Interpretative guidelines for neuropsychiatric measures with dichotomously scored items.

Abstract

Neuropsychiatric measures consisting of dichotomously scored items are commonly used in clinical assessment. After summing these items, clinical guidelines frequently recommend cutoff scores to determine the presence or degree of a particular attribute, such as depression. However, blind application of such cutoffs neglects whether the total score is significantly different from chance. This confounding problem is illustrated using the Geriatric Depression Scale (GDS), and recommendations for interpreting the degree to which a GDS score significantly exceeds chance are presented. Specifically, GDS scores between 11 and 20, inclusive, were found not to differ significantly from chance (p > 0.05), assuming a random response pattern. The importance of supportive clinical evidence of depressive symptomatology is increased for scores in this range. These guidelines will be helpful in using such measures with patients who may vary with respect to response accuracy, and in assessing possible incomplete effort or random responding.