[2] Class 20-21: Case-control Studies Flashcards
Why are observational studies lower on pyramid than interventional?
They cant prove causation like interventional can
Increasing evidence for obs study
Case reports/series-> Ecological-> Cross-sectional-> Case-control-> Cohort
Obs studies allowing researcher to be a passive observer of natural events occurring in individuals w/ the Dz/ condition of interest
Case-control studies
People who don’t have the Dz of interest
Control
Supplies info. About the expected baseline risk-factor profile in the population from which the cases are drawn
The control group
How are group-assignments made?
Based on Disease Status
When is a case control study useful?
When studying a rare disease or investigating an outbreak
What type of randomization is used in case control studies?
IT’s not! Subjects are randomly selected but they’re not randomized!
Obs studies and random selection:
Sometimes the Dz population is so high that you only need a fraction of the population. Randomly pick from that population.
Case-controls commonly generate an ____________ for each, then an _____________ as a measure of asso.
Odds of exposure; odds ratio (OR)
2x2 table Where is Dz
On top of the column
What conclusion can be drawn from case-control studies?
Strong, weak, no association
Know how to calculate an odds ratio
Odds-exp. of interest / odds-placebo
Reasons to use case-control design (4):
- Unable to force group allocation [randomize]
- limited resources
- rare Dz of interest/little known
- can be retrospective/prospective
Case-control design retro or prospective?
Interventional?
Both;
Interventional can only be prospective
Case-control studies are typically conducted in a ______________ fashion
Retrospective
Strengths of case-control study (in general) [6]:
- can assess multiple exposures of one outcome
- useful w/ rare Dz
- Determining asso.
- Less expensive
- useful for ethical limitation w/ interventional
- useful for long induction/latent period of Dz
Determining selection of cases :
Objectively, consistently, accurately, and w/ validity
Best: clinically-supportable/definable criteria
WHo determines selection of cases?
The investigator
Labeling patients _________ is ideal
Correctly
The risk of ________________ subjects into wrong group is always present
Misclassifying
All else being =[same grp] the outcome if something didn’t occur
Counterfactual outcome for smokers estimated by non-smokers
Req’s assumption of exchangeability
Comparability w/ respect to all other determinants of outcome.
Counterfactual theory
What is the only thing investigators want to be different about the 2 groups?
Dz of interest
What is the most difficult part of selection of cases?
The control selection
To assess for the presence of an asso btw exposure and known condition of interest by selecting non-Dz ppl from the sample population which produces cases.
Goal of control selection
What is a major determinant in whether any conclusion is valid?
The way the controls are selected
The way the controls are selected can affect(2):
Internal validity
Selection bias
Best way to select the control study population:
Make the groups as close as possible except the presence of the Dz
If the exposure has no effect, what is the odds ratio:
1
Controls must be selected irrespective of:
Exposure status
Why would you select control irrespective of exposure status?
You could accidentally cause or mask an asso. If you get that involved
Control group can come from several sources (3):
Population
Institutional/organizational/provider
Spouse/relative/friends
Illnesses of control should be:
Unr/t exposure being studied
Why study relatives?
Genetic/environmental, socio-economic
Why were neighbors selected for control in MO heat wave of 1980?
What other factors were accounted for and should be for this and other controls?
They were good pick b/c theyre in the same climate;
Gender, +/- 5 years of age
Systematic ‘control-search’ pattern
An easier way to select controls;
Participated in same event and ‘at-risk’
What is this termed in control selection?
Outbreak-sources of controls:
Study in which we only know the outcome and we’re looking for an association w/ the exposure
Case-control study
An individual can actually function as both an exposed individual AND an unexposed individual in the SAME study.
Subjects are their own controls during the other times they don’t have the acute change in risk
Case-crossover
What is the only case-control design able to adequately ate,pt to address the issue of temporality
Case-crossover design
Case-control studies that are conducted after or as a subsequent study to a prospective cohort study.
Can take ppl from that study and use in either control or case categories
Nested Case-control
3 techniques to selecting controls for nested case-control studies
Survivor
Base
Risk-set
Selection from Sample of non-diseased individuals at the END of the study period for nested case-control studies
Survivor sampling
Selection from Sample of non-diseased individuals at the START of the study period for nested case-control studies
Base sampling
Selection from Sample of non-diseased individuals DURING the study period at the SAME TIME when case was diagnosed for nested case-control studies
Risk-set
R/t to the way subjects are chosen for study- usually more concerning for control group, but less concerning for case-crossover
Common bias in case-control studies
Selection bias
2 Common bias types in case-control studies
Selection bias
Recall bias
R/t the amount / specificity that cases or controls recall past events DIFFERENTLY
More commonly cases more likely to recall past exposures and levels of exposure
Recall bias
Cases can be matched to controls at ratios from 1:1-5:1
Can be done on individual or group
Matching
Matches subjects based on specific patient-based characteristics
Useful for controlling confounding characteristics
Individual matching
Proportion of cases and proportion of controls w/ identical characteristics are matched
Req’s cases be selected 1st
Ex: 41% cases are male=41% of controls will be male
Group matching
What is something that we want to be careful not to match?
Anything that might be a risk factor should not be matched
