study of high-risk children at three years of age. Journal of the American Academy of Child and Adolescent Psychiatry 52: 300–308.

29.

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.

30.

Ozonoff

Heung

Byrd

. (2008) The onset of autism: patterns of symptom emergence in the first years of life. Autism Research 1: 320–328.

31.

Ozonoff

Iosif

A-M

Baguio

. (2010) A prospective study of the emergence of early behavioral signs of autism. Journal of the American Academy of Child and Adolescent Psychiatry 49: 256–266.

32.

Ozonoff

Iosif

A-M

Young

. (2011) Onset patterns in autism: correspondence between home video and parent report. Journal of the American Academy of Child and Adolescent Psychiatry 50: 796–806.

33.

Ozonoff

Williams

Landa

(2005) Parental report of the early development of children with regressive autism: the delays-plus-regression phenotype. Autism 9: 461–486.

34.

Pandey

Verbalis

Robins

. (2008) Screening for autism in older and younger toddlers with the Modified Checklist for Autism in Toddlers. Autism 12: 513–535.

35.

Paul

Chawarska

Volkmar

(2008) Differentiating ASD From DLD in Toddlers. Perspectives on Language Learning and Education 15: 101–111.

36.

Pickles

Simonoff

Conti-Ramsden

. (2009) Loss of language in early development of autism and specific language impairment. Journal of Child Psychology and Psychiatry 50: 843–852.

37.

Reilly

Eadie

Bavin

. (2006) Growth of infant communication between 8 and 12 months: a population study. Journal of Paediatrics and Child Health 42: 764–770.

38.

Reilly

Wake

Bavin

. (2007) Predicting language at 2 years of age: a prospective community study. Pediatrics 120: e1441–1449.

39.

Rozga

Hutman

Young

. (2011) Behavioral profiles of affected and unaffected siblings of children with autism: contribution of measures of mother-infant interaction and nonverbal communication. Journal of Autism and Developmental Disorders 41: 287–301.

40.

Rutter

Bailey

Lord

(2003) The Social Communication Questionnaire. Los Angeles, CA: Western Psychological Services.

41.

Schoen

Paul

Chawarska

(2011) Phonology and vocal behavior in toddlers with autism spectrum disorders. Autism Research 4: 177–188.

42.

Sebat

Lakshmi

Malhotra

. (2007) Strong association of de novo copy number mutations with autism. Science 316: 445–449.

43.

Semel

Wiig

Secord

(2003) Clinical Evaluation of Language Fundamentals, Australian Version. 4th ed. San Antonio, TX: The Psychological Corporation.

44.

Shumway

Thurm

Swedo

. (2011) Brief report: symptom onset patterns and functional outcomes in young children with autism spectrum disorders. Journal of Autism and Developmental Disorders 41: 1727–1732.

45.

Siperstein

Volkmar

(2004) Brief report: parental reporting of regression in children with pervasive developmental disorders. Journal of Autism and Developmental Disorders 34: 731–734.

46.

Stefanatos

(2008) Regression in autistic spectrum disorders. Neuropsychology Review 18: 305–319.

47.

Szatmari

MacLean

Jones

. (2000) The familial aggregation of the lesser variant in biological and nonbiological relatives of PDD probands: a family history study. Journal of Child Psychology and Psychiatry 41: 579–586.

48.

Thurm

Manwaring

Luckenbaugh

. (2014) Patterns of skill attainment and loss in young children with autism. Development and Psychopathology 26: 203–214.

49.

Ukoumunne

Wake

Carlin

. (2012) Profiles of language development in pre-school children: a longitudinal latent class analysis of data from the Early Language in Victoria Study. Child: Care, Health and Development 38: 341–349.

50.

Veness

Prior

Bavin

. (2012) Early indicators of autism spectrum disorders at 12 and 24 months of age: a prospective, longitudinal comparative study. Autism 16: 163–177.

51.

Veness

Prior

Eadie

. (2014) Predicting autism diagnosis by 7 years of age using parent report of infant social communication skills. Journal of Paediatrics and Child Health 50: 693–700.

52.

Wechsler

(1999) Wechsler Abbreviated Scale of Intelligence. New York: The Psychological Corporation.

53.

Wetherby

Prizant

(2002) CSBS DP Manual: Communication and Symbolic Behavior Scales: Developmental Profile. Baltimore, MD: Paul H. Brookes.

54.

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.

55.

Wiig

Secord

Semel

(2006) Clinical Evaluation of Language Fundamentals Preschool (Australian Edition). 2nd ed. Marrickville, NSW, Australia: Harcourt Assessment.

56.

Young

Merin

Rogers

. (2009) Gaze behavior and affect at 6 months: predicting clinical outcomes and language development in typically developing infants and infants at risk for autism. Developmental Science 12: 798–814.

57.

Zwaigenbaum

Bryson

Rogers

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