EXAM QUESTIONS 4 ANNOTATED KEY

pg. 1 1 Johns Hopkins University Fundamentals of Epidemiology

280.350

Spring 2024

EXAM QUESTIONS 4 ANNOTATED KEY

Questions 1 - 3 are based on the following data:

Incidence of Disease Z by Exposure to Factors X and/or Y Factor X Factor Y Incidence Rate of Disease Z Absent Absent 15.0 Present Absent 30.0 Absent Present 45.0 Present Present

• Assuming the effects of factor X and factor Y on disease Z are additive, which value best

fills in the empty cell in the above table?

A. 35.0

B. 45.0

C. 60.0

D. 90.0

E. 120.0

Assuming additive effects, the total combined risk of Z for those exposed to factors X and Y is equal to background risk (unexposed to both X and Y) + attributable risk of Z due to factor X only + attributable risk of Z due to factor Y only: 15.0 + (30.0 – 15.0) +

(45.0 – 15.0) = 15.0 + 15.0 + 30.0 = 60.0.

• Assuming the effects of factor X and factor Y on disease Z are multiplicative, which value

best fills in the empty cell in the above table?

• 35.0
• 45.0
• 60.0
• 90.0 1 / 3

pg. 2 2

• 120.0

Under the multiplicative model, the total combined risk for those exposed to factors X and Y is equal to background risk (unexposed to both X and Y) x relative risk of Z for factor X only x relative risk of Z for factor Y only = 15.0 x (30.0/15.0) x (45.0 / 15.0) = 15.0 x 2.0 x 3.0 = 90.0.

• If an incidence rate of 75.0 for Disease Z was observed in the empty cell in the table

provided, how would you interpret it under the additive model?

• Antagonism, but no interaction between factor X and factor Y
• Antagonistic interaction between factor X and factor Y
• Synergism, but no interaction between factor X and factor Y
• Synergism, but factor X doesn’t modify the effect of factor Y
• Synergistic interaction between factor X and factor Y

Since 75.0 (observed) is greater than 60.0 (expected under additive assumption from question 4) there is a synergistic interaction between factor X and factor Y. Another way to look at it: if the missing incidence rate was 75.0 then the effect of Y on Z would be different for different levels of X (without X: effect of Y is 30.0 (45.0 – 15.0) and with X: effect of Y is 45.0 (75.0 – 30.0)). This is precisely the definition of interaction: when the effect of a factor on an outcome is different for different levels of a third variable.Of course, you still need to know it’s synergistic as opposed to antagonistic but this gives the interaction (yes/no) piece.

Question 4 is based on the information given below:

Mortality rates from circulatory diseases per 100,000 women-years by cigarette smoking habits and use of oral contraceptives

Smoking Status Used oral contraceptives Never used oral contraceptives Non-smokers 13.8 3.0

Smokers 39.5 8.9

• Based on the results shown in the table above, which one of the following is true of the

women studied?

• Use of oral contraceptives contributes independently of smoking to the

risk of dying from circulatory diseases

• Smokers who never used oral contraceptives have a higher mortality rate of

dying from circulatory disease than nonsmokers who used oral contraceptives

• / 3

pg. 3 3

• Smokers who used oral contraceptives have a lower mortality rate of dying from

circulatory disease than smokers who did not use oral contraceptives

• Use of oral contraceptives increased the relative risk of dying from circulatory

diseases to a greater extent in smokers than in nonsmokers

• Smoking increased the relative risk of dying from circulatory diseases to a

lesser extent among those who never used oral contraceptives than among those who used oral contraceptives.

Compare oral contraceptive users to non-users among non-smokers. 13.8 vs. 3.0 indicates a 13.8/3.0 = 4.6x higher mortality rate in non-smoking oral contraceptives users compared to non-smoking non-oral contraceptives users. B is incorrect since 8.9 < 13.8. C is incorrect since 39.5 > 8.9. D is incorrect since in smokers the relative risk of dying comparing OC to non-OC = 39.5 / 8.9 = 4.4 and in non-smokers the relative risk is 13.8 / 3.0 = 4.6. E is incorrect since among those using oral contraceptives the relative risk of dying comparing smokers to non-smokers is 39.5/13.8 = 2.86 and among those not using oral contraceptives the relative risk is

8.9/3.0 = 2.97.

• A study of all women aged 35 years old in a large industrial state found that the

annual cumulative incidence of new cases of cervical cancer in women who used oral contraceptives was 0.0005 and that it was 0.0003 in those who did not use oral contraceptives. On the basis of these data, the inference that taking oral

contraceptives causes cervical cancer is:

• Incorrect because a proportion was used when a rate was required

• Incorrect due to failure to distinguish between incidence and prevalence

• Incorrect because the two groups may differ on other relevant factors

• Incorrect due to failure to adjust for possible differences in age distributions of

users and non-users

• Correct because the two groups were not different with respect to any

confounding variables.

For instance, women who use oral contraceptives may have sex more often and with more partners than those not using oral contraceptives; these two factors may also be associated with HPV acquisition and subsequent development of cervical cancer.

Questions 6 and 7 are based on the following information:

Sugar-sweetened beverages (SSBs) are the single largest source of added sugar and the top source of energy intake in the U.S. diet. Investigators conducted a review to evaluate

• / 3