DEVELOPMENT AND EVALUATION OF THE CENTRAL AFRICA DEPENDENCY SCALE (CA-D) FOR DEMENTIA DIAGNOSIS IN OLDER ADULTS WITH ITEM RESPONSE THEORY: THE EPIDEMCA STUDY

Abstract

Despite the worldwide ageing of population and the associated burden of age-related diseases, dementia remains understudied in Sub-Saharan Africa. The aim of the EPIDEMCA programme is to estimate and compare prevalence of dementia and related syndromes in urban and rural areas in Central Africa. Yet, few validated and adapted tools exist to assess the several dimensions essential for the diagnosis of dementia and cognitive disorders. The aim of the current study was to evaluate and validate the Central African dependency scale (CA-D), an empirically adapted scale from Katz and Lawton scales. An initial 16-item scale of daily activities adapted to the African context was completed by 301 participants with low cognitive performances to evaluate their level of dependency. Before psychometric evaluation, IRT model assumptions were tested using confirmatory and exploratory factor analyses (CFA and EFA). Items were then evaluated using full information item factor analysis on polytomous response data under the item response theory paradigm. Differential item functioning (DIF) across groups (e.g. age, gender, countries) was analyzed and reliability was assessed based on Cronbach's alpha. Unidimensionality was supported by CFA and EFA tests (CFI=0.992, TLI=0.990, first factor at 73% and a ratio of 24.33) after the removal of 6 items. The 10-item scale (FIELDS, OCCUP, MARKET, FESTI, CEREMONI, FINANCES, MOVE, HOME, MEAL and LAUNDRY) fit the IRT model, showed a good coverage of the CA-D scale (threshold parameters range: -2.56 to 3.82) with good item discrimination properties (discrimination item parameters: 1.40 to 4.08) and a high reliability (Cronbach's alpha=0.92). No DIF were found due to gender, age, education level or cognitive status, but the analyses showed DIF for geographical purposes (country) with an effect of rural and urban environment. These results provide good support for the reliability and internal validity of the CA-D scale as a promising tool for dependency assessment in the elderly population in the diagnosis of dementia in Central Africa context. Further research need to investigate the scale external validity and scale calibration to ensure its generalizability in African studies.