Authors: De Cassai et al.
Anesthesiology, September 25, 2025. DOI: 10.1097/ALN.0000000000005776
This meta-research analysis evaluated nearly three decades of randomized controlled trials (RCTs) in anesthesiology and pain medicine to determine how often—and why—authors perform statistical testing of baseline characteristics between study groups. Despite clear methodological guidance discouraging this practice, the analysis of 2,453 trials revealed that almost half (48.3%) still reported such tests. Roughly one in five of these studies (19.2%) reported at least one statistically significant difference between groups, yet most of those findings were likely false positives given the expected rate of random variation. Interestingly, trials with larger author teams were less likely to perform baseline testing, while those testing more variables were predictably more likely to find “significant” differences.
The authors conclude that null-hypothesis significance testing of baseline characteristics continues to be misused and misinterpreted, suggesting persistent misunderstanding of randomization principles. They advocate for stronger editorial enforcement of CONSORT standards and better education among investigators to eliminate this outdated practice and enhance methodological rigor across anesthesiology research.
What You Should Know
-
Nearly half of anesthesia RCTs still perform unnecessary baseline significance testing.
-
The rate of significant baseline findings was actually lower than expected by chance, highlighting misunderstanding rather than bias.
-
Larger, more collaborative author groups tend to follow better statistical practices.
-
Education and editorial oversight remain essential to modernize reporting standards.
References
Altman DG, Bland JM. Statistics notes: Variables and their interactions. BMJ. 2003;326(7393):219.
Senn S. Baseline testing in randomized clinical trials: Significance tests, differences, and “chance.” Stat Med. 1994;13(17):1715–1726.
Thank you to Anesthesiology for publishing this important methodological study.