Epidemiological Study Design I Flashcards
What is a cross-sectional study?
. Gather all information at same point in time- ‘snapshot’
. In many cases, no clear exposure and outcome- just looking at how common something is (e.g. surveys, opinion polls)
. Typically done on a sample of population
How can results from a cross-sectional study be made valid?
Samples must be free from bias, so findings are generalizable
What kind of bias are cross-sectional studies prone to?
. Selection bias (sample not representative) and information/observation bias (measurement error)
. Most prone to selection bias via non-response
Give the advantages of cross-sectional studies
Quick (no follow up needed), cheap, and simple
Give the disadvantages of cross-sectional studies
. Not useful for rare exposures or diseases
. Can’t assess causation because don’t know whether outcome or exposure came first
What is a cohort study?
. Define groups based on exposure
. Groups should be representative samples of all the subjects involved with the relevant exposure in a common base population
. Follow patients forward in time looking for outcomes
E.g. British Doctors Study (Effects of Smoking)
What are the advantages of cohort studies?
. Know that exposure occurred before the disease and measured it before, therefore measurement of exposure unlikely to be biased by the disease
. Can easily look at multiple outcomes because an individual with one exposure can develop multiple diseases
. Can give direct assessment of incidence in the cohorts and work out absolute risk
. Low risk of bias in recall of exposure or outcome because information gathered prospectively
. Excellent design for rare exposures
What are the disadvantages of cohort studies?
They are generally slow and expensive (have to follow people up and wait for outcomes to arise), not great for rare outcomes
What is a case-controlled study?
. Define groups based on outcome
. Select cases with outcome being studied and controls who don’t have the outcome (disease)
. Samples of patients with disease and controls must be from same base population
. Look at patients backwards in time
What are the advantages of case-controlled studies?
. Good for rare outcomes/diseases, especially those with a long latency between exposure and disease
. Cheap and quick because not waiting for outcomes to arise
. Can investigate multiple exposures
. No need to follow up because looking back in time
What are the disadvantages of case-controlled studies?
. Prone to bias
. Can only study one outcome
. Poor for studying rare exposures
. Can’t usually directly calculate incidence rates
Describe the biases present in case-controlled studies.
. Selection bias (Berkson’s bias)- Difficult to use base population to select representative controls (i.e. if want to study certain outcome, would usually go to a hospital to find people with this outcome/disease, but this isn’t a base population
. Recall bias- Individuals often have to recall their exposures, which can lead to recall bias (recall things in a skewed way due to outcome experienced e.g. My baby has a congenital disease, maybe it’s because I did take quite a lot of paracetamol), especially if people are aware of a hypothesis being tested
What are ecological studies?
. Look at the association of an outcome and an exposure at a population level
. Compare populations and look at whether populations with greater levels of an exposure tend to have higher or lower levels of a certain outcome
What are the advantages of ecological studies?
. Quick, cheap, easy
. Good for generating hypotheses to examine on an individual level (give general idea before zooming in to investigate individual cases)
What are the disadvantages of ecological studies?
. Different areas might have different methods of obtaining data, which can lead to ascertainment bias
. Different areas may have differences in the measurement of exposures
. Generally not possible to correct confounding factors because can’t measure them in this kind of study
. (Key fault of this study): Relationships found at a population level don’t always hold true at an individual level. To assume so is termed ‘ecological fallacy’
