Epidemiology Final Exam
1. A study of neurological impairment and aluminum is being planned in a random sample of Northwestern Ontario residents aged 60 to 75.
After reviewing the literature, the investigators feel that an appropriate estimate of the relative risk is 2.0 for those with some aluminum exposure versus those with none. A dichotomous exposure is assessed using a biological marker of aluminum exposure obtained from finger nail clippings.
a) What sample size would be required to have 90% power with a significance of 0.05, if 30% of this population will be classified as exposed and it is expected that 20% of these will be classified as having neurological impairment?
b) If 40% of your sample was male and outcome and exposure rates were equal in each sex, what would be the power to detect the anticipated risk in males (using your sample size from above)?
c) The use of biomarkers to estimate exposure can be prohibitively expensive. Exposure to aluminum products occurs in various occupations, through use of aluminum cooking utensils and to a small degree in drinking water treated with aluminum compounds.
Describe one exposure assessment method (other than biomarkers) which could be used to examine the relationship between aluminum exposure and neurological impairment. Please also briefly describe the study setting(s) in which this method can be used (i.e., prospective / retrospective cohort, case-control, ecologic, etc.).
2. You are designing a case-control study to evaluate the association between benzene exposure and leukemia. Incident cases of leukemia will be identified from the records of the Ontario Cancer Registry. It is estimated that 99% of incident cancers occurring among Ontario residents are captured in this registry system.
a) Describe two sources of controls for this study and contrast these control groups in terms of the appropriateness of each for this study.
b) The following table presents the number of cases and controls by age and benzene exposure. Evaluate the relationship between benzene and leukemia in these data and provide an interpretation of the findings. What is the role of age?
|
Exposed to benzene
|
No benzene exposure
|
|
Age
|
Cases
|
Controls
|
Cases
|
Controls
|
|
5-14
|
67
|
57
|
76
|
66
|
|
15-24
|
138
|
142
|
40
|
50
|
|
25-34
|
186
|
51
|
28
|
10
|
|
35-44
|
206
|
115
|
28
|
23
|
|
45-54
|
104
|
174
|
12
|
43
|
|
55-64
|
32
|
211
|
2
|
75
|
3. In a study measuring use of antihypertensive drugs, two methods of ascertaining exposure were employed, a medical chart review and a subject interview.
In a review of the medical charts of 220 women, 50 indicated some use of antihypertensive drugs. During an interview, 40 of the 220 women reported use of anti-hypertensives. For 30 women, both sources indicate use of antihypertensive drugs, and there was no indication of any use of these drugs from either source of information for 160 women.
Evaluate the agreement between the two methods of obtaining information on antihypertensive drug use.
4. A population screening program for HIV infection among pregnant women has been initiated in Ontario. This screening program targets all pregnant women in Ontario, yet only 20% of pregnant women are currently being screened. You are interested in determining predictors of participation in the screening program (i.e., uptake of the screening program).
You have access to data on all women who have undergone a screening test between Jan 1, 2014 and Dec 30, 2016. Basic demographic information collected on these women includes age, ethnicity (5 categories), marital status, urban or rural residence location and highest level of education attained.
What study design was used to collect the data? Describe any additional data which you need and your analysis strategy for determining the independent effects of the demographic factors described above in predicting uptake of the screening program.
5. A brief written questionnaire concerning hearing deficits was used in the assessment of 125 school children (age 6 to 10). All of the children were also given a medical hearing examination which classified them with essentially no error as to whether they had a hearing deficit.
The following table summarizes the results of these tests on the 125 children:
|
|
Medical Exam
|
|
|
Deficit
|
No Deficit
|
|
Written Test
|
No Deficit
|
6
|
87
|
|
Deficit
|
18
|
14
|
Evaluate and describe the appropriateness of using the questionnaire as an inexpensive substitute for the definitive but costly medical exam in a future epidemiologic study aimed at assessing the prevalence of hearing deficits in school children (aged 6 to 10).
6. Twenty four individuals were followed prospectively for up to 60 days to monitor the incidence of adverse reactions to a drug. The results were as follows:
|
Observation
|
Day
|
Outcome
|
Age (Y is <60 years; O is > 60 years)
|
|
1
|
3
|
Adverse reaction
|
O
|
|
2
|
4
|
Adverse reaction
|
Y
|
|
3
|
7
|
Dropped out of study
|
O
|
|
4
|
7
|
Dropped out of study
|
O
|
|
5
|
11
|
Adverse reaction
|
O
|
|
6
|
12
|
Adverse reaction
|
Y
|
|
7
|
14
|
Adverse reaction
|
O
|
|
8
|
14
|
Adverse reaction
|
Y
|
|
9
|
16
|
Adverse reaction
|
O
|
|
10
|
17
|
Dropped out of study
|
O
|
|
11
|
18
|
Dropped out of study
|
O
|
|
12
|
19
|
Adverse reaction
|
Y
|
|
13
|
19
|
Adverse reaction
|
O
|
|
14
|
21
|
Dropped out of study
|
O
|
|
15
|
34
|
Adverse reaction
|
Y
|
|
16
|
30
|
Adverse reaction
|
O
|
|
17
|
44
|
Adverse reaction
|
O
|
|
18
|
46
|
Adverse reaction
|
Y
|
|
19
|
51
|
Dropped out of study
|
O
|
|
20
|
54
|
Dropped out of study
|
Y
|
|
21
|
60
|
No reaction
|
Y
|
|
22
|
60
|
No reaction
|
Y
|
|
23
|
60
|
No reaction
|
Y
|
|
24
|
60
|
No reaction
|
O
|
a) Create a table which contains the components used in the product-limit estimate (Kaplan- Meier Survival estimate) of having an adverse drug reaction (ignoring age).
b) Calculate the product-limit estimate of the probability of having no reaction by the 30th day (ignoring age).
c) Briefly describe the strategy which you would use to compare survival without an adverse reaction according to age. Discuss any bias you may be concerned about and how it may be addressed.
7. Depression may be a risk factor for mortality. This condition is more common in young people who are, of course, less likely to die than older people. A study using data from Statistics Canada found an unadjusted hazard ratio of 1.0, suggesting no association between depression and mortality. However, with adjustment for age and sex, the hazard ratio doubled.
a) How can you explain this effect?
b) When adjusted for other variables that may predict mortality (such as smoking), the association between depression and mortality again disappeared. How would you explain this in terms of confounding?
8. The attached paper describes a single-group prospective inception cohort pilot study of markers associated with poor recovery after mild traumatic brain injury in older adults. In the concluding paragraph, three broad hypotheses for further study are listed. You've been hired to develop a CIHR research proposal for the fully powered cohort study. Identify and rationalize one specific research aim that could be hypothesized from these pilot results that you will address in your protocol and briefly identify the population you propose to study, the measures (exposure, outcome, and any important potential confounders or effect modifiers) you will use, and the analytic strategy you will use to address your specific research aim. Be sure your proposed analysis is capable of assessing and controlling for any potential confounding factors. Also, conclude with a final paragraph highlighting the strengths and weaknesses of your proposal. Please no more than 1 single-spaced page or 2 double-spaced pages to answer this question.