Quantifying the Impact of Co-Housing on Murine Aging Studies

Area of Science:

• Gerontology
• Biostatistics
• Animal Models

Background:

• Preclinical lifespan studies frequently assume independence of outcome data from co-housed animals.
• Treatments applied to entire housing units (e.g., cages) can induce correlations in animal outcomes.
• This violates the independence assumption common in statistical analyses.

Purpose of the Study:

• To investigate the impact of intra-class (intra-cage) correlation on preclinical lifespan studies.
• To evaluate the validity of the independence assumption in real-world lifespan study data.
• To explore analytical methods for handling correlated data and assess their limitations.

Main Methods:

• Analysis of three published and two unpublished mouse lifespan studies (>20,000 observations).
• Examination of intra-cage correlation, especially for frailty-associated traits.
• Simulations to assess statistical biases from intra-cage correlations.

Main Results:

• The independence assumption was found to be invalid in the analyzed lifespan studies.
• Intra-cage correlations were particularly pronounced for frailty-related traits.
• Standard variance components models (linear mixed models) have limitations in handling these correlations.

Conclusions:

• Intra-cage correlations are a significant factor in preclinical lifespan studies that must be addressed.
• Failure to account for these correlations can lead to statistical biases, affecting study power and reproducibility.
• Appropriate analytical tools are needed to accurately analyze data from co-housed animals.