Psychometric evaluation of the insomnia severity index in 570,295 Chinese adolescents: A bifactor item-response theory analysis

Insomnia is a prevalent concern among adolescents, but accurately measuring its severity remains challenging. The Insomnia Severity Index (ISI) is widely used to assess insomnia symptoms, yet its psychometric properties have not been thoroughly evaluated in Chinese adolescents. This study addresses this gap by applying bifactor modeling and item-response theory (IRT) analysis to a large sample of 570,295 Chinese adolescents. Confirmatory factor analysis (CFA) identified a three-factor structure with correlated factors as the best-fitting model (CFI = 0.996, TLI = 0.991, RMSEA = 0.035, SRMR = 0.010). Further bifactor CFA revealed that a two-factor model (excluding item 4) provided a superior fit (CFI = 0.999, TLI = 0.997, RMSEA = 0.021, SRMR = 0.004), with the general insomnia severity factor explaining 69% of the common variance. The general factor captured variances related to nighttime sleep difficulties (e.g., trouble falling asleep or staying asleep) and daytime impairments (e.g., fatigue, irritability). IRT analysis demonstrated that the ISI exhibited high reliability and discrimination across moderate to high levels of insomnia severity, although reduced reliability was observed at the extreme ends of the scale. Gender differences showed that females had higher insomnia scores than males (Cohen's d = 0.12), while boarding students exhibited significantly higher insomnia severity compared to day students (Cohen's d = 0.30). These findings underscore the ISI's reliability and validity for measuring insomnia severity among Chinese adolescents, highlighting its utility as a valuable assessment tool for both research and clinical practice. The results provide important insights into adolescent sleep health and suggest potential applications for targeted interventions aimed at mitigating insomnia symptoms in this population.