Limited longitudinal research examining developmental changes in visuospatial working memory (WM) among children and adolescents with autism spectrum disorder (ASD) has prompted our investigation.

We assessed 123 autistic children and adolescents and 145 typically developing controls (TDC) using the Cambridge Neuropsychological Test Automated Battery at baseline (Time 1 [mean age ± SD]: ASD: 13.04 ± 2.86; TDC: 11.53 ± 2.81) and 2–9 years later (Time 2: ASD: 18.08 ± 3.17; TDC: 16.41 ± 3.09) to measure changes of visuospatial (working) memory over time. The linear mixed model was used to compare the differences between ASD and TDC and estimate the effect of changes over time, age, ASD diagnosis, and interactions of Time×Age×ASD. The overall Age×ASD effect was calculated in the spline regression.

Autistic children and adolescents exhibited significantly poorer performance on all spatial tasks and some visual tasks than their TDC counterparts at Time 1 and Time 2, after adjusting for sex, age, attention deficit/hyperactivity disorder (ADHD), and full-scale intelligence quotient. There was an overall improvement from Time 1 to Time 2 across all tasks with significant Age×Time interactions. Significant Age×ASD interactions were observed in the delayed matching to sample, pattern recognition memory (PRM), spatial span (SSP), and spatial working memory (SWM) tasks with no significant Time×ASD interactions. In the quadratic nonlinear model, Age×ASD interactions were significant in PRM and SSP.

Despite significant improvements during the follow-up period, autistic children and adolescents continue to experience persistent deficits in SWM, with a weaker age-related improvement in visuospatial WM than TDC.