Lecture 9 (Effect Modification)

Question 1

P-value for heterogeneity

Answer 1

-Prob that heterog across strata could be due to chance if Ho is true

Question 2

Effect modification (def)

Answer 2

• Variation in magnitude of asso across strata of another exposure
• Equivalent Qs: Is OR greater among X? Are ORs hetero across strata?

Question 3

Effect modification (assessing)

Answer 3

• Stratify by potential effect modifier
• Calculate RR/OR for each stratum
• Assess degree of hetero across strata (10% or higher means its an effect modifier)
• Calculate p value

Question 4

Hoidrup et al: effect modifiers of HRT + hip fracture risk

Answer 4

• Effect modifiers: smoking
• Maybe (not stat sig): physical activity
• Not effect modifier: BMI
• Unlikely to be explained by selection bias, loss to follow up, info bias, detection bias