Adaptation of the Aphasia Impact Questionnaire-21 into Greek: A Reliability and Validity Study

Abstract

The impact of aphasia on the everyday life of Greek-speaking people with aphasia (PWA) is often underestimated by rehabilitation clinicians. This study explores the adaptation and psychometric properties of the Greek (GR) version of The Aphasia Impact Questionnaire-21 (AIQ-21-GR) to address this issue. The aim of this study is to determine the reliability and validity of the Greek version of the AIQ-21. The AIQ-21-GR was administered to 69 stroke survivors, 47 with aphasia and 22 without aphasia. The data were analyzed to determine reliability and validity. Content validity was based on the Consensus-based Standards for the selection of health Measurement Instruments guidelines. The AIQ-21-GR shows high levels of reliability and validity. The results confirmed high scores of internal consistency (Cronbach’s α = 0.91) and indicated good known—groups validity (Mann–Whitney U = 202, p < 001). Content validity achieved high scores with an overall median score of 4 [Q25 = 4, Q75 = 5]. The psychometric properties of the AIQ-21-GR support the reliability and validity of the tool for investigating the impact of aphasia on the quality of life of Greek-speaking PWA. The AIQ-21-GR can be used for setting functional goals in collaboration with PWA and as a patient reported outcome measure for functional communication training.

The development of the PAOLI (People with Aphasia and Other Layperson Involvement) guideline for reporting Patient and Public Involvement (PPI) in aphasia research

“…The research team consisted of 7 individuals: a leading investigator, 2 PPI partners and 4 communication partners. The rst author (MC), a senior speech and language therapist practising in aphasia rehabilitation, with previous PPI experience in aphasia research1 served as the lead investigator [7,9,14]. The two PPI partners are co-authors AK and JRS.…”

“…AK holds a Master of Science (MSc) in Social Research Methods from the University of Sussex and was enrolled in doctoral studies in Social Care but dropped out after her stroke event. She has previous experience as a PPI partner on research projects, and has co-authored publications from the research [9,14]. JRS was Professor of English Linguistics at the University of Lausanne and is now retired.…”

The Bayesian One-sample t-Test Supersedes Correlation Analysis as a Test of Validity

“…Commonly a correlation of ρ > 0.9 or ρ < -0.9, is considered an acceptable validity cut-off in psychometry, while a correlation between 0.5 > ρ > -0.5, typically indicates very poor validity (Taherdoost, 2016; see also Charalambous, Phylactou, Kountouri, et al, 2022;Charalambous, Phylactou, Elriz, et al, 2022;Hilari et al, 2003Hilari et al, , 2018.…”

“…In this scenario, a correlation analysis between the two scales (Figure 2B), indicates an almost perfect linear relationship (Pearson's ρ = .999, BF10 = 7.30 x 10 143 ). Considering how correlations have been previously used to test validity (e.g., Charalambous, Phylactou, Kountouri, et al, 2022;Charalambous, Phylactou, Elriz, et al, 2022;Drevon et al, 2017;Hilari et al, 2018;Yaşar et al, 2021), we should assume that the scale with the systematic error is a valid measure, since the correlation is almost perfect, despite the fact that in every measure by the faulty scale there is a constant difference from the actual weight between 4.9 kg and 5.1 kg. In other words, the results from this correlation analysis are misleading and can be misinterpreted as evidence of high validity, even though the two tools provide different observations.…”

“…Such analyses concern the calculation of differences between the slopes and/or the intercepts of the observed correlations (see , for an example), compared to the expected (e.g., perfect) correlation (Figure 2C). However, such analyses are considered exploratory and are often omitted (see Charalambous, Phylactou, Kountouri, et al, 2022;Charalambous, Phylactou, Elriz, et al, 2022;Drevon et al, 2017;Hilari et al, 2018;Yaşar et al, 2021), since they do not offer the benefits of applying a simple statistical test as described above (e.g., Blanchard et al, 2018;Myung & Pitt, 1997; see also Wagenmakers et al, 2010).…”