Scenario 1

Research question: What is the prevalence of bulimia in adolescent girls in the Northern Territory?

Study design and methods: Cross sectional study. Participants recruited from waiting rooms of General Practice surgeries in Darwin. Information collected via a face to face interview.

1 What are the strengths and weaknesses of a cross sectional study?
2 Do you think a cross sectional design is the best type of study to answer this research question? Explain your answer
3 What are 2 possible selection biases you need to consider in this study? Explain your answer
4 How will you try to minimise these biases?
5 Do you need to consider confounding in this study? Explain your answer

Scenario 2

Research question: Is socioeconomic status associated with the risk of childhood injury?

Study design and methods: Cross sectional study. Participants identified via signs advertising the study at a community immunisation centre. Information collected via internet based survey.
1 Do you think a cross sectional design is the best type of study to answer this research question? Explain your answer
2 What is 1 possible selection bias you need to consider in this study? Explain your answer What is 1 possible measurement bias that you need to consider in this study? Explain your answer
3 Other than age and sex, what is possible confounder that you need to consider in this study? Explain your answer
4 How will you try to minimise these biases and confounders?

Scenario 3

Research question: Does referral to a drug and alcohol support worker decrease subsequent emergency department presentations related to intoxication?
Study design and methods: Randomised control trial of intoxicated patients presenting to a large emergency department. Patients will be randomized and either receive a referral to a drug and alcohol support worker (in addition to any other required treatment) or no referral.
1 What is the study factor and outcome factor in this study?
2 Do you think a randomised control trial is the best type of study to answer this research question? Explain your answer
3 How does the design of a randomised control trial minimise confounding?

PART 2: Calculating and interpreting epidemiological measures:

Read through each scenario and answer the related questions. For calculations show all workings and round your answers to 1 decimal point.

6. A cohort study was conducted to investigate the relationship between completion of higher education and IV drug use in a population based sample of adults. The cohort was 3000 people, of whom 2000 had completed higher education. Of those who had completed higher education, 154 had used IV drugs, while among those who had not completed higher education, 111 had used IV drugs.

Set up a two by two table and calculate the following measures for the relationship between higher education and IV drug use.

- Relative risk
- Percent attributable risk
- Percent population attributable risk

Interpret all of your findings in your own words

7. A case control study was carried out to investigate the risk of brain cancer in children who had grown up within a 10 kilometre radius of a power station. Fifty two children who had been diagnosed with brain cancer were enrolled, and 107 children who had not been diagnosed with brain cancer (controls) were enrolled. Fifteen of the cases had grown up within a 10 kilometre radius of a power station, while 32 of the controls had grown up within a 10 kilometre radius of a power station.

Set up a two by two table and calculate the odds ratio for living near a power station and brain cancer. Interpret the findings in your own words.

Part 3: Choose your own research question and design

Think of a question that interests you in public health, that you could investigate via an analytical epidemiological study. State your research question and then:

1 State the study design you would choose and justify why you would choose this type of study design

2 State the measure of association that you would use, and justify why you would choose this

3 List 2 potential selection biases and how you would attempt to control for these

4 List 1 potential measurement bias and how you would attempt to control for this

5 List 2 potential confounders and how you attempt to minimise/control for these