Early Human Development 91 (2015) 271–276

Contents lists available at ScienceDirect

Early Human Development journal homepage: www.elsevier.com/locate/earlhumdev

Screening for autism spectrum disorder in very preterm infants during early childhood Peter H. Gray a,b,⁎, Dawn M. Edwards a,c, Michael J. O'Callaghan a,d, Kristen Gibbons b a

Growth and Development Unit, Mater Mothers' Hospital, South Brisbane, Queensland, Australia Mater Research Institute, The University of Queensland, Mater Health Services, South Brisbane, Queensland, Australia Dept. of Social Work, Mater Children's Hospital, South Brisbane, Queensland, Australia d Dept. of Paediatrics and Child Health, University of Queensland, Mater Children's Hospital, South Brisbane, Queensland, Australia b c

a r t i c l e

i n f o

Article history: Received 27 October 2014 Received in revised form 13 February 2015 Accepted 20 February 2015 Keywords: Infant premature Autism spectrum disorder Modified Checklist for Autism in Toddlers Child development Child behaviour

a b s t r a c t Aim: The aim of the study was to screen very preterm infants for autism spectrum disorder (ASD) with comparisons to a group of term controls. The study also aimed to identify maternal and neonatal risk factors, development and behaviour associated with a positive screen in the preterm group. Method: Preterm infants born ≤30 weeks gestation and term infants were recruited at two years of age. The mothers were posted the questionnaires and completed the Modified Checklist for Autism in Toddlers (MCHAT), the Child Behaviour Checklist (CBCL) and the Depression, Anxiety and Stress Scales (DASS). Previously collected data from the mothers at 12 months — the Edinburgh Postnatal Depression Scales (EPDS) were analysed. The children had neurodevelopmental assessment including the Bayley-III. Infants positive on MCHAT screen had an M-CHAT follow-up interview by phone and then were assessed by a developmental paediatrician as indicated with a diagnosis of autism being made on clinical judgement. Results: 13 (13.4%) of the 97 preterm infants screened positive on the M-CHAT compared to three (3.9%) of the 77 term infants (p = 0.036). On follow-up interview, three of the preterm infants remained positive (one was diagnosed with autism) compared to none of the term infants. The preterm infants who screened positive were born to younger, non-Caucasian mothers and were of lower birth weight and had a higher incidence of being small for gestational age (SGA). The infants had lower composite scores on Bayley-III and had more internalising and externalising behaviours on the CBCL. The mothers had more emotional problems on the DASS and higher scores on the EPDS. On multivariate analysis, SGA, greater internalising behaviours and higher EPDS scores remained statistically significant. Conclusions: A positive screen on the M-CHAT occurs more commonly in very preterm infants than those born at term. Internalising behaviours and maternal mental health are associated with a positive screen in the preterm cohort. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Autism spectrum disorder (ASD) is a major psychiatric disorder seen in childhood. The diagnosis is clinical and based on specific social, language and behavioural characteristics. The reported prevalence of ASD is 6–10/1000 children [1,2]. ASD is generally a lifelong Abbreviations: ASD, autism spectrum disorder; M-CHAT, Modified Checklist for Autism in Childhood; CRIB, Clinical Risk Index for Babies; GMFCS, Gross Motor Functional Classification System; CBCL, Child Behaviour Checklist; DASS, Depression, Anxiety, Stress Scale. ⁎ Corresponding author at: Newborn Services, Mater Mothers' Hospital, Raymond Tce, South Brisbane, Queensland 4101, Australia. Tel.: +61 731638250; fax: +61 731631435. E-mail addresses: [email protected] (P.H. Gray), [email protected] (D.M. Edwards), [email protected] (M.J. O'Callaghan), [email protected] (K. Gibbons).

http://dx.doi.org/10.1016/j.earlhumdev.2015.02.007 0378-3782/© 2015 Elsevier Ireland Ltd. All rights reserved.

disorder associated with substantial co-morbidities affecting development, learning and behaviour, along with long term effects on the individual and family quality of life. Its aetiology is unknown, though it is considered to be a disorder of brain development [3] with genetic and environmental factors playing a role [4]. The strength of the relationship between preterm birth and ASD is controversial. In a meta-analysis on perinatal and neonatal risk factors for autism, Gardener et al. [5] found that preterm birth was not associated with autism, though there was a positive association between low birth weight and the risk of autism. When very preterm infants of b 30 weeks gestation were assessed at the age of 7 years, 4.5% of the cohort were diagnosed with ASD [6], while 8% of children born b26 weeks gestation were diagnosed with the condition at 11 years [7]. With early identification of children with ASD and subsequent intervention, there is evidence of improvement in cognitive performance and language skills [8,9]. Thus,

