case-control study
- observation studies allowing researcher to be a passive observer of natural events occurring in individuals with the DISEASE/CONDITION OF INTEREST (CASES) who are compared with people who DO NOT HAVE THE CONDITION OF INTEREST (CONTROLS)
- group-assignments based on DISEASE STATUS
- useful when study a rare disease or investigating an outbreak
- subject land in study groups based on self-selection by natural occurrence (have the disease or not)
- commonly generates an Odds of exposure for each, then and Odds Ratio as the measure of association
-already know disease column totals (A+C and B+D)
case-control study flowchart
population
- disease/outcome (Cases) -disease/outcome (controls)
- exposed vs non-exposed -exposed vs non-exposed
reasons to select a case-control design
- unable to force group allocation (randomize) due to ethics
- limited resources (time/money/subjects)
- the disease of interest is rare and little is known about its associations/causes
- prospective exposure data is difficult/expensive to obtain and/or very time inappropriate
- usually done in retrospective fashion
strengths of case-control studies
- good for assessing MULTIPLE EXPOSURES OF ONE OUTCOME
- useful when diseases are rare
- useful in determining associations (not causation) due to confounders, recall bias, etc.
- less expensive than interventional trials and prospective cohort studies
- useful when ethical issues limit interventional studies
- useful when disease has a long induction/latent period
weaknesses of case-control studies
- can’t demonstrate causation
- can be impacted by unassessed confounders
- retrospective: can’t control for other exposures or potential changes in amount of study-exposure during study frame
- can be impacted by various biases –> selection and recall biases
- limited by available data (information from records may be missing)
selection of cases in a case control study
defined by the investigator using accurate, medically-reliable, efficient data sources; applied to all study participants objectively and consistently
case definition
misclassification (differential/non-differential) is possible
case-control: control group
as close to “case” group minus the disease/outcome
exchangeability = comparability with respect to all other determinants of outcome
controls must be selected irrespective of exposure status
most difficult part of case-control study is
control selection
goal is: to assess for the presence of an association between exposure and known condition of interest by selecting non-disease individuals from the sample population which produced the “cases”
expectation is the controls represent the baseline risk of exposure in the general or reference population
where can control groups come from in a case-control study
‘population’: can be randomly selected, not randomized
institutional/organizational/provider/payer: e.g. if cases are centered around blood cancer, controls could come from same cancer department in hospital but could differentiate in type of cancer
spouse/relatives/friends: genetic, environmental, soci-economic similarities
outbreak-sources of controls
participated in same event
cases and controls of the following study: association between use of substances that contain caffeine and the risk of crash in long distance commercial vehicle drivers
cases: long-distance truck drivers who were recently involved in a crash regardless of caffeination
controls: long-distance truck drivers who had not been in a crash regardless of caffeination
how can one individual function as both an exposed individual and an unexposed individual in the same study
individual can be associated with an outbreak investigation with multiple exposures. e.g.: studies on exposure to different food at same gathering
in a case-control study, if an individual functions as both an exposed individual and an unexposed individual in the same study, its called what?
case-crossover
only case-control design that attempts to address Temporality
nested case-control studies
case-control studies conducted after, or out of, a prospective previous study-type (cohort or interventional study)
-subjects in cohort study, ultimately developing disease/outcome, are defined as cases for the subsequent case-control study
selection of controls used for Nested Case-Control studies
survivor sampling: sample of non-diseased individuals (survivors) at end of study period
base sampling: sample of non-diseased individuals at start of study period
risk-set sampling: sample of non-diseased individuals during study period at same time when case was diagnosed; when building case-control, your controls are at the same level of risk that the cases were
common biases in case-control
selection bias: way subjects are chosen for the study (usually more concerning for control selection; less concerning during case-crossover)
recall bias: related to amount/specificity that cases or controls recall past events differently (more likely that cases are more likely to recall past exposures and levels of exposure)
matching
making sure whatever type of individual, or numerically the number of people in each of our observational groups are equal; can match in a 1:1 ratio or higher
- trying to build in exchangeability concept
- higher ratios used to average variability and prevent outliers
individual matching
matches individuals based on specific patient-based characteristics
useful for controlling confounding characteristics
group matching
requires cases be selected first
proportion of cases and proportion of controls with identical characteristics are matched
e.g. 41 percent of cases are male, so 41 percent of controls are male
