A methodological consideration in the construction of actuarial interpretation systems

Area of Science:

• Child and Adolescent Psychiatry
• Behavioral Science
• Psychometrics

Background:

• Actuarial interpretive systems often rely on medical records for criterion data.
• Previous studies have not systematically compared data collection methods for these systems.

Purpose of the Study:

• To compare adjective frequencies from medical records versus direct checklist completion in behaviorally disturbed youth.
• To identify potential biases in medical record data for developing actuarial systems.

Main Methods:

• Two samples (n=79 each) of behaviorally disturbed children and adolescents were used.
• One sample's data derived from medical record review; the other from direct checklist completion post-evaluation.
• Adjective frequencies on a 100-item checklist were compared between samples.

Main Results:

• Significant differences in adjective frequencies were found for over one-third of checklist items.
• Syndrome-specific behaviors showed minimal differences.
• Less diagnostically specific, but clinically relevant, characteristics were under-reported in the medical record sample.

Conclusions:

• Medical record data may lead to under-reporting of important clinical characteristics.
• This has implications for the accuracy of actuarial systems, retrospective studies, and empirical classification systems.
• Systematic review of potential correlates is needed when using medical records.