A hazard ratio (HR) compares the hazard of the event between two groups at any moment in time.
Why it matters
It is the main effect measure of Cox regression, so you must read it correctly.
Key ideas
- HR = 1: no difference
- HR > 1: higher hazard (events happen sooner)
- HR < 1: lower hazard (protective)
- Always give the 95% CI
- HR is a ratio of rates, not of probabilities
- Assumes the ratio is constant over time
Real-life research example
Treatment vs control: HR = 0.60 (95% CI 0.42 to 0.86). At any time, treated patients have about 40% lower hazard of dying.
When to use
- Report with every Cox model
When NOT to use
- Don't say "60% longer survival"
- Don't ignore the CI
Mini example
HR = 0.60 → hazard is 40% lower; HR = 1.8 → hazard is 80% higher
Quick takeaway
HR < 1 protects; HR > 1 harms; 1 means no difference.
Try it yourself
If a Cox model gives HR = 1.5 (95% CI 0.9 to 2.4) for smoking, write one sentence interpreting it.
Model answer
Smokers had 1.5 times the hazard, but the CI includes 1, so the data are compatible with no difference and with up to a 2.4-fold higher hazard.