Related Experiment Videos
Beta, or type II error in psychiatric controlled clinical trials.
Journal of Psychiatric Research
|January 1, 1985
Summary
Many psychiatric clinical trials lack sufficient sample size, risking the detection of significant treatment improvements. This underpowered research may obscure important therapeutic differences, impacting patient care.
Area of Science:
- Psychiatry
- Clinical Trials
- Statistical Power
Background:
- Many psychiatric controlled clinical trials report non-significant findings.
- The adequacy of sample sizes in these trials is often questionable.
- Small sample sizes can lead to underpowered studies, failing to detect true treatment effects.
Purpose of the Study:
- To re-examine 52 "negative" psychiatric controlled clinical trials (1980-1984).
- To assess if sample sizes were sufficient to detect a 20% or 50% improvement in therapeutic response (reduction in non-response rate) with high probability (>0.90).
Main Methods:
- Re-analysis of 52 psychiatric controlled clinical trials published between 1980 and 1984.
- Calculation of the risk of missing true differences in non-response rates (20% and 50%).
- Estimation of 90% confidence intervals to assess potential true differences.
Main Results:
- 44 trials had >10% risk of missing a 20% difference in non-response rates.
- 25 trials had >10% risk of missing a 50% difference in non-response rates.
- 47 trials could have missed a potential 20% difference; 30 could have missed a 50% difference.
Conclusions:
- Small sample sizes in psychiatric trials pose a significant risk of overlooking substantial therapeutic improvements.
- Underpowered studies may obscure clinically important differences between treatments.
- Increased attention to statistical power and sample size is crucial in psychiatric research to ensure reliable findings.