Abstract
Depression is common in patients following stroke and has been found to be related to the degree of functional disability, recovery and engagement in rehabilitation. Consequently, screening for depression is crucial in this group in order to facilitate the delivery of appropriate psychological support. The current study sought to determine key psychometric properties of the Chinese version of the Geriatric Depression Scale (GDS) in this group. Three versions of the GDS were evaluated, these being the 30-item original measure (GDS-30), the short-form version comprising 15 items (GDS-SF) and a recently developed innovative four-item version (GDS-4). Confirmatory factor analysis revealed the GDS-30 and GDS-SF to offer an acceptable fit to data suggesting utility of these measures for screening. However, the GDS-4 offered a poor fit to the data, suggesting this measure was an inadequate measure of depression in this clinical group. Further, though GDS-30 and GDS-SF measures revealed good internal consistency, the performance of the GDS-4 was marginal. However, all GDS-derived measures demonstrated excellent convergent and divergent validity. It is concluded that the GDS-30 is a useful and appropriate screening instrument in this group. Further, the GDS-SF shows promise in terms of development as a multidimensional measure of depression that may have predictive and outcome monitoring potential. The psychometric shortcomings of the GDS-4 strongly suggest that this measure is unsuitable for screening in this clinical group.
Keywords:
Acknowledgements
This investigation was supported by the Health Care and Promotion Fund, Hong Kong. We would like to thank two anonymous reviewers for their very helpful and constructive comments regarding an earlier version of this manuscript.
Notes
1 Since both the GDS-SF and the GDS-4 contain items from the GDS-30, we administered the full GDS-30 and then derived the GDS-SF and GDS-4 items and scales for analysis.
2 The GDS-4 was observed to have moderate positive skew (1.03). To safeguard against an impact of possible violation of multivariate normality regarding this measure, the correlational analysis was also run using the non-parametric Spearman's rho test. This non-parametric analysis revealed findings entirely consistent with those of the parametric analysis.
3 Post-hoc GDS-4 uni-dimensional model fit indices: CFI = .97, TLI = .95, RMSEA = .12.
4 Only the GDS-30 and GDS-SF are currently available in a standard format.