Attrition bias happens when people drop out of an epidemiology study over time, leaving a group that is no longer representative. In field trials, that can distort the apparent effect of an intervention.
Attrition bias is the error you get in Intro to Epidemiology when participants leave a study in a way that changes the results. It is not just “missing data.” The problem is that the people who stay may be different from the people who drop out, and those differences can line up with the outcome you are trying to measure.
In a field trial, this shows up when researchers follow people over time to see whether an intervention works in the real world. Maybe the people who feel worse are more likely to quit a vaccine follow-up, or the people who improve fastest stop returning for check-ins. Either way, the final sample no longer looks like the original group.
That matters because the study’s conclusion can drift away from reality. If sicker participants drop out, the remaining group may make the treatment look better than it really is. If people who are doing well are the ones who disappear, the intervention can look weaker than it is. The direction of the bias depends on who leaves and why.
Attrition bias is especially tricky when the reason for leaving is related to the exposure, treatment, or health outcome. A study with lots of dropout is not automatically invalid, but the dropout pattern has to be examined. In epidemiology, you always want to ask: did the loss happen randomly, or did it change the makeup of the study population?
Researchers try to limit this by keeping contact with participants, using reminders, offering incentives, and tracking follow-up carefully. When dropout still happens, they may compare the people who left with the people who stayed, or use an intention-to-treat approach so the original randomization is not completely lost. The main idea is simple: if the sample changes over time, the result may change too.
Attrition bias matters in Intro to Epidemiology because it can make a field trial look more convincing or less convincing than it really is. Since field trials often measure outcomes over weeks, months, or longer, you are not just looking at who enrolls. You are also looking at who makes it to the end, and whether that final group still reflects the original population.
This term also helps you think like an epidemiologist when you read a study summary or results section. If a trial reports a large dropout rate, you should immediately ask whether the missing participants were similar to the ones who remained. If not, the estimate of effectiveness, risk, or disease prevalence may be biased.
Attrition bias connects directly to study validity. A trial can have randomization, a control group, and good data collection at the start, but still weaken if follow-up breaks down. That is why dropout is not just a housekeeping issue. It changes interpretation, especially in public health research where the goal is to understand what happens in a whole community, not just the most compliant participants.
You will also see this term when comparing different ways to handle incomplete data. If a class asks why one study’s conclusion seems stronger than another’s, attrition is often part of the answer. High follow-up quality usually makes findings easier to trust.
Keep studying Intro to Epidemiology Unit 7
Visual cheatsheet
view galleryFollow-up Rate
Follow-up rate tells you how many participants are still being observed at later points in the study. A low follow-up rate often sets up attrition bias, because the people who remain may not match the people who left. When you read a field trial, follow-up rate is one of the first clues that dropout may be affecting the results.
Randomization
Randomization helps make groups comparable at the start of a field trial, but it cannot fully protect a study from attrition later on. If one group loses more participants than the other, the balance created by random assignment can break down. That is why researchers still have to watch dropout after the groups are formed.
Selection Bias
Selection bias and attrition bias both involve groups that stop representing the population you care about. The difference is timing: selection bias is about who gets into the study, while attrition bias is about who leaves during it. In an epidemiology class, they are often discussed together because both can distort conclusions.
primary endpoint
The primary endpoint is the main outcome a field trial is designed to measure, such as infection rates or symptom reduction. Attrition bias becomes a problem when dropout changes who is available to contribute data for that endpoint. If many participants with certain outcomes leave early, the primary endpoint can be pushed up or down in a misleading way.
A quiz or case-analysis question may give you a field trial with a high dropout rate and ask what kind of bias threatens the results. Your job is to spot that attrition is not random missing data, but a systematic change in who remains in the sample. Then explain how that could make the intervention seem more effective, less effective, or just less reliable.
You might also be asked to propose a fix. A strong answer mentions keeping follow-up strong, comparing dropouts with completers, or using intention-to-treat analysis when randomization was part of the design. If a data table or graph shows fewer participants at each time point, use that pattern to support your reasoning.
Selection bias happens before or at the start of a study, when the people recruited are already not representative. Attrition bias happens after the study has begun, when participants drop out and change the makeup of the sample over time. Both can distort results, but they come from different points in the research process.
Attrition bias happens when dropout changes who is left in a study, so the final sample is no longer comparable to the original one.
In field trials, attrition can make an intervention look better or worse depending on which participants leave and why they leave.
A high follow-up rate lowers the risk of attrition bias, but researchers still have to check whether the people lost to follow-up differ from the people who remain.
Randomization helps at the start of a study, but it does not stop bias from appearing later if one group loses more participants than the other.
When you see dropout in an epidemiology question, ask whether the missing data are likely to be random or related to the outcome being measured.
Attrition bias is the distortion that happens when participants drop out of a study and the remaining sample is no longer representative. In Intro to Epidemiology, this is a big issue in field trials and other studies that follow people over time. The bias matters because the final results may reflect who stayed, not the full group that started.
It can make the trial’s outcome look stronger or weaker than it really is. If participants who are more likely to have a certain outcome leave the study, the data left behind become skewed. That can change the estimated effect of a vaccine, treatment, or prevention program.
No. Selection bias happens when the study sample is distorted at recruitment or entry. Attrition bias happens later, when people drop out during follow-up. They are related because both affect representativeness, but they happen at different stages of the study.
They try to keep participants engaged with reminders, incentives, and multiple follow-up methods. They also compare dropouts with completers to see whether the missing people look different. In some studies, they use intention-to-treat analysis so the original group assignment still shapes the final analysis.