Uncertainty in heterogeneity estimates in meta-analyses

By John P. A. Ioannidis, N. Patsopoulos, Evangelos Evangelou. British medical journal, 2007

1,194 citations recorded by the source corpus.

Subfields: Cardiology and Cardiovascular Medicine, Statistics and Probability, Surgery

View Paper: https://www.semanticscholar.org/paper/e35c73c549bc754e73b1b7f211ba16625124492e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cardiology and Cardiovascular Medicine

1. Paul T. von Hippel
2. Pim Cuijpers
3. W. Viechtbauer
4. John P. A. Ioannidis
5. J. Bowden
6. Georgia Salanti

In Statistics and Probability

1. Georgia Salanti
2. Julian P. T. Higgins
3. John P. A. Ioannidis
4. Dan Jackson
5. Paul T. von Hippel
6. Ian R. White
7. Thomas A. Trikalinos
8. W. Viechtbauer

In Surgery

1. Paul T. von Hippel
2. John P. A. Ioannidis
3. Georgia Salanti
4. Julian P. T. Higgins
5. Dan Jackson
6. Thomas A. Trikalinos
7. W. Viechtbauer
8. Joseph Lau

Citations in the corpus

• The Hartung-Knapp-Sidik-Jonkman method for random effects meta-analysis is straightforward and considerably outperforms the standard DerSimonian-Laird method (2014, background): “Some detectable heterogeneity is present in about half of meta-analyses of clinical studies [15-18].”
• The Benefit of Web- and Computer-Based Interventions for Stress: A Systematic Review and Meta-Analysis (2017, positive): “We calculated 95% confidence intervals around I 2 [47] using the noncentral approach based on chi-square within the heterogi module for Stata [48].”
• Large-Scale Replication Projects in Contemporary Psychological Research (2017, background): “For this and related reasons, the Type I error of the significance test of zero heterogeneity is often inflated above the nominal size α and power is low (Huedo-Medina et al., 2006; Ioannidis et al., 2007).”
• Estimates of heterogeneity (I2) can be biased in small meta-analyses (2014, background): “In meta-analyses from the Cochrane Library, for example, half of the values are less than 21% (1).”
• Conducting quantitative synthesis when comparing medical interventions: AHRQ and the Effective Health Care Program. (2011, background): “In addition, the 95% CI for I statistic should also be provided, whenever possible, to reflect the uncertainty in the estimate [48].”
• Replication in genome-wide association studies (2009, neutral): “When the Q test is underpowered, the I 2 metric has large uncertainty and this can be readily visualized by computing its 95% confidence intervals [30].”
• Internet and Computer-Based Cognitive Behavioral Therapy for Anxiety and Depression in Youth: A Meta-Analysis of Randomized Controlled Outcome Trials (2015, methodology): “We calculated 95% confidence intervals around I2 [39] using the non-central chi-squared-based approach within the heterogi module for Stata [41].”
• Why Most Discovered True Associations Are Inflated (2008, background): “problem, an evaluation of the whole Cochrane Library shows 1011 systematic reviews that have at least one meta-analysis with at least 4 studies.(95) Selecting the largest meta-analysis in each of these reviews, 256 of the 1011 meta-analyses have”