Pinto-Martin et al’s study of whether a broad-band developmental screen, i.e., “Parents’ Evaluation of Developmental Status” (PEDS®), detects children with probable autism spectrum disorder (ASD) is fraught with problems in methodology and analysis. Although the authors report administering PEDS® by interview in 7% of cases, it is not evident that interviews were also provided if parents failed to answer any PEDS® questions in writing. The very low reported rates of parents with non-predictive concerns (13% in the study in question, versus 20% to 28% across various PEDS®’ standardization studies) suggests PEDS® was not administered correctly. The apparent lack of adherence to standardized directions raises concerns about whether reported frequencies on other PEDS® paths and on the M-CHAT itself reflect legitimate results. That PEDS® has established sensitivity and specificity to social-emotional problems, language impairments, learning and intellectual disabilities, also casts doubt on the study’s methods, results, and interpretation.

In addition, the authors used an outdated scoring protocol for viewing the relationship between PEDS® performance and possible ASD. In a much larger discriminant validity study on 427 children administered both PEDS® and the M-CHAT,4 three or more parental concerns among social-emotional, behavioral, fine motor, gross motor, or receptive language development was 78% sensitive to problematic M-CHAT scores, while fewer than three such concerns was 75% specific. That the researchers excluded most of the above concerns from their analysis only to conclude that PEDS® does not capture such issues as social interactions and play behaviors, obfuscates the thoroughly established value of listening to parents. The authors were personally informed of scoring changes for PEDS® more than two years ago when they presented findings on 66 children presumably included in the current study.5 At the time, their results were markedly different: PEDS® was found to have 78% sensitivity to M-CHAT failures and 26% specificity, confirming, as is consistent with other research,3-4 that PEDS® identifies children at risk for other types of disabilities and that a different scoring paradigm is needed to refine referrals for ASD evaluations. That Pinto-Martin et al have continued in their refusal to re-evaluate their data, is completely bewildering.

The value of viewing unique performance patterns for children with ASD has also been found on the Ages and Stages Questionnaire.6 In a presentation at the October 2008 Annual Meeting of the Society for Developmental-Behavioral Pediatrics, children with ASD had high rates of deficits in the personal-social, problem-solving, communication, and in the motor domain to a lesser extent. This illustrates the importance of viewing more than just pass/fail results when attempting to discern ASD from other developmental disorders, and is a recommendation that makes sense given the complex characteristics of children on the spectrum.

Although we agree with Pinto-Martin et al in their recommendation that an ASD specific screen should be deployed periodically alongside broad-band screens, many providers do not find time to use measures such as the M-CHAT. Pinto-Martin et al’s study could have contributed much to early detection of ASD in primary care settings if they had administered and correctly scored the broad-band screen they researched and cross-validated prior research by viewing performance patterns. Instead, the researchers proffer only the erroneous conclusion that broad-band screens “will miss a substantial proportion” of those at risk for ASD. This conclusion is not supported by their study. We urge the authors to review their data, deal with the probability that some test administrations were confounded by literacy and/or language barriers, apply current scoring paradigms, reanalyze their results and conclusions, and/or retract their manuscript.

References