Reference no: EM132168269

HOMEWORK 1

Question 1. You have developed a new screening test for a new disease, Disease Z. The prevalence of Disease Z in a general population of 10,000 people is 5%. Of the people who actually have Disease Z, 98% screen positive for the disease. Of the people who actually do not have the disease, 6% screen positive.

a. Set up the 2x2 table that reflects the screening test performance.

b. Calculate the specificity of the test and interpret its meaning.

c. Calculate the positive predictive value (PPV) of this test and interpret its meaning.

d. You use this screening test in a clinic population of 10,000 people in which the prevalence of Disease Z is 15%. Assuming that the sensitivity and specificity do not change, set up the 2x2 table that reflects the screening test performance in this clinic population.

e. Calculate the PPV in this clinic population and compare it to the PPV when the general population was screened. If your result is different from what you calculated in Part C above, why is your result different? Refer specifically to the numbers in the table.

Question 2. What are three measures that are used to evaluate whether a screening program has been successful or effective in terms of health or disease? List and describe these three measures.

Question 3. What type of bias can arise due to different recruitment procedures of cases vs controls? Explain this bias in your own words.

Question 4. Describe and compare non-differential and differential misclassification in dichotomous variables. In other words, explain what happens to the study groups in a case-control study in terms of numbers of study subjects when there is non-differential misclassification vs differential misclassification.

Question 5. You are investigating the role of smoking in Disease X. It has been previously established that heavy alcohol use is a cause of Disease X. You collect data on alcohol use so that you can stratify your analysis to measure the association of smoking with Disease X by heavy alcohol use. Below are the two stratum-specific 2x2 tables by alcohol use.

Heavy alcohol use

	Dz+	Dz-
Smoking+	35	14
Smoking-	15	75

Non-heavy or no alcohol use

	Dz+	Dz-
Smoking+	10	18
Smoking-	40	93

a. What is the crude or overall odds ratio for the association between smoking and Disease X in this study? Create the 2x2 table based on the two stratum-specific tables above.

	Dz+	Dz-
Smoking+
Smoking-

b. What are the stratum-specific odds ratios for the association between smoking and Disease X in this study?

c. In the relationship of smoking with Disease X, is there evidence of confounding by heavy alcohol use? Why or why not? In your answer, refer to the crude association of smoking with Disease X and stratum-specific estimates of association.

HOMEWORK 2

1. You are an investigator performing a case-control study of the effect of safety goggles in preventing eye injuries in tool and die workers. The most severely injured workers are unable to take part in the study due to the severity of their injuries (among both goggle wearers and non-goggle wearers). The data you collect is presented below.

Goggle Use	Any Eye Injury		No Eye Injury
	Included	Excluded*
Goggles	21	7	79
No Goggles	30	10	23

*Excluded due to severity of eye injury

a. Create the appropriate 2x2 table and calculate the odds ratio for eye injury, excluding the most severely injured individuals.

b. Create the appropriate 2x2 table and calculate the odds ratio for any eye injury that you would get if you included the most severely injured individuals.

c. Did selection bias occur here? Why or why not?

2. A group of investigators want to look at the question of whether smokers are at increased risk for heart disease. One of the investigators notes that smokers are more likely to have yellow fingernails than non-smokers. The investigator suggests that the analyses be stratified by the presence of yellow fingernails to avoid confounding by this variable.

Below are the data for the overall cohort and stratified by the presence of yellow fingernails.

Total Cohort
Smoker	Heart Disease	No Heart Disease
Yes	24	76
No	10	90

Yellow fingernails
Smoker	Heart Disease	No Heart Disease
Yes	12	38
No	2	18

No Yellow fingernails
Smoker	Heart Disease	No Heart Disease
Yes	12	38
No	8	72

For yellow fingernails to potentially confound the relationship between smoking and heart disease, yellow fingernails must be associated with the exposure (smoking) and be a risk factor for the outcome (heart disease).

a. Are yellow fingernails associated with smoking (exposure)? Using the data above, create a 2x2 table for these two variables and calculate the relative risk to determine whether yellow fingernails are really associated with smoking. (Hint: Must remove heart disease to look only at the association between yellow fingernails and smoking.)

b. Are yellow fingernails associated with heart disease (outcome)? Create a 2x2 table for these two variables and calculate the relative risk to determine whether yellow fingernails are really associated with heart disease. (Hint: Must remove smoking to look only at the association between yellow fingernails and heart disease.)

c. Calculate the relative risks for smoking and heart disease in the overall cohort and stratified by the presence of yellow fingernails (2x2 tables provided above).

d. Are the yellow-finger-stratum-specific estimates of relative risk any different from the relative risk for the cohort as a whole?

e. Do yellow fingernails confound the relationship between smoking and heart disease? Why or why not?

3. Individuals with Parkinson's disease (PD) have difficulty with movement, and may be at increased risk for falling due to their disease. People with PD are also often depressed, and risk of falling may be further increased by the use of antidepressant medications. Hip fractures are a much-feared consequence of falls in older individuals, and individuals who sustain fractures frequently die as a consequence of surgery or prolonged immobilization. You assemble the data below in a prospective cohort study that seeks to evaluate risk factors for hip fracture in older individuals:

Total Cohort
PD	Hip fracture	No hip fracture
Yes	23	401
No	10	565

Antidepressant use
PD	Hip fracture	No hip fracture
Yes	19	220
No	6	160

No antidepressant use
PD	Hip fracture	No hip fracture
Yes	4	181
No	4	405

a. Are PD patients more likely to use antidepressants than those without PD?

b. Is antidepressant use associated with an increased risk of hip fracture?

c. Calculate the relative risks for Parkinson's disease and hip fractures in the overall cohort and stratified by antidepressant use.

d. Are the antidepressant-stratum-specific estimates of relative risk different from the relative risk for the cohort as a whole?

e. Does antidepressant use confound the relationship between Parkinson's Disease and hip fractures? Why or why not?