Bias and confounding are apparent obstacles which clinical trials face. These factors can influence the results obtained, causing the reliability of the test to be questioned. To perform a good clinical trial study bias and confounding must be prevented early on to ensure the reliability of the data obtained, which can influence the success of the trial. Unfortunately, it is not easy to have completely unbiased trial results because errors can happen at all levels of the study. In this section bias and confounding will be explored. We will look into the different forms of bias and how they can be prevented. Also, confounding will be discussed together with ways to deal with it or prevent it.

Confounding vs. bias:

The observation is valid with a confounding compared to bias. There is clearly a relation between the confounder and disease. Bias creates an association that is not true, confounding describes an association that is true, but potentially misleading. Confounding is caused by a misinterpretation of data and is not the cause of incorrect data attainment.

Bias

Bias is a systematic error in the study that can lead to the distortion of true treatment effects. Bias can arise in different stages of the trial such as:

-       During the clinical trial design

-       During the research conduct

-       When collecting result data

In contrast to random errors, bias cannot be resolved with a larger study population.

The two common types of bias that can occur during a study are:

-       Selection bias

-       Information bias

Selection bias

Selection bias comes from any error in selecting the participants and/or factors affecting the study participation. By doing so, the study population is not representative of the targeted population of the investigated drug or treatment. This often occurs when the selection of the study population is not randomized⁷.

Example:

In a case-control study of smoking and chronic lung disease, the association of exposure (smoking) with disease will tend to be weaker if controls are selected from a hospital population (smoking causes many diseases resulting in hospitalization) than if controls are selected from the general public. Here, hospital controls do not represent the prevalence of exposure in the community from which chronic lung disease arise. The exposure-disease association has been distorted by the selection of an unrepresentable control group.

One way to prevent selection bias is by performing a randomized double-blind or single blind study. When a randomized trial is performed correctly, this ensures that, on average, treatment groups will be well balanced for both known and unknown factors, thus ensuring an unbiased valuation of the treatment effect.

Information bias

Information bias, also known as misclassification, is a common source of bias that affects the validity of the research. This problem arises from the method that is utilized to obtain or confirm study measurements. These measurements can be obtained by experimentations or observations.

Common forms of information bias are:

-       Observer bias

-       Recall bias

Observer bias

Observer bias is a form of information bias caused by study investigators incorrectly ascertaining or recording data from study participants.

Recall bias

Recall bias often occurs in case-control or retrospective cohort studies, where it is essential for participants to evaluate exposure variables retrospectively well using e.g., a self-reporting method. These studies are dependent on the participants' memory to obtain correct measurements regarding the study. Occasionally, the study participants can mistakenly provide responses that depend on his/her ability to recall past events. This causes recall bias, as a result of recall error⁸.

For example:

In a case-control study, cases are often more likely to recall exposure to risk factors than healthy controls. For example, mothers of children with birth defects are more likely to remember certain risk factors such as drugs they took during pregnancy differently than mothers of healthy children. Thus, true exposure might be underreported in healthy controls and overreported in the cases.

To overcome recall bias, it is crucial to recognize cases where recall errors are more likely to happen. Recall bias was found to be related to a number of factors:

-       The length of the recall period

-       Characteristics of the disease under investigation (e.g. acute or chronic)

-       Sample characteristics (e.g. age)

-       Study design (e.g. duration of the study)

Solutions proven to be useful are, for example:

-       Selecting an appropriate recall period

e.g., a short recall period minimizes the chances of recall bias as participants might remember this period more accurately than a longer recall period.

-       Objective data attainment

e.g., instead of obtaining data from the participants self, try obtaining data from the hospital or pharmacy concerning patient’s exposure to certain risk factors

-       Comparable data attainment

e.g., for a case-control study concerning certain exposure and the chances of having cleft-lip at birth, use as control group mothers with children having another form of birth defect. By doing so, both the control and case group are likely to recall certain exposure to risk factors during their pregnancy.