3rd Edition of Preventive Medicine World Conference 2027

Abstract

Background: Sleep duration has been associated with liver cancer, but such associations may reflect reverse causation from pre-existing hepatic, alcohol-related, or cardiometabolic disease. We examined self-reported and accelerometer-derived sleep duration in relation to incident liver cancer in the UK Biobank.

Methods: We analysed 497,917 participants with self-reported sleep duration and 99,136 with wrist accelerometry. Sleep was classified as short (<7 h/day), normal (7-8 h/day, reference), or long (>8 h/day). Incident liver cancer was ascertained by national cancer-registry linkage (ICD-10 C22); hepatocellular carcinoma (HCC) used C22.0. Cox models estimated hazard ratios (HRs), with prespecified 5-year lag, multiple-imputation, competing-risk, HCC-restricted, type 2 diabetes (T2D) interaction, E-value, and cumulative morbidity-exclusion analyses.

Results: Over a median 12.5 years, 594 liver-cancer events occurred in the self-reported cohort and 62 (over 9.1 years) in the accelerometry cohort. Self-reported short and long sleep were weakly associated with liver cancer (short HR 1.21, 95% confidence interval [CI] 1.00-1.45; long HR 1.26, 0.97-1.63), with modest E-values (1.71 and 1.83). T2D was strongly associated (HR 2.91, 2.42-3.51; E-value 5.27) but did not modify sleep associations (P for interaction = 0.481). Short-sleep estimates attenuated after 5-year lag (HR 1.06) and multiple imputation (HR 1.07), approaching the null after cumulative exclusion of baseline liver and alcohol-related morbidity (HR 1.03 and 1.01); accelerometer-derived estimates were directionally similar but imprecise.

Conclusions: Abnormal sleep duration was not robustly associated with incident liver cancer after accounting for reverse causation and baseline morbidity, and is better interpreted as a marker of hepatic, alcohol-related, and cardiometabolic vulnerability than an independent risk factor