Revisiting Western Aphasia Battery-Revised in the Context of Mild Impairment

This study aims to evaluate the diagnostic accuracy, sensitivity, and specificity of the Western Aphasia Battery-Revised Aphasia Quotient (WAB-R AQ) cutoff of 93.7 of 100 in detecting atypical discourse observed on a picture description task and to identify the optimal AQ value for detection of aphasia.

A total of 171 English-speaking adults with a diagnosis of acute ischemic stroke completed both the WAB-R and a description of either the Cookie Theft or the Modern Cookie Theft within 2 weeks of symptom onset as part of an ongoing longitudinal observational study of poststroke recovery of language. Picture descriptions were used to measure the quantity of content and the informational efficiency in syllables, which were compared to normative values. Receiver operating characteristic analysis was used to determine the relationship between AQ and aphasic discourse and to determine the optimal AQ for detecting aphasic discourse.

The current AQ cutoff score was accurate in detecting aphasic discourse in 75% of cases (sensitivity = 77%; specificity = 64%). Most inaccurate classifications were due to the nearly one in five patients who demonstrated aphasic discourse but were deemed not aphasic by the WAB-R (false negatives). The optimal AQ cutoff was 96.7, calculated by maximizing Cohen's kappa (.38). This cutoff increased classification accuracy from 75% to 83% and reduced false negatives from 19% to 9%.

Increasing the WAB-R AQ cutoff to 96.7 is an easily implementable adjustment that improves the detection of aphasia.