Predictive validity of the standardized infant NeuroDevelopmental assessment (SINDA) to identify 4–5 year-old children at risk of developmental delay in a low-risk sample

Abstract

BackgroundEarly detection of developmental problems is important as it allows for early intervention. Previous studies, in high-risk infants, found high predictive values of atypical scores on the Standardized Infant NeuroDevelopmental Assessment (SINDA) for later neurodevelopmental disorders (i.e., cerebral palsy, intellectual disability). AimsThe present study explored SINDA's predictive values to identify risk of developmental delay at 4–5 years. Study designCohort study. Subjects786 low-risk Dutch children (367 boys; median gestational age: 40 (27–42) weeks; mean birth weight: 3455 (SD 577) grams). Outcome measuresThe SINDA was assessed at 2–12 months and risk of developmental delay was assessed using the Ages and Stages Questionnaire (ASQ) at 4–5 years. SINDA's predictive values were determined for five ASQ domains and the total ASQ score for children at risk of marked (all ASQ domains deviant) and any (one or more ASQ domains deviant) developmental delay. ResultsPresence of one atypical SINDA scale score showed low to moderate sensitivities (12–88 %, depending on the SINDA scale and ASQ domain involved), moderate to high specificities (66–94 %), low positive predictive values (PPVs; 3–16 %), and high negative predictive values (NPVs; 95–100 %) for children at risk of marked and any developmental. Presence of multiple atypical SINDA scale scores predicted deviant ASQ domains slightly better (sensitivities = 11–62 %, specificities = 90–98 %, PPVs = 6–30 %, and NPVs = 95–100 %). ConclusionsIn low-risk infants, SINDA's predictive value is low for detecting children at risk of marked and any developmental delay at 4–5 years, as reflected by the low sensitivities. One of the explanations is the relatively low prevalence of developmental delay in low-risk populations. This might have consequences for the application of the SINDA in general healthcare settings (e.g. child health clinics), but further studies are needed to draw this conclusion.