Back to dashboard
Randy Neil Biostats 27: USMLE Biostats - Effect Size, Drug Ad Information
Episode Notes
USMLE purpose: Interpret effect size and drug-ad statistics without getting distracted by abstract formatting.
How to use this page
- Identify the outcome and comparator.
- Check the point estimate and confidence interval.
- Ask whether the CI crosses the no-effect value.
- Translate effect size into clinical meaning.
Episode metadata
| Field | Details |
| Episode | Randy Neil Biostats 27 |
| Topic | Effect size and drug ad information |
| Runtime | 9 min |
| Published | 2024-02-04 |
| Source | Open YouTube video |
One-liner
Drug-ad questions usually test whether the result is statistically significant, clinically meaningful, and supported by the actual data.
Drug-ad interpretation table
| Data shown | No-effect value | Significance rule |
| Relative risk / odds ratio / hazard ratio | 1 | CI crossing 1 = not significant |
| Mean difference | 0 | CI crossing 0 = not significant |
| P value | 0.05 | p < 0.05 usually significant |
| NNT | Lower is usually stronger | NNT = 1 / ARR |
Board-style decision workflow
- Identify the statistic: RR/OR/HR, mean difference, p value, ARR, or NNT.
- Check the no-effect value: 1 for ratios, 0 for differences.
- Check statistical significance: CI must not cross the no-effect value.
- Check clinical meaning: statistically significant does not always mean clinically meaningful.
Anki-style rapid recall
For RR/OR/HR, what CI value means “no difference”?
For a mean difference, what CI value means “no difference”?
Zero.
What is NNT?
Number needed to treat = 1 / absolute risk reduction.
Practice Question
A drug has RR 0.82 with 95% CI 0.60–1.10. What is the best interpretation?
- A) Significant risk reduction
- B) Significant risk increase
- C) Not statistically significant
- D) NNT is 0.82
Reveal answer & explanation
Answer: C) Not statistically significant
The confidence interval crosses 1, so the result could represent no difference.
Extra practice
A trial reports a mean difference of -2.5 with 95% CI -5.0 to 0.8. Is the result statistically significant?
Reveal answer & explanation
No. For mean differences, the no-effect value is 0. Because the CI crosses 0, the result is not statistically significant.
Quick Reference Summary
| Statistic | No-effect value | If CI crosses it |
| RR / OR / HR | 1 | Not statistically significant |
| Mean difference | 0 | Not statistically significant |
| NNT | Calculated from ARR | Requires meaningful absolute risk difference |