A Comparison Among 5 Methods for the Clinical Diagnosis of Fetal Alcohol Spectrum Disorders

Despite the prevalence of fetal alcohol spectrum disorders (FASD) and the importance of accurate identification of patients, clinical diagnosis may not be consistent across sites due to the heterogeneous nature of FASD and the characteristics of different diagnostic systems used. Here, we compare 5 systems designed to operationalize criteria recommended for the diagnosis of effects of prenatal alcohol exposure (PAE). We determined the extent of consistency among them as well as factors that may reduce intersystem reliability. Compared are: Emory Clinic, Seattle 4-Digit System (Diagnostic Guidelines for Fetal Alcohol Spectrum Disorders: The 4-Digit Diagnostic Code, Seattle, WA, University Publication Services, 2004), Centers for Disease Control and Prevention (Fetal Alcohol Syndrome: Guidelines for Referral and Diagnosis, Department of Health and Human Services, Centers for Disease Control and Prevention, Atlanta, GA, 2004), Canadian Guidelines (CMAJ, 172, 2005, S1), and the Hoyme Modifications (Pediatrics, 115, 2005, 39).

Subjects were 1,581 consecutively registered patients applying for evaluation at a university-based clinic treating alcohol and drug-exposed children. Records of the multidisciplinary evaluation (pediatric, social, psychological) were abstracted. Diagnostic criteria for all 5 systems were applied, and patients were diagnosed according to each of the systems. We compared results using Cohen's Kappa to evaluate the extent of agreement.

Percent of individuals diagnosed with FASD ranged from 4.74% (CDC) to 59.58% (Hoyme). Examination using Cohen's Kappa found modest agreement among systems, particularly when individual diagnoses, Fetal Alcohol Syndrome (FAS), partial FAS (pFAS), and Alcohol-Related Neurodevelopmental Disorder (ARND) were used. Examination of diagnostic criteria found almost perfect agreement on growth (weight; height), with limited overlap for physical features (palpebral fissures, hypoplastic philtrum, upper vermillion) and for neurobehavioral outcomes. Child's race and age influenced agreement among systems, with African American and older children more frequently diagnosed.

Results suggest problems in convergent validity among these systems, as demonstrated by a lack of reliability in diagnosis. Absence of an external standard makes it impossible to determine whether any system is more accurate, but outcomes do suggest areas for future research that may refine diagnosis.