Quiz 4: Epidemiology - Case-Control Study

A PDF of this quiz is available in the Online Library-you should use this to work on your answers and then submit online.This is not a timed exam but answers need to be submitted by the end of the appropriate time period listed on the Course Schedule.Please remember that this quiz is subject to the School's Academic Ethics Code,and should be completed individually without any help from current or previous students.You may,however,consult your course notes and lectur slides during the quiz. Once the time window for the quiz has ended (and all students have taken the quiz),you will be sent an email containing all questions,answers,and your individual responses.

1. Suppose we are researching whether studying Intermediate Epidemiology is associated with snow blindness.We conduct a case-control study amongst Hopkins students and determine that there is a protective association between our exposure and snow blindness.We are concerned about confounding and ask students about whether they went skiing during a recent blizzard.We find those who went skiing were more likely to develop snow blindness and were much less likely to be studying IntEpi.Based on these associations,do you anticipate the association between studying for IntEpi and snow blindness would change after adjusting for skiing?

a. Yes,the association without accounting for skiing would be overestimated; after adjustment it should become farther from 1.0

b. Yes,the association without accounting for skiing would be underestimated;after adjustment it should become closer to 1.

c. No,the association should not change and will remain 0.34 after adjustment for skiing.

d. The change in the estimate will depend entirely upon the p-value for the association between skiing and snow blindness.

e. I don't know I was skiing instead of studying my IntEpi readings and am having difficulty reading the screen because of snow blindness

2. For the design and the data given above.

(a) what measure of association could be calculated to describe the relationship between obesity and cancer?

(b) Calculate and

(c) interpret this measure.

3. The table is organized to allow an evaluation of effect measure modification.Which of the following statements identifies and justifies the method of assessment for effect modification as either joint effects or heterogeneity of effects?

1. The table is set up to assess effect modification using a joint effects method because all of the IRRs are >1.0.
2. The table is set up to assess effect modification using a heterogeneity of effects method because the confidence interval for the IRR of positive HIV status alone includes the null value of 1.0.
3. The table is set up to assess effect modification using a joint effects method because there is one common reference category,with IRRs calculated for positive HIV status alone, mother’s comorbid illness alone, and both positive HIV status and mother’s comorbid illness.
4. The table is set up to assess effect modification using the heterogeneity of effects method,because the IRR for positive HIV status is smaller than the IRR for mother’s comorbid illness.

4. Using the data in the table,explain why there is,or is not,evidence of effect measure modification between maternal HIV and comorbidity illness for childhood mortality?

5. Assuming the above DAG reflects the true causal association between these variables,which of the following would be confounding variable for the association between lung cancer and socioeconomic status?

6. In Livetalk,we discussed the Benfante paper.Briefly discuss how the methods used for identifying those who died in the Benfante etal.paper enabled the investigators to assess for the potential of selection bias in mortality rates.

7. Which of the following statements is true about misclassification?

a.Misclassification is a type of selection bias.

b.Misclassification is only relevant to the measurement of exposures, not to the measurement of other variables (e.g., outcomes, confounders).

c.Differential misclassification is preferable because the bias is always towards the null.

d.Non-differential misclassification is preferable because the bias can go in either direction.

e.Non-differential misclassification is preferable because the bias is always towards the null 1 point.

8. Provide one advantage of active outcome ascertainment and one advantage of passive outcome ascertainment.