Risk ratios, odds ratios and rates in epidemiology

Problem

Students often use risk ratio, odds ratio and rate ratio as if they mean the same thing. They do not. These measures answer related but different questions. Risk concerns probability over a defined period. Odds compare the probability of an event with the probability of no event. Rates account for person-time. Misinterpreting these measures can seriously distort the meaning of epidemiological findings.

• Odds ratios are interpreted as risk ratios.
• Rates are reported without person-time.
• Risk is discussed without specifying the time period.
• Students confuse absolute and relative measures.
• Large relative effects may correspond to small absolute differences.
• Logistic regression output is reported without explaining odds.
• Hazard ratios are confused with risk ratios or rate ratios.

Intuition

Epidemiological effect measures describe how often outcomes occur and how that occurrence differs between groups. Risk is intuitive because it is a probability. Odds are less intuitive but arise naturally in logistic regression and case-control studies. Rates are useful when people are followed for different amounts of time. Relative measures compare groups proportionally, while absolute measures describe actual differences in occurrence.

• Risk is the probability of an event in a defined period.
• Odds compare event probability with non-event probability.
• Rate measures events per unit of person-time.
• Risk ratio compares risks between groups.
• Odds ratio compares odds between groups.
• Rate ratio compares rates between groups.
• Absolute differences are often easier to interpret for public health impact.

Method

The correct measure depends on the study design, outcome frequency and follow-up structure. Cohort studies often allow risks and risk ratios to be estimated. Case-control studies usually estimate odds ratios. Studies with variable follow-up time often use rates or hazard-based methods. Reports should always define the numerator, denominator and time period.

• Step 1: Define the event clearly.
• Step 2: Define the population at risk.
• Step 3: Define the time period.
• Step 4: Decide whether cumulative risk can be estimated.
• Step 5: Use risk ratio when comparing probabilities over a defined period.
• Step 6: Use odds ratio for case-control designs or logistic regression.
• Step 7: Use rates when person-time differs across participants.
• Step 8: Report absolute risks or risk differences where possible.
• Step 9: Interpret relative measures with confidence intervals.
• Step 10: Avoid translating odds ratios into risk ratios when the outcome is common.

Working

Suppose 20 out of 100 exposed people develop disease and 10 out of 100 unexposed people develop disease over one year. The risk is 20% in the exposed group and 10% in the unexposed group. The risk ratio is 2, meaning the exposed group has twice the risk over that period. The risk difference is 10 percentage points, which describes the absolute difference.

• Risk in exposed group = 20 / 100 = 0.20.
• Risk in unexposed group = 10 / 100 = 0.10.
• Risk ratio = 0.20 / 0.10 = 2.
• Risk difference = 0.20 - 0.10 = 0.10.
• Odds in exposed group = 20 / 80 = 0.25.
• Odds in unexposed group = 10 / 90 = 0.111.
• Odds ratio is approximately 2.25.
• When the outcome is common, odds ratio and risk ratio can differ noticeably.

Limitations

Relative measures can sound dramatic even when absolute risks are small. Odds ratios are especially easy to misinterpret. Rates require accurate person-time measurement. Observational comparisons can be confounded, and effect measures should not be interpreted causally without appropriate design and assumptions.

• Risk requires a defined time period.
• Risk ratios may not be available in some designs.
• Odds ratios can exaggerate risk ratios when outcomes are common.
• Rates require reliable person-time data.
• Relative measures do not show baseline risk.
• Confounding can distort all effect measures.
• Causal interpretation requires more than an effect measure.

Discussion

A strong epidemiology report should present both relative and absolute measures when possible. For example, a treatment may halve the risk, but if baseline risk is very low, the absolute benefit may be small. Students should explain what the numerator, denominator and time period are, and should interpret effect measures in context.

• Report the event definition.
• Report the denominator or population at risk.
• State the time period for risk.
• Use person-time when reporting rates.
• Report confidence intervals for effect measures.
• Distinguish relative and absolute effects.
• Avoid interpreting odds ratios as risk ratios.

How this connects to learning

Use the guide as a bridge between theory and application.

A resource guide should not replace a full course or live teaching session. Instead, it helps you organise your thinking. Use it to identify what you understand, what feels unclear, and what questions you should ask before applying a method to real data.