Journal of Autism and Developmental Disorders

, Volume 40, Issue 11, pp 1389–1402 | Cite as

Onset Patterns Prior to 36 Months in Autism Spectrum Disorders

  • Luther G. Kalb
  • J. K. Law
  • Rebecca Landa
  • Paul A. Law
Original Paper

Abstract

The present study investigated differences among children with three different patterns of autism symptom onset: regression, plateau, and no loss and no plateau. Cross-sectional data were collected from parents of children aged 3–17 years with an autism spectrum disorder (n = 2,720) who were recruited through a US-based online research database. Parental report of developmental characteristics was assessed through a parent questionnaire, and current autism symptoms were measured via the Social Responsiveness Scale and Social Communication Questionnaire. Multivariate analyses indicated that children with regression had a distinct developmental pattern marked by less delayed early development. However, following regression, these children evinced elevated autism symptom scores and an increased risk for poorer outcomes when compared with the other onset groups.

Keywords

Autism Pervasive developmental disorders Regression Symptom Onset pattern 

Autism is a complex neurobehavioral syndrome that is behaviorally defined and qualitatively revealed through disruption in the development of social, language, and communication skills, and the presence of restricted interests and/or stereotyped behaviors (American Psychiatric Association [APA] 2000). Kanner’s (1943) seminal paper described autism as a disorder where symptoms emerge shortly after birth; however, observational data during the past several decades suggest there are several different patterns of autism symptom onset. At present, the relationship between symptom onset pattern and outcome in children with an autism spectrum disorder (ASD) remains unclear. The aim of the current study was to address this gap in the literature by examining developmental, educational-behavioral, and diagnostic outcomes among children with three different patterns of autism symptom onset prior to 36 months: regression, plateau, and no loss and no plateau.

According to the International Classification of Mental and Behavioural Disorders, tenth edition (ICD-10; 1992, revised 2007) (World Health Organization 1992) and the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSMIVTR; 1994, revised 2000) (American Psychiatric Association 2000), the pathognomonic signs of pervasive developmental delay must be present prior to 3 years of age to meet the criteria for an autism diagnosis. While the majority of children with an ASD display symptoms by 36 months of age (Short and Schopler 1988; Volkmar et al. 1985), current research suggests that symptoms emerge in three different patterns (Ozonoff et al. 2008). The most common of these involves an early manifestation of symptoms in the first year of life, as described by Kanner (1943). Counter to intuition, several investigators have reported that children with early detectable social-communication deficits are not necessarily more severely affected than those with autism whose symptoms become apparent later in development (e.g., Landa et al. 2007; Werner et al. 2005).

The second onset pattern, termed developmental- or autistic regression, involves a loss of previously acquired skills. Regression among children with an ASD can be generally defined as a loss of previously acquired social, communication, and/or motor skills prior to 36 months of age (Kobayashi and Murata 1998; Kurita 1985; Lord et al. 2004; Luyster et al. 2005; Rogers and DiLalla 1990). Some parents report regression after “near typical” development; however, a preponderance of the evidence shows that children with regression often display developmental abnormalities prior to the loss (Kurita 1985; Rogers and DiLalla 1990; Werner et al. 2005). It remains unclear, however, as to whether there are differences in outcomes between children whose development is noticeably delayed before the regression and those who regress after gaining more advanced skills such as phrase speech.

A growing body of research suggests regression can affect one or several domains of development. The domains of loss are typically classified as language, social, and motor; cognitive loss is still under dispute. Both Goldberg et al. (2003) and Hansen et al. (2009) report one-third to half of the children they studied with regression had both social and language losses, while a fifth to over one-third presented with social losses alone, respectively. Few studies have specifically examined children who lost motor skills, as this form of regression is quite rare. While a few studies suggest there are minimal differences between children who lose different or overlapping skills (e.g., Lord et al. 2004; Luyster et al. 2005), further research is required to validate these findings.

Numerous studies have reported that children with regression have poorer outcomes in verbal, social, and behavioral domains than children with nonregressive ASD (Bernabei et al. 2007; Hoshino et al. 1987; Kurita 1985; Kurita et al. 1992; Rogers and DiLalla 1990; Short and Schopler 1988; Wilson et al. 2003). In contrast to those findings, a number of other authors have reported minimal to no differences between affected children with and without developmental regression (Chakrabari and Fombonne 2001; Davidovitch et al. 2000; Lord et al. 2004; Werner et al. 2005). Despite conflicting findings, parent-based retrospective reports, coupled with findings from several other empiric studies, consistently show that regression occurs in a third (Goldberg et al. 2003) to half (Davidovitch et al. 2000) of children with an ASD. However, estimates decline as low as 15–18% when a strict definition of regression is used (Kobayashi and Murata 1998). The timing and pattern of onset for regression also vary across children. Rogers (2004) posits that 65% of families report a gradual regression, making the identified age of skill loss more difficult, while other studies report the mean age of loss to be at around 19–21 months (Davidovitch et al. 2000; Goldberg et al. 2003). It is generally accepted that regression, which rarely occurs before 12 months of age, generally takes place (~50%) between the 1st and 2nd years, with the remaining proportion occurring between the 2nd and 3rd birthdays (Landa 2008).

The final onset pattern involves milestone achievement followed by a plateau or developmental halt. First described by Siperstein and Volkmar (2004), children in this group generally display only mild delays until approximately 2 years of age, when they experience a gradual to abrupt developmental arrest that restricts further advancement of skills. These children may achieve the most basic social communication skills; however, these emerging abilities fail to develop into more advanced skills (Landa 2008). These children are of particular interest to this study because little is known about this onset type.

The overarching goal of the present study was to investigate differences in developmental characteristics and outcomes among children grouped by symptom onset pattern: regression (hereafter referred to as the skill loss group); plateau; and no loss and no plateau. More specifically, the first goal was to examine group differences in parent reported milestone achievement and in current ‘outcomes’ involving autism symptom severity, autism diagnosis, presence or absence of cognitive impairment and phrase speech, and educational placement and supports. The second goal was to examine differences, in the above-stated outcomes, within the skill loss group based on parent reported degree of regression severity (i.e., mild, moderate, or severe) and type of skill most severely affected by the regression (social, language, or motor). The third aim was to examine variations in outcome in two subgroups of children with skill loss: those whose parents reported phrase speech and no concerns prior to the loss and those with early developmental abnormalities that predated the skill loss. Our final aim was to report the characteristics and outcomes of children who exhibited a developmental plateau.

