5. Cohort Studies Flashcards
What is ‘person-years’?
The sum of total time of everybody followed up in a study.
What is a requirement of the individuals recruited for cohort studies?
They have to be outcome-free.
How are people classified in a cohort study?
Exposed and unexposed.
How can the different exposure groups of a cohort study be compared?
By calculating an incidence rate ratio.
IRR = IR of exposed / IR of unexposed.
What are the advantages of a cohort study compared to routinely available data?
You can study exposures that aren’t routinely collected. More detailed information may be obtained on outcomes. More data can be collected about potential confounding factors.
What is the difference between concurrent and prospective cohort studies?
Concurrent studies start immediately whereas prospective cohort studies delay the starting date to ensure those involved in the trial are still outcome free.
What are historical/ retrospective cohort studies?
They recruit outcome free individuals and still follow them but they also use historical studies to get more data.
What would be calculated for an internal comparison and what would be calculated for an external comparison of cohort groups?
An incidence rate ratio compares internal sub groups and standardised mortality relates can be calculates for an external comparison, to the reference population.
What are the limitations of external comparisons in cohort studies?
There may be limited data available for the reference population, there may be no incidence data only mortality data, the reference population and study population may be picked under selection bias.
What are the advantages of cohort studies?
They’re detailed enough so a range of different outcomes and rare exposures can be studied.
What are the disadvantages of cohort studies?
They’re large and expensive, they take a long time to complete (although historical cohorts take a shorter amount of time), high risk of losses to follow-up which may leave a survivor bias, can’t always be used for rare outcomes and there is a difficulty with confounding.
What two groups would the ideal study look at?
Two groups that are identical apart for the exposure being investigated, comparing like with like.
