Characteristics of Autism Spectrum Disorder Surveillance Cases without a Community Diagnosis: Missouri Autism and Developmental Disabilities Monitoring Network, 2006-2008

Background:  The Autism and Developmental Disabilities Monitoring (ADDM) Network is population-based surveillance system that identifies children with autism spectrum disorders (ASD) in multiple areas of the United States by reviewing evaluations (from birth through 8 years of age) contained in medical and/or educational records of children in a defined population.  The ADDM methods do not require a documented diagnosis of ASD to assign an ASD case status, therefore some ADDM ASD cases do not have a documented ASD diagnosis in their surveillance record.

Objectives:  To characterize Missouri ADDM surveillance ASD cases without a documented ASD diagnosis.

Methods:  The Missouri ADDM site’s surveillance area consists of 5 counties in the metropolitan St. Louis area. This analysis was limited to Missouri (MO) ADDM ASD cases from surveillance years 2006 and 2008. We created a dichotomous variable indicating whether or not an ASD case had a documented ASD diagnosis by a community provider. Univariate analyses were conducted to compare ASD cases with and without at community ASD diagnosis on various demographic, surveillance, and diagnostic variables. A χ² test was used for categorical variables. The Wilcoxon Scores test was used to compare distributions of continuous variables for the two groups. We created a logistic regression model to generate adjusted odds ratios for select variables.

Results: A total of 678 8 year-old children met the MO ADDM case definition for.  124/678 (18.2%) of these cases had no abstracted community diagnosis.  Only 66/124 (53%) had any mention of ASD on an abstracted evaluation. There were no differences between ASD cases with a community diagnosis vs.  those without  on gender, race, age at first evaluation, presence of general developmental delay or social delay  before 36 months. Cases without a community diagnosis were more likely to have fewer abstracted evaluations (median of 3 vs. 5, p<.0001),  a comorbid ADDM classification of cerebral palsy (CP) (7.3% vs. 2.0%, p=0.002), and an ASD-NOS rather than Autism ADDM case classification (32.3% vs. 21.7%, p=0.012).  ADDM cases without a community diagnosis were also less likely to have a developmental language (χ² =8.089, df=1, p=0.005)or  play delay(χ² =4.100, df=1, p=0.043)  documented before the age of 3 years. The final logistic regression model (modeling probability of no community diagnosis) contained the following variables: age at first eval (OR= 0.984, 95% CI, 0.973,0.994), number of abstracted evalutions (OR= 0.830, 95% CI, 0.766,0.900), co-morbid cerebral palsy (OR= 3.964, 95% CI, 1.541,10.200), documented regression (OR= 0.461, 95% CI, 0.256,0.831), and presence of language delay prior to 3 years (OR= 0.547, 95% CI, 0.333,0.899). 

Conclusions: Nearly 20% of MO ADDM ASD cases did not have a documented community. The odds of having documented regression or a language delay before 3 years were approximately 50% less for ASD cases without a community diagnosis suggesting perhaps less severe ASD symptomology. Additionally, the odds of co-morbid CP was nearly 4 times higher for cases without a documented diagnosis. This suggests that ASD may be under-diagnosed in children with other developmental disabilities like CP.