EPI540_Midterm2_2024-1

pdf

School

University Of Georgia *

*We aren’t endorsed by this school

Course

MISC

Subject

Statistics

Date

Apr 3, 2024

Type

pdf

Pages

Uploaded by MagistrateJellyfish7163

EPI 540: Epidemiologic Methods II Midterm II - Spring 2024 Read all questions carefully – if a question has multiple parts, please be sure to answer each part. 1. You may refer to your course notes and textbooks while taking this exam; Students may NOT work with others in any capacity to complete the midterm exam. Students should not discuss the exam (conversation or any written communication) with anyone else during the period 7 AM and 11:59 PM EST on Monday, April 1, 2024. Honor code enforcement of this exam provision is in effect. 2. The 4-hour time limit begins as soon as you start the exam and will continue running even if you navigate away from the page. Canvas saves answers automatically every 7 seconds, which helps avoid losing your answers if your internet connection goes out. You can change your answers at any time and the new responses will be saved. Upon submission, Canvas saves everything that is included in the form. 3. For questions requiring calculations, show all of your work and round your answers to specified decimal places . Only round your final answer and not during intermediate steps. 4. The TAs will be monitoring questions submitted via email throughout the day. If you have a question during the exam, please see the table below and contact the appropriate TA based on time of day. Time TA Email 7am - 10am Courtney courtney.babb@emory.edu 10am - 1pm Tsion tarmidi@emory.edu 1pm - 3pm Kelsey kelsey.coy@emory.edu 3pm - 6pm Charis cwiltsh@emory.edu 6pm - 9pm Jeffery jeffery.osei@emory.edu 9pm - 12am Mohammed mohammed.siddiqui@emory.edu 5. As always, please remember that our classroom values of respect and academic integrity , along with the RSPH Student Honor and Conduct Code Links are in place in this course. Violations of the Student Honor and Conduct Code will be referred to the Associate Dean for Student Affairs for investigation.

1. Please type your name in the text box below to affirm your compliance with the RSPH Student Conduct and Honor Code. 2. OPTIONAL: If there were any questions where you felt that you needed to make an assumption in order to respond, please note those assumptions here. Please be sure that these are clearly labeled so that we know which questions they correspond to. The professors and TAs will do their best to take these into account.

Part I (19 points total) 3) True or False?(2 points each) Statement True False A positive interaction contrast means that positive- interdependence types are more prevalent than negative types in the study population. When the interaction contrast equals zero, there must be no person in the study population with negative interdependence or positive interdependence potential outcome types. Statistical tests of homogeneity should not be used alone to determine whether effect measure modification is present. Case-control studies provide measures of association that are inherently always more biased than cohort studies. The case-control odds ratio estimates the risk ratio when the disease is rare regardless of the control sampling strategy used. 4) Control Selection (1 point each) Validity Characteristic True False Controls should never have or get the outcome under study. The probability of being sampled as a control should not depend on exposure status. Ideally, had controls developed the outcome under study, then they would have been identified as a case in the study. Under risk set sampling of controls, all controls must be undiseased at the follow-up time when they are selected as controls. 5) Case-crossover designs (1 point each) True or False: This exposure-disease relation would be well-suited to study using a case- crossover design True False Changes in daily ambient temperature related to onset of acute stroke Gene variants related to onset of acute stroke Changes in daily ambient temperature related to onset of obesity, which is a chronic disease

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Use of mobile telephones while driving related to injury in automobile accident Use of mobile telephones while driving related to chronic back pain Part II (29 points total) In the Journal of the National Cancer Institute , Perneger examined the interaction between cigarette smoking and gender as causes of bladder cancer. The following table shows the data upon which the analysis was based. 6) Complete the shaded cells of this table. Round to one decimal. Express rate differences per 100,000 persons. (1 point each, 10 points total) Gender Men Women Smokers Ever Never Ever Never Cases 981 199 259 75 Person-years 3,340,000 1,660,000 2,755,000 2,245,000 Stratum-specific Rate ratio 1 reference 1 reference Common reference Rate Ratio 1 reference Stratum-specific Rate difference 0 reference 0 reference Common Reference Rate difference 0 reference 7) Is there effect measure modification on the ratio scale? Choose yes or no and support your choice with an explanation in 1-2 sentences. (2 points) 8) Is there effect measure modification on the difference scale? State yes or no and explain your reason in 1-2 sentences. (2 points) 9) Calculate the interaction contrast. Round your answer to one decimal place and report per 100,000 person-years. Show your work . (3 points) 10) Calculate the relative excess risk due to interaction (RERI). Show your work. Round to one decimal. (2 points)

