2.5 - 6: Cohort vs Case control analysis

Question 1

How are cohort studies analysed?

Answer 1

• Binary exposures used to tabulate data
• RR, RD, OR and corresponding CIs calculated (for p1 vs p0)

Question 2

How are case-control studies analysed?

Answer 2

• Binary exposures -> p1 = P(D[E)
• p2 = P(D[E-bar)
• Construct OR
• Note: If disease is rare, OR = approx. p1/p0 and the odds ratio approximates the relative risk

Question 3

How do the exposure and disease odds ratios compare?

Answer 3

• They are equal
• (Try proving…)

Question 4

Increasing precision of estimate for disease odds ratio:

Answer 4

• Increase number of controls
• i.e. b and d in tabular form

Question 5

Issues in selecting cases for rarer diseases:

Answer 5

• Number of available cases is often limited -> may be a complete enumeration of the diseased
• Confounding?

Question 6

Issues in selecting controls:

Answer 6

• Generally much more readily available
• Will likely require screening; don’t want to inadvertently include a diseased individual
• Beyond about 5 controls/case, benefit stops being valuable
• Controls may be a simple random sample from the disease or may be matched to the cases -> matching is used to control the confounders