Jufang Li1, Linda Denise Oakley2, Roger L Brown2, Yun Li3, Maiyun Ye4, Yong Luo5. 1. The First Affiliated Hospital of Chongqing Medical University, Chongqing, PR China; Wenzhou Medical University, School of Nursing, Wenzhou, PR China. 2. University of Wisconsin-Madison School of Nursing, Madison, WI, USA. 3. The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, PR China. 4. The Second Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, PR China. 5. The First Affiliated Hospital of Chongqing Medical University, Chongqing, PR China. Electronic address: luoyong1998@163.com.
Abstract
BACKGROUND: We reported the factor structure and psychometric properties of a new measure of early symptoms of Post-Stroke Depression (PSD). METHODS: Cross-sectional survey methods were used to administer the measure to hospitalized post-stroke patients (N=410) in southeast China, 7-30 days after mild to moderate stroke. Factor structure of the measure was evaluated using exploratory factor analysis (EFA) with first and second order confirmatory factor analysis (CFA). Measurement reliability of each factor and the total measure was assessed using Cronbach alpha coefficient, item-total correlation, item-subscale correlation and the composite coefficient. Discriminant validity was tested using the estimated correlation matrix and average variance extracted (AVE). RESULTS: The EFA extracted a theoretically consistent, clinically interpretable, 29-item, 6-factor model for early symptoms of PSD (dull, guilt, low, wakefulness, emotional, and nervous). A first order CFA retained the 6-factors but deleted 3 underperforming items. The results of a second-order CFA for a 6-factor, 26-item model showed acceptable model fit (χ(2)/df=2.25, CFI=0.973, TLI=0.970, RMSEA=0.055 and WRMR=1.168) with acceptable reliability and discriminant validity. LIMITATIONS: The study survey methods and purposive sampling procedures resulted in a clinically less homogenous final sample. Separate evaluation of predictive validity, criteria validity, test-retest reliability, and invariance to patient psychosocial characteristics of the measure is planned. CONCLUSIONS: Our measure detected early symptoms of PSD in primarily first-ever stroke patients, an average of 11.07 days post stroke, well within the 14 day peak benchmark.
BACKGROUND: We reported the factor structure and psychometric properties of a new measure of early symptoms of Post-Stroke Depression (PSD). METHODS: Cross-sectional survey methods were used to administer the measure to hospitalized post-strokepatients (N=410) in southeast China, 7-30 days after mild to moderate stroke. Factor structure of the measure was evaluated using exploratory factor analysis (EFA) with first and second order confirmatory factor analysis (CFA). Measurement reliability of each factor and the total measure was assessed using Cronbach alpha coefficient, item-total correlation, item-subscale correlation and the composite coefficient. Discriminant validity was tested using the estimated correlation matrix and average variance extracted (AVE). RESULTS: The EFA extracted a theoretically consistent, clinically interpretable, 29-item, 6-factor model for early symptoms of PSD (dull, guilt, low, wakefulness, emotional, and nervous). A first order CFA retained the 6-factors but deleted 3 underperforming items. The results of a second-order CFA for a 6-factor, 26-item model showed acceptable model fit (χ(2)/df=2.25, CFI=0.973, TLI=0.970, RMSEA=0.055 and WRMR=1.168) with acceptable reliability and discriminant validity. LIMITATIONS: The study survey methods and purposive sampling procedures resulted in a clinically less homogenous final sample. Separate evaluation of predictive validity, criteria validity, test-retest reliability, and invariance to patientpsychosocial characteristics of the measure is planned. CONCLUSIONS: Our measure detected early symptoms of PSD in primarily first-ever strokepatients, an average of 11.07 days post stroke, well within the 14 day peak benchmark.