Methods

Participants

Data were collected from a US-based online research database, the Interactive Autism Network (IAN). This unique research mechanism is designed to foster collaboration between the autism community and investigators while assisting with overcoming traditional barriers to research. IAN, launched in April 2007, is now the nation’s largest online autism research effort, with more than 10,000 children with an ASD and more than 20,000 of their family members registered to participate.

Our analysis was conducted using data extracted on November 16, 2009 for participants 3–17 years of age (M = 8 years) with completed primary history questionnaires and autism symptom measures (n = 3,680). Of all registered respondents with completed forms, those who reported a diagnosis of fragile X syndrome (n = 37), or tubular sclerosis (n = 3) were excluded; Rett syndrome is an exclusion criterion for registering with IAN, similar to how a diagnosis of ASD is requisite for registration. As the goal of our study was to examine onset patterns in the first 3 years of life, children of parents who reported developmental concerns after their child reached the age of 3 years (n = 432 or 7%) were excluded (max = 13 years). We chose this approach due to the concern that the late onset group may be disproportionately affected by recall bias and ultimately, these children require an in-depth analysis that is beyond the scope of the present study.

Participants were classified into the following onset groups: skill loss (SL), plateau (PL), and no loss and no plateau (NLNP). The SL group was comprised of children whose parents reported a noticeable loss in a previously acquired skill. These parents also responded to questions regarding which skill was most affected by the loss (social, language, or motor), the subjective severity of the loss (mild, moderate, or severe), and age at which they first noticed the loss. Consistent with methods employed by previous studies of regression (Kurita 1985; Lord et al. 2004), children with a reported loss of skill after 3 years of age (n = 57) were removed from the analysis. Of these 57 children, five parents reported a diagnosis of Child Disintegrative Disorder (CDD). We chose not to include these children in our analysis given the focus of this investigation was on ASD onset patterns in the first 3 years of life, the very small sample size, and previous reports in the literature suggest that children with CDD are different from those with autism in both clinical presentation and developmental course (Volkmar and Cohen 1989).

The PL group consisted of children whose parents reported that their child had experienced “a development plateau or halt such that he/she stopped gaining new skills but retained previously acquired skills” and who provided the age at which they first noticed this plateau. The remaining children were classified into the NLNP group (i.e., children who did not meet our criteria for the SL or PL groups). Lastly, to screen for ASD, we used a cutoff score of ≥15 on the Social Communication Questionnaire (see the Social Communication Questionnaire subsection) as recommended by Berument et al. (1999). The final study sample totaled 2,720 participants.

Measures

IAN Questionnaire

The IAN Project data collection consists of multiple topic-specific forms that were authored by the IAN Research team in collaboration with other researchers and families. All families completed the initial registration and then were invited to complete additional questionnaires that included a profile on each affected child. Independent and dependent variables from these questionnaires included items about demographics, diagnosis, developmental milestones, and education. For current ASD diagnosis, parents reported their child as having (1) autism or autistic disorder, (2) Asperger syndrome (AS), (3) pervasive developmental disorder–not otherwise specified (PDD–NOS), or (4) other ASD (reported as generic “autism spectrum disorder” or “pervasive developmental disorder”). For intellectual disability status, participants were categorized as intellectually disabled (ID) if parents have (a) ever reported a diagnosis of “mental retardation” or (b) an IQ score of <70. For educational outcomes, to be included in the analysis a child had to be enrolled in a public or private school (not enrolled, n = 45; home schooled, n = 59) and in grades pre-K through grade 12 (>grade 12, n = 12). Items then were dichotomized to calculate Odds Ratios; for further explanation of an Odds Ratio, see “Data Collection and Analysis”. For instance, parents were asked how much time their child spent in a special education classroom each day. Participants were coded as a 1 if parents indicated the child spent most or all of each day in a special education classroom or a 0 if the child spent most or all of the time in a regular education classroom; a similar procedure was used for use of a one-to-one aide.

Social Communication Questionnaire

Originally developed as the Autism Screening Questionnaire, the Social Communication Questionnaire (SCQ; Western Psychological Services, Los Angeles) is a widely used, valid, and reliable autism-screening tool (Berument et al. 1999) based on the DSMIVTR criteria and the Autism Diagnostic Interview–Revised (Western Psychological Services, Los Angeles). The SCQ consists of 40 dichotomous items that are summed to produce a total score, ranging from 0 to 39 for verbal children and from 0 to 33 for nonverbal children. A t score of 15 or higher is highly suggestive of ASD. In the original standardization sample, Berument et al. (1999) reported means of 23.08 (SD = 8.07), 7.03 (SD = 7.01), and 17.03 (SD = 4.09) for children with autism, atypical autism, and Asperger syndrome, respectively. The first item on the SCQ, responsible for guiding the scoring algorithm and cut-off scores, was used to determine the presence of phrase speech. The SCQ Lifetime version was offered to all IAN participants.

Social Responsiveness Scale

The Social Responsiveness Scale (SRS) is a 65-item parent/teacher report that measures autism symptomatology (Constantino et al. 2003). In contrast to the SCQ, the quantitative approach of the SRS allows for dimensional measurement of autistic traits including social and communication impairment and stereotyped/repetitive behaviors. Each SRS item rates the frequency of a particular behavior on a Likert scale (0–3); total scores range from 0 to 195, with higher scores indicating an increase in symptom severity. Psychometric studies of the SRS indicate strong reliability and validity (Constantino et al. 2003; Pine et al. 2006) and Constantino et al. (2003) report means of 117.4 (SD = 29.9) and 111.0 (SD = 27.4) on the parent report form of the SRS for children with autistic and Asperger/PDD-NOS disorders.

Data Collection and Analysis

All survey data entered by parents are maintained in the IMRS© (MDLogix, Baltimore). Electronic consent was elicited from participating families under the auspices of the Johns Hopkins Medicine Institutional Review Board (#NA_00002750). Analyses were performed using STATA 9.2 (College Station, TX) on the database.

Chi-square and ANOVA analyses were used to examine demographic and developmental differences among onset groups for categorical and continuous variables. Subsequent analyses using logistic regression (logit) were employed to develop ORs for categorical variables (e.g., ASD diagnosis). Odds Ratios were used to examine the likelihood of an event occurring in one onset group (e.g., SL group) compared with the odds of it occurring in another (e.g., NLNP) while accounting for the variance attributed by potentially confounding variables such as age, gender, and intellectual disability. To note, the NLNP, mild regression, and language loss groups served as reference categories in computations. For continuous dependent variables (e.g., autism symptom scores), data were modeled using a linear model. To adjust for multiple comparisons and avoid spurious results, a Bonferroni correction was applied.

