1. ". . . the relative risk estimates of a large number of studies were contradictory but predominantly around unity (10)." (p616,c1) Which one of the following statements best restates the authors’ meaning? (Choose one best answer and, as always, provide a brief explanation.)

A. Various relative risks were observed, but most indicated no or a weak association
B. Relative risk estimates varied, but most indicated a 100% increase in risk
C. Opposing relative risks were observed, but most indicated either a 100% increase in risk or a 100% decrease
D. Relative risk estimates for the predominant (largest) studies were about 1.0
E. Relative risk estimates were mostly between -1.0 and +1.0.

  *2. Briefly explain how the inclusion of passively exposed women in the reference group could mask the effects of active smoking. (60 words maximum)

  3. ". . . all other studies included predominantly postmenopausal women." (p617,c1) In the present (Kropp and Chang-Claude) study, about how many of the women from whom passive smoking data were collected were premenopausal? Estimate the number (as an integer) and, as always, provide the basis for your answer.

  4. What is the key information that the control group is intended to provide? (30 words maximum)

  5. Which of the following statements about the control group selection procedure is (are) TRUE and which is (are) FALSE?.

a. Having two controls (rather than one) for each case reduces bias.

b. Having two controls (rather than one) for each case results in narrower confidence intervals.

  6. Which of the following statements about the control group selection procedure is (are) TRUE and which is (are) FALSE?.

a. Having four controls (rather than two) for each case would increase precision.

b. Having two controls (rather than one) for each case produces stronger associations (odds ratios further from the null).

  7. Many case-control studies select their controls from among hospital patients with conditions other than the one under study. Why would such a control group be unadvisable for the present research question? (40 words maximum)

  8. In 1992-1995, the percentage of cases participating was greater than the percentage of controls, but in 1999-2000 participation was greater for the controls (both relative to the originally identified cases and controls and relative to cases and controls who participated in 1992-1995). What accounts for this reversal in participation percentages between cases and controls? Support your answer with information from the paper? (30 words maximum)

  9. The authors report that two large prospective studies failed to find an association between smoking and breast cancer risk but note that "these authors based their analyses in part on current exposure status data obtained up to 14 years preceding the diagnosis of breast cancer." (p617,c1) (100 words maximum for both parts combined)

a. What is an important scientific (i.e., not logistical) advantage that this case-control study has over a conventional cohort study for obtaining exposure information on passive smoking in relation to breast cancer risk?

b. What is an important scientific (i.e., not logistical) advantage that a conventional cohort study has over a case-control study for obtaining exposure information on passive smoking in relation to breast cancer risk?

10. What should the investigators have done if 10 women in the control group had developed breast cancer between the time of completing their first questionnaire and re-interview?

11. "After exclusion of current smokers from the analysis, risk estimates decreased with years since cessation from 1.64 for having stopped smoking 1-9 years ago to around unity for 10 or more years." (p620c2) Show the calculation of the crude OR that corresponds to the 1.64 referred to in the quoted sentence.

Questions 12-13: "Wells (17) proposed that the inconsistency of past results on the association between smoking and breast cancer risk was attributed to the inclusion of passively exposed women into the reference group, thus masking the effects of active smoking."? (p616, c1).

12. What percentage of non-active smoker cases (i.e., cases who did not themselves smoke) were exposed to passive smoke?" What was the corresponding percentage for controls? Show the calculation.

13. Using data from Table 2, calculate and compare the following (crude) odds ratios. Then briefly explain whether they are consistent with the proposal that inclusion of passively exposed women in the reference group could mask the effects of active smoking. Show the calculations and brief explanation. (120 words maximum including calculation)

a. OR for breast cancer comparing ever active smoking (regardless of passive smoking exposure) to neither active nor passive smoking exposure

b. OR for breast cancer comparing ever active smoking (regardless of passive smoking exposure) to never active smoking (with or without passive smoking exposure)

c. Are these ORs consistent with the proposal?