Chapter 11

Measuring Interactions
  1. Explain why the mere observation that not every cigarette smoker gets lung cancer implies that cigarette smoking interacts with other factors in causing lung cancer.

  2. From the data in Table 11–2, estimate the proportion of stroke cases among hypertensive women who use oral contraceptives that is attributable to the causal role of oral contraceptives.

  3. In an analysis of the effect of oral contraceptives on stroke based on the data in Table 11–2, suppose that you were interested in the oral-contraceptive effect and wished only to control for possible confounding by hypertension using stratification. What would be the stratum-specific risk ratio estimates for oral contraceptive use for the two strata of hypertension? In an ordinary stratified analysis, why is there a separate referent category in each stratum?

  4. Show that if there is an excess over a multiplicative effect among those with joint exposure to two causes, there will also be an excess over an additive effect.

  5. The investigators of the study described in Table 11–2 claimed that women who faced increased risk from one risk factor ought to avoid additional risk from another risk factor, regardless of whether the two factors interacted in the causation of the disease. Does this suggestion make sense? What would it imply about seat belt use for women who take oral contraceptives?

  6. List reasons why the study of biologic interaction is more difficult than the study of the effects of single factors.

Website Terms and Conditions and Privacy Policy
Please send comments or suggestions about this Website to