Item-Level Positive Predictive Value of MCHAT-R/F in Younger and Older Toddlers

Background: The Modified Checklist for Autism in Toddlers, Revised with Follow Up (M-CHAT-R/F) is a screener for autism spectrum disorders (ASD) for children aged 16-30 months and uses the same scoring algorithm for all ages. A few studies have considered the M-CHAT-R/F’s performance in both younger and older age groups within the validation window, finding higher screen positive rates in children younger than 20 months and higher positive predictive value (PPV) and sensitivity at older ages (Sturner et al., 2017; Sturner et al., 2017). Given variable performance at different ages, it is valuable to understand where differences arise at the item level; that is, which items on the M-CHAT-R/F best discriminate between ASD and other diagnoses at each age group. These data may have implications for improving scoring at younger ages and better understanding age of symptom onset.

Objectives: The present study aims to determine items on MCHAT-R/F that best discriminate between ASD and other diagnoses or no diagnosis in both younger and older toddlers.

Methods: Participants were 384 children ages 16-30 months, evaluated as part of a larger study on the early detection of ASD. Children who screened positive on the M-CHAT-R/F at their 18 or 24-month pediatric well-child care visit were offered a free diagnostic evaluation. Item level PPV and frequency of items endorsed were determined in children aged 20-months and younger (n=234; 149 ASD, 85 other diagnosis or no diagnosis) and older than 20 months (n=150; 87 ASD, 63 other diagnosis or no diagnosis).

Results: In both age groups, items endorsed in more than 50% of ASD cases include items 6, 7, 16, 17, 18 (pointing, joint attention, understanding directions). In younger toddlers, items with highest PPV include 1 (joint attention; PPV=0.648), 10 (response to name; PPV=0.646), 15 (imitation; PPV=0.574), 2 (wondered if deaf; PPV=0.556), and 8 (social interest in peers; PPV=0.548). In older toddlers, items with highest PPV include 18 (understands directions; PPV= 0.849), 15 (imitation; PPV= 0.774), 2 (wondered if child might be deaf; PPV=0.759), 10 (response to name; PPV=0.758), 4 (motor activity; PPV=0.750). PPV for all items was higher in the older vs. younger group.

Conclusions: Results revealed lower PPV for M-CHAT-R/F items in younger toddlers than older toddlers. The most frequently endorsed items in the ASD group did not differ between the younger and older groups; however, items that best discriminated between ASD and other diagnoses were different between age groups, except for items 2 (wondered if deaf), 15 (imitation), and 10 (response to name). Items regarding interest in peers and joint attention are more effective than other items in discriminating ASD in younger children. Items regarding understanding directions, eye contact, and social referencing are more effective than other items in discriminating ASD in older children. Results suggest that different items may be more indicative of ASD in younger and older toddlers. Despite higher false positive results in younger toddlers vs. older toddlers, it is crucial to continue screening at the 18-month checkup, as early screening facilitates early diagnosis and ASD-specific early intervention.