272

P.H. Gray et al. / Early Human Development 91 (2015) 271–276

screening of very preterm infants could be important as they are considered to be at high risk for ASD. The Modified Checklist for Autism in Toddlers (M-CHAT) has been used widely as a screening tool for both term [10] and preterm populations. Positive screening has been found in 21–41% of very preterm infants [11–13]. The children in these studies, however, had a relatively high rate of developmental impairment which may have led to a falsely high positive M-CHAT screen rate [14]. Therefore, it is recommended that a follow-up interview be conducted [15], followed by a formal diagnostic assessment if indicated. The aim of the present study was to screen very preterm infants for ASD using the M-CHAT with comparisons made with a group of term controls. The infants with a positive screen had a follow-up interview, with diagnostic evaluation to establish a diagnosis as indicated. The study also aimed to identify maternal and neonatal risk factors, developmental outcomes and other behavioural and family characteristics associated with a positive screen in the preterm group. 2. Methods 2.1. Participants The mothers of the very preterm (gestation ≤30 weeks) and term infants recruited in the current study had participated in a study of parenting stress, when their infants were 4 and 12 months (corrected for prematurity) for the preterm group [16,17]. The preterm infants born between June 2007 and February 2009 had been managed in the Neonatal Unit at the Mater Mothers' Hospital, Brisbane. The mothers of the preterm infants were approached to join the Parenting Stress study when their infants were stable and were receiving care in the Special Care Nursery. Mothers with multiple pregnancies more than twins, mothers with twins where one twin died, mothers with a baby with a major congenital abnormality or were not expected to survive to hospital discharge and mothers who were not English speaking were excluded from the study. Consent was obtained prior to discharge. The control group of infants were born at term (≥37 weeks gestation), also at the Mater Mothers' Hospital. In a similar time frame to the recruitment of the mothers of preterm infants, the term mothers were also approached while in hospital, but for the most part consent was obtained after discharge home. Prior to the second birthday (corrected for prematurity for the preterm cohort) of the infants, the mothers who participated in the Parenting Stress study were contacted inviting them and their infants to participate in the current study related to the screening for ASD in early infancy. Of the 124 preterm infants recruited into the parenting study, 97 participated in the ASD screening study. Of the 120 term infants recruited into the parenting study, 77 participated in the current study. The study was approved by the Mater Health Services Human Research Ethics Committee, with all mothers giving written consent for the participation of their child in the study. 2.2. Procedures Perinatal data including maternal demographic variables, pregnancy complications and neonatal morbidities were obtained at hospital discharge. Small for gestation age (SGA) was defined as having a birth weight b 10th percentile by gestational age using Australian National Data [18]. Cranial ultrasonography was performed on the preterm infants at 5–7 days of age, 21–28 days of age and at 34–36 weeks (corrected for prematurity) for infants with birth weight b 1000 g, with additional scans as clinically indicated. Ultrasound abnormalities including peri-intraventricular haemorrhage, cystic periventricular leukomalacia and cerebellar haemorrhage were recorded. Bronchopulmonary dysplasia was diagnosed with an ongoing requirement for supplemental oxygen after 36 weeks corrected for prematurity. The Clinical Risk Index for Babies (CRIB)

[19] was also calculated for each baby. When the children were 12 months of age, the mothers had completed the Edinburgh Postnatal Depression Scale (EPDS) [20] as previously described [17]. The EPDS is a 10-item self-report screening tool for depression after child birth. It may be used for mothers 12 months after delivery and beyond [21]. Scores ≥12 indicate probable depression. At 2 years (corrected for prematurity for the preterm cohort) the infants underwent a neurological examination, with cerebral palsy being diagnosed according to standard criteria [22]. Those with cerebral palsy had a Gross Motor Functional Classification System (GMFCS) [23] assessment. The GMFCS is a standardised classification system which grades gross motor skills in a child with cerebral palsy and categorises children into five different levels. The emphasis is on sitting and walking, with advancing levels indicating more severe limitations in motor function. The Bayley Scales of Infant and Toddler DevelopmentIII (Bayley-III) [24] were administered by trained examiners who were blinded to whether the children were preterm or term. The Bayley-III assessment examines the cognitive, language (receptive and expressive communication) and motor (gross and fine motor) abilities of children during infancy. The standardised norm for each of the three components of the Bayley-III is 100 (standard deviation [SD], 15). 2.3. Questionnaires These were posted to the mothers and were completed at home. They were returned by mail or brought along to the clinic at the time of the clinical assessment. 2.3.1. Modified-Checklist for Autism in Toddlers [10] This is a 23 item checklist that was developed as a screening test for symptoms of ASD in which parents report on child behaviours (yes/no). There are six items that are considered to be ‘critical’. When any three items or two or more ‘critical’ items are failed the child is considered to have screened positive with further assessment being indicated. The positive predictive value for the M-CHAT has been reported as 0.36 [25]. 2.3.2. Modified-Checklist for Autism in Toddlers — follow-up interview [12] This was performed by telephone interview with the mothers for the children who screened positive on the M-CHAT, enabling clarification of the mothers' responses to the failed behaviours using relevant specific examples. Using the 2-stage screen, the M-CHAT has been shown to be reliable with a positive predictive value of 0.74 [25]. Only the three children who remained positive on the M-CHAT interview were assessed by experienced developmental paediatricians at the Mater Children's Hospital. A diagnosis of pervasive developmental disorder was made on clinical grounds following interview with the mother and medical assessment of the child, according to DSM-IV criteria. Additional multidisciplinary assessments were undertaken as clinically indicated, though no formal additional assessments were consistently performed. Clinical judgement by an experienced clinician is considered to be the ‘gold standard’ for autism diagnosis [26]. 2.3.3. Child Behaviour Checklist (CBCL) 1.5–5 [27] This is a 100 item parent report that measures problem behaviours of pre-school children. Problem items are converted into syndrome scale scores which can be assigned a corresponding T-score from which internalising and externalising problem scales may be derived. The internalising problem scale is comprised of items from withdrawn, somatic and anxious/depressed behaviours, while the externalising problem scale is comprised of items from delinquent and aggressive behaviours. 2.3.4. Depression, Anxiety, Stress Scale (DASS) [28] A 42 item self-report questionnaire designed to measure the emotional states of depression, anxiety and stress was completed by the

