Accuracy of formal tests for diagnosing mild aphasia: An application of evidence‐based medicine

Abstract

Background: Normal elderly and mildly aphasic individuals may exhibit similar impairments in language comprehension and expression. Accurate differential diagnosis is essential for providing a prognosis, focusing treatment, and justifying reimbursement for services.Aims: We employed the principles of evidence‐based medicine to examine the accuracy of two general language tests (Porch Index of Communicative Ability; Western Aphasia Battery) and two functional communication tests (Communication Activities of Daily Living; American Speech‐Language‐Hearing Association's Functional Assessment of Communication Skills for Adults) for diagnosing mild aphasia, beyond individual clinical expertise.Methods & Procedures: The test performance of 10 mildly aphasic patients was compared with that of 10 non‐brain‐injured (NBI) adults. An operational definition of aphasia was applied as a reference standard for correct diagnosis. Pre‐test and post‐test diagnostic probabilities were compared, using the likelihood ratio as an index of accuracy.Outcomes & Results: In our sample, obtained positive likelihood ratios ranged from 3.00 to 6000.60, and the post‐test probability of a correct positive diagnosis ranged from 91% to 100%. However, the pre‐test probability of a correct positive diagnosis was already high (70–100%), because information necessary to diagnose correctly was available to the clinician prior to formal test administration. Thus, an overall score derived from subsequent administration of a formal test added, at best, moderate improvement over individual clinical expertise. The tests may prove more important for clinicians uncertain of a patient's diagnosis or for diagnosing “borderline” patients whose symptoms are unclear. For pre‐test probabilities between 40% and 60%, administration of one of the formal tests we examined may increase or decrease diagnostic probability by as many as 60 percentage points.Conclusion: Results indicate that overall scores derived from these tests are accurate but may or may not be important in confirming the presence or absence of mild aphasia, depending on the pre‐test probability of a positive diagnosis. Suggestions for testing the validity of our results within a wider spectrum of patients are provided.