Lecture 7: Bias and confounding

Question 1

What is Bias in biostatistics?

Answer 1

Bias is any systematic error in design/implementation of a study
•Systematic error in selecting participants
•Systematic error in data collection

Question 2

Wha is the effect of bias?

Answer 2

It creates the tendency to overestimate or underestimate a measure of association.

Question 3

What are the types of bias?

Answer 3

Random error

Systematic error:

• information bias
• selection bias

Question 4

What is Selection bias?

Answer 4

• Errors introduced when the study population does not represent the target population
• Systematic difference between those included in a study and those who are not

Question 5

How do you reduce selection bias?

Answer 5

The most effective method to minimise selection bias in intervention studies is random allocation to treatment and control groups.

Question 6

What is Volunteer or Non- response bias?

Answer 6

Typically occurs in surveys/observational studies, where individuals decide whether or not to participate.

Question 7

What is the Healthy worker effect ?

Answer 7

Occurs in occupational cohort studies or in studies comparing working to non-working populations.

The effect:

• Workers have lower rate of disease than non-workers
• Lower mortality in the employed population compared to the general population.
• Comparison of workers to non-workers may underestimate the health effects of occupational exposures

Question 8

What is Neyman’s bias .

Answer 8

Neyman Biasis a selection bias where the very sick or very well (or both) are erroneously excluded from a study.

Question 9

What are the effects of Neyman bias?

Answer 9

• Excluding patients with severe disease will make treatment effective
• Excluding patients who have recovered will make conditions lookmoresevere.

It affects studies on long-term conditions e.g. HIV or tuberculosis. Less of a problem with acute conditions.

Question 10

What is Attrition bias?

Answer 10

Loss to follow up

It can occur in Cohort studies and randomized controls trials (RCT’s) which involve following people over a certain period of time

Those that are lost to follow up may be different from those who complete the study

Question 11

What is information (measurement) bias?

Answer 11

Information bias is a systematic error in exposure assessment or outcome assessment

Biased data (even with random sample)

Question 12

What is Recall bias?

Answer 12

Error due to differences in accuracy or completeness of recall to memory of past events or experiences.

Respondents memory vary according to whether they have experienced the outcome
- Cases with disease are more likely to recall the possible cause of their illness than healthy people.

Question 13

What is Interviewer bias?

Answer 13

Interviewer bias occurs when there are systematic differences in the way information is collected for the groups being studied

Question 14

How can interviewers influence respondents’ answers?

Answer 14

Structure of questions – e.g. putting emphasis on questions, use of gestures.

The interview situation

Question 15

What is Observer bias?

Answer 15

Occurs when observers are not blinded to exposure or disease status

Observers may have preconceived expectations of what they should find in an examination

Example: Radiologist aware of a patient’s smoking status may look more critically at abnormalities in X-ray.

Question 16

What is the Hawthorne effect?

Answer 16

Alteration in behaviour when subjects know they are being studied.

Question 17

How does Controlling for bias occur?

Answer 17

• Ensure sample selection is random
• Measure and control all variables related to health behaviour
• Set up strict guidelines for data collection:
  Use standardised questionnaires, train interviewers and observers, preferably use more than one observer.
• Institute Blinding/masking where appropriate:
  Blinding prevents interviewers or observers (and some times subjects) from knowing case/control status, treatment allocation, or exposure status of study subjects.
• Use an appropriate method for data collection
  E.g. for socially sensitive questions, such as alcohol and drug use or sexual behaviours, use a self-administered questionnaire instead of an interviewer.
• Build in methods to minimise loss to follow-up:
  Recruit people who can easily be tracked
  Use of incentives

Question 18

Define confounding

Answer 18

Lack of comparability: when certain background factors differ between groups being compared.

A mixing of effects: A situation in which the apparent effect of an exposure on risk is explained by other factors, resulting in a distortion of the true relationship

Question 19

What is lack of comparability in confounding?

Answer 19

when certain background factors differ between groups being compared.

Question 20

What is a mixing of effects in confounding?

Answer 20

A situation in which the apparent effect of an exposure on risk is explained by other factors, resulting in a distortion of the true relationship

Question 21

What are Confounders?

Answer 21

Factors or variables that cause confounding

Question 22

What are the conditions to be a Confounder?

Answer 22

Associated with the exposure

Associated with the outcome

Not on the causal pathway between the outcome and exposure

Question 23

How to deal with confounding in design and statistical analysis?

Answer 23

Randomisation

Matching

Restriction

Stratification

Multi-variate regression analysis

Question 24

Bias refers to errors in the design or conduct of a study that can result in false conclusions

True or false?

Answer 24

True

Question 25

Selection bias can occur when the study sample does not reflect the population of interest

True or false?

Answer 25

True

Question 26

Information bias can occur when the method used in collecting study data could lead to inaccurate responses or measurements

True or false?

Answer 26

True

Question 27

Confounding can be minimised at ?

Answer 27

Design stage- randomisation

Analysis stage- stratification, regression analysis