The Limitations of Using the Pittsburgh Sleep Quality Index to Assess Athletes' Sleep Quality: Evidence from...

This study aimed to assess the structural validity of the Chinese version of the Pittsburgh Sleep Quality Index (PSQI) among Chinese professional athletes and examine its test-retest reliability and convergent validity across different timeframes.

581 Chinese professional athletes participated. Exploratory and confirmatory factor analyses were conducted on the Chinese version of the PSQI. Test-retest reliability was assessed over 2 weeks, 1 week, and 2-3 days within a 1-month timeframe. Additional reliability analysis over a 2-day interval was conducted within a 1-week timeframe. Convergent validity was assessed using Chinese versions of the Insomnia Severity Index (ISI), the Athlete Sleep Screening Questionnaire (ASSQ), and actigraphy. A 1-month tracking was conducted, with weekly completion of the PSQI using a one-week timeframe, supplemented by assessments in the second and fourth week using two-week and one-month timeframes. Relationships between weekly results and those over two weeks and one month examined, along with convergent validity, using sleep diary and actigraphy.

The PSQI exhibited a two-factor structure (sleep quality and sleep efficiency), with good model fit (CFI = 0.960, AGFI = 0.924, TLI = 0.925, RMSEA = 0.085). Test-retest reliability was satisfactory for intervals of one week or more (r = 0.721 ~ 0.753). Using a one-week timeframe, the total score and two dimensions exhibited good reliability (r = 0.769 ~ 0.881), but only the total score and sleep quality showed high correlations with ISI and ASSQ (r = 0.701 ~ 0.839). Throughout the tracking, monthly responses correlated well with the most recent weeks (r = 0.732 ~ 0.866).

The PSQI demonstrates a two-factor structure in Chinese athletes, with sleep quality being predominant. Test-retest reliability within a one-month timeframe is unstable, suggesting a one-week timeframe performs better. Distinguishing between the two dimensions, employing shorter timeframes, and incorporating objective measures are recommended.