Selection bias
what is selection bias?
Occurs when there is a systematic difference between the people who are included in a study and those who are not, or when study and comparison groups are selected inappropriately or using different criteria
what is bias
any systematic error in an epidemiological study that results in an incorrect estimate of the association between exposure and risk of disease
what can we do about bias
selection and information bias can only be controlled during the design and data collection phases of a study
how do we recruit people into epidemiological studies?
volunteers - although people who volunteer may be different from people who don't volunteer which can cause bias, random selection - can give a more representative sample
what are three steps to be considered with bias
who does the bias affect? how does it affect them? what effect does it have on the measure of association? (is it under or over-estimated?) and you should clearly specify the direction in relation to the null value
what must be considered in cross sectional studies for selection bias
who entered the study, if the sample is representative of the source population, response rate
to control for selection and information bias at the planning stages of a study what must investigators do
identify the potential sources of bias in the proposed study, identify possible ways to minimise these potential biases
how can selection bias occur in RCTs
can be introduced if people selecting participants into the study knew which groups the people would go into and based their selection or non-selection on that in some way, need concealment of allocation, there is potential for loss to follow up
selection bias in case-control studies
can be susceptible to selection bias as participants are selected on the basis of their outcome status and the exposure and outcome have already occurred, if the selection is in some way dependent on their exposure status then bias can occur, cases and controls are also selected independently which can result in selection bias
what happens when not everyone selected agrees to participate
can create problems and bias if the people who participate are different to those who don't
what are the potential biases in the selection of the controls in case control studies
if controls are not representative of the population which gave rise to the cases in place or time, if the inclusion/exclusion criteria is different between cases and controls, if non-participation of eligible controls occurs so it is important to ensure high participation of controls
what are the potential biases in the selection of cases in case control studies
if the identification or participation is related to exposure status, selection bias can be minimised by ensuring high participation, clearly defining the population of interest and having a reliable way of ascertaining all cases or a representative sample of cases
how can selection bias occur in cross-sectional studies
if the sample was not representative of the source population you need to consider if the prevalence is likely to be over-estimated, under-estimated or not affected
how can bias occur from people leaving the study
is the reasons people leave the study are related to the exposure and outcome then bias can occur, want to minimise loss to follow up
when systematic error occurs findings of the study could be
over-estimated, under-estimated or not affected
how can selection bias occur in cohort studies
potential for loss to follow up which can cause selection bias if loss to follow up is related both to the exposure and the outcome, if the comparison group is selected separately from the exposed group eg the healthy worker effect