Results

Demographic Differences

Descriptive demographic and developmental statistics are presented by onset type in Table 1, by severity of the regression in Table 2, and by type of skill most severely affected by the loss in Table 3. There were some demographic differences between onset pattern groups including age (at form completion) and race. In particular, the NLNP group was older (M = 8.36 years) than the PL (M = 7.81 years) and SL (M = 7.99 years) groups (p < .05). Similarly, parents who reported a mild regression (M = 7.25 years) were younger than those with moderate (M = 7.77 years) or severe (M = 8.59 years) loss of skills (p < .05). Moreover, there were also racial differences between onset groups, with parent reporting a loss of skills being disproportionally Asian and Black/AA (p < .05). Parents who reported that language was most severely affected by the loss were also more likely to be Asian and less likely to be White (p < .05). As specified below, these differences in age and race were accounted for as covariates within individual models.
Table 1

Characteristics of ASD onset groups, 3–17 years

 

All participants

No loss or plateau (NLNP)

Plateau (PL)

Skill loss (SL)

p

N (%)

2,720

1070 (39)

469 (17)

1181 (44)

<.01abc

Mean age at form completion (years)

8.10

8.36

7.81

7.99

<.05ab; Nsc

Gender

N (%)

 Male

2,245 (83)

876 (82)

377 (80)

992 (84)

Nsabc

 Female

475 (17)

194 (18)

92 (20)

189 (16)

 

Race

N (%)

 Black/AA

137 (5)

42 (4)

21 (4)

74 (6)

Nsa; <.05bc

 White

2,457 (90)

992 (93)

428 (91)

1037 (88)

Nsa; <.001bc

 Asian

83 (4)

21 (2)

12 (3)

50 (4)

Nsa; <.05bc

 Other

43 (1)

15 (1)

8 (2)

20 (2)

Nsabc

Ethnicity

N (%)

 Hispanic

206 (7)

82 (8)

30 (6)

94 (8)

Nsabc

 Not Hispanic

2,514 (93)

988 (92)

439 (94)

1087 (92)

 

Diagnosis

N (%)

 Autism

1,365 (50)

428 (40)

225 (48)

712 (60)

 

  Adjusted OR

 

Ref

1.35

2.23

<.01a; <.001b

  CI

  

1.08–1.69

1.88–2.64

 

 Asperger

434 (16)

270 (25)

70 (15)

94 (8)

 

  Adjusted OR

 

Ref

.55

.25

<.01a; <.001b

  CI

  

.41–.74

.19–.32

 

 PDD-NOS

580 (21)

254 (24)

112 (24)

214 (18)

 

  Adjusted OR

 

Ref

.99

.70

Nsa; <.001b

  CI

  

.76–1.28

.57–.86

 

 Other

341 (13)

118 (11)

62 (13)

161 (14)

 

  Adjusted OR

 

Ref

1.18

1.24

Nsab

  CI

  

.85–1.65

.96–1.60

 

 Age at diagnosis (years)

 

4.35

3.82

3.25

<.001abc

Developmental characteristics (months)

First concerns

16.18

14.88

17.10

16.98

<.001ab; Nsc

First steps

14.10

14.65

14.55

13.43

Nsa; <.01bc

First words

17.84

21.16

20.12

14.30

Nsa; <.001bc

Phrase speech

37.28

35.46

37.53

38.96

Nsac; <.001b

Toilet trained

49.50

47.66

48.13

51.77

Nsa; <.001b; <.05c

Age at 1st noticed plateau or regression

 

24.06

19.53

 

Social Responsiveness Scale

Mean (SD)

 Total raw

110.32 (27.5)

108.21 (26.9)

108.60 (28.6)

112.92 (27.5)

Nsa; <.001bc

 T-score

88.25 (14.0)

87.29 (13.8)

87.61 (14.7)

89.38 (13.9)

Nsa; <.001bc

 Awareness

14.33 (3.8)

14.16 (3.7)

14.05 (3.9)

14.60 (3.8)

Nsa; <.001bc

 Cognition

21.02 (5.6)

20.72 (5.6)

20.98 (6.0)

21.30 (5.5)

Nsa; <.001bc

 Motivation

16.36 (5.8)

15.98 (5.7)

15.91 (5.9)

16.90 (5.8)

Nsa; <.001bc

 Stereotypies

21.26 (6.6)

20.91 (6.5)

20.80 (6.9)

21.77 (6.4)

Nsa; <.001bc

 Communication

37.35 (10.2)

36.44 (9.9)

36.86 (10.2)

38.36 (10.2)

Nsa; <.001bc

Social Communication Questionnaire

Mean (SD)

 Total

24.83 (5.5)

24.13 (5.5)

24.54 (5.6)

25.57 (5.5)

*<.05a; <.001bc

 Social

9.18 (3.2)

8.84 (3.2)

8.83 (3.1)

9.63 (3.2)

Nsa;<.001bc

 Communication

7.62 (2.3)

7.50 (2.3)

7.74 (2.4)

7.69 (2.3)

<.01a; <.01b; Nsc

 Stereo

6.20 (1.5)

6.04 (1.6)

6.19 (1.6)

6.34 (1.4)

Nsa; <.001b; <.05c

Phrase speech

SCQ item1 N (%)

 No

544 (20)

143 (13)

70 (15)

331 (28)

 

 Yes

2,176 (80)

927 (87)

399 (85)

850 (72)

 

 Adjusted OR

 

Ref

1.05

2.43

Nsa; <.001b

 CI

  

.77–1.44

1.95–3.04

 

Intellectual disability

N (%)

 No

2,259 (83)

893 (84)

407 (87)

959 (81)

 

 Yes

461 (17)

177 (16)

62 (13)

222 (19)

 

 Adjusted OR

 

Ref

.81

1.20

Nsab

 CI

  

.59–1.11

.966–1.5

 

Special education classroom setting

N (%)

 No

941 (40)

467 (50)

169 (43)

305 (31)

 

 Yes

1,394 (60)

474 (50)

226 (57)

694 (69)

 

 Adjusted OR

 

Ref

1.41

1.85

<.05a; <.001b

 CI

  

1.04–1.91

1.45–2.35

 

1:1 Aide

N (%)

 No

1,221 (49)

542 (54)

205 (47)

474 (44)

 

 Yes

1,293 (52)

457 (46)

234 (53)

602 (56)

 

 Adjusted OR

 

Ref

1.43

1.63

<.01a; <.001b

 CI

  

1.14–1.91

1.36–1.96

 

aNLNP versus PL

bNLNP versus SL

cPL versus SL

* Significant difference found after adjustment for age and race

Ns not significant

Table 2

Characteristics of skill loss group by severity

 

Mild

Moderate

Severe

p

N (%)

230 (19)

495 (42)

456 (39)

