Evidence Based Medicine Flashcards
3 types of observational study?
case control
cohort
cross sectional
Why do an observational study?
• Hypothesis generating (good to gather information to provide framework for RCT)
• RCTs are expensive and carry a risk as delivering intervention
• Study of rare events
Ethics
Describe a case control study?
- Cases have disease of interest and controls do not have the disease (age and sex matched with the cases)
- Direction of research inquiry is backwards in time
- Then decide who was most exposed to chosen risk factor
- Can look at multiple factors or risks- construct 2x2 table
- Calculate odds in diseased ad not diseased
- Odds ratio of 1 means there is no impact
Describe 4 types of bias in case control studies?
• Selection bias: cases may not represent exposure distribution in all cases in the source population and controls may not represent exposure distribution in those without disease
• Observer bias: knowledge of case/ control status may influence data collection, should be standardised collection and blind researchers
• Recall bias: cases and controls recall prior exposures differently so need to minimise period of recall and measure exposure data objectively (medical notes, third party verification of exposure information)
Survivor bias: If exposure is rapidly fatal then the people you are able to recruit will not be representative as survivors. You may then be capturing a characteristic that leads to survival.
Describe advantages and disadvantages of case control studies?
Case Control Advantages
• Rapid and cheap
• Ideal for rare disease or outcomes
• Useful for diseases with long latent periods
• Can simultaneously examine a large number of potential exposures
Weakness of Case Controls
• Bias
• Temporal relationships can be difficult to establish
Cannot compare incidence rates
Describe cohort studies?
• Selecting sample based on characteristics then following them up over time
• So direction of research inquiry is forward in time
• Used to look at impact of infrequent/ unusual exposure e.g. impact of chernobyl
• Multiple outcomes related to infrequent exposure
Measure disease incidence (CANNOT do this with case control as the people already have the disease)
What type of study can measure disease incidence?
Cohort studies
Describe the two types of cohort study? Advantages disadvantages of each?
A prospective study starts it now and goes forward into the future. A retrospective study is based on information in the past, establish cohort based on something previously ( a cohort of children who all sat an exam at particular time and looked at lots of things) Retrospective still following in future it’s just you are using info from the past
* Prospective: More time consuming, expensive, defined cohort detailed exposure records with standardised measurement * Retrospective: faster answers, don’t have to employ people, less control over the info
Cohort vs control study advanatages and disadvantages?
Advantages of cohort vs case control
• Temporality
• No recall bias
• Measures incidence
Disadvantages of cohort vs case control • Required large investment of time resources and finance if prospective • Requires large sample sizes • Los to follow up bias • Inefficient for rare diseases Uncontrolled confounding
Describe interpretation of results of cohort?
• RR> 1exposure Is harmful
• < 1 exposure in protective
• 1=nothing
• RR= 3 would mean 3 fold increase increase in risk of outcome in exposed compared to unexposed
• Logarithmic scale so never goes below 0.
• 0.8=0.8 20% decrease
1.5=1.5 50% increase
Describe interpretation of results of case control?
• Calculate odds in diseased ad not diseased
Odds ratio of 1 means there is no impact
Risk by not risk in present diseased and not diseased
Then divided diseased by not diseased result
Literally the probability of that risk factor with the disease
Describe cross sectional studies?
• Carried out a single point in time • Most frequently is a survey • E.g. census, crime and justice survey • Particularly good for estimating point prevalence • Quick and cheap to conduct • Cannot establish causation • Not a lot in medical research Some in public health