Predictive Validity of the Modified Checklist for Autism in Toddlers (M-CHAT) Born Very Preterm

Abstract

Objective To examine the predictive validity of the Modified Checklist for Autism in Toddlers (M-CHAT) administered at 24 months of age for autism spectrum disorder (ASD) diagnosed at 10 years in a U.S. cohort of 827 Extremely Low Gestational Age Newborns (ELGAN) followed from birth. Study design We examined the sensitivity, specificity, positive predictive value and negative predictive value of the M-CHAT in predicting an ASD diagnosis at age 10 based on gold-standard diagnostic instruments. We then assessed how these predictive parameters were affected by sensorimotor and cognitive impairments, and socioeconomic status (SES), as well as emotional/behavioral dysregulation at age 2. Results Using standard criteria, the M-CHAT had a sensitivity of 52%, a specificity of 84%, a positive predictive value (PPV) of 20%, and the negative predictive value (NPV) of 96%. False positive and false negative rates were high among children who had hearing and vision impairments. High false positive rates were also associated with lower SES, motor and cognitive impairments, and emotional/behavioral dysregulation at age 2. Conclusions Among extremely preterm children with ASD, almost one-half were not correctly screened by the M-CHAT at age 2. Sensorimotor and cognitive impairments, SES, and emotional/behavioral dysregulation contributed significantly to M-CHAT misclassifications. Clinicians are advised to consider these factors when screening very preterm toddlers for ASD.

Comparison of Autism Screening in Younger and Older Toddlers

“…The possibility that hyperacusis as an ASD symptom may tend to appear after 18 months in children with ASD requires investigation. The three items that were found to have the highest failure rates in younger toddlers in this study are the same items which tend to have the lowest specificities for predicting autism outcomes in other studies including large samples of M-CHAT negative children providing item data (Kim, Joseph & Faiser, 2016; Srisinghasongkrami, et. al., 2016).…”

Stockholm preterm interaction-based intervention (SPIBI) - study protocol for an RCT of a 12-month parallel-group post-discharge program for extremely preterm infants and their parents

“…Another neuropsychiatric disorder overrepresented among EPT children is autism spectrum disorder (ASD), which is diagnosed in 17% of EPT children. In some studies, up to 29% of EPT children screen positive according to ASD observation protocols [24][25][26]. Moreover, several skills that are important for school success, such as mathematical [27] and linguistic abilities [28,29], are negatively affected by extreme prematurity; preterm-born students in general do not perform at the same level as their term classmates in school [30][31][32].…”

Should the Gross Motor Function Classification System be used for children who do not have cerebral palsy?

“…Among the 118 papers, 76 did not mention that the GMFCS was developed specifically for use with children with CP or discuss its measurement properties . Twenty noted that the GMFCS was created for children with CP but did not justify its application to their condition/clinical description .…”

“…In 38 studies, the five GMFCS levels were collapsed into two or three broader groups (e.g. levels I, II, and III combined under the title of ‘ambulatory’ participants and levels IV and V combined under the title of ‘non‐ambulatory’ participants) . In one study involving children with CP, closed‐head injury, learning disability, or developmental delay, merging of levels (i.e.…”

“…This was to accommodate a lack of spontaneous running or jumping in people with visual deficits. Another study described GMFCS level I as representing ‘normal’ mobility, while 15 studies used a GMFCS level of less than I or GMFCS level ‘0’ to reflect ‘normal’ development or no gross motor dysfunction. One retrospective study used a ‘possible level I’ when they could not decide between ‘level 0’ and level I …”