P.H. Gray et al. / Early Human Development 91 (2015) 271–276

mothers. The scores for the items on the questionnaire were summed for each of the three syndromes. The summed scores were then converted into Z scores using the normative values according to the manual [28]. Those with scores with a Z score of N 1 were considered to have moderate expression for each of the three syndromes. 2.4. Sample size calculation To calculate the sample size, it was anticipated that there would be a positive screening rate on the M-CHAT of ≈ 25% and 10% in the preterm and term populations respectively, based on the studies of Limperopolous et al. [11] and Robins et al. [10]. Thus with a type I error of 0.05 and a power of 80%, a sample size of 100 participants in each group would be required to statistically demonstrate this proposed difference. It was expected that these numbers of participants would be available for analysis in the proposed project.

Table 1 Maternal demographic and infant characteristics for the preterm and term groups.

Maternal age — mean (SD) Paternal age — mean (SD) Multiple pregnancy, n (%) Assisted reproduction, n (%) Vertex presentation, n (%) Caesarean section, n (%) Parity, 1st baby n (%) Married/de facto relationship, n (%) Ethnicity, White/Caucasian n (%) Public insurance status, n (%)

3. Results Questionnaires were returned from all 144 mothers (at a mean [SD] of 25.1 [2.3] months for the preterm group; mean [SD] 24.7 [2.4] months for the term group). The preterm group consisted of 80 mothers who gave birth to 97 babies (34 twins). The term group consisted of 64 mothers who gave birth to 77 babies (26 twins). Table 1 outlines the maternal and infant characteristics of the preterm and term infants. The clinical characteristics of the pregnancy and neonatal morbidities in the preterm cohort are shown in Table 2. There was no difference in gender or gestational age distribution between the preterm infants that took part in the present study and those who were eligible but the mother declined to give consent, though the birth weight was greater (1266 g [SD 363 g] v 1072 g [326 g]; p = 0.009) for the infants for whom consent was not obtained. For the term group, the infants for whom consent was not obtained to participate in the study were of similar gender and birth weight, but were of greater gestational age (39.3 [SD 1.26] v 38.6 [SD 1.4] weeks; p = 0.006) than those infants that took part. Table 3 shows the results of the follow-up assessment. On neurological assessment, six of the preterm group had cerebral palsy, all being GMFCS Stage I. While none of the preterm children had hearing impairment, one of the term children had congenital hearing impairment requiring hearing aids. Of the preterm group, the Bayley scales could not be performed in six infants, either because of distance to the recruitment centre or behaviour difficulties during the testing. Of the term group, the Bayley could not be completed for six infants, either because of distance to the recruitment centre or behavioural difficulties during the testing, while further six infants failed to attend appointments. The results indicated that the preterm group had significantly lower Bayley-III scores in cognitive, language and motor performance. On the CBCL, significantly higher scores for the scales of internalising and

Preterm

Term

n = 80

n = 64

30.32 (5.35) 32.21 (6.31) 17 (21) 14 (17) 50 (63) 50 (62) 60 (59) 76 (95) 71 (89) 43 (54)

31.67 (4.72) 32.92 (4.88) 14 (13) 6 (9) 55 (86) 31 (48) 52 (53) 61 (95) 62 (97) 27 (42)

0.11 0.45 0.89 0.15 0.003 0.091 0.37 0.93 0.068 0.17

59 (74) 24 (30)

59 (92) 39 (61)

0.004 0.001

Educational level n (%) Schooling complete Tertiary qualification

2.5. Statistical analysis Descriptive characteristics, including maternal demographics and infant characteristics comparing the very preterm and term groups are presented. Bivariate analysis was undertaken to compare the preterm and term groups. Bivariate analysis was also performed to compare the preterm children who screened positive and those who screened negative on the M-CHAT. To further elucidate the independent factors associated with a positive M-CHAT screen for the preterm group, multivariate logistic regression analyses adjusting for related observations due to twin births and including maternal and infant characteristics were performed. Variables with a p value b 0.1 were entered into the model, once it was confirmed that multicollinearity between the variables did not exist. Statistical analyses were deemed significant at the 0.05 level, with all analysis performed in StataSE version 10.1 (StataCorp Ltd., College Station, Texas).

273

Male infant, n (%) Birth weight (g) — mean (SD) Weight b 1000 g, n (%) Gestation (weeks) — mean (SD) Born b 28 weeks, n (%)

n = 97

n = 77

51 (53) 1072 (326) 44 (45) 27.6 (2.0) 42 (43)

42 (55) 3315 (583)

p value

0.76

38.6 (1.4)

SD, standard deviation.

externalising behaviours were found for the preterm group compared to the term children.

