QUICK-REFERENCE BOX — Relative vs Absolute
- Relative measures (RR, OR, RRR) describe the strength of an association — and can look impressive even when the real benefit is tiny.
- Absolute measures (ARR, NNT) describe the clinical impact in a real population.
- Always ask for the absolute numbers before being wowed by a relative reduction.
- NNT = 1 / ARR — how many patients you treat to prevent one bad outcome (round up).
- Case-control studies give odds ratios, not relative risks.
1. Overview
The Core Idea
Measures of effect quantify how much an exposure or treatment changes the risk of an outcome. Relative measures express the size of the association; absolute measures express what that means for real patients. Reporting only relative measures is the classic way to make a small effect sound large.
2. Relative Measures
RR, OR, RRR
- Relative risk (RR) = risk in exposed / risk in unexposed. RR = 1 → no association; >1 → harm; <1 → protection. Used in cohort studies and RCTs.
- Odds ratio (OR) = odds of exposure in cases / odds in controls. Used in case-control studies; approximates RR when the outcome is rare.
- Relative risk reduction (RRR) = ARR / control event rate — the proportional reduction; can look large even when ARR is small.
3. Absolute Measures
ARR
- Absolute risk reduction (ARR) = control event rate − treatment event rate — the actual reduction in risk.
- ARR reflects the real-world benefit and depends on baseline risk (higher-risk patients gain more).
4. Number Needed to Treat / Harm
NNT & NNH
- NNT = 1 / ARR — patients treated to prevent one additional bad outcome (always round up).
- NNH = 1 / absolute risk increase — patients exposed to cause one additional harm.
- A smaller NNT means a more clinically valuable treatment.
5. Worked Example
| Measure | Calculation | Result |
|---|
| Control event rate | — | 10% |
| Treatment event rate | — | 5% |
| Relative risk (RR) | 5% / 10% | 0.5 |
| Relative risk reduction (RRR) | (10% − 5%) / 10% | 50% |
| Absolute risk reduction (ARR) | 10% − 5% | 5% |
| Number needed to treat (NNT) | 1 / 0.05 | 20 |
The Lesson
A "50% relative risk reduction" sounds dramatic, but the absolute reduction is only 5% and you must treat 20 patients to prevent one event. Same data, very different impression.
6. Interpreting Them
At the Bedside
- Always look for absolute numbers alongside relative ones.
- Remember NNT depends on baseline risk — the same drug helps high-risk patients more.
- Check the confidence interval around the estimate (see 15.4) — it conveys precision.
7. Key Pearls
High-Value Points
- RR/OR describe strength; ARR/NNT describe clinical impact.
- NNT = 1/ARR (round up); smaller NNT = more useful treatment.
- OR ≈ RR only when the outcome is rare.
- A big RRR with a small ARR is the classic way to oversell a therapy.
- NNT falls (benefit rises) as baseline risk increases.
8. Common Mistakes to Avoid
Misreadings & Better Practice
Frequent errors with effect measures.
| Mistake | Why it's wrong | Better reading |
|---|
| Quoting RRR alone. | Hides small absolute benefit. | Report ARR and NNT too. |
| Treating OR as RR in a common outcome. | OR overstates RR when outcome is frequent. | Use RR from cohort/RCT data. |
| Ignoring baseline risk for NNT. | NNT is not universal. | Interpret NNT for the relevant risk group. |
9. Board-Style High-Yield Summary
Key Takeaways
- RR = risk ratio (cohort/RCT); OR = odds ratio (case-control), ≈ RR when outcome rare.
- ARR = control − treatment event rate; RRR = ARR / control rate.
- NNT = 1/ARR (round up); NNH = 1/ARI.
- Relative measures can exaggerate; always weigh absolute measures.
- NNT depends on baseline risk.
10. References
- 1.Barratt A, et al. Tips for learners of evidence-based medicine: relative risk reduction, absolute risk reduction and number needed to treat. CMAJ. 2004.
- 2.Sedgwick P. Relative risks versus odds ratios. BMJ.
- 3.Guyatt G, et al. Users' Guides to the Medical Literature.
Back to Clinical Guideline Hubs