Factor structure of the Chinese version of the Pittsburgh Sleep Quality Index in breast cancer patients

Abstract

Objective The Pittsburgh Sleep Quality Index (PSQI) is used extensively to assess subjective sleep disturbance in cancer populations. Although previous studies on the PSQI suggested a better fit for a two- or three-factor model than the original one-factor model, none accounted for the indicator-specific effect between sleep duration and habitual sleep efficiency. This study evaluated the PSQI’s dimensionality and its convergent validity with cancer-related psychopathological states in female breast cancer patients. Methods The PSQI was administered to 197 women with breast cancer. Confirmatory factor analysis examined the relative fit of one-, two-, three-, and revised one-factor models. The PSQI’s convergent validity was evaluated via bivariate correlations between the PSQI factor scores and measures of anxiety, depression, fatigue, pain, and quality of life. Results Confirmatory factor analyses showed an adequate fit for the revised one-factor model with the PSQI global score as the overall index of sleep disturbance. Although the revised one- and two-factor solutions showed statistically equivalent model fits, the one-factor model was selected due to utility reasons. The severity of sleep dysfunction that the PSQI global score represented was positively correlated with anxiety, depression, fatigue, pain, and reduced quality of life. Conclusion The results support the PSQI’s original unidimensional structure, demonstrating that the PSQI global score is a valid and parsimonious measure for assessing and screening sleep dysfunction in cancer patients.