Study designs Flashcards
For a qualitative type of question such as “why do young people have unprotected sex?”, what is the best type of study design to answer this question?
Cross- sectional survey - questionnaire - can have issues with people providing false answers
Qualitative study - often used initially to inform design of questionnaire so could be done before the qualitative study
What is an ecological study?
correlate measure of death / illness (y-axis) against an exposure (x-axis)
- uses groups / populations as the units of study
- measure of association is a correlation coefficient
What is a cross-sectional study?
Prevalence study
- surveys
- info on individual exposure and disease collected simultaneously
- repeat overtime for a time series
What is a cohort study?
Start with 2 populations differentiated by their exposure to a specific factor - followed up over tie to see if they have developed range of disease measure
- measure of association is relative risk
- compare risk of disease in exposed group with unexposed group
- can calculate the incidence because you have a population
What is a case-control study?
Start with cases of disease and controls without disease and ask questions about their exposure to a range of risk factors (opposite to cohort)
- measure of association is odds ration - can’t do relative risk because you are not starting with a population
What is a controlled clinical trial (special cohort) / RCT?
Randomise exposure (new treatment vs old treatment)
Sometimes use a placebo
prevents people that have chosen to have a specific exposure
- measure of association is relative risk
What is the best evidence?
A good systemic review and this is for all types of questions
For an interventional effectiveness type of question such as “Does texting messages to young people help them to stop smoking?”, what is the best type of study design to answer this question?
RCT
- the advantage is that any confounding factors can be evenly distributed - best to do this in the design phase. This is highly beneficial because there may be additional unknown confounding factors that will hopefully be randomly distributed
For a risk factors/causal factor type of question such as “Were specific risk factors such as maternal smoking and bed sharing more likely to be present for babies experiencing SIDS than for other babies of the same age?”, what is the best type of study design to answer this question?
Case-control - these types of studies are best for rare cases
- a key challenge would be matching the groups effectively
For an active intervention/risk factor of question such as “What effects do HRT and moderate alcohol intake have on the development of cancer in older women?”, what is the best type of study design to answer this question?
HRT can be seen as an active intervention or a risk factor
Cohort study
For a qualitative/intervention type of question such as “What is the value of family therapist when a family member has had a stroke?”, what is the best type of study design to answer this question?
Non-randomised control trial - could compare 2 areas, one which offers this service and another that doesn’t
What are the advantages of case-control studies?
- rare disease/outcomes
- rare side effects
- retrospective so no loss to follow up
- range of possible risk factors
- small, quick, cheap and easy - data is already there
What are the disadvantages of case-control studies?
Only look at exposure of one disease
recall bias
retrospective
can’t calculate incidence - have to do odds ratio
What are the advantages of cohort studies?
- multiple outcomes because you are following up over time
- assess rare exposures
- no recall bias
- large sample size and maintain temporal relationships
What are the disadvantages of cohort studies?
- expensive
- time consuming
- loss to follow up
- prone to measurement bias - people my measure things differently
- groups may not be representative - may pick based on ease of follow up