<.01ab; Nsc

Mean age at form completion (years)

7.25

7.77

8.59

<.05abc

Gender

N (%)

 Male

192 (83)

417 (84)

383 (84)

Nsabc

 Female

38 (17)

78 (16)

73 (16)

Nsabc

Race

N (%)

 Black/AA

15 (6)

26 (5)

33 (7)

Nsabc

 White

202 (88)

442 (89)

393 (86)

Nsabc

 Asian

7 (3)

20 (4)

23 (5)

Nsabc

 Other

6 (3)

7 (2)

7 (2)

Nsabc

Ethnicity

N (%)

 Hispanic

18 (8)

46 (9)

30 (7)

Nsabc

 Not Hispanic

212 (92)

449 (91)

426 (93)

 

Diagnosis

N (%)

 Autism

116 (50)

273 (55)

323 (71)

 

  Adjusted OR

Ref

1.22

2.49

Nsa; <.001b

  CI

 

.89–1.67

1.78–3.48

 

 Asperger

27 (12)

46 (9)

21 (5)

 

  Adjusted OR

Ref

.70

.27

 

  CI

 

.41–1.17

.15–.50

Nsa; <.001b

 PDD-NOS

53 (23)

102 (21)

59 (13)

 

  Adjusted OR

Ref

.87

.50

 

  CI

 

.59–1.26

.33–.76

Nsa; <.01b

 Other

34 (15)

74 (15)

53 (11)

 

  Adjusted OR

Ref

1.04

.81

Nsab

  CI

 

.67–1.63

.50–1.30

 

 Age at diagnosis (years)

3.51

3.47

2.94

Nsa; <.001bc

Developmental characteristics (months)

First concerns

16.10

16.88

17.53

Nsabc

First steps

13.62

13.63

13.12

Nsabc

First words

15.15

14.80

13.30

Nsabc

Phrase speech

35.94

39.14

40.43

Nsabc

Toilet trained

50.05

49.56

55.53

Nsa; <.01b; <.05c

Age at 1st noticed regression

20.14

19.56

19.19

Nsabc

Social Responsiveness Scale

Mean (SD)

 Total raw

110.57 (26.9)

110.01 (26.8)

117.25 (28.2)

Nsa; <.001bc

 T-score

88.29 (13.7)

87.96 (13.5)

91.48 (14.3)

Nsa; <.001bc

 Awareness

14.30 (3.4)

14.19 (3.7)

15.18 (4.0)

Nsa; <.001bc

 Cognition

21.04 (5.7)

20.91 (5.4)

21.85 (5.5)

Nsabc

 Motivation

16.51 (5.9)

16.44 (5.5)

17.58 (5.9)

Nsa; <.05bc

 Stereotypies

21.04 (6.4)

21.23 (6.4)

22.72 (6.4)

Nsa; <.01bc

 Communication

37.68 (9.8)

37.24 (9.9)

39.92 (10.5)

Nsa; <.01bc

Social Communication Questionnaire

Mean(SD)

 Total

24.49 (5.3)

25.15 (5.3)

26.58 (5.6)

Nsa; <.001bc

 Social

8.90 (3.1)

9.26 (3.0)

10.41 (3.2)

Nsa; <.001bc

 Communication

7.42 (2.3)

7.62 (2.2)

7.90 (2.3)

Nsabc

 Stereo

6.34 (1.3)

6.39 (1.5)

6.28 (1.5)

Nsabc

Phrase speech

SCQ item1 N (%)

 No

44 (19)

97 (20)

190 (42)

 

 Yes

186 (81)

398 (80)

266 (58)

 

 Adjusted OR

Ref

1.11

3.94

Nsa; <.001b

 CI

 

.75–1.67

2.65–5.85

 

Intellectual disability

N (%)

 No

195 (85)

419 (85)

345 (76)

 

 Yes

35 (15)

76 (15)

111 (24)

 

 Adjusted OR

Ref

.94

1.47

Nsab

 CI

 

.59–1.47

.95–2.28

 

Special education classroom setting

N (%)

 No

70 (37)

151 (36)

84 (22)

 

 Yes

120 (63)

270 (64)

304 (78)

 

 Adjusted OR

Ref

1.29

2.33

Nsa; <.001b

 CI

 

.83–2.01

1.44–3.75

 

1:1 Aide

N (%)

 No

98 (47)

214 (47)

162 (40)

 

 Yes

109 (53)

246 (53)

247 (60)

 

 Adjusted OR

Ref

1.14

1.52

Nsa; <.01b

 CI

 

.81–1.6

1.06–2.17

 

aMild versus moderate

bMild versus severe

cModerate versus severe

Ns not significant

Table 3

Characteristics of skill loss group by type of skill most severely affected

 

Language

Social

Motor

p

N (%)

902 (78)

198 (17)

57 (5)

<.001abc

Mean age at form completion (years)

7.97

7.84

8.35

Nsabc

Gender

N (%)

 Male

766 (85)

163 (82)

47 (83)

Nsabc

 Female

136 (15)

35 (18)

10 (17)

 

Race

N (%)

 Black/AA

58 (6)

8 (4)

5 (9)

Nsabc

 White

782 (87)

184 (93)

51 (89)

<.05a; Nsbc

 Asian/Asian

46 (5)

3 (1)

1 (2)

<.01ab; Nsc

 Other

16 (2)

3 (2)

0 (0)

Nsabc

Ethnicity

N (%)

 Hispanic

69 (8)

16 (8)

6 (10)

Nsabc

 Not Hispanic

833 (92)

182 (92)

51 (90)

 

Diagnosis

N (%)

 Autism

555 (61)

109 (55)

31 (54)

 

 Adjusted OR

Ref

.77

.76

Nsab

 CI

 

.56–1.05

.44–1.31

 

 Asperger

58 (6)

23 (11)

12 (21)

<.01a; <.001b

  Adjusted OR

Ref

1.96

3.89

 

  CI

 

1.16–3.32

1.90–7.96

 

 PDD-NOS

169 (19)

33 (17)

9 (16)

 

  Adjusted OR

Ref

.86

.80

Nsab

  CI

 

.57–1.30

.38–1.66

 

 Other

120 (14)

33 (17)

5 (9)

 

  Adjusted OR

Ref

1.30

.64

Nsab

  CI

 

.85–2.98

.25–1.64

 

 Age at diagnosis (years)

3.17

3.56

3.99

<.01ab; Nsc

Developmental characteristics (months)

