What factors go into the design of a cohort study?
Timing, selection of cohort and comparison groups, measurements of exposure, ensuring follow-up and measurements of outcomes.
When do you start a prospective cohort study? Historical/retrospective?
Prospective = start study prior to outcome. Retrospective/historical = look at data collected before the outcome and trace records leading up to the outcome that already occurred.
What are the pros and cons of a retrospective cohort study?
Pros = cheap and fast, efficient with long developing diseases. Cons = data may be inadequate and you're stuck with it.
What are the pros and once of a prospective cohort study?
Pros = better exposure and confounder data, less vulnerable to bias. Cons = expensive, lengthy and not efficient for long-developing diseases.
When might you do a special cohort study?
With rare exposures (occupation, dietary habits, lifestyle, medical treatments). You find a group of people with the particular exposure of interest.
When might you use a general cohort study?
With common exposures among the general population.
What is your ideal population for your study?
Unexposed are similar to exposed in all aspects except for the exposure, free of outcome at start of study, and had the opportunity for exposure of interest.
What are different ways you can choose possible comparison groups?
Internal comparison (Framingham, divide single cohort into exposed and unexposed). External comparison (Occupational, compare an entire exposed population to an entire unexposed population). Compare exposed cohort with general population data.
What is a draw-back to comparing your exposed group to the general population?
Healthy worker effect. You are comparing people who are healthy enough to work to a population that probably has more sick people in it to start with.
What type of study was Framingham?
Prospective, geographic, internal comparison of exposed and unexposed groups.
How were the exposures measured in the Framingham heart study?
Direct physical exams every 2 years, questionnaires and pre-existing record evaluation.
How were outcomes measured in the Framingham study?
Analysis of death certificates, health records, questionnaires, and medical exams for new coronary events (MI, angina pectoris, stroke and other CV disease).
What conditions must be met in order for a risk factor to become a cause?
Temporality, Consistency, Strength of Association, Dose-Response, Plausibility, Reversibility, Specificity and Analogy.
How can you adjust your prospective cohort study if you think of new exposures you want to include?
Start a new cohort or go back in time to look and see if they had the new exposure of interest (a nested case-control study within your cohort study).