Problems in Using Diagnosis in Child and Adolescent Mental Health Services Research

Abstract

This paper presents results from a three-part study on diagnosis of children with affective and behavior disorders. We examined the reliability, discriminant, and predictive validity of common diagnoses used in mental health services research using a research diagnostic interview. Results suggest four problems: a) some diagnoses demonstrate internal consistency only slightly better than symptoms chosen at random; b) diagnosis did not add appreciably to a brief global functioning screen in predicting service use; c) low interrater reliability among informants and clinicians for six of the most common diagnoses; and d) clinician diagnoses differed between sites in ways that reflect different reimbursement strategies. The study concludes that clinicians and researchers should not assume diagnosis is a useful measure of child and adolescent problems and outcomes until there is more evidence supporting the validity of diagnosis.

Improving Mental Health Services: A 50-Year Journey from Randomized Experiments to Artificial Intelligence and Precision Mental Health

“…Other studies have demonstrated that the diagnosis of depression alone is not sufficient for treatment selection; additional information is required (Iniesta et al 2016 ). Moreover, others have shown that diagnostic categories overlap and are not mutually exclusive (Bickman et al 2012c ). In practice, medication is prescribed according to symptoms and not diagnosis (Waszczuk et al 2017 ).…”

Our need for certainty in an uncertain world: the difference between special education and inclusion?

“…The unreliability of human practice, an issue raised by Kauffman and Sosso and Gallagher, is also particularly important. One issue touched on above is the unreliability of the categories, diagnosis and assessment (Vislie, 2003; Bickman et al, 2012; Frances & Widiger, 2012; BPS – DCP, 2013; Rix, 2015). Another is that for an intervention that emerges from scientific method to be true to itself it cannot propose a hypothesis about an intervention and effectively prove its validity in isolation from the huge range of complexities that exist in moving the intervention into an endlessly variable everyday.…”