Bias

Systematic error, or 'bias' is of particular importance in any epidemiological investigation, and should be avoided wherever possible. Biases will reduce the validity of any results obtained, whether it be by overestimating or underestimating the frequency of disease in a population or the association between an exposure and disease. The forms of bias covered here can only be minimised through careful study design and execution - they cannot be accounted for in the analysis. Although confounding is considered by many authors as a form of bias, it can be accounted for during analysis, and so is covered separately.

Bias can be introduced into a study through the selection of participants (selection bias), or through errors made in the classification of measurement of exposures/outcomes of interest (information bias). These biases may also be classified as differential or non-differential, depending on whether the likelihood of bias is affected by an individual's exposure to factors of interest, or experience of the outcome of interest.

Validity

As mentioned earlier, the validity of an estimate is a measure of how well it can be extrapolated to the population in question (or, in the case of external validity, to other populations). The same concept applies to analytic studies, although the validity here relates to the ability to estimate the association between an exposure and an outcome of interest. Both selection and information biases can reduce the validity of an estimate.

Due to different investigation aims, it should be noted that introduction of biases into descriptive analytic studies differs from that in analytic studies. In descriptive studies, any process which results in the selection of a population which is not representative of the source population can be said to bias the results of the study. However, in the case of analytic study, the representation of the source population is less important , as long as the measure of association between the exposure and outcome is not affected. Therefore, both exposure and outcome should be considered - if selection is associated with both of these (for example, if selection of 'diseased' individuals was affected by their exposure category), then bias will result. If selection is 'only' associated with 'either' exposure status or outcome, then bias will not result.

Selection bias

Selection bias affects inclusion of individuals in the study and results in the study sample not being representative of the source population. This may occur in the initial selection process, or may be a result of nonresponse or losses to follow-up during the study period. The mechanisms by which selection bias can enter a study will depend on the study design:

Descriptive studies

In these studies, lack of compliance ('non-response') may result in selection bias, as participation amongst selected individuals is very rarely 100%. It is for this reason that compliance should be maximised as much as possible, through the use of reminders or incentives to participate.

Cross sectional studies

Non-response is also the main route of selection bias entering these studies,

Case-control studies

Case-control studies are particularly susceptible to selection bias. The main source of selection bias in these studies is through the selection of the control group. It is of vital importance that the control group comes from the same population as the case group (that is, if they happened to experience the outcome of interest during the study period, they would have been classified as a case instead of a control), and that there is no association between exposures of interest and selection as a control. The selection of the cases can also result in selection bias - in particular, in the case of hospital-based studies, where cases are selected from hospitals (as the population from which these individuals were drawn from makes selection of controls very difficult).

Cohort studies

Selection bias in cohort studies is generally less likely than in case-control studies, as selection for participation in the study generally precedes the development of the outcome of interest. However, non-response may result in some biases, and losses to follow-up during the study may also introduce selection bias. For example, a study may be conducted in order to investigate the effect of farm biosecurity on introduction of disease. It is plausible that farmers with poor biosecurity will be less likely to remain in the study, meaning that these farms individuals (which may be more likely to experience disease problems) will be lost from the study.

Information bias

Information bias results from errors in measurement or classification of exposures or outcomes of interest amongst the individuals included in the study. In the case of analytic studies, this may be classified as differential or non-differential. whereas the effect of differential bias cannot be predicted.

Non-differental information bias/Measurement error

Non-differential bias occurs when the chance of bias is not affected by the group the individuals belong to. This type of bias in analytic studies will tend to reduce the strength of any association present, and will increase the probability of a type II error. Errors in measurement (known as measurement error in the case of continuous variables, and misclassification bias in the case of binary or categorical variables) is a common example of non-differential bias - for example, if scales are not correctly calibrated, they will incorrectly record the weight of all animals weighed, regardless of their 'exposure' status.

Differential information bias

Differential bias occurs when the chance of bias is different for the different groups being compared, and may strengthen or weaken the estimated strength of association in an analytic study. For example, Boxer dogs may be more likely than other dog breeds to be diagnosed as having mast cell tumours (even if they do not have them), due to a postulated breed predisposition.