Heterogeneity

Heterogeneity refers to differences among studies pooled in a meta-analysis: their effect estimates vary more than random sampling error would predict. It can arise from clinical differences (different patient groups, doses, or settings) or methodological differences (different designs or quality). Statisticians quantify it with tests and with the I-squared statistic, which estimates the percentage of variation across studies due to real differences rather than chance; higher values indicate more inconsistency. When reading a meta-analysis, look at the reported heterogeneity and the forest plot: if study results point in different directions or scatter widely, a single pooled number can be misleading and may hide subgroups that respond differently. Ask whether the authors explored the sources, for instance through subgroup analysis, and whether they chose a random-effects model to account for the variation. For example, if trials in older and younger patients show opposite effects, reporting only the combined average would obscure a difference that matters for interpretation.