Implementation of Screening in Pediatric Primary Care

Background: Since the American Academy of Pediatrics (2006) recommended universal toddler screening for autism spectrum disorder (ASD) coupled with broad developmental screening and surveillance, surveys of pediatric providers report increased use of standardized tools such as the Modified Checklist for Autism in Toddlers (M-CHAT) during toddler well child care visits. However, although many providers value the contribution of screening to pediatric care, providers often encounter barriers to conducting systematic screening (Silver et al., 2017; Miller et al., 2011) and to using standardized algorithms to guide referral decisions. For example, in a study exploring nationwide implementation of developmental screening, King and colleagues (2010) noted that as many as 80% of the children within the targeted age ranges were screened. However, in some practices, as few as 27% of children who screened positive were referred for further evaluation. Beyond these survey results, there is limited literature examining fidelity of developmental screening in community practice.

Objectives: The current study utilized a brief interview to evaluate pediatric providers’ implementation of universal, standardized, high-fidelity screening during toddler well child care visits.

Methods: Pediatric primary care providers (n = 27) in community practice in three geographic regions were interviewed about screening practices. Providers completed the interview in order to determine eligibility for a study, although they were not aware of the criteria for inclusion or exclusion. One open-ended question asked how providers identify ASD in their patients, and seven questions asked about specific aspects of screening administration, scoring, interpretation, and subsequent referrals for ASD evaluation. Responses were coded to indicate degree of concordance with universal, standardized, high-fidelity procedures using a 3-point scale to indicate no, partial, or full fidelity.

Results: Intraclass correlation between coders was excellent (ICC = .908). The majority of providers reported using screening tools to identify ASD symptoms (85.2%), using all items in a given tool (75%), and using the tool universally, rather than for children with prior concerns (82.6%). However, of the 24 providers using the M-CHAT or M-CHAT-Revised, only 12.5% reported using the Follow-Up algorithm. Furthermore, few providers (14.3%) followed referral algorithms based on screening results; most used clinical judgment to determine need for referrals to an ASD evaluation, rather than referring all screen positive children. Only one provider (3.7%) reported complete adherence to universal, standardized, high-fidelity screening.

Conclusions: Although most providers reported using screening tools to identify ASD risk in toddlers, many combine screening results with clinical judgment to guide referrals for ASD evaluations. This practice deviates from high-fidelity implementation of screening, and may impede the earliest identification of ASD. Future research aimed at understanding barriers to universal screening and testing of appropriate implementation methods could support dissemination of screening practices.