3.1. Prevalence of positive M-CHAT screens Thirteen (13.4%) (95% confidence intervals [CI] 7.9%–21.7%) of the preterm group had a positive screen on the M-CHAT compared to three (3.9%) (95% CI 0.9%–11.3%) of the term infants (p = 0.036) (Table 3). No preterm infant with a positive screen was diagnosed with cerebral palsy. When the two infants who had a cognitive score on the Bayley-III of less than 2 SD below the standardised mean (b70) were excluded from the analysis, the rate of positive screens on the M-CHAT was 11.6%. The mothers of all infants (preterm and term) with a positive screen were contacted by phone for a follow-up interview. Three of the preterm infants had features of ASD, of whom on evaluation by a developmental paediatrician, one was diagnosed with autism. No term infant was positive on follow-up interview.

Table 2 Clinical characteristics of the pregnancy and neonatal morbidities in the preterm cohort. Clinical characteristic

n = 97 (%)

Preterm labour Maternal preeclampsia Maternal antepartum haemorrhage Chorioamnionitis Antenatal steroids — complete course CRIB score — mean (SD) Apgar, 5 min — mean (SD) Bronchopulmonary dysplasia Home oxygen programme Postnatal steroids Late onset infection Peri-intraventricular haemorrhage (P-IVH) P-IVH grade 3 or 4 or cysts Cerebellar haemorrhage Retinopathy of prematurity (ROP) ROP treatment Length of stay, days — mean (SD)

57 (59) 16 (16) 14 (14) 43/93 (46) 68 (70) 3.18 (3.16) 8.10 (1.70) 16 (16) 11 (11) 6 (6) 16 (16) 9 (9) 6 (6) 1 (1) 51 (53) 4 (4) 77.2 (24.1)

CRIB, Clinical Risk Index for Babies; SD, standard deviation.

274

P.H. Gray et al. / Early Human Development 91 (2015) 271–276

Table 3 Follow-up outcomes, scores on the Modified Checklist for Autism in Toddlers and Child Behaviour Checklist for the preterm and term groups at 2 years of age.

Cerebral palsy, n (%) Visual impairment Hearing impairment, n (%) Bayley-III Cognitive — mean (SD) Language — mean (SD) Motor — mean (SD) M-CHAT Median (range) Abnormal, n (%) M-CHAT follow-up interview Abnormal, n (%) Child Behaviour Checklist Internalising — mean (SD) Externalising — mean (SD)

Preterm

Term

n = 97

n = 77

6 (6.2) 0 0 n = 91a 100.87 (15.39) 102.78 (17.39) 98.46 (14.21)

0 0 1 n = 65a 109.53 (15.72) 109.11 (12.82) 108.4 (13.18)

1 (0–10) 13 (13.4)

0 (0–4) 3 (3.9)

3 (3.1)

0

46.92 (11.33) 50.92 (11.12)

43.44 (9.73) 45.76 (8.96)

Table 4 Maternal demographic, perinatal and birth factors for the preterm infants who were screened using the Modified Checklist for Autism in Toddlers.

p value

0.035 –

0.0008 0.0138 0.0001

0.036

0.0337 0.001

SD, standard deviation; M-CHAT, Modified Checklist for Autism in Toddlers. a 6 preterm infants and 12 term infants for whom questionnaires were returned did not attend for Bayley-III assessment.

3.2. Risk factors for positive M-CHAT screens for the preterm cohort Clinical, developmental and behavioural risk factors for a positive M-CHAT screen were examined, with the results summarised in Table 4. Maternal age was lower in the group that screened positive. White/Caucasian mothers were more to have children who screened negative. While there were no group differences in the gestational age between those who screened positive and negative, those who screened positive were of lower birth weight (p = 0.047). Additionally, those who screened positive had a higher incidence of being SGA (p = 0.014). None of the neonatal factors considered — gender (p = 0.76), CRIB scores (p = 0.46), late onset infection (p = 0.15), bronchopulmonary dysplasia (p = 0.99), need for home oxygen (p = 0.16), administration of postnatal steroids (p = 0.89), results on cranial ultrasound scans (p = 0.16), retinopathy of prematurity (p = 0.08), laser treatment for retinopathy of prematurity (p = 0.49) or length of hospital stay (p = 0.29) were found to be risk factors for a positive screen on bivariate analysis. While cerebral palsy was not a risk factor for a positive screen, the results of the Bayley-III assessments indicated that those with a positive screen had lower cognitive scores (p = 0.022), lower language scores (p = 0.014) and lower motor scores (p = 0.006). Additionally, the mean scores for the two broad based scales on the CBCL, internalising and externalising were both significantly higher for those with positive screens (p b 0.001 and p = 0.005, respectively). On the DASS, the bivariate analysis results indicated that the mothers of the M-CHAT positive infants had higher scores and more mothers had scores in the range considered to represent at least moderate expression; (Table 4). The results of the EPDS, which was performed at 12 months revealed a significant association with the M-CHAT (OR, 4.11; 95% CI, 1.02–16.42). Multiple logistic regression analysis was performed for a positive MCHAT screen with those risk factors with a p value b 0.1 on bivariate analysis — maternal and paternal age, ethnicity and birth weight, SGA, gender and retinopathy of prematurity. The only factor that was found to be statistically significant was SGA (OR 4.58 [95% CI, 1.09–19.31], p = 0.038). Multiple regression analysis was also performed with the results of the Bayley III (cognitive, language and motor composite scores) together with the domains of internalising and externalising behaviours from the CBCL being entered into the model. Only internalising behaviours with an odds ratio (OR) of 1.21 (95% CI 1.06–1.37; p = 0.004) was

