Parent Reported Regression in Autism Spectrum Disorder: Characteristics, Early Development and Later Outcomes.

Background: Previous retrospective research on regression in ASD has produced conflicting results caused by divergent definitions and methodologies. Although development prior to regression is assumed to be typical, evidence has been found for early delays. Besides, mixed results have been reported concerning whether children with regression experience better or worse long-term outcomes. These inconsistencies explain the shift towards prospective studies of infants at risk which provide evidence for declining trajectories for the majority of children with ASD. However, as there are no practical alternative approaches to retrospective parent report, it remains the most commonly used method to study early development. Different strategies to improve this method need to be developed since examining regression in ASD could be of significant diagnostic and therapeutic value.

Objectives: Our first objective is to explore the characteristics of onset patterns reported by parents. In the present retrospective study, different strategies (e.g., use of developmental records) have been applied to deal with parent report reliability. Second, our goal is to confirm whether early delays are already present prior to regression. Third, we investigate outcomes in terms of severity of ASD-symptoms, language and motor skills.

Methods: Parents have reported on the early development of 100 children with ASD (M age=7.57y,SD=1.95,range=3-11y,71% boys) through a questionnaire (EDQ) and interviews (ADI-R+RSQ). Current outcomes have been examined using parent (SCQ,SRS-2,N-CDI) and teacher (SRS-2) reports as well as standardized (WNV,CELF,ABC-M-2) and observational (ADOS-2) measurements. Group differences between children who have ever experienced a regression (ASD-R;n=36), or a regression before 36 months (ASD-R≤36M;n=24), or no regression (ASD-NR;n=64) have been mutually investigated.

Results: Regression concerning language and social skills before 36 months has been reported for 24% of the children and 12% of the children lost skills after 36 months. Four percent showed plateau only, mainly in language skills, starting at a mean age of 30 months.
During early development, ASD-R and ASD-R≤36M have shown a similar amount of social atypicalities and repetitive and stereotyped behaviours as compared to ASD-NR. However, ASD-R have shown significantly less early communication skills compared to ASD-NR (t(98)=2.593,p=<.013,d=.53). Kaplan-Meier survival analyses have revealed no significant differences between the three age groups for their first steps and words, but a significant difference has been detected for first sentences between ASD-NR and ASD-R≤36M (24 vs. 33 months;χ²(1)=6.640,p=<.05).
With regard to later ASD characteristics, ASD-R≤36M and ASD-R have shown significantly more restricted and repetitive behaviours as compared to ASD-NR (U(88)=1,226.50,p=<.002,d=.1.03 and U(100)=1,629.00,p=<.002,d=.73, respectively). Lastly, no significant differences have been found between ASD-NR, ASD-R≤36M and ASD-R concerning later cognitive, language and motor skills.

Conclusions: Generally, children with and without regression seem to develop in a similar way prior to regression. However, children with regression show less communication skills: a finding that may be a valuable predictor. Further, children that regress, turn out to experience a more deleterious outcome later in life, characterized by early language delays and more repetitive and stereotyped behaviour. The findings of the present retrospective study merit further investigation through prospective longitudinal studies with high-risk siblings of children with ASD.