EIP - Bias - Week 3 Flashcards
When deciding on the specifics for a given intervention, what five factors should be given consideration to?
Cost of the intervention
Sample size of the intervention group
Sample size of the control group
Length of the study
All the overheads associated with the study
Define selection bias. When does it occur?
A bias that occurs when subjects selected into a trial habe characteristics different to those intended.
Occurs during selection.
Define referral bias. What kind of bias does it fall under? When does it occur? Why does it occur? Describe what patients with this bias are like.
It is a special kind form of selection bias, and is due to the recruitment of subjects from referral centres.
These patients tend to be more severely affected by the disease in question and this causes a bias in the result different to the general population.
Occurs during selection.
Define spectrum bias. What kind of bias is it similar to? How does it occur? When does it occur?
Very similar to referral bias. Has a similar effect on the external validity of the trial.
Occurs when subjects are recruited from patients with classic or severe symptoms of the disease.
Occurs during selection.
Define membership bias. How does it occur? What two study categories is it more likely to affect? When does it occur?
This bias occurs if the groups chosen for the intervention and control are selected from different subgroups and have different health/socio-economic characteristics etc.
More likely to affect cohort studies and case-control studies.
Occurs during selection.
Describe baseline assessment. How does good randomisation fit in with this process?
Assess and record the factors that might affect the outcomes for each subject.
These can be compared to assess whether you have good randomisation.
How is baseline assessment and comparison usually done in randomised controlled trials?
Typically is first listed in a table (usually Table 1) then comparisons are made in the discussion
Can adjustments be made t data analysis if required by basline assessment and comparison, or are studies prematurely ended if important differences are found between trial groups?
Adjustments can be made in retrospect
It is never as good as having similar groups to begin with
What is the key method to reducing bias and when should it be done? What does this process achieve? What can be done to demonstrate whether it has worked?
Randomisation
Should be done after enrolment
Generates groups that are similar in all characteristics except the intervention
Statistical tests can be done to demonstrate whether it has worked
If a statistical test has been done to demonstrate randomisation, does this mean there is no possible confounding? Explain.
No, 1 in 20 of these comparisons will be statistically significant at p<0.05 simply by chance
Define confounding.
It is the presence of certain variables that influence the effect you are studying
Can confounding be mitigated by limiting the subject’s characteristics using multiple exclusion criteria? Explain.
No, this would limit the external validity of the study and narrow the usefulness of the results
Describe fixed allocation for randomisation.
Assigns the subjects to an intervention depending on a pre-specified probability (ie a coin toss)
Define block randomisation. and what it is used for.
Used to ensure more equal numbers across groups.
Order in which the intervention is assigned is randomly allocated with each block of subjects.
which studies is stratified randomisation used in and does it achieve equal distribution in all baseline factors across all sample sizes?
Used in smaller studies
Randomisation may not achieve equal distribution of baseline factors if the sample size is small