First concerns

16.96

17.96

14.65

Nsab; <.05c

First steps

13.11

13.99

16.65

Nsa; <.01b; <.05c

First words

13.29

16.81

19.13

<.01ab; Nsc

Phrase speech

38.94

37.08

38.87

Nsabc

Toilet trained

51.83

50.89

52.32

Nsabc

Age at 1st noticed regression

19.22

21.04

19.70

<.001a; Nsbc

Social Responsiveness Scale

Mean (SD)

 Total raw

111.03 (27.6)

117.50 (26.9)

124.72 (24.4)

<.01a; <.001b; Nsc

 T-score

88.37 (13.9)

91.79 (13.7)

95.33 (12.5)

< .01a; <.001b; Nsc

 Awareness

14.40 (3.8)

14.97 (3.6)

16.12 (3.9)

Nsac; <.01b

 Cognition

20.96 (5.6)

22.14 (5.4)

23.02 (5.2)

<.05ab; Nsc

 Motivation

16.59 (5.8)

17.55 (5.6)

19.21 (5.9)

<.05a; <.01b; Nsc

 Stereo

21.39 (6.4)

22.67 (6.5)

24.46 (6.5)

<.05a; <.01bc

 Communication

37.69 (10.3)

40.18 (10.0)

41.91 (8.3)

<.01ab; Nsc

Social Communication Scale

Mean(SD)

 Total

25.30 (5.5)

25.85 (5.3)

27.82 (4.7)

Nsa; <.01bc

 Social

9.53 (3.2)

9.66 (2.3)

10.70 (2.7)

Nsa; <.001bc

 Communication

7.60 (2.3)

7.76 (2.3)

8.42 (2.1)

Nsa; <.001bc

 Stereo

6.29 (1.4)

6.46 (1.5)

6.54 (1.4)

Nsabc

Phrase speech

SCQ item1 N (%)

 No

267 (30)

43 (22)

14 (25)

 

 Yes

635 (70)

155 (78)

43 (75)

 

 Adjusted OR

Ref

.65

.82

<.01a; Nsb

 CI

 

.45–.95

.44–1.54

 

Intellectual disability

N (%)

 No

728 (81)

162 (82)

51 (89)

 

 Yes

174 (19)

36 (18)

6 (11)

 

 Adjusted OR

Ref

.98

.45

Nsab

 CI

 

.65–1.58

.19–1.1

 

Special education classroom setting

N (%)

 No

229 (30)

53 (32)

16 (35)

 

 Yes

572 (70)

110 (68)

29 (65)

 

 Adjusted OR

Ref

.94

.84

Nsab

 CI

 

.59–1.48

.39–1.82

 

1:1 Aide

N (%)

 No

365 (44)

76 (43)

23 (46)

 

 Yes

464 (56)

99 (57)

27 (54)

 

 Adjusted OR

Ref

1.07

.88

Nsab

 CI

 

.76–1.52

.48–1.58

 

aLanguage versus social

bLanguage versus motor

cSocial versus motor

Ns not significant

Developmental Characteristics

ANOVA models were used to examine developmental differences among groups. Results from the Bonferroni test for corrections revealed that first concerns arrived over 2 months later for the SL and PL groups compared to NLNP group (p < .001). First steps and words were reported earlier for the SL group compared with the PL and NLNP group (p < .01). Phrase speech was achieved over 3 months later for the SL group then the NLNP group (p < .001). Toilet training was also achieved 3 and 4 months later for the SL group compared to the PL (p < .05) and NLNP (p < .001) groups, respectively. Mean age at onset of plateau was 24.06 months, and onset of regression was 19.53 months.

A similar pattern of later first concerns, phrase speech and toilet training, and earlier first words, steps, and age at noticed loss was observed for those with more severe regression, though only toilet training reached statistical significance (p < .05). Children whose parents endorsed motor abilities as the skill most affected by the loss reported earlier first concerns compared to the social loss group and later first steps compared to children who lost social and language skills (all p < .05). Children whose parents identified language as the skill most affected by the loss had earlier first words compared to rest of the SL group and earlier age at first noticed regression compared to those with social losses alone (all p < .01).

Autism Symptom Severity Scores

The primary dependent variable examined was autism symptom severity score measured via the SRS and SCQ. While controlling for race and age and after a Bonferroni adjustment, results from the multiple linear model indicated that children in the SL group had higher SRS total raw scores (M = 112.92, SD = 27.5) compared with children in the PL (M = 108.60, SD = 28.6) and NLNP (M = 108.21, SD = 26.9) groups (p < .001). Children in the SL group (M = 25.57, SD = 5.5) also had higher SCQ scores than the NLNP (M = 24.13, SD = 5.5) and PL (M = 24.54, SD = 5.6) groups (p < .001). Although no differences in SRS scores were discovered between the PL and SL groups, once demographic differences were controlled for children in the PL group had significantly higher SCQ scores compared to those in the NLNP group (p < .05).

Regression severity and type of skill most affected by the loss were also associated with higher SRS scores. Children in the severe regression group had higher SRS raw scores (M = 117.25, SD = 28.2) than those in the moderate (M = 110.01, SD = 26.8) and mild (M = 110.57, SD = 26.9) regressive groups (p < .001); an identical pattern was observed for SCQ scores. Children whose parents reported social (M = 117.50, SD = 26.9) and motor (M = 124.74, SD = 24.4) skills were most affected by the loss also reported higher SRS scores compared to the language loss group (M = 111.03, SD = 27.6, p < .01); SCQ scores were only elevated for child with motor losses. These data indicate a higher degree of autism symptom severity for children who had a regression and especially for those with a severe and/or primarily motor skill loss. Means and standard deviations for t-scores and sub-scale scores for both the SRS and SCQ are reported in Tables 1, 2, and 3.

Development of First Words

To graphically display differences in language attainment by onset pattern groups, a nonparametric Kaplan–Meier procedure was used to estimate the proportion of children who achieved first words across age. Notably, we graphed the trend of skill achievement as opposed to failure because this function is more intuitive and easier to interpret. If a child had not acquired first words by the time of the survey, they were considered a censored observation (17%). Figure 1 shows that, at 18 months, roughly 30% more children with regression achieved words before the NLNP and PL groups (p < .001).
Fig. 1

Achievement of first words by onset pattern. Note: Data displayed by Kaplan–Meier curves

Diagnosis

A strong association between onset pattern and parent reported ASD diagnosis was uncovered. Children in the SL (OR = 2.23, CI.95 = 1.88, 2.64) and PL (OR = 1.35, CI.95 = 1.08, 1.69) groups were more likely to be diagnosed with autistic disorder compared to the NLNP group (p < .01). While children in the SL group were less likely to be diagnosed with PDD-NOS (OR = .70, CI.95 = .57, .86), both PL (OR = .55, CI.95 = .41, .74) and SL (OR = .25, CI.95 = .19, .32) groups were far less likely to be diagnosed with Asperger syndrome than those in the NLNP group (all p < .01). Significant differences among the three groups also were observed for age at diagnosis with the SL group receiving the diagnosis earliest (M = 3.25 years), followed by the PL group (M = 3.82 years), and lastly, the NLNP group (M = 4.35 years) (p < .001).

