Parent screening questionnaires to detect cognitive and language delay at 2 years in high-risk infants: an analysis from the Victorian Infant Collaborative Study 2016–2017 cohort

Objective To determine the accuracy of two developmental screening questionnaires to detect cognitive or language delay, defined using the Bayley Scales of Infant and Toddler Development-Third Edition (Bayley-III), in children born extremely preterm (EP: <28 weeks’ gestation) or extremely low birth weight (ELBW: <1000 g). Design Prospective cohort study. Setting State of Victoria, Australia. Patients 211 infants born EP/ELBW assessed at 2 years’ corrected age (mean 2.2, SD 0.2). Main outcome measures Cognitive and language delay (<−1 SD) on the Bayley-III. The screening questionnaires were the Parent Report of Children’s Abilities-Revised (PARCA-R) and the Ages & Stages Questionnaires Third Edition (ASQ-3). Results The PARCA-R performed better than the ASQ-3, but neither questionnaire had substantial agreement with the Bayley-III to detect cognitive delay; kappa (95% CI): PARCA-R 0.43 (0.23, 0.63); ASQ-3 0.15 (−0.05, 0.35); sensitivity (95% CI): PARCA-R 70% (53%, 84%) ASQ-3 62% (47%, 76%); specificity (95% CI): PARCA-R 73% (60%, 84%) ASQ-3 53% (38%, 68%). When both tools were used in combination (below cut-off on at least one assessment), sensitivity increased to 78% (60%, 91%) but specificity fell to 45% (29%, 62%). Similar trends were noted for language delay on the Bayley-III, although kappa values were better than for cognitive delay. Conclusions Neither screening questionnaire identified cognitive delay well, but both were better at identifying language delay. The PARCA-R detects delay on the Bayley-III more accurately than the ASQ-3. Sensitivity for detecting delay is greatest when the PARCA-R and ASQ-3 were used in combination, but resulted in lower specificity.