Is Earlier Better? The Relationship between Age When Starting Early Intervention and Outcomes for Children with Autism Spectrum Disorder: A Selective Review
Table 3
Results of studies and functioning areas showing beneficial effects of earlier starting age of intervention.
Study
Results showing significant predictive contribution of age of starting EI
Unique variance estimated for age of starting EI
Interaction effects that included starting age as predictor
Outcome variables affected by starting age
Cognitive (nonverbal)
Adaptive
ASD severity
Language
Social
Studies with primarily preschoolers
Perry et al., 2011
Performed separate regressions for eight outcome variables: age contributed significantly for cognitive scores, autism severity, socialization, and motor skills when the baseline of the same variable at Time 1 was held constant. In further analyses of subgroups, such as those no longer having an ASD at outcome, results showed these participants, when first entering EI, tended to be younger and had higher IQs and less severe autism
N/A
N/A
Yes
N/A
N/A
N/A
N/A
Kasari et al., 2012
For the outcome of spoken vocabulary at age 8 years, child age at the beginning of the intervention, joint attention ability, play level, and treatment group assignment all contributed significantly. “On average, the children gained a standard score of 1.1 (SE = .3) in spoken vocabulary ability per month that they enter the treatment earlier”
Child age at the beginning of the study predicted by itself 14% of the variability of spoken vocabulary ability at age 8 ()
N/A
No
N/A
N/A
Yes
N/A
Smith, Klorman, & Mruzek, 2015
Performed multiple regressions for each outcome measure, separately for Year 1 and Year 2. Controlling for other predictors, younger age at intake predicted higher outcome for MSEL and, marginally, for VABS and ADOS scores
N/A
Both younger and older children made more progress in Year 1 than in Year 2, but children who were younger when starting made even more progress in Year 1 than those who were older
Yes
No
No (Year 1)
N/A
No
Yes
Yes
Yes (Year 2)
N/A
Yes
Eapen, Crncec, & Walter, 2016
Controlling for all other predictors (baseline IQ, autism severity, and adaptive behaviors) in a linear regression, initial age accounted for a significant amount of the variance for autism severity as measured by parent-rated SCQ, but not for other outcomes
N/A
N/A
No
No
Yes
No
Yes
Studies spanning above and below three years of age
Flanagan, Perry, & Freeman, 2012
Performed regression analyses with one Time 2 outcome variable (IQ). Age × Treatment Group interaction term was significant (see column 3)
Used in an interaction term, age accounted for 13% of the variance
Being younger was an advantage only if children were in the Intensive Behavioral Intervention group rather than the treatment as usual group
Yes
N/A
N/A
N/A
N/A
Pellicano, 2012
At Time 2, all participants chosen for review had IQs > 80, but one group was determined to keep an ASD diagnosis and the other was deemed nonspectrum. These two groups did not differ on initial symptom level, language levels, or weekly hours of intervention. They did significantly differ on age of starting EI in that the nonspectrum group started earlier
N/A
N/A
N/A
N/A
Yes
N/A
N/A
Virues-Ortega, Rodriguez, & Yu, 2013
Regression analysis of predictors fitting to an established trajectory curve showed that total intervention time was the best predictor, but the second best was starting age, and after that, starting level of skills depending on outcome variable
N/A
N/A
No
Yes
N/A
Yes
Yes
Hedvall et al.,2015
From a larger group, 30 children who gained the most and 23 who lost the most over the intervention period were compared. Five predictors, including age at entry, when tested individually, showed significant relationships to outcome (adaptive behavior scores). However, results suggested younger children were more impaired. In a logistic regression of these predictors, only Time 1 cognitive level contributed independent variance
N/A
N/A
No
No
No
No
No
Vivanti et al., 2016
Intervention led to an increase in verbal scores for both younger and older age groups, but younger children made more gains. The regression analysis suggests that children who are older and have a lower language level (<13 months age equivalent) at entry made the least gains in Verbal DQ)
N/A
A significant Age Group × Time interaction showed that children in the younger group experienced comparatively larger gains after intervention (Nonverbal)
No (Nonverbal DQ)
No
No
Yes (Verbal DQ)
N/A
Studies with infants and toddlers (all <3 years old)
Itzchak & Zachor, 2011
78 children were combined from an earlier study (ABA vs. eclectic intervention) to examine predictors of outcome apart from treatment. Predictors included child variables such as age as well as the environmental variables of mother’s age and education. When age was included in one of the regression analyses, with the change score in MSEL from Time 1 to Time 2 as the cognitive outcome, it was a significant predictor
When age was entered at the third step, it accounted for 3% () of the unique variance in the model predicting cognitive change
N/A
Yes
No
N/A
N/A
N/A
Rogers et al.,2012
Regression procedure using Time 2 IQ as outcome showed age of starting intervention and number of intervention hours significantly predicted better outcomes for control and intervention groups combined
N/A
N/A
Yes (Nonverbal DQ)
N/A
N/A
N/A
N/A
Anderson, Liang, & Lord, 2014
Two groups (ASD with IQ > 70 vs. those who lost ASD diagnosis by age 19) were compared on Time 1 adaptive scores to determine that they did not differ. In this way, this baseline ability feature was held constant. Those who no longer had ASD were more likely to have had intervention between ages 2 and 3 years compared to those in the cognitively able ASD group
N/A
N/A
N/A
N/A
Yes
N/A
N/A
Tiura et al., 2016
Growth curve analysis was used to investigate predictive relationships to longitudinal changes of the ABA treatment outcomes. Relationships between Time 1 variables showed age of entry positively related to communication, cognitive, and adaptive levels. Age did not predict outcome levels of these variables. There was a nonsignificant trend for younger age to predict growth scores, however
N/A
N/A
N/A
No
N/A
No
No
Vivanti et al, 2018
The predictive role of starting age was examined for each of the two treatment groups (N = 22 each) for several distal and proximal child outcome variables. Partial correlations and linear regression showed that starting age was significantly and independently associated with NVQ such that the younger the starting age, at Time 1, the higher the NVQ at Time 2. This outcome was not achieved for any of the four other domains examined