Maternal age — mean (SD) Paternal age — mean (SD) Married/de facto relationship, n (%) Ethnicity, White/Caucasian, n (%) Public insurance status, n (%) Assisted reproduction, n (%) Educational level, n (%) Schooling complete Tertiary qualification Preterm labour, n (%) Preeclampsia, n (%) Antepartum haemorrhage, n (%) Chorioamnionitis, n (%) Antenatal steroids, n (%) Parity — 1st baby, n (%) Singleton, n (%) Vertex presentation, n (%) Caesarean section, n (%) Male infant, n (%) Birth weight (g) — mean (SD) Gestation (weeks) — mean (SD) Small for gestational age, n (%) Apgar – 5 min – mean (SD) Cerebral palsy, n (%) Bayley – cognitive CS – mean (SD) Bayley – language CS – mean (SD) Bayley – motor CS – mean (SD) CBCL – internalising – mean (SD) CBCL – externalising – mean (SD) DASS — depression, n (%) DASS — anxiety, n (%) DASS — stress, n (%) EPDS ≥ 12, n (%)

M-CHAT

M-CHAT

Positive

Negative

p value

n = 13

n = 84

27.62 (5.36) 29.42 (4.78) 12 (92) 9 (69) 8 (62) 2 (15)

31.07 (5.49) 32.89 (6.26) 81 (96) 78 (93) 41 (49) 17 (20)

0.041 0.074 0.44 0.017 0.39 0.69

10 (77) 2 (15) 7 (54) 4 (31) 2 (15) 5 (38) 13 (100) 10 (77) 8 (62) 7 (54) 9 (69) 6 (46) 910 (378) 27.2 (2.2) 5 (38.5) 8.62 (1.12) 0 92.1 (18.9) 90.75 (17.2) 86.75 (15.6) 60.15 (7.73) 59.46 (12.39) 5 (38.5) 5 (38.5) 6 (46.2) 4 (33.3)b

60 (71) 25 (30) 50 (60) 12 (14) 12 (14) 38/80 (48) 82 (98) 50 (60) 55 (65) 50 (60) 57 (68) 45 (54) 1097 (312) 27.6 (2.0) 9 (10.7) 8.02 (1.77) 6 (7.1) 103.4 (14.4) 104.6 (16.8) 100.3 (13.2) 44.84 (10.49) 49.58 (10.37) 9 (8.3)a 10 (12.2)a 12 (14.6)a 9 (10.8)b

0.68 0.28 0.7 0.15 0.92 0.55 1.0 0.36 0.86 0.667 1.0 0.089 0.048 1.0 0.014 0.24 0.71 0.022 0.014 0.006 0.0002 0.005 0.022 0.030 0.015 0.034

M-CHAT, Modified Checklist for Autism in Toddlers; CS, Composite Scale; CBCL, Child Behaviour Checklist; DASS, Depression, Anxiety, Stress Scales; EPDS, Edinburgh Postnatal Depression Scale. a 2 of the 84 mothers who infants did not screen positive on the M-CHAT did not complete the DASS. b 2 mothers (1 in each group) of the preterm infants did not complete the EPDS.

independently significantly associated with a positive screen on the M-CHAT. A further multiple regression analysis was performed with the results of child behaviours on the CBCL (internalising and externalising) and the maternal health variables (DASS and EPDS) being entered into the model. As there was significant collinearity between the DASS Depression scale and the EPDS, the DASS Depression scale was not included in the analysis. The results indicated that the independent variables associated with a positive M-CHAT screen were internalising behaviours (p = 0.006) and results on the EPDS (p = 0.023). 4. Discussion The results of the present study indicated that 13% of very preterm children born ≤ 30 weeks gestation in early childhood had a positive screening test for ASD on the M-CHAT. This was significantly greater than a contemporaneously recruited group of term infants. Follow-up interviews revealed that 3% of the total preterm cohort screened positive following detailed questioning of the mothers. Only 1% of the preterm children were diagnosed with autism, though because numbers are small, confidence intervals are wide. Nevertheless, this is quite different to the study of Johnston et al. [7], who found the incidence of autism to be 8% in a population of preterm children born b26 weeks gestation at 11 years of age. Previous studies using the M-CHAT for

