Relative Risk Versus Odds Ratio, and Why They Stop Agreeing When an Outcome Is Common

A relative risk compares the chance of an event between two groups, while an odds ratio compares the odds of that event between the same groups, and those are not the same arithmetic. When the event is rare, the two numbers sit almost on top of each other, which is why people treat them as interchangeable. When the event is common, they pull apart, and an odds ratio can look far more dramatic than the underlying change in chances. Reading one as if it were the other is a quiet way a true result gets oversold. This article is general education for interpreting evidence, not medical advice; any decision about your own care belongs in a conversation with a qualified clinician.

Why trust this walk-through? Dr. Damon Tojjar is a physician-scientist whose peer-reviewed research reports effect sizes in exactly these units. His meta-analysis of ethnic differences in insulin sensitivity, published in Diabetes Care, pooled effects across many studies, and pooling forces a researcher to be precise about which measure is being combined, since the two answer different questions.

What is the difference between relative risk and an odds ratio?

Start with the building blocks. A chance is the number of events divided by everyone in the group. Odds are events divided by non-events. If 20 people in a group of 100 have an event, the chance is 20 in 100, but the odds are 20 to 80, which reduces to 1 to 4. Chance and odds describe one situation in two currencies.

Relative risk is the ratio of two chances, one group over another. An odds ratio is the ratio of two odds. Both equal one when the groups are alike, both rise above one when the first group has more events, and both fall below one when it has fewer. That shared shape is exactly why they get mistaken for each other.

The divergence lives in the denominators. Relative risk divides by everyone, so the ratio stays anchored no matter how common the event becomes. An odds ratio divides events by non-events, and as the event gets more common the pool of non-events shrinks, which stretches the odds and pushes the ratio further from one. The rarer the event, the smaller that distortion.

A neutral worked example with plain numbers

I will use round invented numbers so that nothing rides on a real study. Picture two groups of 100 people each. Ten have the outcome in the comparison group; twenty have it in the other.

Read the table as relative risk first. The chance is 20 in 100 against 10 in 100, so the relative risk is 2.0. The outcome is twice as common in the second group.

Now read the same table as odds. The comparison group sits at 10 to 90; the other at 20 to 80. The odds ratio is (20 over 80) divided by (10 over 90), which works out to about 2.25. Same table, and the odds ratio already sits above the relative risk of 2.0, because one in five is common enough to bend the odds.

Why the gap widens as the outcome gets common

Push the example toward a common outcome and the split grows. Suppose 50 in 100 have the outcome in one group and 25 in 100 in the other. The relative risk is a tidy 2.0 again. The odds tell a louder story: 50 to 50 against 25 to 75, an odds ratio of about 3.0. The chances doubled, yet the odds ratio reports a tripling.

A reader who hears "three" and pictures "three times as likely" has been misled, not by a wrong number but by the wrong frame for it. The odds ratio of 3.0 is correct as a statement about odds. It is not a statement about chances, and most people instinctively translate any ratio above one into chances. That is why reading odds ratios as risk ratios stays safe for rare events and grows risky for common ones.

Why studies report odds ratios at all

If odds ratios can mislead, a fair question is why so much of the literature leans on them. The reasons are technical and mostly sound. Some designs begin by selecting people who already have the outcome and comparing them with people who do not, and those cannot estimate a chance directly, because the ratio of cases to non-cases was set by the investigators rather than observed in nature. The dominant model for yes-or-no outcomes also produces odds ratios as its native output. The error enters at the translation step, when one gets described in plain language as though it measured chances.

How each measure can mislead a reader

An odds ratio misleads when a common outcome makes it drift far above the relative risk and a confident voice reports it in the language of likelihood. The number is accurate and the sentence around it is wrong, the hardest kind of error to catch, because nothing in the figure itself looks off.

A relative risk misleads in a different way, one it shares with any ratio. It strips out the baseline. A relative risk of 2.0 says nothing about whether the chance moved from one in a thousand to two in a thousand or from one in four to one in two. Both are a doubling, yet only one would change a decision about treatment.

So the two measures fail at different seams. The odds ratio asks you to respect that odds are not chances; the relative risk asks you to demand the absolute numbers underneath. Neither flaw is a reason to discard the measure.

How to read either number in under a minute

A few habits do most of the work. Find out which ratio you have, a risk ratio or an odds ratio, because the label changes how literally you can take it. When you see an odds ratio, ask how common the outcome was, since a common outcome means the figure overstates the change in chances. And whichever ratio you are handed, ask for the two plain chances behind it, because a baseline turns a floating ratio into something you can weigh.

The model to keep is small. Relative risk compares chances; an odds ratio compares odds; they agree when the outcome is rare and separate when it is common. Hold that distinction, and most of the overstatement around effect sizes loses its power.