Aim In an attempt to clarify the debate surrounding the diagnostic validity of childhood disintegrative disorder (CDD), we systematically reviewed its characteristics and compared it with autism spectrum disorder (ASD). Method Four databases were searched (PubMed, PsycINFO, Embase, and Web of Science). Included articles had participants with CDD, as defined by symptoms present in the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision and the International Classification of Diseases, 10th Revision. Comparison groups were those with ASD and ASD with regression. Case studies were excluded. Results Twenty articles, comprising 96 participants with CDD (80 males, 16 females), were included. Most studies were cross‐sectional. The prevalence of CDD was 1.1 to 9.2 per 100 000, with a mean age at regression of 3 years 2 months (SD 1y 1mo), with a range of 2 years to 7 years. In addition to core CDD symptoms, most had intellectual impairment, anxiety, challenging behaviours, and regressed in toileting skills. Participants with CDD and ASD shared core diagnostic and extra‐diagnostic features. However, participants with CDD seemed to have more severe symptoms and a different symptom profile, including apparently typical development before regression, faster regression, more affective symptoms, and more global developmental deficit. Possible genetic and autoimmune neurobiological mechanisms were identified. Interpretation There is limited high‐quality evidence describing the aetiology and outcomes of CDD. However, given the qualitative and prognostic differences between ASD and CDD, we recommend that future diagnostic criteria should distinguish late‐onset regression.

SummaryAltered neural connectivity in neurodevelopmental disorders is likely subtle, meaning that neuroimaging literature studying development has produced heterogeneous findings. A recent study, published in this issue, illustrates the translational potential of functional connectivity magnetic resonance imaging findings as a biomarker for attention-deficit hyperactivity disorder and autism spectrum disorder. Importantly, it highlights the overlap between disorders, emphasising the need for transdiagnostic and dimensional approaches in neurodevelopment.Declaration of interestNone.