P.H. Gray et al. / Early Human Development 91 (2015) 271–276

preterm infants have demonstrated a much higher prevalence of positive screens: 21–41% [11–13], though those of Kuban et al. [12] and Moore et al. [13] were in extremely preterm infants. No follow-up interviews were performed in those studies and there was no attempt to make a definitive diagnosis of ASD. The previous screening studies of preterm infants had a much higher incidence of neurodevelopmental disability, which as has been suggested may have resulted in problems with test interpretation [10]. In the study of Kuban et al. [12] in which screening for ASD was undertaken in extremely low gestational age newborns, when the children with cognitive impairment were excluded, 10% were positive, which is very similar to the 11.6%, that was found in our study. It is also of importance that the mean CRIB score of our preterm cohort is similar to that reported previously for a group of infants born ≤30 weeks gestation [29], suggesting that the preterm infants in the present study are representative of the very preterm population. Of note, just 3.9% of the term children were positive on the initial M-CHAT screen. This is considerably lower than the 9.7% reported by Robins et al. [10]. Some other studies however, have indicated somewhat lower positive screening rates; 6.5% in the study of Pandey et al. [30] and 4.7% in a study of children who were screened at less than 48 months [31]. We are not aware of any study in Australia in which screening of low-risk children for ASD using the M-CHAT has been undertaken. Even though our results need to be interpreted with caution, given the low numbers of children screened, it may be that the rate of positive screens in the Australian population is lower than similar populations elsewhere or just may reflect differences in sample sources. The infants in the current study were identified at birth while the infants in the studies that have reported higher rates of positive screening were recruited from infants attending doctors' clinics [15,32]. On bivariate analysis in the present study, younger maternal age was associated with a positive screen for ASD, though this was no longer seen following regression analysis, with similar findings being reported by Stephens et al. [33]. Similarly, in our study on bivariate analysis, nonWhite/Caucasian mothers were more likely to have preterm children with a positive screen, with this risk factor not being statistically significant in the regression model. This data needs to be treated with caution as it is unknown whether or not these families were multilingual, which may have influenced the results. However, in an Australian study, mothers' country of birth being outside Australia was found to be a risk factor for autism [34]. Previous studies using the M-CHAT to screen preterm infants for ASD have found both birth weight and gestational age to be significant risk factors [11,13]. In the study of Stephens et al. [33] on screening for ASD for children born at b 27 weeks gestation, positive screens were associated with lower birth weight but not lower gestational age, in accordance with the present study. We also found the infants born SGA had an increased risk of a positive M-CHAT screen. This variable was not examined in some previous studies, but Moore et al. [13] found that there was an association for infants with lower birth weight standard deviation scores and a higher risk of positive M-CHAT screens. This is probably not surprising as an increased risk of autism has been described in preterm SGA infants of 23–31 weeks gestation [35]. No perinatal or neonatal factor (apart from SGA) was found to be significantly associated with a positive screen on the M-CHAT. This differs from the findings of Moore et al. [13] in a study of infants born ≤26 weeks gestation who reported that severe bronchopulmonary dysplasia, administration of postnatal steroids, late onset bacteraemia and cranial ultrasound abnormality were associated with positive M-CHAT screens. Limperoupoulos et al. [11], found in their study somewhat different risk factors associated with a positive M-CHAT; chorioamnionitis, acute intrapartum haemorrhage, illness severity on admission together with abnormal MRI studies. In our study, we found no association for pregnancy factors or illness severity on admission. We did not have

275

the opportunity to perform MRI scans on our preterm cohort. However, a rigorous schedule of cranial ultrasound scanning was performed with a relatively low incidence of abnormal findings. Only one infant had cerebellar haemorrhage, which is in accordance with our previous experience [36]. There is increasing evidence of a link between prematurity, cerebellar haemorrhage and autism [37]. However, because of our low incidence, we cannot make comment on a possible relationship between cerebellar haemorrhage and ASD. The results of the CBCL indicated that internalising behaviours were independently associated with abnormal M-CHAT scores. In the only previous study that examined behavioural symptoms in a group of preterm infants screened using the M-CHAT [11], internalising behaviours were also found to be highly correlated with abnormal M-CHAT scores. Given that internalising behavioural problems are considered to be present in a child that directs feelings and emotions inwards, with the child being shy, wanting to be alone and feeling unloved, it is not surprising that the infants who screened positive on the M-CHAT also displayed a tendency to have more internalising problems. It is quite well established that maternal health problems are associated with autism [38–40]. Accordingly, while it would seem that this has not been examined in previous studies of screening preterm infants, we found that maternal psychological symptoms as evidenced by the results on the DASS were associated with a positive score on the M-CHAT. Maternal depression assessed when the children were 12 months of age was also associated with a positive screen at 2 years of age, though only the result of the EPDS was found to be an independent variable on multivariate analysis. It is unclear, however, whether the maternal depression is directly associated with the positive screen or alternatively may reflect a situation whereby a mother with mental health issues may report more behaviour problems in their children [41]. A limitation of our study is the relatively low numbers of preterm infants that were assessed, and hence real differences between risk factors for the infants who screened positive and those that screened negative may have been missed due to a type II error. Furthermore, the number of infants recruited was fewer than anticipated. However as the incidence of positive M-CHAT screens in the term group was less than expected, statistical differences between the preterm and term groups were found. Not all the preterm infants were formally assessed for ASD and hence there may have been some false negatives with additional children with ASD who were missed. A proportion of the mothers and infants who took part in the parenting study did not consent to participate in the current study. Importantly, however a strength to the study is that all mothers who consented for their infants to take part in the current study completed the M-CHAT questionnaire. Additionally, to our knowledge, it is the first study of preterm infants that used the follow-up interview component of the M-CHAT, with further information being obtained from all mothers followed by a diagnostic evaluation for ASD. Moreover, we had a control group of term children for comparison purposes to enable an enhanced understanding of the results from the preterm children, especially as there are no previously published Australian data for M-CHAT screening. In conclusion, in the present study of very preterm infants, the prevalence of a positive screen on the M-CHAT was significantly higher than in a contemporaneously screened group of term infants. On follow-up the rate of autism was 1%. Born SGA was the only perinatal or neonatal factor that was associated with a positive screen in the preterm group, though there was an independent association with internalising behaviours. Importantly, maternal mental health may be associated with the results of the M-CHAT screening. While, the prevalence of ASD was relatively low and indeed little different to that described in population studies, we would agree with others [42] that early screening for very preterm infants with follow-up assessments as appropriate is important in the early identification of these children. This may enable early intervention for those infants diagnosed with ASD which has the potential to improve long-term outcomes [43,44].

