A deeper dive into biostatistics: fixed-effect and random-effects models in meta-analysis

Ultimately, which model to use depends on the circumstances. Generally, the random-effects model is often the more appropriate model, as it does capture uncertainty resulting from heterogeneity among studies.

When there are too few studies to obtain an accurate estimate of the between-studies variance, the fixed-effect model remains very appropriate. For instance, in the scenario of a researcher being limited to a high-quality study with a large sample size and a low-quality study with a small sample size, a fixed-effect model will assign a greater weight to the larger, better-quality study.

Ultimately, conclusions that can be derived from meta-analyses are very much reliant on the source quality of the studies included. By itself, a meta-analysis can most certainly not compensate for poorly designed studies that meet inclusion criteriae.

Therefore, it is not an uncommon finding that authors describe the level of evidence as ‘weak’ and call for more, better quality research to be able to better answer the question at hand. While not the conclusion one would hope for an honest status check of the quality of available scientific literature can present an important finding in itself.

When, however, existing studies show prove themselves to be well-designed and credible, applying the preferred statistical model for the analysis of collated results is necessary in order to reach the methodologically correct conclusion. I hope that this deeper foray into the wonderful and growing world of metanalyses was helpful to the reader.

References and further reading:

1. Joseph R. Dettori, Daniel C. Norvell, Jens R. Chapman. Fixed-Effect vs Random-Effects Models for Meta-Analysis: 3 Points to Consider. Global Spine J. 2022: 12(7) 1624-1626.
2. Borenstein M, Hedges LV, Higgins JP, Rothstein HR. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res Synth Methods. 2010;1(2):97-111.
3. Nunna RS, Ostrov PB, Ansari D, et al. The Risk of Nonunion in Smokers Revisited: A Systematic Review and Meta-Analysis. Global Spine J. 2022;12(3):526-539.
4. Dettori JR, Norvell DC, Chapman JR. Seeing the Forest by Looking at the Trees: How to Interpret a Meta-Analysis Forest Plot. Global Spine J. 2021;11(4):614-616.
5. OrthoEvidence. Fixed- vs. Random-Effects Models: 5 Tips to Get a Better Understanding. OE Original. 2019;2(112). Available from: https://myorthoevidence.com/Blog/Show/48