Attributable proportion due to interaction (AP) is the fraction of an outcome that can be credited to two exposures working together beyond their separate effects. In Intro to Epidemiology, it is used to describe additive interaction in disease risk.
Attributable proportion due to interaction (AP) is an epidemiology measure that tells you how much of the disease burden among people exposed to two factors is due to the interaction between those factors, not just their separate effects. It is usually discussed when you want to know whether the joint effect is bigger than what you would expect if the exposures simply added up.
In Intro to Epidemiology, AP sits inside the topic of effect modification and interaction. The basic idea is that one exposure can change how another exposure behaves. If smoking and asbestos exposure together produce more lung cancer than you would predict from either exposure alone, AP helps describe the extra burden caused by that joint effect.
A useful way to think about AP is that it focuses on the excess part of the risk that comes from the combination. If the two exposures are working independently, AP is zero or close to zero. If they reinforce each other, AP becomes positive, which suggests synergy on an additive scale.
That additive scale matters. Students often mix up interaction with just “two things happening at once,” but AP is more specific than that. It asks whether the combined effect creates more disease than expected from adding the individual effects. That is different from multiplicative interaction, which compares ratios rather than risk differences.
You usually see AP in stratified tables or regression output when the course is covering interaction terms. A common setup is comparing groups with exposure A only, exposure B only, both exposures, and neither exposure. From those results, AP helps summarize how much of the combined risk can be attributed to the interaction itself.
Because AP is about interpretation, not just calculation, the context of the study matters a lot. Observational studies can be affected by confounding, bias, or poor measurement, so a high AP does not automatically prove causation. It does, however, give you a sharper picture of how exposures may be acting together in a population.
AP matters because Intro to Epidemiology is not just about whether a risk factor raises disease risk, but about whether two risks together create a different public health story than either one alone. That matters when you are deciding which group needs the most prevention effort or which combined exposure deserves attention in a report or class case study.
It also helps you separate simple association from interaction. A disease can be associated with two exposures without those exposures changing each other’s effect. AP gives you a way to talk about the extra disease burden tied to the combination, which is a more useful idea when you are thinking about screening, prevention, or workplace hazards.
This term also shows up when the course asks you to interpret subgroup differences. For example, if one exposure raises risk more in people who already have another risk factor, AP helps explain why the combined group looks worse than expected. That kind of reasoning is central to effect modification and to making sense of stratified data in public health.
Keep studying Intro to Epidemiology Unit 8
Visual cheatsheet
view galleryEffect modification
Effect modification is the broader idea that the effect of one exposure changes across levels of another variable. AP is one way to describe that change when you are focused on the extra burden from the joint exposure. If the effect truly differs by subgroup, you will often see it in the pattern that AP summarizes.
Interaction
Interaction is the umbrella term for when two exposures combine in a way that is not just a simple add-up of their separate effects. AP is one way to measure that joint effect on an additive scale. When you read a table with multiple exposure categories, interaction is the concept behind the comparison.
Relative Excess Risk due to Interaction (RERI)
RERI is another additive interaction measure that looks at excess risk from the joint exposure. AP is related, but it expresses that excess as a proportion rather than a raw difference. If you see both in the same problem, RERI helps you see how much extra risk there is, while AP shows what share of the combined effect is due to interaction.
Causal inference
AP is most meaningful when you are thinking carefully about whether the observed joint effect reflects a real causal pattern. Causal inference pushes you to ask whether the interaction is likely to be genuine or whether confounding, bias, or measurement error might explain it. That makes AP a interpretive tool, not just a numerical result.
A quiz question or data interpretation item may give you a 2 by 2 or 4-group table and ask whether the joint effect suggests interaction. Your job is to spot whether the combined exposure does more than the separate exposures and then explain AP in plain language as the proportion of outcome due to that extra joint effect. If a regression output includes an interaction term, you may need to say what it means for disease risk in the exposed group. In a short-answer response, use AP to describe the share of the outcome tied to the combination, not just the individual risks. If a scenario asks about prevention, connect a positive AP to targeting both exposures together rather than treating them as separate problems.
Population-attributable risk looks at how much disease in the whole population could be prevented if an exposure were removed. AP is narrower because it focuses on the extra outcome due to the interaction between two exposures among the jointly exposed group. One is about population burden from an exposure, the other is about synergy between exposures.
Attributable proportion due to interaction (AP) tells you how much of a health outcome is due to two exposures acting together, beyond their separate effects.
AP is part of effect modification and interaction, so it is most useful when a course question asks whether the combined exposure creates extra risk.
A positive AP suggests synergy on an additive scale, while an AP near zero suggests the two exposures are not interacting much in terms of added risk.
You usually interpret AP from stratified data or regression models, then explain what the joint exposure means for disease burden in a population.
AP is a description of the data, not automatic proof of causation, so confounding and bias still matter when you interpret it.
AP is the fraction of an outcome that can be traced to the interaction between two exposures, rather than to either exposure alone. It is used when you want to describe the extra disease burden created by a joint effect on an additive scale. In plain terms, it asks how much worse the combined exposure is than the separate exposures would predict.
Not exactly. Effect modification is the broader pattern where the effect of one exposure changes depending on another variable. AP is one way to quantify that pattern, especially when you want to describe the proportion of risk due to their interaction. So effect modification is the concept, and AP is one measurement you might use.
Population-attributable risk deals with disease burden in the whole population if an exposure were removed. AP is about the combined effect of two exposures and the part of the outcome that comes from their interaction. If you are comparing them, think population burden versus synergy between exposures.
You usually use AP when you are given exposure groups and asked to interpret whether the joint exposure adds extra risk. The key move is to explain the combined effect in words, then connect it to interaction or effect modification. If the data show a strong joint effect, AP helps you say how much of that effect is due to the exposures working together.