276

P.H. Gray et al. / Early Human Development 91 (2015) 271–276

Funding The study was funded by the Mater Mother's Research Centre through a grant from the Mater Foundation. Conflict of interest The authors declare that there is no conflict of interest with regard to this manuscript. Acknowledgement We thank the team at the Growth and Development Unit, Mater Mothers' Hospital for performing the medical and developmental assessments. References [1] Duchan E, Patel DR. Epidemiology of autism spectrum disorders. Pediatr Clin North Am 2012;59:27–43. [2] Atladóttir HÓ, Henriksen TB, Schendel DE, Parner ETI. Autism after infection, febrile episodes and antibiotic use during pregnancy: an exploratory study. Pediatrics 2012; 130:e1447–54. [3] Courchesne E, Mouton PR, Calhoun ME, Semendeferi K, Ahrens-Barbeau C, Hallet MJ, Barnes CC, Pierce K. Neuron number and size in prefrontal cortex of children with autism. JAMA 2011;306:2001–9. [4] Hallmayer J, Cleveland S, Al Torres, Phillips J, Cohen B, Torigoe T, Miller J, Fedele A, Collins J, Smith K, Lotspeich L, Croen LA, Ozonoff S, Lajonchere C, Grether JK, Risch N. Genetic variability and shared environmental factors among twin pairs with autism. Arch Gen Psychiatry 2011;68:1095–122. [5] Gardener H, Spiegelman D, Buka SL. Perinatal and neonatal risk factors for autism:a comprehensive meta-analysis. Pediatrics 2011;128:344–55. [6] Treyvaud K, Ure A, Doyle LW, Rogers CE, Kidokoro H, Inder TE, Anderson PJ. Psychiatric outcomes at age seven for very preterm children: rates and predictors. J Child Psychol Psychiatry 2013;51:181–91. [7] Johnson S, Hollis C, Kochar P, Hennessy E, Wolke D, Marlow N. Autism spectrum disorders in extremely preterm children. J Pediatr 2010;156:525–31. [8] Warren Z, McPheeters ML, Sathe N, Foss-Feig JH, Glasser A, Veenstra-Vanderweele J. A systematic review of early intensive intervention for autism spectrum disorders. Pediatrics 2011;127:1303–11. [9] Koegel LK, Koegel RL, Ashbaugh K, Bradshaw JA. The importance of early identification and intervention for children with or at risk for autism spectrum disorders. Int J Speech Lang Pathol 2014;16:50–6. [10] Robins DL, Fein D, Barton M, Green JA. The Modified Checklist for Autism in Toddlers: an initial study investigating the early detection of autism and pervasive developmental disorders. J Autism Dev Disord 2001;31:131–44. [11] Limoeropoulos C, Bassan H, Sullivan NR, Soul JS, Robertson RL, Moore M, Ringer SA, Volpe JJ, du Plessis A. Positive screening for autism in ex-preterm infants: prevalence and risk factors. Pediatrics 2008;121:758–65. [12] Kuban KC, O'Shea TM, Allred EN, Tager-Flusberg H, Goldstein DJ, Leviton A. Positive screening on the modified Checklist for Autism in Toddlers (M-CHAT) in extremely low gestational age newborns. J Pediatr 2009;154:535–40. [13] Moore T, Johnson S, Hennessy E, NI Marlow. Screening for autism in extremely preterm infants: problems in interpretation. Dev Med Child Neurol 2012;54:514–20. [14] Luyster RJ, Kuban KC, O'Shea TM, Paneth N, Allred EN, Leviton A. ELGAN Study Investigators The Modified Checklist for Autism in Toddlers in extremely low gestational age newborns: individual items associated with motor, cognitive, vision and hearing limitations. Paediatr Perinat Epidemiol 2011;25:366–76. [15] Robins DL. Screening for autism spectrum disorders in primary care settings. Autism 2008;12 [537–6]. [16] Gray PH, Edwards DM, O'Callaghan MJ, Cuskelly M. Parenting stress in mothers of preterm infants during early infancy. Early Hum Dev 2012;88:45–9. [17] Gray PH, Edwards DM, O'Callaghan MJ, Cuskelly M, Gibbons K. Parenting stress in mothers of very preterm infants — influence of development, temperament and maternal depression. Early Hum Dev 2013;89:625–9. [18] Roberts CL, Lancaster PA. Australian national birthweight percentiles by gestational. Med J Aust 1999;170:114–8. [19] International Neonatal Network. The CRIB (clinical risk index for babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. Lancet 1993;342:193–8.

