Epidemiology and causal inference · BEGINNER LESSON
Negative control
A negative control is a variable or comparison that is not supposed to have a real causal link with the exposure or outcome of interest, but shares the same kind of bias problems. Researchers use it to look for bias or sometimes to help correct it.
What it is and what is measured
In epidemiology, a negative control is not a treatment itself. It is a variable or comparison chosen because it should not be causally connected to the main exposure or outcome, but it may still be affected by the same bias structure. The measured result is whether the negative control shows an unexpected association. If it does, that can signal bias in the main analysis. Negative controls are often divided into negative-control outcomes, which should not be caused by the exposure, and negative-control exposures, which should not affect the outcome. [0,1,2]
[1] Negative controls and how to use them: a guidance paper - PMC [2] Negative controls: Concepts and caveats - PMC [3] Selecting the most informative positive and negative controls for self ...
How it works
The idea comes from experimental checks: if a condition should not produce a true causal effect, then any observed link may be due to other factors. In causal research, negative controls are used to detect unmeasured confounding, selection bias, information bias from misclassification, and type I error, which means a false positive result. Some methods also use them to help correct bias. This is useful only when the negative control really shares the bias structure of the main question. [0,1,2]
[1] Negative controls and how to use them: a guidance paper - PMC [2] Negative controls: Concepts and caveats - PMC [3] Selecting the most informative positive and negative controls for self ...
How to interpret results
A null result means no association was seen between the negative control and the variable of interest, but that does not prove there is no bias. The sample may be too small, two opposing biases may cancel out, or the control may not match the bias structure well enough. A non-null result can suggest bias, but it does not prove which bias caused it. Researchers should decide in advance what counts as a null finding. [0,1]
[1] Negative controls and how to use them: a guidance paper - PMC [2] Negative controls: Concepts and caveats - PMC
Limitations
Negative controls can miss some problems, especially bias that pushes an estimate toward the null, such as misclassification. They may also fail when the control is rare, when there is poor overlap in bias structure, or when the study has time-varying confounding in designs such as the self-controlled case series, a method that compares periods within the same people. So negative controls are evidence tools, not proof tools. [0,1,2]
[1] Negative controls and how to use them: a guidance paper - PMC [2] Negative controls: Concepts and caveats - PMC [3] Selecting the most informative positive and negative controls for self ...
SEE THE IDEA
How a negative control is used
The primary causal question being studied.
A comparison that should not have a real causal link to the main question.
Both the control and the main question are influenced by similar bias sources.
A signal that bias may be affecting the main analysis.
If the negative control shows an association when none should exist, the result suggests bias may also affect the main exposure-outcome analysis.
A negative control works by checking for bias in a comparison expected to be non-causal.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.