Confidence intervals for the between-study variance in random-effects meta-analysis using generalised heterogeneity...

Area of Science:

• Statistics
• Biostatistics
• Medical Research Methodology

Background:

• Confidence intervals for between-study variance are crucial for random-effects meta-analyses.
• Existing methods, while statistically sound, often produce overly wide and uninformative intervals.
• Quantifying uncertainty in meta-analysis estimates is essential for reliable interpretation.

Purpose of the Study:

• To introduce a novel strategy for computing narrower confidence intervals for between-study variance.
• To maintain nominal coverage probability while reducing interval width.
• To enhance the informativeness of meta-analysis results.

Main Methods:

• Utilizing generalized heterogeneity statistics with unequal tail probabilities for interval calculation.
• Allocating a greater tail probability (e.g., >2.5%) to the upper bound of the confidence interval.
• Validating the approach through real-world examples and extensive simulation studies.

Main Results:

• The proposed method significantly reduces the width of confidence intervals for between-study variance.
• Shorter intervals improve the precision of estimates, aiding sensitivity analyses for average effects.
• Unequal tail probabilities maintain the desired coverage probability.

Conclusions:

• Using unequal tail probabilities offers a practical improvement for confidence intervals of between-study variance.
• The '1-4% split' is recommended for practical application, allocating more probability to the upper bound.
• The 'width-optimal' interval warrants further investigation for its utility.