The article by Gerta Rücker, Guido Schwarzer, James R Carpenter, and Martin Schumacher highlights the limitations of using I² as a measure of heterogeneity in meta-analysis. I² is interpreted as the percentage of variability in treatment estimates due to heterogeneity rather than sampling error. However, the authors argue that I² is not a suitable measure for assessing clinically relevant heterogeneity because it increases with the number of patients included in the studies. This is due to the fact that I² depends on the precision of the studies, which is proportional to their size. As study sizes increase, the confidence intervals become smaller and the heterogeneity, measured by I², increases. This can lead to misleading conclusions about the presence of clinically relevant heterogeneity. The authors conducted simulations under the random effects model to demonstrate this phenomenon. They found that as the precision of the studies increases, I² increases rapidly, approaching 100%. This was also observed in a sample of 157 meta-analyses. The authors also note that I² is not a suitable measure for comparing the statistical heterogeneity of meta-analyses with different numbers of studies. Instead, they argue that r², which describes the underlying between-study variability, is a more appropriate measure for this purpose. The article also discusses the interpretation of I² and its relationship with other measures of heterogeneity. It highlights the importance of considering the clinical relevance of heterogeneity when deciding whether to pool studies in a meta-analysis. The authors conclude that I² should not be used as the sole criterion for this decision, as it can be misleading when study sizes are large. Instead, the decision should be based on the clinical relevance of any heterogeneity present.The article by Gerta Rücker, Guido Schwarzer, James R Carpenter, and Martin Schumacher highlights the limitations of using I² as a measure of heterogeneity in meta-analysis. I² is interpreted as the percentage of variability in treatment estimates due to heterogeneity rather than sampling error. However, the authors argue that I² is not a suitable measure for assessing clinically relevant heterogeneity because it increases with the number of patients included in the studies. This is due to the fact that I² depends on the precision of the studies, which is proportional to their size. As study sizes increase, the confidence intervals become smaller and the heterogeneity, measured by I², increases. This can lead to misleading conclusions about the presence of clinically relevant heterogeneity. The authors conducted simulations under the random effects model to demonstrate this phenomenon. They found that as the precision of the studies increases, I² increases rapidly, approaching 100%. This was also observed in a sample of 157 meta-analyses. The authors also note that I² is not a suitable measure for comparing the statistical heterogeneity of meta-analyses with different numbers of studies. Instead, they argue that r², which describes the underlying between-study variability, is a more appropriate measure for this purpose. The article also discusses the interpretation of I² and its relationship with other measures of heterogeneity. It highlights the importance of considering the clinical relevance of heterogeneity when deciding whether to pool studies in a meta-analysis. The authors conclude that I² should not be used as the sole criterion for this decision, as it can be misleading when study sizes are large. Instead, the decision should be based on the clinical relevance of any heterogeneity present.