Lecture 6; Introduction To Cohort Study Designs Flashcards
Remember you must treat both groups the same when you carry out a cohort study
.
How do you carry out a cohort study?
Get two groups; those who are exposed and those who are unexposed and then wait to see an outcome.
(Compare and contrast exposed and unexposed for potential outcomes)
This is a concurrent or prospective style cohort study
What is the difference between a retrospective cohort study and a prospective cohort study?
Retrospective (historical) cohort study- find exposed and find unexposed, from a while ago (ie the exposure was in the past) and you follow them for potential outcomes
- involve collecting follow-up data from the past. (Follow-up data starts in the past but doesn’t have to end in the present time, you can continue to follow the groups)
- recruit outcome free individuals and classify their exposure status and subsequent outcome status using historical records.
- follow-up -> count p-y and d using historical records
Prospective cohort study- find exposed and unexposed (those exposed now) and follow them for the potential outcomes
What would be an ideal study?
Basic method of comparing ‘like with like’
Two identical groups differing only in the exposure of interest
Differences can then reasonably be attributed to that exposure
Why cant the ‘ideal study’ ever be carried out?
It is impossible to find two identical groups of people differing only in exposure of interest. The exposure may be linked to other factors
Ie smoking is linked to alcohol consumption, occupation, ethnicity, socio-economic group, exercise levels etc
Seeing as an ‘ideal study’ isn’t possible, what is the 2 next best things?
1)Using randomisation
Eg a randomised controlled trial
2) cohort study (measure and record any non-identical features)
What do you need to check for when finding people for a cohort study?
- the recruits must be outcome-free
- classified into exposed and unexposed groups
What is the null hypothesis for a ratio?
1
Because you divide one number by the same number?*
What are the advantages of a cohort study?
- can study exposures and personal characteristics that are not routinely collected
- can obtain more detailed information on outcomes and exposures
- can collect additional data on potential confounding factors
- enable details and prospective assessment of exposure, outcomes and confounders (historical cohorts are not as good for this)
Why would you start a study and then wait some time before actually performing the follow-up?
To establish that those in the ‘unexposed’ group definitely don’t have the outcome
What do all cohort studies have in common?
They all involve prospective follow-up;
- recruit outcome-free individuals and classify them according to their exposure status
- follow-up -> count p-y and d
In what ways are cohort studies better than case-control studies?
- better at studying a range of different outcomes
- better at studying a rare exposure
- better air establishing that exposure(s) precede outcome(s)
- better for conditions that fluctuate with age- randomly or systematically
What are the disadvantages of cohort studies?
- usually large and resource intensive
- take a long time (historical cohorts takes less time)
- rigorous definitions of outcome and exposure can require experience and sometimes intensive/invasive investigation
- risk of high number of losses to follow up -> survivor bias (those who remain in study differ from those that left)
- results take a long time-> potential ethical dilemmas and can become politically charged
- not good for rare outcomes (would get too few cases)
- difficulty with confounding- esp. unknown confounders
What ratios can you calculate in a cohort study?
Odds
And incident rate
What is the odds ratio a measurement of? How does this differ from the IRR
Pure measurement of association
IRR includes time of exposure (which is why you can only calculate this when doing a cohort study)
