A genetics headline reporting that risk has doubled leaves a crucial question unanswered: doubled from what? Relative risk describes a comparison. Absolute risk describes the probability of an outcome over a specified period. Researchers need both to communicate the scale of a finding and its possible clinical relevance. The National Cancer Institute explains these measures in its overview of cancer risk measurement.
Put the baseline beside the multiplier
Consider a hypothetical study in which the five-year disease risk is 1% in a reference group and 2% in a comparison group. The risk ratio is 2: the comparison group has twice the risk, or a 100% relative increase. The absolute difference is one percentage point, equivalent to 10 additional cases per 1,000 people over five years.
Now imagine risks of 10% and 20%. The risk ratio is still 2, but the absolute difference is 10 percentage points, or 100 additional cases per 1,000. These are arithmetic illustrations, not findings about a particular gene. They show why an identical relative association can have very different practical implications.
Keep denominators and timeframes consistent. Comparing a five-year risk in one group with lifetime risk in another does not produce an interpretable risk ratio. NCI defines absolute risk through a specified population and observation period, and relative risk through the ratio of risks in two groups. NCI risk definitions.
Identify what the study actually estimated
Before translating a headline, inspect the outcome, reference group and statistical measure. Developing cancer, dying from cancer and developing a second cancer are different outcomes. A percentage is incomplete without that context.
Also check whether the reported estimate is a risk ratio or an odds ratio. Case-control studies select participants according to disease status and commonly estimate odds ratios. Odds compare an event with its absence; risks compare events with everyone at risk. The measures should not simply be relabelled. NCI explanation of study measures.
A polygenic percentile is not a disease probability
A polygenic score combines information across many genomic variants. Its percentile describes its position within a reference distribution. A score at the 90th percentile does not mean a 90% chance of developing disease.
The score alone does not supply a baseline disease probability or a time horizon. NHGRI illustrates this using people of very different ages with the same score: their disease probabilities can differ despite their shared genetic ranking. Its educational overview also distinguishes genetic association from causation. NHGRI explanation of polygenic scores.
Converting a score into an absolute prediction requires an appropriate risk model and evidence that its predictions work in the intended population. NCI identifies independent validation and assessment of model performance as central parts of absolute-risk modelling. This supplies the numerical foundation for the catalog article on polygenic scores, calibration and clinical context. NCI absolute-risk modelling overview.
Inherited cancer risk still needs a time horizon
High-impact pathogenic variants do not remove the need for careful framing. NCI reports that more than 60% of women with an inherited harmful BRCA1 or BRCA2 variant develop breast cancer during their lifetime, compared with about 13% of women in the general population. These are population estimates of lifetime risk, not predictions of when an individual will develop cancer. NCI BRCA fact sheet.
Lifetime risk cannot be read as risk during the next year. Nor should a broad population estimate be presented as a precise personal forecast. In research reports, preserve the estimated range and describe the population and follow-up supporting it.
Carry uncertainty into the clinical interpretation
A useful report should state the absolute risks alongside relative estimates, identify the comparison population, and retain uncertainty intervals where available. When translating a model, ask whether it has been independently validated and whether predicted probabilities agree with observed outcomes in the intended setting. NCI describes validation and monitoring of calibration as essential model-development activities. NCI model evaluation.
Finally, identifying increased risk does not establish that acting on a new genetic predictor improves health. Established care requires evidence appropriate to the intervention; research associations alone cannot supply it. Screening can produce false positives, overdiagnosis and procedural harms, so clinical usefulness depends on benefits and harms as well as prediction. NCI screening evidence and limitations.