Analytic methods for individually randomized group treatment trials and group-randomized trials when subjects belong...

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Public Health Research

Background:

• Group-randomized trials (GRTs) and individually randomized group treatment trials (IRGTs) involve participants receiving treatment in groups.
• Positive correlations in outcome measurements are expected when participants receive group treatment.
• A common challenge arises when participants receive treatment through multiple groups, complicating analysis.

Purpose of the Study:

• To evaluate analytical methods for group-randomized and individually randomized group treatment trials where participants receive treatment from multiple groups.
• To identify the most reliable statistical approach for handling complex group structures in intervention research.

Main Methods:

• Simulation studies were conducted to compare various analytical strategies.
• A mixed-effects model incorporating random effects for both individual and group levels was investigated.
• The performance of analytical approaches was assessed based on Type I error rates and statistical power.

Main Results:

• A mixed model with random effects for both groups consistently protected against inflated Type I error rates.
• This approach demonstrated a moderate loss of power only when intraclass correlations were substantial.
• Constraining variance estimates to be positive and using the Kenward-Roger adjustment for degrees of freedom enhanced power while maintaining nominal Type I error rates.

Conclusions:

• Mixed models with appropriate random effects are recommended for analyzing data from group-randomized and individually randomized group treatment trials with multiple treatment groups.
• The proposed method, including positive variance estimates and the Kenward-Roger adjustment, offers a robust solution for maintaining statistical integrity in complex trial designs.