Public Health

Question 1

Screening test for lung cancer with increased 5 year survival in the group who were screened,
but still die at the same age as the non-screened group. What type of bias accounts for this?

Answer 1

Lead time bias

Question 2

Best method to reduce information bias?
1. Measure a 95% confidence interval
2. Measure the p value
3. Double-blinded study
4. Multicenter study
5. Placebo-controlled
6. Randomisation
7. Exclusion criteria
8. Logistic regression analysis

Answer 2

Double-blinded study

Question 3

Mammogram has a sensitivity of 0.8 and a specificity of 0.9. In a population of 1000, the
prevalence of breast cancer is 2%. How many false negatives will there be if this population is
screened?

Answer 3

4

Question 4

A girl who was contact traced with an STI. How to prevent SPREAD in the community of
Chlamydia (or something)
1. Contact tracing
2. Condoms
3. Counsel on safe sex
4. Treat with azithromycin

Answer 4

Contact tracing

Question 5

Study looking at hospital and census data for rheumatic fever, seeing what decile they were and
calculating ARF incidence rates
1. Cross sectional study
2. Every other type of study possible
3. Retrospective cohort
4. Cross-sectional
5. Case series
6. Ecological study
7. Nested case control

Answer 5

Retrospective cohort

Question 6

PPIs given as RCT for functional dyspepsia. Showed a statistically sig difference for crude RR. BUT
then adjusted for age and shit, RR no longer stat sig. What was the problem with crude RR?

Answer 6

Confounding (age etc are confounders)

Question 7

Cervical cancer screening - what is the sensitivity

 got cancer|no cancer

+ve 10 , 90
-ve 5 , 895

Answer 7

10/15 = 66.6%

Question 8

1000 patients, 10% actually had ankle fractures. Sensitivity of Ottawa ankle rules is 0.98. How
many of this population would this test have missed?

Answer 8

2

Question 9

Screening picks up more slowly growing cancers which are less aggressive leading to
higher 5 year survival rates. What is this type of bias?
1. Length bias
2. Lead time bias
3. Overdiagnosis

Answer 9

Length bias

Lead-time bias:
The earlier we diagnose a disease, the longer patients will appear to survive, when in fact we just started counting earlier.

Length-time bias:
Patients with slowly progressing diseases are more likely to be detected through screening

Question 10

89/100,000 got flu over winter. What is this measuring

Answer 10

Incidence as over time compared to prevalence at any point

Question 11

Case control study on breast cancer and exposed to pesticide spray. Whats good about case controls?

1. Good for rare outcomes
2. Good for rare exposures

Answer 11

Rare outcomes

Cohort is good for rare exposures as you select based on this

Question 12

Population has 6 in 100 absolute risk of ischaemic heart disease. Smokers have 50% greater relative risk of getting that disease. A drug reduces risk in the smoker group by one third.

What is the absolute risk for smokers developing the disease if they are taking the drug?

Answer 12

6

Question 13

5 year study done about cardiovascular events and aspirin, there is 1000 people in each
arm. What is the number needed to treat for aspirin:

    Aspirin , No aspirin MI      1     ,      2 No MI     99     ,  98

Answer 13

100