Related Experiment Videos
How much of the placebo 'effect' is really statistical regression?
Statistics in Medicine
|October 1, 1983
Summary
Statistical regression to the mean explains patient improvement, not always the placebo effect. Modern trials show minimal placebo effect when regression is controlled, highlighting the need for cautious interpretation of clinical outcomes.
Area of Science:
- Biostatistics
- Clinical Trial Design
- Evidence-Based Medicine
Background:
- Statistical regression to the mean suggests patients with abnormal test results tend to improve over time.
- Improvements in patients receiving placebos are often attributed to the placebo effect, but may be due to regression to the mean.
- Older clinical trial designs were more susceptible to regression to the mean, potentially inflating placebo effects.
Purpose of the Study:
- To investigate the extent to which statistical regression to the mean influences patient improvement in clinical trials.
- To differentiate between true placebo effects and improvements attributable to regression to the mean.
- To re-evaluate the interpretation of patient improvements in clinical practice and research.
Main Methods:
- Analysis of historical and modern clinical trial data, comparing placebo-treated groups.
- Examination of changes in biochemical test results in patients selected for high abnormality.
- Theoretical and empirical estimation of improvement due to regression to the mean for various biochemical markers.
Main Results:
- Modern clinical trials with designs mitigating regression to the mean showed no significant improvement in placebo-treated patients (median change 0.3%).
- Regression to the mean can account for substantial improvements in biochemical tests, with theoretical estimates ranging up to 26% and empirical estimates up to 37.3%.
- The median estimated improvement from regression to the mean for biochemical tests was approximately 10%.
Conclusions:
- Most patient improvements attributed to the placebo effect are likely instances of statistical regression to the mean.
- Clinical trial designs must adequately control for regression to the mean to accurately assess treatment effects.
- Clinicians and researchers should exercise caution when interpreting patient improvements, avoiding assumptions about inherent healing powers or strong placebo effects.