Abstract
The Clinical Assessment of Confusion-A (CAC-A) is an observational checklist developed for practicing nurses to measure the presence and level of confusion in hospitalized adults. In a previous study, the following dimensions were found using principal factor analysis: cognition, general behavior, motor activity, orientation, psychotic/neurotic behavior, and two uninterpretable factors. A replication study was conducted to evaluate the validity of a statistically derived model for confusion suggested by the CAC-A. Data from a sample of 566 nurses were analyzed. Three theoretically justified statistical models for the structure of confusion were estimated and compared using a model selection approach to covariance structures analysis: a single-factor unidimensional model, an orthogonal six-factor model, and an oblique six-factor model similar to the structure suggested in the development study. The oblique six-factor model provided the best fit in the predictive sense and was the most satisfactory from a theoretical perspective.