Performance of a two-item sleep quality measure (PSQI-2): a comprehensive evaluation in a multiethnic cohort (MESA Study)

Objective To evaluated the abbreviated two-item Pittsburgh Sleep Quality Index (PSQI-2) against the full PSQI in a multi-ethnic cohort. Methods We analyzed data from 2,237 participants from the MESA Sleep Ancillary Study. The full PSQI was adapted by integrating actigraphy data for sleep duration, latency, and efficiency components, while maintaining the original seven-component structure scored 0–3. The PSQI-2 was derived from two components: sleep duration (questionnaire-based) and subjective sleep quality. Validation analyses included correlation analysis; ROC curves for three PSQI cutpoints (> 5, > 7, >10) with sensitivity/specificity calculations, Bland-Altman analysis for agreement, bootstrap internal validation, and logistic regression for demographic, clinical, and sleep-related covariates. Results Poor sleep quality was prevalent (65.6% by PSQI > 5; 65.7% by PSQI-2 > 1). The PSQI-2 showed strong correlation with the full PSQI (r = 0.520, p < 0.001), consistent across gender and age subgroups. Both measures identified similar risk patterns: Black and Hispanic participants had higher odds of poor sleep, and obesity, sleep disorders, daytime sleepiness, and evening chronotype consistently increased poor sleep odds. The PSQI-2 demonstrated good discriminant validity across PSQI cutpoints (AUC: 0.785 for > 5, 0.748 for > 7, 0.750 for > 10), with sensitivity ranging from 71.8–86.8% and specificity from 46.3–79.1%. Conclusion The PSQI-2 shows strong validity and consistent performance with the full PSQI, effectively identifying poor sleep quality and associated factors. Its brevity makes it suitable for large-scale studies and clinical screening.