Every oncology drug approval is built on a number: the hazard ratio. When the FDA reviews a new cancer therapy, the HR from the pivotal trial is the statistic that most directly answers the question, "Does this drug work?" Yet hazard ratios are routinely misinterpreted — by patients, journalists, and sometimes clinicians. This guide clarifies what the number actually tells you, what it does not, and how to use it when evaluating the therapies tracked on PipelineEvidence.
What a Hazard Ratio Is (and Is Not)
A hazard ratio compares the rate at which events (death, progression, recurrence) occur in one group relative to another. If a trial reports HR = 0.70 for overall survival, it means that at any point during the study, patients receiving the experimental treatment were experiencing death at 70% the rate of patients in the control arm — or equivalently, a 30% reduction in the rate of death.
The critical nuance: an HR is a rate ratio, not a probability ratio. An HR of 0.70 does not mean "30% of patients are cured" or "you have a 30% better chance of surviving." It means the instantaneous rate of the event is 30% lower in the experimental arm at any given moment, averaged over the study period.
This distinction matters because the actual survival benefit in absolute terms depends on the baseline risk. A 30% rate reduction in a disease where median survival is 6 months translates to a different absolute benefit than the same HR in a disease where median survival is 5 years.
A Practical Scale for Interpreting HRs in Oncology
Confidence Intervals: The Error Bars You Cannot Ignore
A hazard ratio is always reported with a 95% confidence interval (CI). The CI represents the range within which the true HR is likely to fall. If the entire CI is below 1.0, the result is statistically significant in favor of the experimental arm. If the CI crosses 1.0, the observed difference could be due to chance.
The width of the CI also conveys precision. A narrow CI like HR 0.65 (0.58–0.73) from a large trial gives you high confidence in the effect size. A wide CI like HR 0.65 (0.38–1.12) from a small study tells you the true effect could range from substantial benefit to no benefit at all.
When evaluating drug pages on PipelineEvidence, you will find links to pivotal trial publications for each approved therapy. Check the CI alongside the HR — the CI is often more informative than the p-value for understanding how certain we should be about the magnitude of benefit.
When the HR Misleads: Limitations to Know
The standard hazard ratio assumes that the treatment effect is constant over time — the proportional hazards assumption. In reality, many cancer treatments have effects that change. Immunotherapy is the classic example: checkpoint inhibitors often show minimal early benefit (the curves overlap or even cross initially) followed by substantial long-term separation. A single HR averages these two phases together, potentially underrepresenting the long-term benefit.
Another limitation: the HR tells you about relative rates, not absolute benefit. A drug with HR 0.60 might improve median PFS from 3 months to 5 months in one disease (2-month absolute gain) or from 12 months to 20 months in another (8-month absolute gain). Always look at the absolute differences in median or landmark survival times alongside the HR.
Finally, be cautious about comparing HRs across different trials. An HR of 0.70 in one trial is not equivalent to HR 0.70 in another — the patient populations, comparator arms, disease settings, and follow-up durations all differ. Cross-trial comparisons require formal indirect comparison methods, not HR-to-HR matching.
Applying This to Real Drug Approvals
Every drug in the PipelineEvidence Drug Library links to the clinical trials that supported its approval. When you visit a drug page, you can follow the ClinicalTrials.gov and PubMed links to find the full publications. Focus on the primary endpoint HR and CI, then look at the Kaplan-Meier curves to understand the pattern of separation.
This combination — HR for the magnitude of effect, CI for certainty, and KM curve for the pattern over time — gives you the complete picture of how well a cancer drug performs. For more on reading those curves, see our companion article: How to Read a Kaplan-Meier Survival Curve.