Lecture 20 Case control study Flashcards
Analytic studies types
observational
- cross sectional
- ecological
- cohort
- case control
intervention
- randomised controlled trial
Limitations of cohort studies
Can be inefficient with rare/slow to develop outcomes
Can be inefficient with transient/acute exposures
case control study
Designed for rare/slow to develop outcomes
Can efficiently examine acute or transient exposures
Cohort vs case-control studies
cohort
- Ascertain exposure status,
then find out outcome(s)
case-control
- Ascertain outcome status, then find out exposure(s)
Randomised controlled trials
Intervene in individual studies
drug trial
Cohort study Can be inefficient with rare/slow to develop outcomes
Follow up long time
Incidence low
Cancers
cohort study Can be inefficient with transient/acute exposures
Brief increase in risk of outcome
Injuries
in case control study
greater odds of exposure (cases)
exposure more common in cases
potential risk factors
In case control study
greater odds of exposure (controls)
Exposure more common in controls
Potential protective factor
Logic of case-control studies
If controls good representation of source population
Ratio of odds of exposure may indicate association between exposure and outcome
In case controlled study the case is
Defined by only one outcome (cohort has multiples)
Define outcome clearly
- Dont want people who like they have outcome but dont
- Introduce bias
- Eg death from alzheimer’s vs dementia
Comprehensive case finding
Prevalent cases
Association of developing disease (outcome) and association of how long have it
Survival or lack of cure with it
in case control
Relevant exposure period
Just before the outcome
in case control
Irrelevant exposure period
Right at the beginning of the follow up
Won’t affect the risk
Case controlled study allows
Look at relevant exposure period
Because we get people who’ve had the outcome and look before it
Odds of exposure in cases
People with exposure / people without exposure
Odds of people in control
People in controls with exposure / people in controls without exposure
Measure of association
Odds ratio (not incidences)
How many times as likely cases are to have the exposure compared to controls
What is the null value for the odds ratio?
Null value for ratios = 1
calculate odds ratio
Odds of exposure in cases / odds of exposures in controls
Relative risks and odds ratios
Interpret exactly same way
OR when disease rare
Will approximate RR
case control study
Can use RR interpretation
Exposed group
Value
Outcome
Comparison group
case control
Null value 1 below
Protective factor
case control
Above null value 1
Risk factor
What happens in a case controlled study
- Identify Source Population
- Identify people with outcome (cases)
- Sample people without outcome from same population (controls)
- Measure exposure prior to outcome in cases and controls
- Compare odds of exposure to calculate measure of association (odds ratio)
Controls don’t have outcome
So when do you measure exposure for controls?
Can’t measure exposure just prior to outcome because by definition dont have it
Index dates
Index dates
Point define when the controls had the outcome
control case
Why use odds?
Can’t calculate prevalence or incidence of outcome
Have selected number of people in study with and without outcome
Measure of association for case-control studies:
odds ratios
Interpreting the odds ratio
Different interpretation to the RR
People with outcome are x times as likely to have the exposure than people without the outcome
interpret RR
People with exposure are x times as likely to develop the outcome as people without the exposure
When disease is rare,
OR approximates the RR (rare disease
assumption)
Can interpret OR just like RR
Use RR interpretation in this course
case control interpretation
Expose group
value (x times as likely)
outcome
comparison group
what are Index dates?
Pretend control had event on same date as case (or close to it)
Case selection:
important points
Defined by outcome, so only one
Clear outcome definition and
identification
Comprehensive case finding
E.g. Death from Alzheimer’s vs
death from dementia.
Routine data collections
Medical records
Case selection:
prevalent cases
Usually try to identify incident cases
Sometimes just recruit people with the outcome (prevalent cases)
Control selection:
important points
Need to represent the exposure distribution in source population
- Good representation of the odds of exposure in source population
Must be capable of becoming a case
Often select multiple controls per case for statistical power
Exposure measurement:
important points
Need to measure exposure period before outcome
Exposure measurement:
important points
Differential recall
Exposure measurement must be comparable
Differential recall
Cases trying to work out what made them sick
Outcome may affect recall ability
Exposure measurement must be comparable
Dead cases vs alive controls
Interviewers may act differently for cases and controls
Cohort vs case-control studies
Cohort
- Ascertain exposure status, then find out outcome(s)
case-control studies
case-control studies
- Ascertain outcome status, then find out exposure(s)
Case-control studies
strengths
- Rare outcome, transient exposures
- Multiple exposures
- Temporal sequencing
- Often comparatively quick and inexpensive
Case-control studies
Limitations
- Usually can only study one outcome
- Difficult to select appropriate control group
- Can be susceptible to selection and recall bias
General population
Ideal
- Most likely to represent the odds of exposure in source population
Hospital controls
- Looking at serious outcome tends to lead to hospital treatment
- Convenient (know where they’re gonna be)
