Vol. I · No. 33
Thursday, August 13, 2026
Issue: Summer · 2026
Established · MMXXVI
— The evidence base for longevity medicine —
Indexed by PubMed · CTG · Cochrane
Editorial team · geroevidence.com
Subscription · app.geroevidence.com

What is a hazard ratio? A physician's guide to reading longevity trial results

"HR 0.62" appears throughout Geroevidence's intervention index. Here is precisely what the number means, why the confidence interval around it matters as much as the point estimate, and where it's easy to misread.

By Geroevidence editorial team·Published July 26, 2026·7 min read
§ READING A FOREST-PLOT-STYLE HAZARD RATIO HR = 1.0 (no effect) EMPA-REG · CV death 0.62 SELECT · MACE 0.80 Hypothetical · null result 0.95 ← Lower risk in treatment arm HR = 1.0 Higher risk in treatment arm →
Fig. 1 — The horizontal line is the confidence interval; the dot is the point estimate. The third example's interval crosses 1.0 conceptually — a null result even with a point estimate below 1.0.
§ The core definition

A hazard ratio compares the instantaneous rate of an event — death, a cardiovascular event, cancer diagnosis — between two groups over the course of a trial. An HR of 0.62 means the treatment group experienced the event at 62% the rate of the control group at any given moment during follow-up; an HR of 1.0 means no difference; an HR above 1.0 means the treatment group had a higher event rate.

This is a relative measure, not an absolute one. An HR of 0.62 sounds dramatic — "38% risk reduction" — but the absolute reduction it represents depends entirely on the baseline event rate in that specific population. In a high-risk population with a 10% baseline event rate, HR 0.62 might mean roughly 4 fewer events per 100 people; in a low-risk population with a 1% baseline rate, the same HR might mean less than 1 fewer event per 100 people.

§ Why the confidence interval matters more than the point estimate

Every hazard ratio Geroevidence reports comes with a 95% confidence interval — for example, EMPA-REG's HR 0.62 (95% CI, roughly 0.49–0.77 for cardiovascular death). This range represents the plausible values consistent with the trial data, given sampling variability. If that interval crosses 1.0, the result is not statistically significant, regardless of how impressive the point estimate looks — a critically important distinction that headline HR numbers alone can obscure.

§ What a hazard ratio doesn't tell you

A hazard ratio assumes proportional hazards — that the relative risk reduction stays roughly constant over the follow-up period, an assumption that can break down with very long or heterogeneous follow-up. It also doesn't by itself convey absolute risk reduction, number needed to treat, or how the result generalizes to a population that looks different from the one enrolled — three separate questions worth asking of any HR before applying it clinically.

§ The clinical takeaway

A hazard ratio is a relative measure of risk over time, only meaningful alongside its confidence interval, its baseline event rate, and the population it was measured in. Geroevidence reports the HR and its source trial together for exactly this reason — the number in isolation is easy to over- or under-interpret.

This information is provided for educational reference only and does not constitute medical advice or a treatment recommendation.

See the full index
View intervention index
How to read trial data → RCT vs. observational study →