[20] Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression Scale. Br J Psychiatry 1987;150:782–6. [21] Moehler E, Brunner R, Wiebel A, Reck C, Resch F. Maternal depression symptoms in the postnatal period are associated with long-term impairment of mother–child bonding. Arch Womens Ment Health 2006;9:273–8. [22] Rosenbaum P, Paneth N, Leviton A, Goldstein M, Bax M, Damiano D, Dan B, Jacobsson B. A report: the definition and classification of cerebral palsy April 2006. Dev Med Child Neurol 2007;109:8–14. [23] Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol 1997;39:214–23. [24] Bayley N. Bayley Scales of Infant and Toddler Development. 3rd ed. San Antonio, TX: The Psychological Corporation; 2006. [25] Kleinman JM, Robins DL, Ventola PE, Pandey J, Boorstein HC, Esser EL, Wilson LB, Rosenthal MA, Sutera S, Verbalis AD, Barton M, Hodgson S, Green J, DumontMathieu T, Volkmar F, Chawsarska K, Klin A, Fein D. The Modified Checklist for Autism in Toddlers; a follow-up study investigating the early detection of autism spectrum disorders. J Autism Dev Disord 2008;38:827–39. [26] Volkmar F, Chawarska K, Klin A. Autism in infancy and early childhood. J Autism Dev Disord 2005;56:315–36. [27] Achenbach TM, Rescoria LA. Manual for ASEBA Preschool Forms and Profiles. Burlington VT: University of Vermont, Research Centre for Children, Youth and Families; 2000. [28] Lovibond SH, Lovibond PF. Manual for the Depression Anxiety Stress Scales. 2nd ed. Sydney, NSW: Psychology Foundation; 1995. [29] Baumer JH, Wright D, Mill T. Illness severity measured by CRIB score: a product of changes in perinatal care? Arch Dis Child Fetal Neonatal Ed 1997;77:F211–5. [30] Pandey J, Verbalis A, Robins D, Boorstein H, Klin AM, Chawarska K, Volkmar F, Green J, Barton M, Fein D. Screening for autism in older and younger toddlers with the Modified Checklist for Autism in Toddlers. Autism 2008;12:613–35. [31] Yama B, Freeman T, Graves E, Yuan S, Karen Campbell M. Examination of the properties of the Modified Checklist for Autism in Toddlers (M-CHAT) in a population sample. J Autism Dev Disord 2012;42:23–34. [32] Chlebowski C, Robins DL, Barton ML, Fein D. Large-scale use of the Modified Checklist for Autism in low-risk toddlers. Pediatrics 2013;131:e1121–7. [33] Stephens BE, Bann CM, Watson VW, Sheinkopf SJ, Peralta-Carcelen M, Bodnar A, Yolton K, Goldstein RF, Dusick AM, Wilson-Costello DE, Acarregui MJ, Pappas A, Adams-Chapman I, McGowan EC, Heyne RJ, Hintz AR, Ehrenkranz RA, Fuller J, Das A, Higgins RD, Vohr BR, for the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Screening for autism spectrum disorders in extremely preterm infants. J Dev Behav Pediatr 2012;35: 535–41. [34] Williams K, Helmer M, Duncan GW, Peat JK, Mellis CM. Perinatal and maternal risk factors for autism spectrum disorders in New South Wales, Australia. Child Care Health Dev 2008;34:249–56. [35] Moore GS, Kneitel AW, Walker CK, Gilbert WM, Xing G. Autism in small- and largefor-gestational-age infants. Am J Obstet Gynecol 2012;206:314.e1–9. [36] Hou D, Shetty U, Phillips M, Gray PH. Cerebellar haemorrhage in the extremely preterm infant. J Paediatr Child Health 2012;48:350–5. [37] Limperopoulos C. Autism spectrum disorders in survivors of extreme prematurity. Clin Perinatol 2009;36:791–805. [38] Daniels JL, Forssen U, Hultman CM, Cnattingius S, Savitz DA, Feychting M, Sparen P. Parental psychiatric disorders associated with autism spectrum disorders in the offspring. Pediatrics 2008;121:e1357–62. [39] Davis NO, Carter AS. Parenting stress in mothers and fathers of toddlers with autism spectrum disorders: association with child characteristics. J Autism Dev Disord 2008;38:1276–91. [40] Feinberg E, Augustyn M, Fitzgerald E, Sandler J, Ferreira-Cesar Suarez Z, Chen N, Cabral H, Beardslee W, Silverstein M. Improving maternal mental health after a child's diagnosis of autism spectrum disorder: results from a randomised clinical trial. JAMA Pediatr 2014;168:40–6. [41] Najman JM, Williams GM, Nikles J, Spence S, Bor W, O'Callaghan M, Le Brocque R, Andersen M, Shuttlewood GJ. Bias influencing maternal reports of child behaviour and emotional state. Soc Psychiatry Psychiatr Epidemiol 2001;36:186–94. [42] Johnson S, Marlow N. Positive screening results on the Modified Checklist for Autism in Toddlers implications for very preterm populations. J Pediatr 2009;154:478–80. [43] Reichow B, Barton EE, Boyd BA, Hume. Early intensive behavioural intervention (EIBI) for young children with autism spectrum disorders (ASD). Cochrane Database Syst Rev 2012;10:CD009260. [44] Fernell E, Eriksson MA, Gillberg C. Early diagnosis of autism and impact on prognosis. J Clin Epidemiol 2013;5:33–43.