Seven-day wrist-accelerometer activity–rest patterns and subsequently recorded inflammatory skin diseases in UK Biobank: a prospective cohort study
Year: 2026
Type: preprint
Abstract: Background Wrist-accelerometer activity–rest measures are increasingly studied as disease risk markers, but their specificity beyond activity volume and their sensitivity to diagnosis source remain uncertain. We examined seven-day activity–rest patterns in relation to subsequently recorded inflammatory skin diseases and assessed their relevance to population health research. Methods We followed 94,406 UK Biobank participants after wrist accelerometry in 2013–2015. Eight activity features across hospital-recorded psoriasis, atopic dermatitis and urticaria formed a 24-test Benjamini–Hochberg false-discovery-rate family. Cause-specific Cox models used adjustment constrained by event counts. Additional analyses examined broader first-occurrence outcomes, temporal landmarks, missing data, competing mortality, measured accelerometer participation, total activity, 24-hour curve shape, absolute risk and incremental discrimination. Results The psoriasis complete-case analysis included 81,582 participants and 304 events. Relative amplitude was nominally inversely associated with hospital-recorded psoriasis per standard deviation (hazard ratio 0.899, 95% confidence interval 0.812–0.995; P = 0.039; q = 0.126). Estimates attenuated after participation weighting (0.941, 0.849–1.043), total-activity adjustment (0.933, 0.829–1.049) and broader ascertainment (0.930, 0.851–1.017). Conditional curve shape did not improve model fit (P = 0.849). Atopic dermatitis associations for least-active five-hour activity and relative amplitude passed false-discovery-rate correction (q = 0.005 and q = 0.011), but relied on 25 hospital events under minimal adjustment and were smaller with broader ascertainment. No hospital urticaria test survived correction. The five-year standardized psoriasis risk difference was − 0.314 per 1,000 (95% bootstrap interval − 0.568 to − 0.013); the optimism-corrected concordance increment after adding relative amplitude was 0.0028 (− 0.0058 to 0.0070). Conclusions Activity–rest associations varied with disease, outcome source, activity volume and measured participation. The findings do not establish a timing-specific prevention target or support clinical screening. Future wearable population studies should separate activity volume from curve shape, validate disease ascertainment and evaluate absolute and predictive value. Clinical trial registration Clinical trial number: not applicable.
Source: Research Square
Authors: Jinrong Zhang, Yim Lai, Yanhui Gao, Wai-kit Ming
Institutions: Jinan University, City University of Hong Kong
Language: English
Topic: Physical Activity and Health
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7225731886