Back
Science

MULTI Consortium Details Methods for Sleep Duration Study Using UK Biobank and Other Cohorts

Generated on: Last updated: 1 sources
View source

Sleep Duration’s Impact on Organ Aging: A Multi-Omics Investigation

The MULTI Consortium, a major initiative integrating multi-organ and multi-omics data, has published the methodology for a comprehensive study examining the associations between sleep duration, organ ageing clocks, proteins, and metabolites.

Study Overview

This large-scale investigation draws on data from six major biobanks and cohort studies:

  • UK Biobank (UKBB)
  • FinnGen
  • Psychiatric Genomics Consortium (PGC)
  • TriNetX
  • Baltimore Longitudinal Study of Ageing (BLSA)
  • Multi-Ethnic Study of Atherosclerosis (MESA)

Key Methodological Details

Sleep Duration Measurement

In UKBB, sleep duration (field ID: 1160) was self-reported via touchscreen questionnaire. Responses ranged from 1 to 23 hours, with values below 3 hours or above 12 hours triggering confirmation. Responses of "Do not know" (-1) and "Prefer not to answer" (-3) were excluded.

Organ Biological Age Gaps (BAGs)

The main analysis included 23 organ biological age gaps:

  • 7 MRI-based
  • 11 protein-based
  • 5 metabolite-based

Statistical Modeling

Generalised additive models (GAMs) were used to model nonlinear associations between sleep duration and BAGs, adjusting for demographics, physiology, and disease status.

The study employed a comprehensive analytic framework spanning proteomics, metabolomics, genetics, and survival analysis.

Proteome and Metabolome Analysis

  • Proteome-wide associations (ProWAS): 2,923 plasma proteins tested
  • Metabolome-wide associations (MetWAS): 327 plasma metabolites tested
  • Correction: Bonferroni adjustment applied

Genetic Analyses

Genome-wide association studies (GWAS) compared:

  • Short sleep (<6 hours) vs. normal sleep (6-8 hours)
  • Long sleep (>8 hours) vs. normal sleep

Analysis was performed using REGENIE software.

Genetic correlations were estimated using LD score regression, linking sleep patterns to 527 disease endpoints from FinnGen and PGC.

Survival and Mediation Analysis

Survival analyses using Cox proportional hazards models assessed risk of incident diseases and all-cause mortality in UKBB.

Mediation analysis via structural equation modelling tested whether MRI-BAGs mediate the relationship between sleep duration and late-life depression subtypes.

Mendelian randomisation investigated causal links between 525 disease endpoints and short/long sleep duration.

Background on Data Sources

  • MULTI Consortium: Approved by Columbia University IRB (AAAV6751); all individual studies obtained ethical approvals and informed consent.
  • UK Biobank: Population-based study with approximately 500,000 participants, recruited between 2006 and 2010.
  • FinnGen: Provided GWAS summary statistics (R9) for over 500,000 Finnish samples.
  • PGC: Provided GWAS summary statistics for psychiatric disorders.
  • TriNetX: Contributed deidentified electronic medical records from over 90 million patients.
  • BLSA and MESA: Provided brain MRI and sleep data for replication analyses.

This study represents a landmark effort to understand how sleep duration influences biological aging across multiple organ systems, using an unprecedented combination of data sources and analytical methods.