Within the SL group, parents who reported the regression as severe were most likely to also report a diagnosis of autism (OR = 2.49, CI.95 = 1.78, 3.48) rather than a diagnosis of PDD-NOS (OR = .50, CI.95 = .33, .76) or Asperger syndrome (OR = .27, CI.95 = .15, .50) compared to children in the mild group (all p < .001). Likewise, children in the severe regression group were the first of those in the children in the SL group, on average, to be diagnosed (M = 2.94 years). No significant differences were observed for the “Other” diagnostic category by onset type or regression severity as well as no diagnostic differences, including age at regression, were found between the mild and moderate groups. Lastly, parents of children in the language loss group reported an earlier age at diagnosis and were less likely to report a diagnosis of Asperger’s disorder compared to children in the motor (OR = 3.89, CI.95 = 1.90, 7.96) and social (OR = 1.96, CI.95 = 1.16, 3.32) loss groups (all p < .01).

Verbal and Cognitive Impairment

Logistic Regression Models were used to examine the association between intellectual disability and current absence of phrase speech (as measured by the first item on the SCQ) and onset pattern, controlling for age and race/ethnicity. Results from the analysis found children in the SL group (OR = 2.43, CI.95 = 1.95, 3.04), especially those with a severe loss (OR = 3.94, CI.95 = 2.65, 5.85) were least likely to have achieved phrase speech compared to children in the NLNP and mild groups (p < .001). Children with primarily social losses were at less risk, than those who primarily lost language, to not have attained phrase speech (OR = .65, CI.95 = .45, .95, p < .01). No significant differences in attainment of phrase speech were found for those who lost primarily motor skills, children in the PL group, or those with moderate losses. We also examined parent report of cognitive disability and no significant differences were detected between onset, regression severity, or skill loss groups.

Educational Supports

To explore the question of increased need for support in an academic setting, ORs were calculated using multiple logistic regression to examine the odds of (1) being placed in a special education classroom setting and (2) having a one-to-one aide. After adjustment for age, gender, race/ethnicity, cognitive delay, school setting, and the opportunity for classroom inclusion, we found that children with regression were more likely to be in a supportive academic setting (OR = 1.85, CI.95 = 1.45, 2.35) and have a classroom aide (OR = 1.63, CI.95 = 1.36, 1.96) than were those in the NLNP group (p < .001). Likewise, the PL group was more likely than the NLNP group to have been placed in a supportive educational setting (OR = 1.41, CI.95 = 1.04, 1.91) and have a one-to-one aide (OR = 1.43, CI.95 = 1.14, 1.91) (p < .05). Within the regression group, the odds of being in a special education setting were more than twice as high if the degree of regression was severe (OR = 2.33, CI.95 = 1.44, 3.75) compared with the mild group (p < .001). There was also an increase in the risk of receiving a one-to-one aide (OR = 1.52, CI.95 = 1.06, 2.17, p < .01) for the severe group. No significant differences were found between children who lost language, social, or motor skills.

Timing of Parental Concerns and Recognition of Skill Loss

A basic descriptive analysis was used to explore the timing of linguistic development, parents concerns, and loss of skills. We found that roughly 35% of parents had concerns about their child’s development before they first noticed a loss and that only 23% of children in the SL group had obtained phrase speech 1 month prior to the loss, although 81% had attained their first words.

To better understand if there were differences in outcomes between children whose development is noticeably delayed before the regression and those who lost more advanced skills, we identified children who had developed phrase speech and whose parents had yet to identify developmental abnormalities prior to the loss (13%; n = 153) and compared them with the remainder of the SL group. We found that children who had attained phrase speech and parents had yet to raise concerns had lower SRS scores (p < .05) and were less likely to report an autistic disorder diagnosis (OR = .67, CI95 = .47, .95) and cognitive disability (OR = .60, CI95 = .37, .96); no other differences were uncovered.

Discussion

Our data suggest there is a relationship between autism onset pattern and developmental, diagnostic, and educational-behavioral outcomes with a cogent association between a loss of skills in the first 3 years and poorer outcomes. These findings were particularly robust for the children whose parents reported the regression as severe.

One of the clearest trends in our data was the less delayed early development in the SL group compared with the PL and NLNP groups. We found that first steps and words arrived within normal limits, and parental concerns surfaced later for children of parents who had reported a regression. The Kaplan–Meier curves, exhibited in Fig. 1, illustrate the striking difference in the rate of early word development between the onset pattern groups, with children in the SL group, especially those with a severe loss, achieving their first words much earlier. Yet, despite less impaired early language development, after the regression, these children were quite delayed in phrase speech.

Regression severity and type of skill loss were also associated with milestone achievement. Children whose parents endorsed motor abilities as the primary skill affected by the regression reported earlier first concerns and later first steps, while children that primarily lost language skills had earlier first words and not surprisingly, age at first noticed regression. Although just outside of statistical significance, but rather interesting, was the relationship between increasing regression severity, later initial concerns, earlier nascent milestone achievement, and the growing gap between time of regression and attainment of complex skills. These findings coupled with similar intuitive developmental differences between onset pattern groups provide new insights and compelling evidence for regression in autism spectrum disorders.

Given the differentiation in milestone achievement, the next logical step was to examine if there was a relationship between these patterns and diagnostic outcomes. As expected, children with regression were more than 2 times more likely to receive an autistic disorder diagnosis, and were between 30 and 75% less likely to receive a PDD-NOS or Asperger diagnosis than children without plateau or regression. Parents who endorsed a developmental plateau reported a similar pattern, except for the diagnosis of PDD-NOS. An acute decrease in the likelihood of PDD-NOS and Asperger’s diagnoses and a strong increase in the probability of an autistic disorder diagnosis were also apparent for those with severe regression compared to children with a mild loss of skills. Not surprisingly, parents of children who reported social and motor skill losses were more likely to report a diagnosis of Asperger’s syndrome than those with language losses. Lastly, earlier age at diagnosis was also associated with plateau, regression, increasing severity of parent’s perception of their child’s regression, and language as the skill most severely affected by the loss. These data strongly suggest that a loss or halt in development is associated with autistic disorder and earlier age at diagnosis.

