Related Experiment Videos
Impact of multiple comparisons in randomized clinical trials.
D G Smith1, J Clemens, W Crede
1Department of Medicine, Temple University School of Medicine, Philadelphia, Pennsylvania 19140.
The American Journal of Medicine
|September 1, 1987
Summary
Most randomized clinical trials (RCTs) perform multiple statistical comparisons, often impairing their results. Further research is needed to address the issue of multiple comparisons in clinical trials.
Area of Science:
- Clinical Research
- Biostatistics
- Medical Statistics
Background:
- Randomized clinical trials (RCTs) are crucial for evaluating medical interventions.
- Confidence in RCT conclusions relies on adherence to methodological and statistical standards.
- The issue of multiple comparisons can compromise statistical validity in RCTs.
Purpose of the Study:
- To assess the attention paid to the problem of multiple comparisons in contemporary RCT analyses.
- To quantify the impact of multiple comparisons on statistical significance in published RCTs.
Main Methods:
- A survey of 67 contemporary randomized clinical trials was conducted.
- The study analyzed the statistical methods used in these trials, focusing on multiple comparisons.
- Criteria for statistical impairment due to multiple comparisons were applied.
Main Results:
- 99% of surveyed RCTs (66 out of 67) performed multiple comparisons, averaging 30 per trial.
- 75% of trials (50 out of 67) had statistical significance impaired by multiple comparisons.
- 22% of trials (15 out of 67) had all comparisons statistically impaired.
Conclusions:
- Contemporary RCTs frequently employ multiple comparisons, often without adequate statistical adjustment.
- The problem of multiple comparisons significantly impacts the validity of statistical findings in a majority of RCTs.
- There is a critical need for further research to clarify and mitigate the effects of multiple comparisons in clinical trial design and analysis.