•  
  •  
 

Article Title

Early Identification of Autism Spectrum Disorders Using Patterns of Service Use Prior to Diagnosis

Publication Date

8-15-2016

Keywords

autism spectrum disorders, diagnosis

Abstract

Background/Aims: Early intervention services in the first few years of life significantly improve outcomes for children with autism spectrum disorders (ASD). Yet there are often delays in the identification and diagnosis of ASD. If health systems were better able to identify children at high risk for ASD and bring them in for comprehensive evaluation earlier, more children could be able benefit from early intervention.

Methods: This study used data from three Kaiser Permanente sites (Northern California, Georgia, Northwest). All children born between Jan. 1, 2000, and Dec. 31, 2009, were included in the study population. All data stored in the electronic medical records up to June 2012 were collected. Children with an ASD diagnosis from an ASD specialist or two or more ASD diagnoses from nonspecialists were considered to have a valid ASD diagnosis. Controls were matched to cases at a 10:1 ratio and were matched on age, birth month and membership in the 12 months prior to the case’s first ASD diagnosis. We used two-part regression models and cluster analysis to examine whether children diagnosed with ASD have different patterns of service use prior to their first ASD diagnosis than children who do not go on to receive an ASD diagnosis.

Results: Standard two-part regression models and cluster analyses identified differential patterns of service use prior to diagnosis for children with ASD compared to children without ASD. Regression models indicated that children with ASD have much greater use of specialty care (e.g. mental health services, neurology) prior to first ASD diagnosis. Cluster analyses provided a more nuanced picture. For example, early-onset ASD cases had a distinctive pattern separate from those who received a diagnosis at an older age.

Conclusion: Using data on patterns of service use prior to the initial ASD diagnosis, it may be possible to identify children who are likely on a trajectory to get an ASD diagnosis. By developing and applying algorithms that can identify children most at risk in the first few years of life, comprehensive evaluation could be done at an earlier age, improving the outcomes for these children.

Share

COinS