Evidence Integration Lab Initiative

RiskTranslate

Turning Hazard Ratios into Clinical Meaning

Converts hazard ratio results from medical studies into baseline-adjusted absolute numbers that clinicians can explain more clearly and patients can understand more easily.

Important disclaimer

RiskTranslate is an educational evidence-translation tool. These outputs are approximation-based and should support, not replace, direct trial data, Kaplan-Meier estimates, clinical judgment, or individualized patient advice.

Why This Matters

Hazard ratios are statistically efficient, but clinically incomplete.

A hazard ratio can summarize the relative separation between event curves over time, but it does not directly tell a clinician how many patients are likely to benefit. Clinical decisions usually depend on absolute differences: how many fewer events occur, in whom, and over what time horizon.

The same hazard ratio can imply a modest or substantial absolute benefit depending on the baseline event rate. That is why ARR and NNT can change meaningfully across populations even when the relative effect remains stable.

Worked example

Baseline risk: 31.2% | HR: 0.56 | Treated risk: 19.4% | ARR: 11.8% | NNT: 8.5 over 14 years

Formula

From relative effect to absolute risk

Rt=1−(1−Rc)HRR_t = 1 - (1 - R_c)^{HR}

Rt: treated event rate at the chosen time horizon.

Rc: control or baseline event rate over the same horizon.

HR: hazard ratio, assumed to represent a proportional hazards effect.

This approximation assumes proportional hazards, a stable hazard structure over the follow-up interval, and an exponential relationship between cumulative risk and hazard.

Survival interpretationHigher survival
Control survivalTreated survival
Interactive Calculator

Estimate absolute treatment effects

Live updating calculations using an exponential approximation under proportional hazards assumptions.

Control risk at the chosen follow-up horizon.

Relative time-to-event effect estimate.

Time horizon for interpretation and NNT.

Advanced precision inputs

Add HR confidence limits to estimate a plausible range for treated risk and ARR. NNT interval estimates are shown cautiously because NNT behaves non-linearly near zero risk difference.

Optional lower confidence limit.

Optional upper confidence limit.

Export calculator report

Generate a PDF from the current calculator inputs and outputs only, including the icon array and report diagrams.

Treated Event Rate

18.9%

Estimated cumulative event rate with treatment over the specified follow-up horizon.

Absolute Risk Reduction

12.3%

The event reduction attributable to treatment after anchoring the relative effect to baseline risk.

Relative Risk Reduction

39.4%

Useful for comparison, but should not be interpreted without the baseline event rate.

Number Needed to Treat

8.1

Average number of patients needing treatment over the chosen time period to prevent one event.

Approximate confidence-informed range

Derived by propagating the reported HR interval through the same transformation.

Treated risk range

23.9% to 14.9%

ARR range

7.3% to 16.3%

NNT range

6.1 to 13.7

Patient Explanation

Plain-language interpretation

Out of 100 patients similar to you, approximately 31 may experience the event over 14 years without treatment. With treatment, approximately 19 may experience the event. This means treatment may prevent about 12 events per 100 patients over 14 years.

Approximately 8 patients would need treatment for 14 years to prevent one event.

Clinical Framing

Interpretive cues for discussion

Use the ARR when speaking about expected benefit in a specific population and time frame.

Use the HR when comparing trial-level relative effects, but avoid presenting it alone to patients because it is rarely intuitive.

Recalculate for different baseline risks when translating evidence across subgroups or practice settings.

Visualization

Absolute effect at the bedside

Publication-style comparison of event burden with and without treatment.

Baseline Sensitivity

ARR changes with baseline risk

The same hazard ratio can yield different absolute effects in lower-risk versus higher-risk populations.

100-Patient View

Icon-array style event burden

Each square represents one patient over 14.0 years.

Without treatment

With treatment

Assumptions & Limitations

Interpret with statistical discipline

Hazard ratio is not the same as a risk ratio and should not be read as a direct percentage risk change.

This transformation is an approximation, not an exact reconstruction of Kaplan-Meier event probabilities.

NNT is time-dependent; changing the follow-up horizon changes its meaning.

Baseline risk strongly influences ARR, even when the relative effect is unchanged.

Competing risks are ignored in this simplified transformation.

Kaplan-Meier estimates or directly modeled absolute risks are preferable when available.

The proportional hazards assumption may fail, especially when curves cross or diverge late.

Confidence intervals for NNT can become unstable when ARR approaches zero.

References

Evidence base

  1. Altman DG, Andersen PK. Calculating the number needed to treat for trials where the outcome is time to an event. BMJ. 1999;319:1492-1495.

    Open source

  2. Schunemann HJ, Vist GE, Higgins JPT, et al. GRADE guidelines 27: how to calculate absolute effects for time-to-event outcomes in Summary of Findings tables and Evidence Profiles. J Clin Epidemiol. 2020;118:124-131.

    Open source

  3. Spruance SL, Reid JE, Grace M, Samore M. Hazard ratio in clinical trials. Antimicrob Agents Chemother. 2004;48(8):2787-2792.

    Open source

  4. Guyatt GH, Oxman AD, Kunz R, et al. GRADE guidelines: going from evidence to recommendations. J Clin Epidemiol. 2011;64(4):395-400.

    Open source

  5. BMJ Evidence-Based Medicine resources on absolute and relative effects, including risk differences, baseline risk, and NNT interpretation.

    Open source
About

RiskTranslate

Developed by Dr. Manjunath, Consultant Nephrologist and Academic Researcher.

RiskTranslate is positioned as an Evidence Integration Lab Initiative focused on translating trial outputs into clinically communicable meaning without flattening the underlying statistical nuance.

Designed for journal club use, bedside explanation, academic discussion, and rapid evidence contextualization.

Disclaimer: RiskTranslate is an educational evidence-translation tool. These outputs are approximation-based and should support, not replace, direct trial data, Kaplan-Meier estimates, clinical judgment, or individualized patient advice.