Bias, Confounding and Causality

Question 1

What are the reasons for real associations between 2 variables?

Answer 1

co incidence
confounding
reverse causality
causal

Question 2

What are the reasons for no real associations between 2 variables?

Answer 2

chance
bias
incorrect analysis

Question 3

A cohort study found that heavy drinkers had a higher rare of lung cancer than all other drinkers. What is the reason for real association?

Answer 3

confounding
heavy smokers tend to be heavy drinkers
smoking leads to lung cancer

Question 4

What is a confounder?

Answer 4

A confounder is a factor that is associated with the exposure and the disease

Question 5

Between 1950 and 1970 sales of ice cream per month were correlated with the number of deaths by drowning per month. What is the reason for real association?

Answer 5

confounding
high temperature is associated with increase in ice cream sales
high temperature is associated with increase in drownings

Question 6

A case control study of stomach cancer patients found that they had a higher antacid usage in the 6 months prior to diagnosis than did the control. What is the reason for real association?

Answer 6

reverse causality

people with undiagnosed stomach cancer tend to have the same symptoms for which antacids are likely to be taken

Question 7

A nested case control study of men who had died from stomach cancer compared with controls who had not, found that they had higher levels of H Pylori infection. What is the reason for real association?

Answer 7

Causal

high rate of H Pylori leads to stomach cancer

Question 8

Between 1920 and 1970 there has been a positive association between the miles of hedges in the UK and the incidence of tb. What is the reason for real association?

Answer 8

Co incidence

Question 9

A survey for leukaemia patients in the 1970s showed that a higher proportion of them had a blood group A compared with the general population. Is the association real?

Answer 9

Survival bias

incidence of disease is similar in all blood groups, but those with blood group A survive longer

Question 10

A case control study of patients with Crohn’s disease showed that they were much more likely to have reported ever eating Corn flakes. Is the association real?

Answer 10

Recall bias
those with Crohn’s disease are more likely to remember everything they ate. most people have eaten corn flakes at some time

Question 11

The proportion of younger compared to old men dying from IHD is higher in India than in England. Is the association real?

Answer 11

Selection bias
there is a higher proportion of young men in India than in Englan and therefore more deaths occur in the younger population

Question 12

The cause of death in women in a clinical trial in 1966 of folic acid during pregnancy was examined 35 years later and the author’s concluded:
“Women taking high doses of folate throughout pregnancy may be more likely to die from breast cancer later in life than women taking no folate”
Is the association real?

Answer 12

Incorrect analysis: Data dredging
breast cancer was not a prior hypothesis
the authors probably looked at lots of outcomes and only reported the significant one

Question 13

What is name of the list of criteria used to decide if something is causal?

Answer 13

Bradford Hill Criteria for Causation

Question 14

What is Bradford Hill Criteria for Causation?

Answer 14

1. Strength of Association
2. Dose response
3. Time sequence
4. Consistency of findings
5. Biological plausibility
6. Coherence of the evidence
7. Reversibility

Question 15

What measurements are used to see how big the association is?

Answer 15

adjusted vs unadjusted estimates