Citation

Reichenheim ME, Paixão CM, Moraes CL. J. Epidemiol. Community Health 2009; 63(11): 878-883.

Affiliation

Institute of Social Medicine (IMS), Rio de Janeiro State University (UERJ), Brazil.

Copyright

(Copyright © 2009, BMJ Publishing Group)

DOI

10.1136/jech.2008.084095

PMID

19622518

Abstract

BACKGROUND: Domestic violence against the elderly (DVAE) is an increasingly recognized problem and its appropriate detection is imperative. As part of an ongoing cross-cultural adaptation process, this study evaluates the construct validity of a Brazilian Portuguese version of the instrument Caregiver Abuse Screen (CASE). METHODS: A cross-sectional study comprising 507 pairs of caregivers/elderly patients was carried out in three geriatric outpatient units. A multidimensional questionnaire was completed for each pair of interviewees covering DVAE, socio-demographic characteristics, and several theory related variables (alcohol misuse, burden, depression, social support, cognitive status, IADL, general health status and living arrangements). Sequential Confirmatory and Exploratory Factor Analyses (CFA/EFA) were implemented to assess construct validity through an in depth exploration of the instrument's dimensional structure. Construct validity was also assessed through relationships with other variables. RESULTS: The initial CFA did not support the postulated two factor solution. A one dimensional solution suggested by an EFA was followed by a CFA, which showed adequate fit (WRMR=0.985, RMSEA=0.056, CFI=0.967 and TLI=0.969), internal consistency ( rho(cr)=0.85) and temporal stability ( kappa(w)=0.77). Yet, convergent validity was not achieved (average variance extracted < 0.50; rho(ve)=0.45). Using Kendall's Tau b correlations, the CASE was positively associated with two other instruments assessing DVAE (H/S-EAST: tau(b)=0.13 and CTS 1 subscales: tau(b)=0.15-0.37); caregiver burden (Zarit BI: tau(b)=0.40); and depression (GDS: tau(b)=0.32). CONCLUSION: In spite of the some remaining dimensionality issues needing refinement and the relatively restricted correlations with expected variables, the CASE may be re affirmed as a promising detection tool for risk of abuse in clinical practice and applied research.

Language: en