Case Control Study Flashcards
How is case control study conducted
1) Consider research question and population
2) Select cases
3) Select controls
4) Measure exposure
5) Data analysis
6) Interpret results
7) Assess possible sources of error
Important to consider picking cases
a. Must be well defined including method used to define case, as homogenous as possible
b. Define population from which cases will be selected, controls selected from same population
c. Decide if incident or prevalent cases used – incident is preferred as prevalent might indicate survival of disease
Where should controls be picked from
Same population as cases (population at risk)
Sources of population
i. Hospital – convenient and cooperative but risk of selection bias as does not reflect distribution among population
ii. Population – more likely to represent population that produced cases but more expensive to collect and less likely to cooperate
How many control groups should there be
d. Should be 1:1 ratio of control to cases
i. Rarely worth more than 4:1 – can be used if not many cases available, increasing the ratio increases statistical power up to this point
Ways to measure exposure
a. Through subjects (interviews), medical records, occupational records, biological samples
What is calculated from case control studies
Only odds ratio - cannot calculate incidence
Common errors in case control studies
a. Chance, bias, confounding
b. Selection bias in selecting cases and controls
c. Information bias in obtaining exposure information (recall bias in particular)
d. Misclassification – form of measurement error in how exposure and outcome are categorised
Types of misclassification bias
i. Called differential if not independent of outcome of study, can cause underestimation or overestimation of association
ii. Non-differential if proportion of misclassified is the same in cases and controls, resulting in underestimate of association
Advantages of matching
o More efficient study
o Useful when limited number of cases
o Useful when complex variables
Disadvantages of matching
o Difficult and expensive to find matched controls
o Cannot evaluate effect of matched variable
o Possibility of overmatching
Types of matching
- Individual
o Control for each case - Group
o Control group as a group are similar to the cases in terms of matched variables
What is odds ratio
Compares frequency of exposure among the cases with frequency of exposure among the controls
Odds ratio formula
Odds ratio = odds of exposure (cases) / odds of exposure (controls)
What is reverse causality
When exposure is a consequence of outcome:
- Important to define exposure time period of interest
