Study design features increase replicability in cross-sectional and longitudinal brain-wide association studies.

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Biostatistics

Background:

• Brain-wide association studies (BWAS) are vital for understanding brain-behavior links.
• Recent studies highlight the need for large sample sizes in BWAS due to small standardized effect sizes (ESs).
• Optimizing study design is proposed as a method to enhance BWAS replicability.

Purpose of the Study:

• To investigate how study design influences standardized effect sizes (ESs) and replicability in brain-wide association studies (BWAS).
• To provide guidance on optimizing study designs for more robust and reproducible BWAS findings.

Main Methods:

• Meta-analysis of 63 longitudinal and cross-sectional magnetic resonance imaging studies from the Lifespan Brain Chart Consortium (77,695 scans).
• Analysis of age effects on brain measures using UK Biobank and ADNI datasets.
• Evaluation of longitudinal sampling schemes on diverse brain-behavior associations in the ABCD dataset.

Main Results:

• BWAS with greater covariate variability and longitudinal designs show larger standardized ESs.
• Specific sampling schemes, like increasing between-subject age variability and adding longitudinal measurements, improve ESs and replicability.
• Common longitudinal models can paradoxically reduce ESs; explicit modeling of between- and within-subject effects is beneficial.

Conclusions:

• Study design critically impacts the magnitude of standardized effect sizes and replicability in BWAS.
• Longitudinal study designs, when optimized through appropriate sampling, can significantly enhance BWAS findings.
• Explicitly modeling between- and within-subject effects is recommended for maximizing BWAS robustness and generalizability.