Long Sleep Duration and Multimorbidity in Older U.S. Adults: A Dose–Response Analysis of NHANES 2017–March 2020

Background/Objectives: Sleep disturbance and chronic-disease multimorbidity are both common in aging populations, yet the dose–response shape of their association in U.S. older adults, and the sleep threshold of clinical relevance, remain unclear. Methods: We analyzed 2998 community-dwelling U.S. adults aged ≥60 years from the NHANES 2017–March 2020 pre-pandemic cycle. Self-reported sleep duration was categorized as short (<6 h), normal (6–8 h), or long (>8 h); multimorbidity was ≥2 of eleven physician-diagnosed chronic conditions. Survey-weighted logistic regression with jackknife variance estimated adjusted odds ratios (aORs), and restricted cubic splines characterized the dose–response shape. Results: Multimorbidity prevalence was 64.0%. The excess risk was confined to sleep of ≥9 h (aOR = 1.65, 95% CI 1.22–2.25); the band immediately above the reference, >8 to <9 h, showed no elevation (aOR = 1.08, 0.67–1.73), and the spline was J-shaped (P-nonlinearity = 0.005). Across the conventional >8 h category, long sleep was associated with multimorbidity (aOR = 1.41, 1.08–1.85, p = 0.014), whereas short sleep was not (aOR = 0.92, 0.58–1.47). This corresponded to an adjusted risk difference of 7.0 percentage points, and it was not materially attenuated by adjustment for a frailty phenotype (aOR = 1.38, 1.00–1.90) or among participants carrying at most one of the three available frailty criteria (aOR = 1.42, 1.03–1.97). The E-value, corrected for this common outcome, was 1.51. Conclusions: Long sleep, particularly ≥9 h, was associated with multimorbidity along a J-shaped dose–response. The association was moderate and the cross-sectional design cannot exclude reverse causation, so long sleep is best interpreted as a marker of accumulated disease burden rather than an actionable risk factor.