Lecture 23 & 24: Case-Control Studies Flashcards
How are case-control studies selected?
-Selection of study subjects is based on disease/ outcome status
-Cases= have the disease or outcome of interest
-Controls= DO not have the disease of interest
THEN
-Look at if each of those groups were Exposed to Not exposed
Why are most case-control studies retrospective?
-Since selection is based on outcome, typically look back to see if the individual was exposed
-If cases have already occurred and can be identified we can conduct the study in a short period of time using hospital records or surveys to collect risk factor info
Ex of retrospective would be: asked about food they ate in the week before (for E. coli and lettuce ex)
What is the objective, sampling and analysis of case-control studies?
Objective: is the E (s) related to the disease (O)?
Sampling: Based on disease status, identify individuals with and without outcome of interest (purposive sampling and collect info on exposure status)
Analysis: compare proportion of individuals exposed and unexposed by disease status (OR commonly used) basically want to know those who had disease had increase/more E than those who did not have disease
What are some advantages of case-control studies?
-Retrospective= often relatively fast and inexpensive
-Best design for investigating the source of an outbreak (bc the disease has already occurred and most of the sick people have already been identified)
-Good for rare diseases (but not rare E that would be cohort)
-Good for identifying multiple risk factors for a single disease (therefore useful for disease of unknown etiology) for ex with covid can look all the risk factors (age, vax status etc) and see if only some are associated with O
What are disadvantages of case-control studies?
-Cant calculate disease incidence (new cases) or prevalence (existing cases) in population, bc select when they are diseased so can’t form exact timeline
-Can study only 1 outcome
-Subject to many different types of bias (selection, info, confounding)
-Case and control selection can be difficult
-Temporal relationship an issue if using prevalent cases
Where do we get cases for case-control studies?
Keep in mind section based on Disease/O
-Hospitalized patients, patients from physicians offices, clinic patients
-Disease registries (ex cancer registers), death certificates
BUT PROBLEMS CAN ARISE
-if just one hospital is used (some things be unique to that hospital so want to use multiple)
-If a tertiary-care (ie referral) clinic is used (bc only used for severe versions of O)
-If prevalent cases are used instead of incident cases
What is the difference between incident cases and prevalent cases?
Incident cases: Newly diagnosed cases, better for establishing causation but slower bc need to wait for cases to arise/ be diagnosed
Prevalent cases: people may have had disease for some time (new+ old cases), faster as cases generally avaliable for study, BUT any risk factors identified may be more related to survival with disease than getting the disease
What are some case definitions and criteria?
-Need to be clear and explicit
-Want to use criteria that minimize likelihood of a true case being missed (criteria are sensitive) and of a non-affected person being falsely classified as a case (criteria are specific)
-Can so this by lab conformation, time frames, geographic locations, and symptom description being as detailed as possible
Where are the source of cases coming from for case-control studies?
Study base: population from which the cases (and the controls) are obtained
-Primary study base: population from which the cases arise can be easily defined (ex residence from Guelph, common in vet med. use farm with good records usually involves explicit list)
-Secondary study base: one or more steps removed from primary population (patients at referral clinicians, hospitals, cancer registry, lab which are easier to find)
What are control selection in secondary base for case-controls?
- Controls should be readily selected from non-cases in the registry (other ill patients, just with different diseases)
AND - Those controls should be selected from diagnostic categories not associated with the E of interest
What are control selection in primary base for case-controls?
-Primary case (non-hospitalized) sources include:
-People living in the community/neighbourhood controls
-School rosters, selective service lists, insurance company lists
-“Best friends list” (if one has O and other doesn’t, might be similar ppl who do similar tings so might be difficult to distinguish)
How are multiple controls used for case-controls?
-Some investigators use multiple controls from different types of sources (ex hospital controls AND neighbourhood controls)
-Each group may have its own biases which need to be take into consideration
-Hopfully get similar results across the groups
