Pioneering a multi-phase framework to harmonize self-reported sleep data across cohorts.

Area of Science:

• Sleep science
• Data science
• Epidemiology

Background:

• Harmonizing and aggregating data across studies is crucial for pooled analyses, enhancing external validation, replicability, and generalizability.
• The multidimensional nature of sleep presents significant challenges for data harmonization and aggregation.

Purpose of the Study:

• To describe and implement a process for harmonizing self-reported sleep data.
• To create a harmonized, pooled multi-cohort sample for sleep research.

Main Methods:

• A multi-phase framework was established: item compilation, domain grouping, item harmonization, and harmonizability evaluation.
• The process was applied to five US cohorts and a separate Rotterdam sample.
• 190 unique sleep items were collected and grouped into 15 domains, leading to 14 harmonized items.

Main Results:

• 14 harmonized sleep items were developed, measuring satisfaction, alertness, timing, efficiency, duration, insomnia, and sleep apnea.
• 13 of 14 items demonstrated moderate-to-high harmonizability, with continuous quantitative items showing higher rates.
• Descriptive statistics revealed consistent features (e.g., wake-up time, duration) and heterogeneous features (e.g., time in bed, bedtime) across samples.

Conclusions:

• The developed process provides guidance for researchers and cohort stewards in sleep data harmonization.
• Further methodological development and broader initiatives promoting common data elements are needed for enhanced future harmonization and aggregation efforts.