BackgroundPrognostic factors from naturalistic treatment studies of children with Autism Spectrum Disorder (ASD) remain largely unknown. We aimed to identify baseline and treatment-related prognostic predictors at 1-year follow-up after Integrative Care Practices (ICPs). MethodsEighty-nine preschool children with severe ASD were given ICP combining nine therapeutic workshops based on children’s needs. Participants were assessed at baseline and during 12 months follow-up with the Psycho-educational Profile-3-R, Children Autism Rating Scale, Parental Global Impression, and the Autistic Behaviors Scale. We assessed prognostic predictors using multivariable regression models and explored treatment ingredients influencing outcome using Classification and Regression Trees (CART). ResultsMultivariable models showed that being a child from first generation immigrant parents predicted increased maladaptive behaviors, whereas play activities had an opposite effect; severity of ASD symptoms and impaired cognitive functions predicted worse autism severity at follow-up; and lower play activities predicted worse parent impression. Regarding treatment effects, more emotion/behavioral interventions predicted better outcomes, and more communication interventions predicted lower autism severity, whereas more education and cognitive interventions had an opposite effect. CART confirmed that more hours of intervention in the emotion/behavioral domain helped classifying cases with better outcomes. More parental support was associated with decreased maladaptive behaviors. Sensorimotor and education interventions also significantly contributed to classifying cases according to outcomes but defined subgroups with opposite prognosis. ConclusionChildren who exhibited the best prognosis following ICPs had less autism severity, better cognition, and non-immigrant parents at baseline. Emotion/behavior interventions appeared key across all outcomes and should be promoted.