Reference no: EM13100415

Part 1

Identify the Role of the Third Variable

For questions 1-5, choose which of the following best describes the third variable (i.e. the variable other than the exposure and the outcome) based on the study results and/or description of the researchers' theory.

A. confounder

B. mediator

C. effect measure modifier

1. Using data from a case-control study, researchers evaluate the association between a history of obesity as measured by BMI and a type of esophageal cancer (adenocarcinoma). They find that obese men have 2.6 times the odds of developing esophageal adenocarcinoma compared to non-obese men. They then hypothesize that obesity causes an increased risk of gastric reflux, which is a risk factor for esophageal adenocarcinoma. Once they adjust for gastric reflux, they find that obese men have 1.3 times the odds of developing esophageal adenocarcinoma compared to non-obese men. In this scenario, what is the third variable, gastric reflux, likely to be?

2. Researchers conduct a case control study to look at the association between cell phone usage (exposure) and brain cancer (outcome). They find, overall, people who use a cell phone regularly have 2 times the odds of having brain cancer compared to people who do not use a cell phone regularly. Once they adjust for age, they find that people who use a cell phone regularly have 3 times the odds of having brain cancer compared to people who do not use a cell phone regularly. In this scenario, what is the third variable, age, likely to be?

3. Researchers conduct a cohort study in Southern Africa to look at the association between attending school and HIV infection in youth. Overall, they find a slightly protective relationship between attending school and HIV risk (RR=0.95). However, after stratifying on gender, they find that attending school for women is strongly protective (RR=0.6) against the risk of HIV, whereas the risk of HIV is slightly increased among men who attend school (RR=1.3). In this scenario, what is the third variable, gender, likely to be?

4. Using data from the Framingham cohort study, researchers evaluate the association between obesity as measured by BMI (exposure) and breast cancer (outcome). They find that obese women have 3.1 times the rate of breast cancer compared to non-obese women. They hypothesize that obesity causes increased levels of estradiol in the blood, thereby increasing the risk of breast cancer. Once they adjust for estradiol level, they find that obese women have 1.9 times the rate of breast cancer compared to non-obese women. In this scenario, what is the third variable, estradiol level, likely to be?

5. Researchers implement a cohort study to look at the association between vitamin D supplementation (exposure) and colon cancer (outcome). They find that people who take daily vitamin D supplementation have half the risk of getting colon cancer compared to people who do not take daily vitamin D supplementation (RR=0.5). When they stratify by how much people exercise, they find no effect of vitamin D supplementation on colon cancer among people who exercise regularly (RR=1) or among people who do not exercise regularly (RR=1). In this scenario, what is the third variable, exercise level, likely to be?

Identify the Type of Bias or Error

A. Selection bias

B. Information bias

C. Random sampling error / lack of precision

D. Confounding

E. Lack of generalizability/external validity in conclusion

6. A randomized controlled trial was implemented to assess whether a human papillomavirus (HPV) vaccine prevents pre-cursors of cervical cancer in women between the ages of 15 and 25 years. HPV is a sexually transmitted infection that is known to cause cervical cancer. The researchers found an incident rate ratio (IRR) of 0.10 (95% CI 0.04, 0.76) comparing precursors of cervical cancer among vaccinated women to women who were not vaccinated. Based on these results, what type of error/bias should the researchers be concerned about?

7. Based on the findings in question 6, the researchers recommend vaccinating girls ages 11-14 years in order to vaccinate girls prior to sexual initiation. The US Food and Drug Administration (FDA) approves the vaccine for girls/women aged 15 years and over, but does not recommend its use in girls under age 15 without additional RCTs that have enrolled younger girls. In this scenario, what type of error/bias is the US FDA concerned about?

8. Researchers implement a case control study to assess whether a diet high in fat content causes breast cancer. Diet is measured by self-report. What type of error/bias should the researchers be concerned about based on what you know about the study so far?

9. For the study in question 8, the researchers selected all cases of breast cancer that were diagnosed at Memorial Sloan Kettering and New York Presbyterian Hospital in 2010. They selected controls from the same hospitals that were diagnosed with gastro-intestinal disorders. What type of error/bias should the researchers be concerned about based on the new information described in this question?

10. These same researchers from question 8 know from a previous study that people who eat high fat diets generally lead less healthy lifestyles then people who eat low fat diets. What type of error/bias should the researchers be concerned about based on this new information?

Part 2. 

Please answer the following 6 questions using full sentences. Carefully show all calculations and interpret results. Be sure to answer all parts of each question for full credit.

11. What does herd immunity mean? How is herd immunity achieved? Please be sure to use the basic reproductive number (R0) in your explanation of how herd immunity is achieved.

12. You decide to implement a case control study that evaluates the association between smoking marijuana and lung cancer. You are worried that the results from your observational study will be confounded by age.

a. Name and describe one way you could attempt to control for confounding by age during the analysis phase. 

b. Name and describe one way you could attempt to control for confounding by age during the study design phase.

