Confounder or Effect Modifier

Excelsior College PBH 321 Page 1

I N S T R

I N S T R U C T I O N S F O R C O MP L E T I N G T H E I N D I V I D U A L A S S I G N ME N T In a short essay format, respond to each of the questions below. Create a Word document and submit your answers to the dropbox for this module. Make sure to show all of your work.

1. You wish to conduct a case-control study of alcohol consumption and the incidence of acute pancreatitis. You interview new cases of pancreatitis about their drinking habits 5 years before their diagnosis, and compare them with normal controls identified with random digit dialing. As part of your investigation, you need to assess whether gender is a potential confounder or effect modifier. The results are shown below for men and women.

Men Women

a) What is the most appropriate measure of association for this study?

b) Using this measure of association, does gender modify the association between alcohol consumption and pancreatitis? Why or why not? (If gender is an effect modifier, then stop with part b. If gender is not an effect modifier, then continue with parts c and d.)

c) If gender is not an effect modifier, does gender confound the association between alcohol consumption and pancreatitis? Why or why not?

d) If gender is not an effect modifier, show how gender either meets or does not meet each of the criteria for confounding.

e) What might be a possible ramification from how controls were selected for this study? Continue to Page 2

Cases Controls > 3 drinks per day

79

101

≤ 3 drinks per day

18

25

Cases Controls > 3 drinks per day

21

25

≤ 3 drinks per day

30

100

Excelsior College PBH 321

2. In a prospective cohort study, persons who wear contact lenses are compared to persons who only wear glasses regarding the incidence of eye infections. Each person is followed for a one-year period with no loss to follow-up. As part of your investigation, you need to assess whether age is a potential confounder or effect modifier. The results are shown below for persons above or below 40 years of age.

≤ 40 years old > 40 years old

Eye No eye infection infection

Contact lenses

20

80

Glasses

10

90

a) What is the most appropriate measure of association for this study?

b) Using this measure of association, does age modify the association between contact lens use and eye infection? Why or why not? (If age is an effect modifier, then stop with part b. If age is not an effect modifier, then continue with parts c and d.)

c) If age is not an effect modifier, does age confound the association between contact lens use and eye infection? Why or why not?

d) If age is not an effect modifier, show how age either meets or does not meet each of the criteria for confounding.

3. In a cross-sectional study, 9,000 people were identified as the source population, and 6,000 agreed to participate. The distribution by exposure and disease status is shown below for both groups.

Source population Study population

D+ D-

E

500

2500

E 500

5500

a) What is the most appropriate relative measure of association to use in this study?

b) Comparing the study population to the source population using this measure of association, is there evidence of selection bias?

c) Explain how selection bias occurred or did not occur by comparing the participation rates of exposed who were ill and healthy and unexposed who were ill and healthy.

d) If there is selection bias, is the bias towards or away from the null? Why?

Page 2

Eye No eye infection infection

Contact lenses

80

320

Glasses

80

720

D+ D-

E

400

2000

E 300

3300

D+ D-

E

650

2850

E 200

3300

D+ D-

E

500

3000

E 250

3250

Excelsior College PBH 321 4. In a cohort study with no loss to follow-up, disease status was misclassified as follows:

Population without Population with misclassification misclassification

a) What is the most appropriate relative measure of association to use in this study?

b) Comparing the study population with and without misclassification using this measure of association, is there evidence of information bias?

c) If there is evidence of information bias, is the bias towards or away from the null? Why? As a reminder: In a short essay format, respond to each of the questions above. Create a Word document and submit your answers to the dropbox for this module. Don’t forget to show all of your work!

Page 3