Cluster-derived profiles of parental reports of Autism and ADHD traits and their associations with diagnoses: EPINED study.

Autism and attention-deficit/hyperactivity disorder (ADHD) frequently co-occur and show substantial phenotypic overlap, challenging categorical diagnostic frameworks. Person-centred approaches may better capture heterogeneity, but the consistency and external validity of these profiles across informants remain unclear. Using data from the EPINED population-based study in Tarragona, Spain, we identified symptom-based clusters from parent reports in preschool-aged (n = 1805) and school-aged (n = 1930) children, examined concordance with teacher-derived clusters, and evaluated associations with diagnoses. Parent-reported cluster analyses yielded a four-cluster solution across both age groups: low/no symptoms (Parent-CP-1/Parent-CS-1), sub-autism + sub-ADHD traits (Parent-CP-2/Parent-CS-2), social communication problems (Parent-CP-3/Parent-CS-3), and ADHD-dominant profiles (Parent-CP-4/Parent-CS-4). Although structures were comparable, parent-teacher agreement was low (κ = 0.03-0.05). Across both age groups, clinical diagnoses were primarily concentrated within ADHD-related behavioural clusters, with other clusters showing lower and more variable representation across diagnostic groups. In preschool-aged children, Parent-CP-4 clusters were associated with autism across both parent (ORs: 4.55-7.46) and teacher-CP-4 (ORs: 6.09-6.51). In school-aged children, only teacher-reported Teacher-CS-4 (ORs: 11.96-20.50) predicted autism diagnoses, while parent-reported behavioural features, higher thought problems and lower rule-breaking behaviour were significant contributors. For ADHD, parents (Parent-CP-4; preschool-aged ORs: 19.74-20.58, Parent-CS-4; school-aged ORs: 25.22-27.63) and teachers (Teacher-CP-2; preschool-aged ORs: 20.80-20.94, Teacher-CS-2; school-aged ORs: 2.66-3.10) were significant contributors, although the parent-cluster effect attenuated to non-significance after inclusion of teacher report. Data-driven profiles explained variance in related clinical measures beyond diagnostic categories. Overall, parent-reported clustering identified meaningful behavioural phenotypes, but cross-informant discrepancies highlight context-dependent symptom expression. The findings underscore the clinical importance of multi-informant and multidimensional assessment approaches to better capture heterogeneity within neurodevelopmental conditions.