Extreme between-study homogeneity in meta-analyses could offer useful insights

Meta-analyses are routinely evaluated for the presence of large between-study heterogeneity. We examined whether it is also important to probe whether there is extreme between-study homogeneity.

We used heterogeneity tests with left-sided statistical significance for inference and developed a Monte Carlo simulation test for testing extreme homogeneity in risk ratios across studies, using the empiric distribution of the summary risk ratio and heterogeneity statistic. A left-sided P=0.01 threshold was set for claiming extreme homogeneity to minimize type I error.

Among 11,803 meta-analyses with binary contrasts from the Cochrane Library, 143 (1.21%) had left-sided P-value <0.01 for the asymptotic Q statistic and 1,004 (8.50%) had left-sided P-value <0.10. The frequency of extreme between-study homogeneity did not depend on the number of studies in the meta-analyses. We identified examples where extreme between-study homogeneity (left-sided P-value <0.01) could result from various possibilities beyond chance. These included inappropriate statistical inference (asymptotic vs. Monte Carlo), use of a specific effect metric, correlated data or stratification using strong predictors of outcome, and biases and potential fraud.

Extreme between-study homogeneity may provide useful insights about a meta-analysis and its constituent studies.