After characterizing the developmental and diagnostic features of our sample, we explored the relationship between onset pattern, symptom severity, and cognitive and verbal impairment. Children in the SL group evidenced elevated autism symptom scores on both of the standardized instruments and a nearly 2.5 fold increased risk for not attaining conversational speech. For children with severe regression, increases in symptom severity and a 4-fold increased risk for not attaining phrase speech are that much more disconcerting. Furthermore, children who reportedly plateaued in development did not show an increase in SRS scores, though their SCQ scores did reach statistical significance once differences in age and race were controlled. The PL group also did not demonstrate an increased risk for phrase speech delay. No differences were found for cognitive disability between any groups under study, although parent report of this outcome may be particularly influenced by reporting biases.

To assess the effect of these impairments, we examined educational setting (special education vs. regular) and the use of a one-to-one behavioral aide within that setting. After dichotomizing these variables and adjusting for numerous covariates, we found that children in the SL group demonstrated an increased risk for being placed in a special education setting rather than a regular classroom setting, and within that classroom they were more likely to have a one-to-one aide, compared with the NLNP group. Again, we found this risk to be further increased for children with severe regression. Although not as strong, we also found this risk to be significantly increased for children with a developmental plateau. No differences in these outcomes were found among children with language, social, or motor losses.

Taken together, these findings provide strong evidence for an association between autistic regression and generally poorer outcomes. Our data also suggests that children with a developmental plateau are at an increased risk, compared to those without a loss or plateau, for requiring additional educational supports. Each of these findings has important implications to the design and priority of clinical intervention.

Another objective of our study was to observe the timing of first parental concerns and acquisition of the child’s first words and phrase speech in the skill loss group to explore differences in outcomes between children whose development is noticeably delayed before the regression and those who lose more advanced skills. It is important to highlight this topic because some authors have required a period of near typical development, including conversational speech, to be present for 1–3 months prior to the skill loss for inclusion to their study. Results from this post-hoc descriptive analysis showed that more than 35% of parents had concerns about their child’s development before they first noticed a loss and that only 23% of children had obtained phrase speech 1 month prior to the loss, although over 80% had attained their first words. Based on these results, we suggest that many of the children in our sample who lost skills displayed atypical developmental nuances prior to the loss and that the majority of these children lost budding words rather than complex-conversational speech. These findings are in line with those reported by Kurita (1985), Goldberg et al. (2003), Luyster et al. (2005) and Landa et al. (2007). To better understand if there were differences between children who were more developmentally advanced before the loss and children who were beginning to show delays per parent report, we identified those children who had developed phrase speech and whose parents had yet to identify developmental abnormalities (13%; n = 153) and compared them with the remainder of the SL group. We found that children who attained phrase speech and had no parental concerns prior to the loss had lower autism symptom scores, and were less likely to report a cognitive disability and an autistic disorder diagnosis; no other differences in outcome were identified. Results from this analysis imply that regression studies that adhere to stringent inclusion criteria are excluding a large proportion of severely affected children. This finding may shed some new light on the discrepant findings between studies of regression in autism.

To our knowledge, this is the first study to examine the relationship between parents’ subjective view of their child’s regression severity and the child’s current functioning. We found that parents who viewed their child’s regression as severe reported less delayed development prior to the loss and noticed that their child’s skill loss occurred at an earlier age. Children in the severe regression group were also more likely to have an autism diagnosis; higher symptom severity scores; and generally poorer verbal and educational outcomes than were those identified as having a mild or moderate regression. Few differences, except for age at noticed regression and diagnosis, were found between the mild and moderate groups. Furthermore, it is important to note that we did not gather qualitative data on what factors influenced parental perception of the skill loss or why they viewed their child’s loss as severe rather than mild. Nevertheless, we speculate that the pattern of less delayed development prior to the loss and earlier noticed age at regression followed by later complex skill achievement, a 4-fold increase risk for not obtaining phrase speech, earlier age at diagnosis, and poorer outcomes provides insight into their selections. These data suggest not all children with regression may be equally affected by the loss and warrant further investigation into what bio-psycho-social factors are associated with regression severity.

Our final objective was to investigate the characteristics of children whose parents reported a developmental plateau. Seventeen percent of parent respondents in this study reported a developmental stasis, which tended to occur at around a child’s 2nd birthday. Our data suggests these children may be at an increased risk for several alarming outcomes. Given this is the first study to report such findings, we hope these data spawn further and more thorough investigation of this onset type.

Parents who did not endorse skill loss or plateau were assigned to the no loss and no plateau group. This heterogeneous group is consistent with the proportion (~40%) and developmental characteristics, including age at first concern, with those previously found in the literature (Volkmar et al. 1985). Researchers have found that this group can be further broken down into early versus late onset (Werner et al. 2005), but the items available to our study were not specific enough to accurately discriminate these sub-phenotypes. Nonetheless, our data support Werner et al.’s (2005) conclusion that earlier detectable signs of ASD do not necessarily equate with increased severity of the disorder.

There are several limitations inherent to our study that must be considered when interpreting the results. First, the cross-sectional retrospective design allows for examination of the independent-dependent relationship only at a single point in time and likely reflects recall biases such as forward telescoping. We examined the effect of this bias by reducing the age limit of our sample to early adolescence and found it did not significantly alter the relationships under study; consequently, we left the natural limit of 17 years in place. All of these issues, along with data based solely on parent report, an ASD diagnosis that is not validated, and the overall quality concerns of online data collection, may represent biases, but we have assumed that these biases are constant throughout our convenient sample. These limitations must be held in light of our study’s sample size.

In sum, we found that children whose parents reported a loss of skills have a distinct developmental pattern marked by earlier first words and delayed phrase speech compared with children without regression. These children also demonstrated elevated autism symptom scores and an increased risk for poorer verbal, educational-behavioral, and diagnostic outcomes. These findings were particularly robust for the children whose parents reported the regression as severe. We also reported characteristics about an under-researched group of children on the autism spectrum who suffer from a developmental plateau or halt. These findings have important implications to theory, policy, and practice.

Notes

Acknowledgments

This research was supported by grants to the last author from Autism Speaks. The funder had no role in determining content. We would like to thank Connie Anderson, Teresa Foden, and Gregory Kalb for their unique contributions to this article.

