Information bias, also called measurement bias, occurs when outcomes are systematically measured and/or analyzed differently, possibly resulting from researchers’ awareness of the groups that participants were assigned to, that leads to biased outcomes and conclusions.
In clinical research, participants are allocated to groups. Each group receives a different intervention to be compared.
Here are some ways information bias can occur:
- The researchers are aware of the intervention that participants in each group receive
- The researchers use different methods to assess the outcomes in each group
- There are systematic measurement errors of the outcomes that are related to the interventions
In these scenarios, the researchers have an unconscious bias when measuring the outcomes that favors one group over the other.
Information bias is not fraud
Information bias should not be confused with fraud. For it to be fraud, researchers would have to deliberately alter the data in favor of one group over the other.
Examples of information bias
Consider the following hypothetical examples of information bias in the outcome.
Researchers know which subjects took which drug
In a preclinical study, researchers compared drug A and drug B on physical activity in mice. They hypothesized that mice that received drug A would be more physically active.
If the researchers were aware which group of mice received drug A, they may have unconsciously interpreted physical activity, which was subjectively measured, in favor of drug A over drug B. Thus, they would have wrongly concluded that drug A resulted in better physical activity.
Had they not been aware of the intervention groups, their unconscious bias could not have affected the outcome measurements.
Researchers studied routine non-opiate painkiller A versus on-demand painkiller B on people with severe discogenic low back pain. They hypothesized the routine painkiller A would decrease pain more than painkiller B.
Subjects are aware of the intervention they received
If the patients were aware they were in the intervention group receiving painkiller A, they may have reported less pain. During the trial, some participants needed additional painkillers. They were allowed to request additional medication.
Administrators have bias
The total count of additional painkillers administered was an outcome in the study measured by a nurse. Let’s say the nurse who administered the additional painkillers believed routine painkillers decreased pain (as hypothesized by the researchers). Let’s also say the nurse was aware of the groups the participants were assigned to. As a result, that nurse may have withheld administration of additional painkillers.
This happened unconsciously without deliberate intent. Because fewer painkillers were administered to the participants in the drug A group, the outcome was biased in favor of drug A versus drug B. This bias ultimately shows up in the research, because less people in the drug A group will have received additional medication.
How to prevent bias: Blinding/masking
Information bias can be mitigated by “blinding” or “masking” the participants, researchers, and outcome assessors. Blinding/masking does not literally mean loss of sight or covering with a mask. It means the knowledge of the interventions is concealed from all parties involved in the research to ensure a clean exposure status.
For example, in Example 2 above, steps can be taken to ensure:
- The study participants are unaware of the pills they receive
- The researchers running the trial and analyzing the outcomes do not know which group of participants received each drug
- The nurses administering the drugs are unaware of the groups
Blinding helps reduce the influence of unconscious beliefs that may bias the results in favor of one group or another.
In some studies, blinding investigators and/or participants is not always possible. Surgical trials where the investigators/doctors performing the surgery must be aware of whether a surgery has been performed.
However, in a majority of studies, blinding is possible and critical to perform to ensure that the outcomes are equally measured and comparable to obtain accurate effect measurements of the results.
Information bias can inhibit the results of a study. Sometimes, researchers or subjects involved don’t even realize they are exhibiting information bias. To safely remove the chance of information, try blinding and masking to conceal all precious information from the study.
Explainer: What is Classification Bias
Explainer: What is Selective Reporting Bias?
Lost in Translation: The Clinical Failure of Preclinical Research
Explainer: What is a Systematic Review and Meta-analysis?