Multiple-outcome meta-analysis of clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Rheumatology

Background:

• Traditional meta-analyses often analyze multiple outcomes from clinical trials separately.
• This can lead to less accurate estimates and missed opportunities for covariate adjustment.
• A joint modeling approach is needed to leverage all available information efficiently.

Purpose of the Study:

• To introduce and apply a novel generalized-least-squares (GLS) regression model for joint analysis of multiple outcomes in meta-analyses.
• To incorporate correlations among outcomes within treatment groups for more accurate effect estimates.
• To facilitate the adjustment for study-level and treatment-arm-level covariates.

Main Methods:

• A fixed-effects generalized-least-squares (GLS) regression model was developed for joint meta-analysis of multiple outcomes.
• The model accommodates studies reporting subsets of outcomes or treatments of interest.
• Covariates such as trial quality, duration, and baseline disease measures were included.

Main Results:

• The proposed GLS method was applied to 44 randomized controlled trials evaluating injectable gold versus auranofin for rheumatoid arthritis.
• The joint model produced moderate changes in coefficient values and slightly smaller standard errors compared to separate outcome analyses.
• Injectable gold demonstrated significantly greater effectiveness than auranofin across all three evaluated outcomes (tender joint count, grip strength, erythrocyte sedimentation rate).

Conclusions:

• The developed GLS regression model provides a robust framework for joint meta-analysis of multiple outcomes.
• This approach enhances estimation accuracy and allows for covariate adjustment, offering a more comprehensive analysis.
• The findings confirm the superior efficacy of injectable gold over auranofin in second-line rheumatoid arthritis treatment.