Epidemiology Multiple Questions Assignment

QUESTION 4: Circle all the descriptions that apply to a confounding variable:
a) Can always be eliminated through careful study design and analysis
b) Is caused by the exposure
c) Is associated with the exposure
d) Is associated with the outcome
e) Can be an effect modifier

QUESTION 5: Circle all of the descriptions that apply to an effect modifying variable:
a. Should be eliminated through careful study design and analysis
b. Is caused by the exposure
c. Is caused by the outcome
d. Impacts the exposure/outcome relationship
e. Can be identified through stratification

QUESTION 6: Pharmacoepidemiology studies can best be described as (choose one):
a. Case-control studies
b. Phase I studies
c. Phase IV studies
d. Meta Analyses

QUESTION 7: True or false. Write T or F.
e. _________ The confidence interval provides information about the precision of an estimate.
f. _________ Administrative data, such as hospital discharge information, are collected for research purposes.
g. ________ Differences in health states related to genetic differences in racial groups are considered health inequities.
h. ________ In constructing an epidemic curve, public health officials will consider both the incidence and prevalence of a disease.
i. ________ Epidemiology is the study of the distributions and determinants of health.
j. ________ An advantage to using the direct method to generate age-adjusted rates is that you can compare rates of several populations directly with each other.
k. ________ If two populations have the same age-specific rates, then their age-adjusted rates will also be the same.
l. ________ Blinding reduces observation bias since outcome determination is not influenced by knowledge of exposure status.
m. ________ If a variable is thought to be an effect modifier, then it is best to try and match on that variable during the participant selection.
n. ________ According to Rothman’s causal pie models, a sufficient cause of disease is a characteristic that must always be present in order for disease to occur.

QUESTION 8: In the “Online social integration is associated with reduced mortality risk” paper by Hobbs et al, the researchers report that the risk of mortality due to diabetes among Facebook users compared to nonusers is 0.72 (95% Confidence Interval: 0.63-0.82). Circle all of the appropriate interpretations:
a. There is a 95% chance that there is a true difference in diabetes mortality risk among Facebook users compared to nonusers.
b. In numerous repetitions of the study, the true risk ratio of diabetes mortality would be between 0.63 and 0.82 95% of the time.
c. There is a 5% chance the risk ratio of 0.72 or lower could have been observed by chance alone.
d. There is a 5% chance that the null hypothesis is true.
e. There is not sufficient power to detect a difference in diabetes mortality risk among Facebook users compared to nonusers.

QUESTION 9: The following hypothetical data are based on the Medicare.gov Hospital Compare website and show three acute care hospitals located in Lexington.
Readmissions for stroke patients Hospital-acquired pneumonia Blood stream infections
Baptist Health 0.40 1.25 1.55
University of Kentucky 1.76 0.86 2.10
Saint Joseph 0.97 0.80 1.25

a. Circle all of the statements below pertaining to the data in the table that are correct.
a. These data provide information on patient safety and quality of care.
b. Hospitals self-report the data used to generate these reports
c. The numbers shown in each of the cells are rates.
d. The numbers shown in each of the cells are ratios.
e. Blood stream infections at all three facilities are higher than the national average.
f. Readmissions for stroke patients at Baptist Health are approximately 40% lower than the national average.
g. The rate of hospital-acquired pneumonia at Baptist Health is 1.45 times that of the University of Kentucky
h. Reported blood stream infections at UK are approximately 110% higher than the national average.

QUESTION 10: UK Healthcare is interested in knowing whether providing Fitbits to their HMO enrollees will result in lower BMIs. In a pilot project, 300 adults were enrolled in the study and received a physical exam and interviewed about their physical activity levels. All enrollees met the following criteria: 1) Aged 35-65; 2) Able to walk for 15 minutes unassisted and without pain; and 3) Not currently using any sort of exercise or fitness tracker but are willing to do so. Half of them were randomly assigned to receive a free Fitbit and instructions for use. One year later the physical exam and interviews were repeated.
A. The researchers found there was no difference in average BMI one year later between the two groups. What could be a possible explanations for this? (circle all that apply)
a. Fitbits do not lead to a reduction in BMI
b. The study did not have sufficient power to detect the reduction in BMI due to Fitbits
c. The sample size was too small to detect a difference in average BMI due to Fitbits
d. Researchers did not appropriate adjust for potential confounders in the Fitbit/BMI relationship
e. Misclassification biased the results towards the null

B. Baseline physical activity is associated both with lower BMI and increased future physical activity. Baseline physical activity levels were assessed through a participant survey. If all participants over-reported their baseline physical activity, this would result in what?
a. Differential misclassification and bias results towards the null
b. Differential misclassification and bias results away from the null
c. Nondifferential misclassification and bias results towards the null
d. Nondifferential misclassification and bias results away from the null

C. The researchers compared self-reported Fitbit usage to what was actually recorded by the device and found that those with the least amount of physical activity had over-reported their usage. How would this impact the results?
a. Differential misclassification and bias results towards the null
b. Differential misclassification and bias results away from the null
c. Nondifferential misclassification and bias results towards the null
d. Nondifferential misclassification and bias results away from the null

D. If only participants who already have healthy BMIs at baseline actually used their Fitbits, how might this impact the results?
a. Bias results towards the null
b. Bias results away from the null
c. This would have no impact on the results

QUESTION 11: A cohort study of employees in a large retail store was used to investigate whether cashiers have detectable levels of the chemical bisphenol A (BPA) in their blood due to handling cash register receipts compared to stock workers who do not handle receipts. The hypothetical data from that study are as follows:
All employees
BPA No BPA
Cashiers 60 40
Stockers 55 105

a. Calculate the risk ratio of BPA for cashiers compared to stockers, and provide an interpretation of the result.

b. BPA can also be found in plastic bottles and cans of soda, therefore the researchers stratified their data based on beverage consumption
Regularly drink out plastic bottle or can Do not drink out of plastic bottle or can
BPA No BPA BPA No BPA
Cashiers 50 20 Cashiers 10 20
Stockers 50 95 Stockers 5 10
Calculate the appropriate measures of association for the stratified data.

