Bias

Question 1

What four biases do you need to be aware of

Answer 1

selection bias
Performance bias
Observer
Attrition

Question 2

What is selection bias

Answer 2

when the sample pop is unrepresentative of the target population

• • how are the patients being recruited? volunteer bias, channeling bias
• • what are baseline characteristics?

and when the two groups are significantly different

• • randomisation? stratification?
• • allocation concealment

Question 3

What is allocation concealment

Answer 3

when the person recruiting patients does not know what the next allocation will be

So they cannot unconsciously select people, removing selection bias

Question 4

How can allocation concealment be done

Answer 4

via central ranomdisation (computer based) or opaque sealed envelope

Question 5

what is allocation bias?

Answer 5

Type of selection bias - when investigators know or predict the intervention the next person will get, and may influence how investigators approach that partipant or what allocation he willl get

Question 6

What is performance bias?

Answer 6

differences that occur due to knowledge of intervention allocation, in either participant or researcher

Question 7

What are ways to reduce performance bias

Answer 7

Blinding (triple blind: participant, investigator, data analyst)

Question 8

What is observation bias

Answer 8

When the way that data is collected or analysed is influenced by the expectations of assessors

OR

differences in measurements between centres

Question 9

How do you reduce observation bias

Answer 9

Pre-specify primary and secondary outcomes
Pre-specify data collection methods
Blinding

Question 10

What is attrition bias caused by

Answer 10

Bias caused by differences between intervention and control groups in terms of numbers of withdrawals from the study

Question 11

How can you minimise attrition bias

Answer 11

use intention to treat analysis

Use Kaplan Meir curves instead of set end points, as different arms drop out at different rates

++ avoid subjective exclusion criteria

Question 12

How do you minimise selection bias

Answer 12

recruitment size
recruitment location, multicentre / single centre
recruitment manner (continuous, non-continous)

allocation concealment
randomisation (although stratification may increase selection bias, as it makes it possible to predict the next allocation)

Question 13

what diagram will represent attrition bias

Answer 13

consort flow diagram

Question 14

what is blinded outcome adjudication?

Answer 14

When the data analyst is blinded to the which group is which

this minimises performance bias

useful in open label study