Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Diagnostic Accuracy of Multidomain Aphasia Batteries by Administration Duration: A Meta-Analysis
Poster Session E, Friday, October 2, 11:00 am - 1:00 pm, Wangari Maathai
Sabina Sharifova1, Laine Cialdella1, E. Susan Duncan1; 1The Graduate Center of the City University of New York
Aphasia serves as an important model for investigating the neurobiology of language. Accurate aphasia assessment tools are essential for reliable characterization of language impairment and establishing valid brain-behavior relationships. This is particularly relevant in the era of large-scale neuroimaging and “big data” research, where brief yet accurate assessment tools may facilitate broader participant inclusion and efficient phenotyping across heterogeneous datasets. Detection of aphasia is also clinically critical for prognosis, treatment planning, and rehabilitation outcomes, although administration time may be limited. Comprehensive tools provide detailed linguistic profiling but often require extended administration time, limiting feasibility in acute-care settings. Shorter screening tools reduce administration burden but may raise sensitivity concerns, particularly for milder impairments. This meta-analysis examined whether longer assessments provide superior diagnostic performance compared to shorter multidomain tools commonly administered in less than 30 minutes. We hypothesized that longer assessments would not necessarily demonstrate superior diagnostic accuracy and that well-designed short-form tools would show comparable clinical and research utility. A systematic literature search was conducted across six databases from inception to September 2025, with an updated manual search through April 2026. Searches included terms related to aphasia and diagnostic performance. Screening and deduplication were conducted in Rayyan for reviewer blinding. Of 5,749 records identified, 78 studies met the inclusion criteria following title/abstract and full-text review, comprising 91 independent data entries. Interrater reliability during full-text screening was substantial (89.5% agreement, Cohen’s κ=0.708). Extracted variables included participant demographics, battery administration time, self-identified length (e.g., brief/rapid vs comprehensive/full), and diagnostic performance metrics (AUC, sensitivity, specificity, accuracy, predictive values, likelihood ratios). Administration time categories were binned as 1-15 min (N=34), 16-30 (N=8), 31-45 (N=4), 46-60 (N=1), 61+ (N=2), and N/A (N=42); categories above 30 minutes were collapsed into a single 31+ group. Battery length classifications included brief (N=53), comprehensive (N=10), and N/A (N=28). Studies with unavailable moderator information were excluded from moderator analyses. Random-effects meta-analyses and mixed-effects meta-regressions examined diagnostic performance. Aphasia batteries demonstrated high overall diagnostic accuracy. Pooled random-effects sensitivity was 0.933 (95% CI [0.909, 0.951]; I²=55.1%) and pooled specificity was 0.925 (95% CI [0.899, 0.945]; I²=79.4%). Substantial heterogeneity supported the use of random-effects models, and Egger’s regression tests suggested possible small-study effects. Continuous administration time significantly negatively moderated sensitivity (QM=6.11, p=.013; β=−0.0115), although the effect size was small and should be interpreted cautiously; specificity was not significantly moderated (QM=0.02, p=.895). Categorical administration time bins were not significant moderators of diagnostic performance for sensitivity (QM=1.77, p=.413) or specificity (QM=0.87, p=.647). Subgroup analyses similarly showed no meaningful differences between brief and comprehensive batteries, with comparable pooled sensitivity estimates (0.936 vs. 0.920; p=.702). Our findings did not demonstrate superior diagnostic performance of comprehensive batteries relative to shorter tools. These results support the use of brief aphasia batteries in brain-behavior research and as practical alternatives in clinical settings where administration time is limited. Limitations of this study include substantial heterogeneity, evidence of potential publication bias, and data distribution bias towards shorter tools. Future analyses will synthesize sensitivity and specificity measures jointly using a hierarchical diagnostic accuracy model.
Topic Areas: Disorders: Acquired, Speech-Language Treatment