C. Based on these findings, what can you conclude about the relationship between beverage consumption, job type, and BPA? (select all that apply)
a. Beverage consumption is a positive confounder in the relationship between job type and BPA
b. Beverage consumption is a negative confounder in the relationship between job type and BPA
c. Beverage consumption is an effect modifier in the relationship between job type and BPA
d. There is no association between job type and BPA

D. Which estimate(s) should be reported?
a. The crude RR
b. The stratum-specific RR
c. The RR adjusted for beverage consumption

QUESTION 12: A study of mother-to-infant transmission of HIV during breastfeeding was conducted in Cameroon. Researchers recruited a group of 500 mother-infant pairs who were breastfed and 500 mother-infant pairs who were formula-fed. All mothers were known to be infected with HIV. Two years later, the children were tested for HIV. 170 of the breastfed infants had seroconverted (were HIV positive), while 110 formula-fed had seroconverted. The 500 breastfed infants contributed 900 person-years of observation and the 500 formula-fed infants contributed 600 person-years of observation.
Infant HIV positive Infant HIV negative Total # infants Total Person-Years
Breast-fed
Formula-fed
Total

A. What kind of study is this?
a. Retrospective cohort
b. Prospective cohort
c. Ambidirectional cohort
d. Case-control

B. Fill in the above table.

C. What is the risk of HIV transmission among each the breastfed and formula-fed infants measured in cumulative incidence?

D. What is the rate of HIV transmission among each the breastfed and formula-fed infants measured in incidence density?

E. Calculate and interpret the risk ratio.

F. Calculate and interpret the incidence density ratio (aka rate ratio).

G. In this study, which is the best estimate of the true association between breastfeeding and HIV transmission?
a. Risk Ratio
b. Incidence Density Ratio (Rate Ratio)
c. Odds Ratio

H. Assume that only the risk ratio (from e) was reported, but that the incidence density ratio (from f) is the “correct” measure of association between breastfeeding and HIV transmission. Which type of error would this be?
a. Type I
b. Type II
c. Neither
d. Both

I. What proportion of the risk of HIV transmission among breastfed babies would we attribute to being breastfed from each of these measures (risk and incidence density)? Provide an interpretation of each [Calculate and interpret the attributable risk for both the risk and incidence density measures]

J. Assuming that only the risk ratio was reported, what would be the estimated Population Attributable Risk % (PAR%) of HIV transmission attributable to breastfeeding if 75% of infants in Cameroon are breastfed? How do we interpret this PAR?

QUESTION 13:
There are many types of rapid point-of-care diagnostic tests available for Hepatitis C virus (HCV) infection. Two tests are described below:
1. Noninvasive saliva test of HCV antibodies which has a sensitivity of 90% and specificity of 94%
2. Rapid finger stick test of HCV antibodies which has a sensitivity of 95% and specificity of 99%

A. If you wanted to ensure that you have the fewest number of false positive HCV tests, which rapid testing protocol would you recommend?
a. Saliva test only
b. Finger stick test only
c. Sequential testing with both
d. Simultaneous testing with both

B. CDC recommends that all “Baby Boomers” be tested for HCV. 10,000 “Baby Boomers” within an HMO are administered both of these point-of-care tests. The prevalence of HCV is estimated to be 10% in this population. Complete the tables below for the saliva and finger stick tests, assuming simultaneous testing.
HCV No HCV
Saliva test positive
Saliva test negative

HCV No HCV
Finger stick test positive
Finger stick test negative

C. What is the net sensitivity and net specificity of the simultaneous testing?

D. How would the use of these tests in a different HMO setting that has a baseline prevalence of HCV of 30% be affected? (circle all that apply)
a. The positive predictive value of both tests would be higher
b. The negative predictive value of both tests would be higher
c. The sensitivity of the saliva test would be higher
d. The specificity of both tests would be higher

QUESTION 14: Due to a recent increase in distracted pedestrian-related car accidents, city officials started a campaign to paint sidewalks to remind pedestrians to be aware of their surroundings at busy intersections. They then compared the occurrences of pedestrian-related car accidents in the 6 months before and after the campaign. Circle all the statements below that are true that pertain to this study.
a. The appropriate measure of association in this study is the cumulative incidence ratio
b. The appropriate measure of association in this study is the incidence density ratio
c. Due to this study design, the researchers do not need to control for any potential confounding variables
d. This is an ecologic study design
e. This is a randomized-controlled trial
f. None of these statements are true

QUESTION 15: Assume that taking vitamin supplements decreases the risk of breast cancer. Will the following variables exaggerate, reduce, or leave unchanged the measures of association (relative risk or odds ratio)?
b. Exercisers take more vitamin supplements and regular exercise decreases the risk of breast
cancer
c. Women with a family history of breast cancer take more vitamin supplements but they have an increased risk of the disease
d. Women who are obese take less vitamin supplements and have a higher risk of breast cancer.
e. Women who drink alcohol take neither more nor less vitamins but have an increased risk of breast cancer.