Test-retest repeatability of the multiple sleep latency test in non-cataleptic hypersomnolence disorders

Differentiating narcolepsy type 2 (NT2) and idiopathic hypersomnia (IH) is challenging as their diagnosis relies on the multiple sleep latency test (MSLT), previously shown to have poor test-retest repeatability. To address this gap we assessed test-retest repeatability of the MSLT in NT2 and IH.

Retrospective study including patients evaluated for excessive sleepiness with more than one PSG-MSLT at Cleveland Clinic (2007-2024). Narcolepsy type 1 and subjects with <6 h of sleep on PSG, <7 h of sleep by sleep logs, <5 MSLT trials, use of sleep-modulating medications within 14 days, and untreated comorbid sleep disorders were excluded. Diagnoses were based on ICSD-3 TR criteria; those not meeting criteria were classified as indeterminate. Test-retest repeatability and correlations were analyzed.

45 patients (71.1% female; mean age 29.1 ± 16.0 years) were included, with inter-test interval of 3.6 ± 3.3 years. On initial MSLT, median mean sleep latency (MSL) was 11.5[5.1,14.2] minutes; 24.4% had ≥2 SOREMPs. On retest, median MSL was 10.6[3.6,14.6] minutes; 28.8% had ≥2 SOREMPs. Diagnoses were unchanged in 71.1% of cases; NT2 showed highest retest repeatability (NT2-75%, IH-60%, indeterminate-74.1%). Changes in diagnosis were attributed to MSL variation alone in 46.1%, SOREMPs alone 23.1% or both 30.8%. Spearman correlation and intraclass correlation coefficients showed moderate agreement for MSL [Spearman's ρ = 0.68 (95% CI:0.48-0.81, p < 0.001), ICC = 0.69(95% CI: 0.5-0.82)] and SOREMPs [Spearman's ρ = 0.54 (95% CI:0.30-0.72, p < 0.001), ICC = 0.52(95% CI: 0.27, 0.70)].

While higher MSLT test-retest repeatability was found compared to prior studies, substantial diagnostic instability remains, highlighting the need to refine NT2 and IH criteria.