Objectives: The goal of this study was to investigate the level of parent and clinician ASD diagnostic agreement. Our secondary goal was to examine several child, family, and clinician factors that may influence this agreement.
Methods: Cross-Sectional data from one-hundred and fifty four children, ages one to seventeen, were used for the study. Parental report of the child’s ASD diagnosis was obtained from a registration form that was completed while parents were consenting to a database project at an outpatient autism clinic.Clinical diagnosis was extracted from a chart review of the patient medical records. Our chart review protocol promoted the methodological strategies proposed by Gilbert and colleagues (1996) that enhance data quality. For instance, we gave emphasis to evaluations that utilized the Autism Diagnostic Observation Schedule (ADOS; Lord, Rutter, DiLavore, & Risi, 1999), were most recent, and were part of an interdisciplinary team evaluation. When the appropriate document was identified, two reviewers independently determined the clinician’s diagnostic conclusion. If these reviewers were discrepant, a master’s level practitioner re-reviewed the medical record and resolved the final diagnosis; although, this was often unnecessary given the inter-rater agreement was 89% (Kappa=.85). Logistic regression models were then employed to examine the association between parental (marital status, low family income), clinician (physician vs. psychologist), and child (age, race/ethnicity) factors and diagnostic agreement.
Results: Overall, 84% of parents agreed with clinicians that their child was on the autism spectrum (sensitivity=75%, specificity=96%, ROC=.85), while 70% of these families reported the same ASD-specific diagnosis. Results from the logistic models revealed that non-white families (OR .45, CI.95 = .21, .95) and a parent (OR .06, CI.95 = .01, .33) or clinician (OR .13, CI.95 = .02, .70) report of “Autism Spectrum Disorder” significantly (p<.05) decreased the likelihood of concordance.
Conclusions: Results from our analysis yielded a moderate to high level of accord concerning the child’s specific ASD diagnosis (e.g., PDD-NOS) and a high level of agreement regarding whether the child has an ASD or not. We also found that non-white families and a diagnosis that was provided outside of the DSM-IV-TR (AMA, 2000) (e.g., “Autism Spectrum Disorder”) increased the risk for dissonance. Although this data holds clinical and research utility, it is presently unclear why parental report did not match the results from the child’s diagnostic evaluation. We suspect this could be a product of poor communication, disagreement, variability in the diagnostic presentation, and recall bias. Implications and new avenues for research will be discussed.