In a clinical trial, participants in one group may be treated differently versus those in the comparative group(s). The different treatment introduces another variable that could account for differences in outcomes between groups.
When this occurs, a bias is introduced; one of the group(s) deviated from the intervention being tested that should ideally be the only variable to account for differences between groups. This is called bias due to deviations from intended intervention.
Bias due to deviations from intended intervention explained
In comparative research studies, two or more groups are tested against each other comparing the effect of one group receiving an intervention and another other group(s) receiving a comparator intervention. To determine if there is a true difference between groups in respect to an outcome(s), the groups must be treated as equally as possible during the study except for the tested intervention.
Bias due to deviations from intended interventions occurs when there are systematic differences in the care provided between the experimental intervention and comparator group(s).
In other words, if one of the groups was treated differently from the group(s) other than the intervention (which should be the only interventional difference between groups), and if the results show that there is a difference between groups in respect to the outcome(s) of the study, the results may be biased because of the unequal treatment provided between the experimental and comparator group(s).
Consider the following hypothetical scenarios:
Examples of bias due to deviations from intended intervention
Example 1
In a multicenter clinical trial, researchers compared the effect of injection A and injection B. The objective was to examine if injection A could prevent infections better than injection B. Care providers in some of the centers thought participants who received injection A had more infections compared to those who received implant B. Consequently, they administered more antibiotics to the injection A group.
Differential care was provided to the injection A group because the different care that each group received was dependent upon whether the participants received injection A or injection B. The results of this study would be biased due a deviation from the intended intervention.
Example 2
Researchers compare stem cells derived from fat versus stem cells derived from bone marrow regarding their ability to decrease pain in patients with spinal disc degeneration (DGD). Adults with DGD were recruited and randomized into two groups, group A that receive stem cells from fat and group B that receive stem cells from bone marrow.
After extracting the stem cells from the patients, the researchers perceived the stem cells derived from marrow were less viable. Consequently, when culturing the stem cells, they provided better nutrition only to the stem cells derived from marrow.
Six months after injecting the stem cells into the patients’ discs, patients in group A had a better response to the treatment with decreased pain in comparison to patients in group B. However, the difference between group A and group B could be due to the nutrition provided to the stem cells rather than the origin of the stem cells. Hence, bias due to deviation from the intended intervention is likely to have prejudiced the results leading to misleading results.
Example 3
Bias due to deviations can manifest from intentional or unintentional actions. Non-pharmacological interventions (i.e. surgery, endoscopy, physiotherapy, or psychotherapy, etc.) are particularly at risk of this bias.
For example, a surgeon may believe the protocol set forth in the study is incorrect, so they change how they operate on patients during the study. It may not be possible for an endoscopist to perform an endoscopic procedure due to an inability to reach the desired location. An anesthesiologist may fail to recognize that a study participant was assigned a specific antibiotic as the interventional drug and may give the standard antibiotic rather than interventional drug.
Final thoughts
In these three examples, bias due to deviations are present regardless of who is responsible for lack of adherence (deviation) to the protocol.
Researchers can mitigate this bias by ensuring that a study protocol is clearly defined, that the only pre-specified difference between groups is the intervention under investigation, that the protocol is understood by the researchers interacting with the participants, and that the protocol is followed exactly as defined.