Extraction and Refinement Strategy for detection of autism in 18-month-olds: a guarantee of higher sensitivity and specificity in the process of mass screening.

BACKGROUND

For early detection of autism, it is difficult to maintain an efficient level of sensitivity and specificity based on observational data from a single screening. The Extraction and Refinement (E&R) Strategy utilizes a public children's health surveillance program to produce maximum efficacy in early detection of autism. In the extraction stage, all cases at risk of childhood problems, including developmental abnormality, are identified; in the refinement stage, cases without problems are excluded, leaving only cases with conclusive diagnoses.

METHODS

The city of Yokohama, Japan, conducts a routine child health surveillance program for children at 18 months in which specialized public health nurses administer YACHT-18 (Young Autism and other developmental disorders CHeckup Tool), a screening instrument to identify children at risk for developmental disorders. Children who screen positive undergo further observation, and those without disorders are subsequently excluded. To study the efficacy of early detection procedures for developmental disorders, including autism, 2,814 children born in 1988, examined at 18 months of age, and not already receiving treatment for diseases or disorders were selected.

RESULTS

In the extraction stage, 402 (14.3%) children were identified for follow-up. In the refinement stage, 19 (.7%) of these were referred to the Yokohama Rehabilitation Center and diagnosed with developmental disorders. The extraction stage produced four false negatives, bringing total diagnoses of developmental disorders to 23 (.8%) - including 5 with autistic disorder and 9 with pervasive developmental disorder - not otherwise specified (PDDNOS). Sensitivity was 60% for autistic disorder and 82.6% for developmental disorders. Specificity for developmental disorders rose to 100% with the E&R Strategy. Picture cards used in YACHT-18 provided a finer screen that excluded some false positive cases.

CONCLUSIONS

An extraction and refinement methodology utilizing child health surveillance programs achieve high efficacy for early detection of autism.