Intermediate Epidemiology

Requirements: short
OPH251 Midterm Exam Fall 2023
This exam is worth a total of 100 points (+12 bonus points). There are 6 questions total, some are worth more points; try to allocate your time accordingly. The BONUS questions are embedded in questions 4, 5, & 6. Answer those questions only if you have time or if you want to make up for a question you think is too hard.
This is an exam. Do not work with anyone else on this exam. Any cheating will result in failure of the exam and notification to the Tufts PH Academic Committee.
Good luck!
Question 1. (15 points)
In the DAG below, the exposure is “SUS” and the outcome is “EGC”. Answer the following questions using this DAG.
What is the minimal set of variables needed to obtain an unbiased estimate of the total effect of SUS on EGC?
If you are only interested in the direct effect, what variables need to be adjusted for to get an unbiased estimate of the direct effect of SUS on EGC? Provide the minimal set of variables (including any potential confounders).
Now, assume that you found out that enrollment of study participants only included those who were HOS=1. Could the results of the study be biased? Explain.

Question 2. (20 points) A study was conducted to determine the relationship between SES and asthma morbidity among African American/Black and Hispanic/Latinx adults with moderate to severe asthma using multidomain SES frameworks and mediation analyses.
What does c represent in the figure above:
Direct effect
Indirect effect
Total effect
Results from the study are provided in the figure below. Using the figure, was there a statistically significant direct effect for SES on poorly controlled asthma.
Calculate the indirect effect for high perceived stress (a mediator) using the beta coefficients provided in the figure.
Calculate the total effect using the direct effect and the indirect through high perceived stress only (as the other two mediating paths were not statistically significant).
Note: the red lines indicate statistically significant associations. Low SES status is a binary variable.

Question 3. (10 points) Use the DAG below to answer the following question.
Based on the DAG what is the minimal set of covariates needed to obtain the unbiased measure of association between diabetes mellitus and cardiac ischemia?
Which of the hazard ratio (you can think of these as rate ratios) in the table below provides the unbiased association between diabetes mellitus and cardiac ischemia? Explain your choice.
Question 4. (10 points). The Mediterranean Healthy Eating, Aging, and Lifestyle study is an observational study aiming to investigate the association between typical lifestyle and dietary habits of the Mediterranean area and non-communicable diseases. Between 2014 and 2015, a cohort of 2044 randomly selected individuals from Catania, in the south of Italy, was recruited for the study. Depressive symptoms were measured on 1572 participants using a depressive short form. Diet was measured with a detailed food frequency questionnaire (which asks about foods eaten over the past year). The associations between total and different types of dietary fats and depressive symptoms observed in this study are reported in Table 2.
What type of measurement error might you expect in the exposure? Explain.
In what direction would you expect the error you describe in (a) to impact the odd ratios in the table.
BONUS question. 5 points. What type of measurement error might you expect in the outcome? Explain.
Question 5. (15 points) Using the table below to answer the questions. The study took place over 5 years.
Calculate and interpret the cumulative incidence (risk) of heart failure (HF) in people with HIV+ and major depressive disorder.
Calculate and interpret the cumulative incidence (risk) of heart failure (HF) in people with HIV- and no major depressive disorder.
Calculate and interpret the cumulative incidence (risk) ratio for heart failure (HF) in people with HIV+/MDD compared to those with neither.
BONUS QUESTIONS (5 points)
Calculate and interpret the incidence rate ratio for heart failure (HF) in people with HIV+/MDD compared to those with neither (use the third column).
Is the incidence rate ratio (d) the same as the cumulative risk ratio in (c)?
Question 6. (30 points) Vitamin D deficiency is a risk factor for arterial hypertension, but randomized controlled trials showed mixed effects of vitamin D supplementation on blood pressure (BP). A Vitamin D Hypertension Trial was conducted in a single-center, double-blind, placebo-controlled study conducted from June 2011 to August 2014 at an endocrine outpatient clinic. 200 study participants with arterial hypertension and 25-hydroxyvitamin D levels below 30 ng/mL were enrolled. Study participants were randomized to receive either 2800 IU of vitamin D3 per day as oily drops (n=100) or placebo (n=100) for 8 weeks. Primary outcome measure was 24-hour systolic BP.
When you look at the Table 1 with baseline characteristics from the two arms (not provided) you notice that vitamin D was statistically higher (at baseline) in the treatment arm than in the control arm.
Explain why this is important and how might that impact your interpretation of the results.
Explain how the researchers might address this in the analysis.
Using the figure above, is there evidence of differential loss-to-follow up by treatment arm? Explain.
Using causal framework language, explain what conditions must apply for selection bias to occur in an RCT study.
Using the description in (d), do you think you need to be concerned about selection bias in this study (and why or why not).
The main finding of this study was summarized in the abstract: “The mean treatment effect (95% confidence interval) for 24-hour systolic BP was −0.4 (−2.8 to 1.9) mm Hg (P=0.712).”
Provide one reasons (in the context of this study specifically) and not already discussed above, why this finding may be different from the true underlying causal association.
BONUS QUESTION (2 points). Provide a SECOND reason (for question 6.f.).

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *