Validation of the Amharic version of Cognitive Assessment Interview (CAI-A) in people with schizophrenia in Ethiopia

Abstract

Assessing cognition with interview-based measures could be a low-resource alternative to traditional cognitive tests. We previously adapted the Cognitive Assessment Interview (CAI) into Amharic (CAI-A) for use with people with schizophrenia (PWS) in Ethiopia. This study examined the convergent and structural validity of the CAI-A in a group of 350 PWS sub-sampled from the Neuropsychiatric Genetics of African Populations – Psychosis (NeuroGAP-Psychosis) study, who fulfilled the inclusion criteria. Data were analyzed using confirmatory factor analysis (CFA), Spearman's correlation coefficient (ρ), multiple regression, and Item Response Theory (IRT). A one-factor solution best fits the items in the tool (factor loadings between 0.58 and 0.79), suggesting structural validity. The total score of the CAI-A moderately correlated with functioning (ρ = 0.44, p < 0.001) and symptom dimensions (ρ between 0.38 and 0.46, p < 0.001), suggesting convergent validity. Multiple regression showed that age (β = −0.06, 95 % CI (−0.12, 0.00), p = 0.044), the duration of illness (β = 0.08, 95 % CI (0.01, 0.14), p = 0.033), and medication side effects (β = 0.35, 95 % CI (0.21, 0.50), p < 0.001) were positively and significantly associated with the CAI-A total score. The IRT analysis suggested that the tool best functions among participants with moderate to severe impairment (difficulty coefficient between 0.05 and 2.73). We found that the CAI-A is a valid tool for use in Ethiopia. The moderate correlation with symptom and functional measures suggested that self-reported cognitive symptoms parallel other symptom dimensions and functional disability. The CAI-A can be used in clinical practice and research activities in PWS in Ethiopia when subjective assessment of cognition is desired.