References

  1. American Psychiatric Association (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision). Washington, DC: American Psychiatric Association.Google Scholar
  2. Bernabei, P., Cerquiglini, A., Cortesi, F., & D’Ardia, C. (2007). Regression versus no regression in the autistic disorder: Developmental trajectories. Journal of Autism and Developmental Disorders, 37, 580–588.CrossRefPubMedGoogle Scholar
  3. Berument, S. K., Rutter, M., Lord, C., Pickles, A., & Bailey, A. (1999). Autism Screening Questionnaire: Diagnostic validity. The British Journal of Psychiatry, 175, 444–451.CrossRefPubMedGoogle Scholar
  4. Chakrabari, S., & Fombonne, E. (2001). Pervasive developmental disorders in preschool children. The Journal of the American Medical Association, 285, 3093–3099.CrossRefGoogle Scholar
  5. Constantino, J., Davis, S., Todd, R., Schindler, M., Gross, M., Brophy, S., et al. (2003). Validation of a brief quantitative measure of autistic traits: Comparison of the Social Responsiveness Scale with the Autism Diagnostic Interview-Revised. Journal of Autism and Developmental Disorders, 33, 427–433.CrossRefPubMedGoogle Scholar
  6. Davidovitch, M., Glick, L., Holtzman, G., Tirosh, E., & Safir, M. P. (2000). Developmental regression in autism: Maternal perception. Journal of Autism and Developmental Disorders, 30, 113–119.CrossRefPubMedGoogle Scholar
  7. Goldberg, W. A., Osann, K., Filipek, P. A., Laulhere, T., Jarvis, K., Modahl, C., et al. (2003). Language and other regression: Assessment and timing. Journal of Autism and Developmental Disorders, 33, 607–616.CrossRefPubMedGoogle Scholar
  8. Hansen, R. L., Ozonoff, S., Krakowiak, P., Angkustsiri, K., Jones, C., et al. (2009). Regression in autism: Prevalence and associated factors in the CHARGE study. Ambulatory Pediatrics, 8, 25–31.CrossRefGoogle Scholar
  9. Hoshino, Y., Kaneko, M., Yashima, Y., Kumashiro, H., Volkmar, F., & Cohen, D. (1987). Clinical features of autistic children with setback course in their infancy. Japanese Journal of Child and Adolescent Psychiatry, 26, 201–212.Google Scholar
  10. Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217–250.Google Scholar
  11. Kobayashi, R., & Murata, T. (1998). Setback phenomenon in autism and long term prognosis. Acta Psychiatrica Scandinavia, 98, 296–303.CrossRefGoogle Scholar
  12. Kurita, H. (1985). Infantile autism with speech loss before the age of thirty months. Journal of the American Academy of Child and Adolescent Psychiatry, 24, 191–196.CrossRefGoogle Scholar
  13. Kurita, H., Kita, M., & Miyake, Y. (1992). A comparative study of development and symptoms among disintegrative psychosis and infantile autism with and without speech loss. Journal of Autism and Developmental Disorders, 22, 175–188.CrossRefPubMedGoogle Scholar
  14. Landa, R. J. (2008). Diagnosis of autism spectrum disorders in the first 3 years of life. Nature Clinical Practice Neurology, 4, 138–147.CrossRefPubMedGoogle Scholar
  15. Landa, R. J., Holman, K. C., & Garrett-Mayer, E. (2007). Social and communication development in toddlers with early and later diagnosis of autism spectrum disorders. Archives of General Psychiatry, 64, 853–864.CrossRefPubMedGoogle Scholar
  16. Lord, C., Shulman, C., & DiLavore, P. (2004). Regression and word loss in autistic spectrum disorders. Journal of Child Psychology and Psychiatry and Allied Disciplines, 45, 936–955.CrossRefGoogle Scholar
  17. Luyster, R., Richler, J., Risi, S., Hsu, W. L., Dawson, G., Bernier, R., et al. (2005). Early regression in social communication in autism spectrum disorders: A CPEA study. Developmental Neuropsychology, 27, 311–336.CrossRefPubMedGoogle Scholar
  18. Ozonoff, S., Heung, K., Byrd, R., Hansen, R., & Hertz-Picciotto, I. (2008). The onset of autism: Patterns of symptom emergence in the first years of life. Autism Research, 1, 320–328.CrossRefPubMedGoogle Scholar
  19. Pine, E., Luby, J., Abbacchi, A., & Constantino, J. N. (2006). Quantitative assessment of autistic symptomatology in preschoolers. Autism: The International Journal of Research and Practice, 10, 344–352.Google Scholar
  20. Rogers, S. J. (2004). Developmental regression in autism spectrum disorders. Journal of Mental Retardation and Developmental Disabilities Research, 10, 139–143.CrossRefGoogle Scholar
  21. Rogers, S. J., & DiLalla, D. (1990). Age of symptom onset in young children with pervasive developmental disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 863–872.CrossRefPubMedGoogle Scholar
  22. Short, A. B., & Schopler, E. (1988). Factors relating to age of onset in autism. Journal of Autism and Developmental Disorders, 18, 207–215.CrossRefPubMedGoogle Scholar
  23. Siperstein, R., & Volkmar, F. (2004). Brief report: Parental reporting of regression in children with pervasive developmental disorders. Journal of Autism and Developmental Disorders, 34, 731–734.CrossRefPubMedGoogle Scholar
  24. Volkmar, F. R., & Cohen, D. J. (1989). Disintegrative disorder or “late onset” autism. Journal of Child Psychiatry and Psychology, 30, 717–724.CrossRefGoogle Scholar
  25. Volkmar, F. R., Stier, D. M., & Cohen, D. J. (1985). Age of recognition of pervasive developmental disorder. American Journal of Psychiatry, 142, 1450–1452.PubMedGoogle Scholar
  26. Werner, E., Dawson, G., Munson, J., & Osterling, J. (2005). Variation in early developmental course in autism and its relation with behavioral outcome at 3–4 years of age. Journal of Autism and Developmental Disorders, 35, 337–350.CrossRefPubMedGoogle Scholar
  27. Wilson, S., Djukic, A., Shinnar, S., Dharmani, C., & Rapin, I. (2003). Clinical characteristics of language regression in children. Developmental Medicine and Child Neurology, 45, 508–514.CrossRefPubMedGoogle Scholar
  28. World Health Organization (1992). International classification of disease-10, Geneva.Google Scholar

Copyright information

© Springer Science+Business Media, LLC 2010

Authors and Affiliations

  • Luther G. Kalb
    • 1
  • J. K. Law
    • 2
    • 3
  • Rebecca Landa
    • 1
    • 4
  • Paul A. Law
    • 2
    • 3
  1. 1.Center for Autism and Related DisordersKennedy Krieger InstituteBaltimoreUSA
  2. 2.Department of Medical InformaticsKennedy Krieger InstituteBaltimoreUSA
  3. 3.Department of PediatricsThe Johns Hopkins University School of MedicineBaltimoreUSA
  4. 4.Department of Psychiatry and Behavioral SciencesThe Johns Hopkins University School of MedicineBaltimoreUSA

Personalised recommendations