biases might occur in this study

Question

A study was conducted. 4,000 adult males working for a large University in any capacity( professors, custodians, etc.) are asked in 1995 if they are smokers. Twenty years later, the researchers look to see if any of the male employees developed lung cancer between 2000-2015.

1. Define the exposure and the outcome.
2. What is the research question? (Neutral!)
3. Is this an experimental or observational study? Explain how you know.
4. If observational, what type of study design is this and how do you know?
5. What biases might occur in this study? How might you avoid them?
6. What results might you expect?
7. If your results show an association between the exposure and the outcome, what other factors do you think need to be taken into consideration before you share these results?
8. What type of study would you suggest next to find stronger and/or additional evidence?

Answer 1

1. Define the exposure and the outcome.

exposure to smoking. Outcome is lung cancer or no lung cancer

2. What is the research question? (Neutral!)

Does smoking increase the risk of lung cancer?

3. Is this an experimental or observational study? Explain how you know.

Observational. It is not experimental because no sort of treatment condition was applied to at least some participants by random assignment.

4. If observational, what type of study design is this and how do you know?

Cohort. The researchers followed a group of people with a common characteristic over time to find how many reach a certain health outcome of interest

5. What biases might occur in this study? How might you avoid them?

6. By selecting this cohort, the study will be biased to this particular demographic.

We should expect that "4,000 adult males working for a large University in any capacity( professors, custodians, etc.)" to show some noticeable differences compared to a random sample of the full adult population, or even a random sample of adults in the city the study was conducted. For example, university professors are, by definition, highly educated and most have a high socio-economic status. Finally, this study is limited to males, even though the US population has slightly more females.

7. What results might you expect?

We would expect a correlation between smokers and lung cancer diagnosis, or exposure and outcome.

8. If your results show an association between the exposure and the outcome, what other factors do you think need to be taken into consideration before you share these results?

We need to consider other carcinogenic risks that this population might be particularly exposed. For example, if the university had used asbestos in its walls or ceiling, then this sample might have been unusually exposed to another known cause of lung cancer.

9. What type of study would you suggest next to find stronger and/or additional evidence?



A case-control study could be better depending on the sample size and selection. A larger cohort study could also improve the results, especially if the sample was more representative of the general population.