Absolute Risk, Relative Risk, and Baseline Risk: A Reader Worksheet for a Health Headline

By MedClinRes.org Editorial Team

A headline that says a treatment “cuts risk by 50%” is almost always reporting a relative risk, not an absolute one. Knowing what that claim actually establishes takes three numbers the headline usually leaves out: the baseline risk the comparison is built on, the absolute difference that baseline produces, and the time frame the risk was measured over. This article walks through how to find or estimate those numbers and gives a short worksheet for applying them to any headline.

Why the Same Study Can Produce Two Very Different-Sounding Numbers

Relative risk compares two rates to each other: one rate divided by the other. Absolute risk difference is the plain subtraction of one rate from the other. Both numbers can come from the exact same data, and both can be accurate, but they create very different impressions.

A methodology reference on field trials of health interventions lays out the arithmetic directly. If the disease rate in one group is r1 and the rate in a comparison group is r2, the relative rate is r1 divided by r2, and the absolute difference is r1 minus r2. The source works through a hypothetical case to illustrate this: a disease incidence of ten cases per thousand in an unvaccinated group, and a vaccine with 50% efficacy, meaning the vaccinated group's incidence is five cases per thousand. The relative figure — a 50% reduction — is accurate. So is the absolute figure — a reduction of five cases per thousand. They describe the same result, but “cuts your risk in half” and “lowers your risk by five cases per thousand” land very differently with a reader.

The Three Numbers a Relative-Risk Headline Usually Skips

Baseline risk

This is the rate of the outcome in the reference or control group before any comparison is applied. Without it, a relative-risk percentage has no anchor. A 50% reduction applied to a common outcome moves the absolute numbers a lot; the same 50% applied to a rare outcome moves them very little.

Absolute difference

This is baseline risk minus the risk in the group being compared (or the reverse, for an increase). It is usually the number closest to what an affected group would actually experience, expressed as cases per some number of people rather than a percentage of a percentage.

Time frame

Every risk figure is tied to a period of observation, and the period changes the number. A reference on U.S. cancer statistics reports its own mortality trends as annual rates over specific, named windows — for example, a decline of 1.7% per year among men from 2018 to 2022, 1.3% per year among women over the same years, and 1.5% per year among children ages 0 to 14 from 2001 to 2022. Three different populations, three different spans, three different rates. A relative-risk figure reported with no stated time frame can't be compared to any of these, or to a later update of itself.

Worked Example: Reading a Relative-Risk Claim

Using hypothetical figures modeled on the structure above: suppose a release states that “Condition X risk drops 50% with daily use of Product Y.” Running that one sentence through the worksheet below:

Population: not stated in the headline — adults over a certain age, people with a specific prior diagnosis, a specific country's population, or something else entirely all change what “risk” means here.
Baseline risk: not stated — if the underlying rate in the comparison group is 10 cases per 1,000 people over the study period, a 50% relative reduction brings it to 5 cases per 1,000.
Comparison: not stated — reduced relative to what: a placebo, no treatment, an older standard approach, or a different dose?
Absolute difference: in this hypothetical, 5 fewer cases per 1,000 people — the number a reader can actually hold onto.
Time frame: not stated — a 50% relative reduction measured over one year and the same reduction measured over ten years describe different realities.
Missing data: all five fields above are absent from the sentence as written. A claim this thin can't be evaluated yet — only logged as incomplete.

Once the baseline risk, comparison, and time frame are filled in from the actual source material — a study abstract, a release's methods section, or the reported confidence interval around the estimate — a relative-risk headline turns into something a reader can actually use: “Among people with X characteristic, Y fewer people per 1,000 experienced the outcome over Z years, compared with the comparison group.”

The Risk-Claim Worksheet

The same fields, formatted to apply to any headline, release, or abstract a reader wants to check:

Population: Who was studied — age range, diagnosis status, geography, and any other defining criteria. If this isn't stated, the claim can't be generalized to a specific reader.
Baseline risk: The rate of the outcome in the reference or control group, and its units (per 100, per 1,000, per 100,000).
Comparison: What the result is relative to — placebo, an older treatment, no intervention, a different population, or a different time period.
Absolute difference: Baseline risk minus (or plus, for an increase) the comparison group's risk, in the same units as the baseline figure.
Time frame: The observation period the rates were measured over, and whether the figure is a single-period rate or an annualized one.
Missing data: A running list of which fields above the source material didn't provide. An unfilled field isn't a flaw in the worksheet — it's information about how complete the original claim actually is.

Population Statistics Are Not a Personal Risk Prediction

Even a fully completed worksheet describes a group, not an individual. The National Cancer Institute's overview of cancer statistics makes this distinction directly, noting that statistical trends are usually not directly applicable to individual patients, even though they are essential for understanding a disease's impact on a population and for guiding broader strategy. The same source reports that an estimated 38.9% of men and women will be diagnosed with cancer at some point during their lifetimes, based on 2018–2021 data. That figure describes lifetime incidence across the whole population it was measured in — it is not a probability assigned to any one person, whose own risk is shaped by factors a population average doesn't capture.

Phrases Worth a Second Look

A few patterns tend to signal that a baseline risk or time frame has been left out of a headline or summary:

A percentage change with no stated comparison group (“risk reduced by X%” without “compared to what”).
A risk or benefit described with no time frame attached.
Language that shifts from a population finding to an individual recommendation within the same sentence.
Round, headline-friendly percentages (50%, double, triple) with no supporting rate figures nearby.

None of these patterns mean a claim is false. They mean the arithmetic needed to interpret it hasn't been shown yet — and the worksheet above is a way to go find it before deciding what the headline actually establishes.