13. In a case control study of the association between illegal drug use and the diagnosis of a sexually transmitted infection (STI), you are concerned that cases and controls will be equally unlikely to truthfully report their drug use.

a. What type of misclassification is this? Be sure to state whether it is differential or non-differential, and whether it is misclassification of the exposure or the disease in your response. Justify your response.

b. How would this misclassification bias the OR you calculate using your study findings? Be sure to state whether you think it would bias the OR toward or away from the null value of 1 in your response.

Question 14: You conducted a cohort study of the association between maternal consumption of water contaminated with manganese during pregnancy, and pediatric impaired motor development at age 4. Manganese is a metal often found in water from very deep wells. You find that the crude risk ratio is 2.0. You are concerned that people who can afford to have deep wells dug (those exposed) may be wealthier, and therefore they may have better nutrition during pregnancy. Poor prenatal nutrition has been implicated as a risk factor for numerous neurological diseases. You stratify your study results on prenatal nutrition as seen in the table below:

	Good Nutrition				Poor Nutrition
Manganese exposure	Impaired Motor Development (D+)	Healthy (D-)	Total		Impaired Motor Development (D+)	Healthy (D-)	Total
Yes (E+)	10	43	53		111	97	208
No (E-)	35	175	210		50	355	405
Total	45	218	263		161	452	613

14a. Calculate the stratum-specific risk ratios for the association between maternal exposure to manganese and pediatric impaired motor development at age 4.

14b. Based on these results alone, does nutrition appear to be a confounder, mediator, or an effect measure modifier? Why? 

Question 15: The following are results from a study to assess the performance of a new rapid HIV test compared to the gold standard ELISA test for HIV.

Result based on new rapid HIV test	ELISA test (gold standard)		Total
	HIV+	HIV-
HIV+	304	523	827
HIV-	52	978	1030
Total	356	1501	1857

15a. Calculate and interpret the sensitivity of the new test.

15b. Calculate and interpret the specificity of the new test. 

15c. Calculate and interpret the positive predictive value of the new test.

15d. Calculate and interpret the negative predictive value of the new test. 

16. A case control study was conducted to determine if screening for early detection of prostate cancer increases survival time. Upon diagnosis of prostate cancer, patients were asked about past screening history and were followed to determine survival time (time between diagnosis and death). The study found that men who had been screened in the past lived an average of 5 years longer than men who had not been screened in the past. Explain one possible explanation for this study result assuming that screening did nothing to increase survival time. 

17A. What is the difference between a mediator and a confounder? Be sure to give an example of each in your response.

17B. How can you tell the difference between a mediator and a confounder based on what you observe in your data? 

18. You are sent to investigate an outbreak of gastrointestinal disorder among Hunter College MPH students. In interviewing the 15 students experiencing symptoms, you discover that all had attended the holiday party which included a buffet. You draw the below histogram based on how many days after the holiday party they attended the party.

Based on the above histogram what is the:

18a. minimum incubation period? 

18b. maximum incubation period? 

18c. median incubation period? 

You get the list of all students who attended the holiday party and interview them about what they ate from the buffet. Below is the table with the data you collect. 

	Sick (D+)	Not Sick (D-)	Attack rate	Attack rate ratio
Ate devilled eggs (E+)	8	15
Did not eat devilled eggs (E-)	7	15
Ate sour cream dip (E+)	12	25
Did not eat sour cream dip (E-)	3	5
Ate chocolate mousse (E+)	12	10
Did not eat chocolate mousse (E-)	3	15

19a. Calculate the food specific attack rates and attack rate ratios for devilled eggs, sour cream dip and the chocolate mousse.

19b. What food do you think was the likely culprit of the gastrointestinal disorders? Justify your response with data.  

Part 3. SPSS

Use SPSS to answer the following four questions. Each question is worth 5 points. You will need to use the database called "hw3data.sav" posted on Blackboard to answer these questions. Be sure to paste the appropriate output table into your response in addition to expressing your response using complete sentences.

Background

You conduct a case-control study, among 80 participants, to determine whether excessive exposure to UV irradiation increases one's risk of skin cancer. You enroll 39 participants with a skin cancer diagnosis and 41 controls (never been diagnosed with skin cancer). But you are concerned that people with xeroderma pigmentosum (a hereditary defect in the enzyme system that repairs UV-induced damage to DNA) may be at an even greater risk of getting skin cancer than those without this defect because the condition is characterized by pigmentary changes, premature skin aging, and greatly increased susceptibility to malignant tumor development. Therefore, based on what you read in the literature, you think xeroderma pigmentosum may be an effect modifier in the relationship between UV irradiation exposure and skin cancer.

20. What is the crude OR and 95% CIs for the association between UV irradiation exposure and skin cancer. How do you interpret this estimate? 

21. What are the stratum-specific ORs and the 95% CI when stratified on xeroderma pigmentosum? How do you interpret these estimates?

22. Do your study results support what you learned in the literature that xeroderma pigmentosum is an effect measure modifier in the relationship between UV irradiation exposure and skin cancer? Describe the procedure you followed to reach your conclusion. Be sure to use the data results to fully support your statements.

23a. Given your findings in questions 20-22, which estimate(s) would you report and why?

23b. Why might your findings be similar/different from what you expected from the literature?

Download:- data.sav