The Hypersomnia Severity Index: reliability, construct, and criterion validity in a clinical sample of patients with sleep disorders

Study Objectives: The Hypersomnia Severity Index (HSI) was designed to assess the severity and impairment of hypersomnolence and has been validated in persons with psychiatric disorders. Little is known about its psychometric properties in clinical samples of patients with sleep disorders. Methods: One hundred fifty-eight patients (aged 44.1 ± 16.4 years, 29.1% male, 19.6% racial/ethnic minority) evaluated at the Behavioral Sleep Medicine program of the Penn State Health Sleep Research and Treatment Center completed the HSI and other patient-reported outcomes. We examined the HSI’s reliability and factorial, construct, and criterion validity. Results: The HSI showed satisfactory internal consistency (α = 0.79). A 2-factor structure, reflecting symptoms (HSI-S) and impairment, explained 56.2% of the variance. Convergent validity with the Epworth Sleepiness Scale was optimal (r = .65) but greater for HSI-S (r = .69) than for impairment (r = .39). Divergent validity was optimal for HSI-S against unrelated measures of sleep effort, reactivity, and incompatible behaviors (r ≤ .02). Construct validity showed higher scores in patients with central disorders of hypersomnolence and lower scores in patients with chronic insomnia disorder compared to those with other sleep disorders; however, these divergent scores were primarily driven by HSI-S rather than impairment. Criterion validity showed that an HSI-S cutoff score ≥ 8 provided the best balance in sensitivity/ specificity (0.82/0.78) to identify central disorders of hypersomnolence (area under the curve, 0.85). Conclusions: The HSI shows satisfactory indices of reliability and validity in a clinical patient sample. Its construct and criterion validity are supported by its divergent association with other patient-reported outcomes and central disorders of hypersomnolence vs chronic insomnia disorder diagnoses and the adequate sensitivity/ specificity of its HSI-S cutoff score to reliably identify central disorders of hypersomnolence.