31341
Utility of the M-CHAT-R for Identifying Autism and/or Developmental Disorder in a Representative Community Sample at the 18-Month Pediatric Visit

Poster Presentation
Saturday, May 4, 2019: 11:30 AM-1:30 PM
Room: 710 (Palais des congres de Montreal)
R. A. Sturner1,2, B. J. Howard3,4, L. Stewart5,6, S. M. Attar4, P. E. Bergmann7 and K. Bet4, (1)Pediatrics, Center for Promotion of Child Development through Primary Care, Baltimore, MD, (2)Pediatrics, Johns Hopkins U School of Medicine, Baltimore, MD, (3)Pediatrics, The Johns Hopkins U Sch. of Medicine, Baltimore, MD, (4)Total Child Health, Baltimore, MD, (5)Mental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, (6)Center for Promotion of Child Development through Primary Care, Baltimore, MD, (7)ForesightLogic, Shoreview, MN
Background:

Autism screening is recommended by the AAP at both 18 and 24-month check-up visits (2006) and M-CHAT-R/F is the most popular screen. However, a USPSTF report (2015) did not recommend universal autism screening, citing inadequate data about current screening tool accuracy, specifically lack of information on children screened negative. Studies suggest that the Positive Predictive Validity of the M-CHAT is lower at 18 months than at 24 months, even with the required follow-up interview (Pandey, 2008) (Sturner, 2017). However, it has been argued that almost all false positive M-CHAT-R/F screens have alternative clinical value in detecting children with developmental disorder (Robins, 2013).

Objectives:

To determine the predictive utility of the M-CHAT-R with Follow Up Interview (M-CHAT-R/F) for concurrent autism and developmental diagnoses in a community pediatric sample at 18 months, including children screening negative.

Methods:

Parents of 11,878 children completed the M-CHAT-R at 18-month pediatric visits (16-20 mo.) via an online system (CHADIS). Children with positive screens (96) and age and practice matched controls (314) were recruited. All children had diagnostic evaluations using the ADOS-2 Toddler Version and Mullen tests with clinical impression of autism considered to be a positive for autism. Developmental disorder was defined as the typical criteria for early intervention services (score >1 ½ SD below the mean on two or more subscales or > 2SD on a single subscale).

Results:

The M-CHAT-R/F had significantly lower sensitivity (0.35 vs 0.73) compared to M-CHAT-R with significantly higher specificity (0.89 vs 0.64) but positive predictive value (PPV) tended to be low (0.37) and not significantly different than M-CHAT-R (0.27), and negative predictive value did not differ (0.88 vs 0.93). 51% of the MCHAT-R/F false positives were developmental delay (DD) positives. However, MCHAT-R/F correctly identified only 19% of the developmental screen (ASQ-3) false negatives as true positives for delay. When the entire group of autism +/or DD (Table 2) is considered slightly more than half are identified by the combination of MCHAT-R/F + ASQ (sensitivity 0.53). Addition of the Follow Up to M-CHAT-R+ ASQ lowers sensitivity (0.75 to 0.53) and NPV (0.80 to 0.74) but increases specificity (0.63 to 0.81) and PPV (0.55 to 0.61).

Conclusions:

We cannot assume that children 16-20 months old passing the M-CHAT-R/F are unlikely to have an autism diagnosis. The low sensitivity to the presence of autism was without the intended balancing benefit of increase in positive predictive value, the rationale for requiring a follow-up interview. If referral for developmental delay in addition to referral for autism is considered desirable, false positives are infrequent. However, the clinical utility of the recommended procedure (M-CHAT-R/F) false positives is considerably less if only children missed by a developmental screen are considered.

See more of: Diagnostic, Behavioral, Sensory and Intellectual Screening and Assessment
See more of: Diagnostic, Behavioral, Sensory and Intellectual Screening and Assessment