Translation, cultural adaptation and psychometric evaluation of the age-friendly cities and communities questionnaire for Sweden

BACKGROUND: The Age-Friendly Cities and Communities Questionnaire (AFCCQ) measures how older people perceive the age-friendliness of their city or community. With 23% of Swedish residents projected to be aged ≥ 65 years by 2040, a culturally adapted and psychometrically validated Swedish version (AFCCQ-SE) is needed to guide age-friendly initiatives nationally. This study aimed to translate the AFCCQ into Swedish and to conduct its cross-cultural adaptation and psychometric validation. METHODS: In Phase I, the AFCCQ-EN was independently forward-translated into Swedish and back-translated into English. Ten older adults and ten academic experts then rated each item's relevance on a 4-point scale. In Phase II, 874 community-dwelling Swedes aged ≥ 65 years (64.9% female; mean age 75.7 ± 5.8 years; 82.8% urban) from two regions completed the AFCCQ-SE. Confirmatory factor analysis (CFA) was used to assess the nine-domain structure against fit criteria and composite reliability (CR) was used to evaluate internal consistency. Rasch analysis examined various aspects of the AFCCQ at the item and rating scale level. RESULTS: I-CVI values of all items ranged from 0.80 to 1.00 (mostly 0.90). CFA supported the proposed nine-domain structure (χ²[194] = 562.9; normed χ² = 2.90; CFI = 0.962; TLI = 0.951; SRMR = 0.0335; RMSEA = 0.047). CR values were acceptable across most domains (0.783-0.937), with two domains marginally below the threshold (0.658-0.699). Rasch analyses showed acceptable rating categories and satisfactory item fit. Urban-rural DIF was found for two transportation items (0.51 & 0.71 logits, Holm adjusted p = 0.000). PCA supported unidimensionality within each domain rather than for the questionnaire as a whole. Twenty-two item pairs showed local dependence (r = 0.16-0.84), 21 of them within a single domain. Aggregating the items into nine testlets largely resolved local dependence (largest residual correlation = 0.10), and the first residual contrast eigenvalue decreased from 3.32 to 1.52 (< 2.0), supporting unidimensionality of the testlet model. CONCLUSIONS: This AFCCQ-SE study, using both CFA and Rasch analysis, provided preliminary support for its psychometric properties in this Swedish sample. Further validation using samples from larger regions and longitudinal study designs is needed to verify these findings.