aetiology case control studies (also rct, cohort and ecological studies)
TRANSCRIPT
AETIOLOGY
Case control studies
(also RCT, cohort and ecological studies)
Objectives
Case control studydescribe structure and analysis interpret and calculate odds ratio identify potential for bias
10 year old girl, URTI Improved after taking aspirin
After several days, sleepy, vomited, lethargic, very confused, liver enlarged, blood aspartate and alanine transferase elevated
?
How assess cause(s)?
Case control study
CASE
CONTROL
EXPOSED
NOT EXPOSED
EXPOSED
NOT EXPOSED
Start
Case control studies
To assess the association between disease and exposure by comparing one group with the disease and the other without “Retrospective” study Disease status known History of exposure is unknown
Reye’s syndrome and medication useHalpin et al. JAMA 1982;248:687-91
All cases of Reyes syndrome in Ohio Dec.1978 - March 1980 (n=97)
Controls - children with similar viral illness from same schools as cases
Use of medication during illness Aspirin taken by 97% of cases vs only 71% of
controls Other medications similar in the 2 groups
Case control
CASE
Reyes syndrome
CONTROL
Recent viral illness
EXPOSED Aspirin
NOT EXPOSEDNo aspirin
EXPOSED Aspirin
NOT EXPOSEDNo aspirin
START
*Hypothetical case control study of Reyes syndrome and aspirin (unmatched)
Cases Controls Total
Aspirin used 190 130 320
No aspirin 10 70 80
Total 200 200 400
Why not relative risk/ risk difference?
"Outcome" Cases Controls Total Exposed
a b
Not exposed
c d
Total a+c b+d a+b+c+d
Odds ratio (OR)= Case / control ratio in exposed
Case / control ratio in non-exposed
= Exposed / non exposed ratio in cases
Exposed / non exposed ratio in controls
= a / b = ad / bc Ξ a /c = ad / bc
c / d b /d
= 190 x 70 / 10 x 130 = 10.2 (p < 0.05)
Interpret OR as for relative risk
Matched case control studies
Each case matched to 1 or more controls for selected confounding variables (usually age and gender, and sometimes other confounders) E.g. for a case aged 4 years, a control aged 4
years is selected Need to use matched analysis
OR = ratio of discordant pairs
Options for each matched pair
Case Exposed Not exposed Exposed Not exposed
Control Exposed Not exposed Not exposed Exposed
Control exposed
Control not exposed
Total
Case exposed
132 57 189
Case not exposed
5 6 11
Total 137 63 200
OR = ratio of discordant pairs= 57 / 5 = 11.4
Applications of case control studies
Etiologic research Vaccination effectiveness Outbreak investigation Occupational exposures Evaluation of screening Evaluation of treatment
Case control Advantages :
Efficiency : cost, time If rare disease Large number of exposures
Disadvantages : Cannot estimate incidence Susceptible to bias
Recall bias and selection bias Not suited for weak association
Strength of study types
ExperimentalRandomised controlled trial
Observational CohortCase-controlCross-sectionalEcologicalCase series
Confounding variable associated with outcome (disease)
and associated with exposure (risk factor)
risk factor
disease
confoundingvariable
Aspirin causes Reyes syndrome? Why important to distinguish between causal and
non causal associations? How distinguish between causal and non causal
associations? Epidemiological framework for causality
Aspirin causes Reyes syndrome?
Epidemiological framework for causality Other evidence?
aspirin no longer recommended for children use declined incidence of Reyes declined (reversibility
criterion)
SEE Causality lecture
Note, in children given aspirin, absolute risk is VERY low.