Bias: Bias is defined as “any process at any stage of inference which tends to produce results or conclusions that differ systematically from the truth”. Bias can arise at three steps of the study: during initial enrollment of the participants, during implementation of the study, and during analysis of the findings. Following types of bias are seen in studies:
i) Selection bias: Selection bias occurs if the study population does not reflect a representative sample of the target population. Randomization reduces selection bias. Sampling bias is a type of selection bias.
Berkson’s bias is a type of selection bias. It depends on the fact that hospitalized patients are more likely to have higher risk exposure and more severe disease compared to the general population. Looking at just hospital data may give the false impression that there is strong association between a risk factor and a disease. This can affect the results of a case-control study. It can be avoided by selecting a comparable control group, for example selecting both case and control groups from hospitalized patients rather than cases from hospitalized patients and controls from the general population.
ii) Attrition bias: It is a type of selection bias that occurs due to uneven loss of follow-up or withdrawal from the study between the exposure and control groups.
iii) Measurement/ information/ observation or classification bias: It occurs from inaccurate gathering or recording of data or variables.
iv) Memory or recall bias: It is a type of measurement bias that is typically seen in case control studies. Subjects in the disease group may remember exposures more accurately compared to the control group.
v) Confounding bias: It occurs when a factor is associated with the outcome but is not a causal factor. Such an associated factor may be wrongly determined to be the causal factor. Randomization decreases the risk of confounding.
vi) Lead time bias: It is related to screening tests and survival. Even if people die at the same time with or without early diagnosis, the disease is detected earlier with a screening test and that may give the impression that patients are living longer, when in actuality they are just getting diagnosed earlier.
vii) Observer bias: It results when the observer is aware of the patient’s treatment group, or due to subjective differences in judgement between observers, variations in collection and analysis of data etc. It can be prevented by blinding or RCTs.
viii) Hawthorne effect: When participants know that they are being watched, they may behave differently. For example hand washing compliance increases when staff know that they are being observed.
ix) Ascertainment bias: It can happen when there is more intense surveillance or screening for outcomes among exposed individuals than among unexposed individuals, or differential recording of outcomes. It is related to observer, sampling, measurement and selection bias.
