TrialCurveHazard ratio interpretation, with examples
A hazard ratio (HR) compares the instantaneous event rate between two groups. HR = 1 means no difference. HR < 1 means fewer events in the treatment group: HR 0.75 is a 25% lower hazard at any given time. HR > 1 means more events: HR 1.30 is a 30% higher hazard. The 95% confidence interval shows the precision. If it excludes 1, the difference is statistically significant at the 5% level.
Reading the number
| Hazard ratio | Meaning (treatment vs control) |
|---|---|
| 0.50 | Hazard halved: 50% lower instantaneous risk of the event |
| 0.75 | 25% lower hazard |
| 1.00 | No difference |
| 1.30 | 30% higher hazard |
| 2.00 | Hazard doubled |
A hazard ratio can't be negative. It is a ratio of two rates, so it ranges from 0 to infinity. Its logarithm, used in meta-analysis, can be negative.
The confidence interval
- HR 0.75 (95% CI 0.62–0.91): the data are consistent with a 9–38% lower hazard. The CI excludes 1, so the result is significant.
- HR 0.90 (95% CI 0.72–1.12): the CI includes 1, so the data are consistent with no difference, a modest benefit or a small harm.
- A wide CI means few events. Look at the numbers at risk and events per arm.
What a hazard ratio is not
- Not a risk ratio at a fixed time. HR 0.5 does not mean half as many patients died by 5 years.
- Not a ratio of medians, except approximately under exponential survival.
- Not meaningful as one number when hazards are non-proportional, for example when curves cross, as often happens with immunotherapy. Report RMST differences (Royston & Parmar 2013) or landmark survival alongside.
Getting a hazard ratio when the paper doesn't report one
You can estimate an HR from a published Kaplan–Meier figure by reconstructing the patient-level data and fitting a Cox model. See the hazard-ratio tool. For your own data, the Kaplan–Meier generator reports the HR, 95% CI and log-rank p.
FAQ
What does a hazard ratio of 0.75 mean?
The treatment group has a 25% lower hazard (instantaneous event rate) than the comparator at any given time, assuming proportional hazards.
What does a hazard ratio greater than 1 mean?
The event rate is higher in the treatment (or exposed) group than in the reference group; for example HR 1.3 is a 30% higher hazard.
Can a hazard ratio be negative?
No. It is a ratio of rates and lies between 0 and infinity; only its logarithm can be negative.
Is a hazard ratio the same as relative risk?
No. Relative risk compares cumulative risk at a fixed time; the hazard ratio compares instantaneous rates over the whole follow-up.
Reconstruct IPD from your own figure. Upload a Kaplan–Meier plot and get curves, numbers at risk, pseudo-IPD, hazard ratios and a validation pack. Three figures a month are free.
Try the live demo Browse trial data API docs