11) Interpret the magnitude and direction of the RERI you calculated in 10. (3 points) 12) Given your calculation of the interaction contrast, and assuming it’s a valid estimate, which of the following statements is true for the study population (select one best answer): (2 points) ____ The proportion of negative interdependence types exceeds the proportion of positive interdependence types. ____ The proportion of positive interdependence types exceeds the proportion of negative interdependence types. ____ The proportion of positive interdependence types equals the proportion of negative interdependence types. ____ The balance of positive and negative interdependence types is unknown. 13) The original investigators interpreted these results as follows: “the risk of bladder cancer (from smoking) may be higher in women than in men.” The Associated Press echoed this opinion under the headline “Women smokers may face a higher risk of bladder c ancer than men.” In the space below, describe (1) how the investigators might have reached that conclusion, and (2) whether and why you agree or disagree. Be sure to consider measures of association (ratio and difference). (5 points)

Part III (19 points total) The chart below shows the follow-up experience of a cohort followed from time zero for up to ten years. Use the chart to answer the questions below. Note that the first ten persons in the chart are exposed and the second ten persons in the chart are unexposed. Assume that all events occur at the end of a year. For example, person 1 was lost to follow-up at the end of year 9, so accumulated 9 person-years of follow-up. 14) What is the rate difference for new disease occurrence, comparing the exposed with the unexposed over ten years? Round to one decimal place. Show your work . (2 points) 15) What is the rate ratio, comparing the exposed with the unexposed over ten years? Round to one decimal place. Show your work . (2 points)

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

16) Select "yes" or "no" for whether each of these 10 persons would be eligible to be selected as a control under each of the control sampling strategies (15 points total; 0.5 points per person per answer). Person number Case-cohort sampling Cumulative sampling Risk set sampling at the time person 7 becomes a case Person 1 Person 2 Person 3 Person 4 Person 5 Person 6 Person 7 Person 8 Person 9 Person 10

Part IV (14 points) The following table shows data from a hypothetical cohort observed for two years. Use the data in the table to answer the questions that follow. Exposed Unexposed Cases 40 25 Non-cases 60 75 Total 100 100 17) Complete the table below to show the expected result if a case-control study was nested in the cohort and (a) controls were sampled from the population at risk by case-cohort sampling, and (b) two controls were sampled for every observed case. Round to the nearest integer. Assume 100% of cases were enrolled. (1 point each; 4 points total) Exposed Unexposed Cases Controls 18) Compute the case-control odds ratio expected to be obtained from this case-control study? Show your work. Round to one decimal place. (2 points) 19) Complete the table below to show the expected result if a case-control study was nested in the cohort and (a) controls were sampled from the population at risk by cumulative sampling, and (b) two controls were sampled for every observed case. Round to the nearest integer. Assume 100% of cases were enrolled. (1 point each; 4 points total) Exposed Unexposed Cases Controls 20) Compute the case-control odds ratio expected to be obtained from this case-control study? Show your work. Round to one decimal place. (2 points)

21) Which sampling strategy led to a case-control odds ratio that is a more valid estimate of the risk ratio obtained from the full cohort? Choose only one. (2 points) _____ Cumulative sampling __ __ Case-cohort sampling _____ Neither sampling design _____ Both sampling designs Part V (12 points) 22) For each scenario, identify the best study design from the list below. (2 points each) A. Matched case-control design B. Case-crossover design C. Case series D. Quasi-experimental design A physician notices that three patients are diagnosed with pneumonia and all stayed at the same hotel at the same time. Persons who do not regularly exercise, and then have a sudden short period of exercise, have increased odds of triggering myocardial infarction in the next hour by 107-fold. A study of the relation between smoking and breast cancer among women carrying BRCA genetic variants discarded 1/3 of the cases from the analysis due to being unable to find matched controls. A volcanic eruption in Iceland was associated with a ~30% higher infant mortality rate as measured in 1784 parish registries from Sweden Childhood dehydration, as measured by hot and dry weather in the first year of birth, is associated with higher adult blood pressure. Part VI (8 points). Please mark the following questions (22-26) as either True or False. 23) Confounding is only controlled by the matching factor(s) in a cohort study if there are no competing risks or loss to follow-up. (2 points) 24) Matching in a cohort study controls for confounding but introduces a selection bias. (2 points) 25) In a cohort study, subjects should not contribute person ‐ time during periods when they are not at-risk of the outcome. (2 points)

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

26) A study subject in a cohort study for whom the outcome was identified during the induction period should be removed from your study. (2 points) 27) In cumulative sampling of controls, controls cannot also be cases. (2 points)

EPI540_Midterm2_2024-1

Related Documents