Case Control studies Flashcards
(36 cards)
what is a case control study
observational study of people with disease of interest and suitable control group of people without disease
= to identify risk factors for an outcome
what is the relationship between a potential risk factor and disease examined by comparing the two groups on?
whether the risk factor is present
how frequent the risk factor is
the amount of the risk factor (e.g dose)
what is a retrospective study
starts by knowing who has the disease and work back to find who was exposed
so starts after onset of disease and looks back at risk factors
what is a prospective retrospective study
cases and controls in case-control study can be accumulated prospectively
as each new case is diagnosed, it is entered into the study BUT still looks back retrospectively from disease to exposure
what is the approach of case controls studies
figure out if diseased cases differ from non-diseases controls in terms of prior exposure
compare frequency of exposure
why cant disease incidence rate be calculated for case-control study
because it doesnt follow a disease-free population over time
what does an odds ratio of 1 indicate
no association between exposure and disease
what does an odds ratio of <1 suggest
potential protective effect
what does an odds ratio of >1 indicate
positive association between exposure and disease, meaning exposure may be a risk factor
what are the essential features of a case-control study
directionality (outcome to exposure)
timing (retrospective for exposure)
rare or new disease - always the design choice if disease is rare
how do you choose cases in case control study
define outcome - standard diagnostic criteria and severity/duration of disease
age range for inclusion
temporal boundaries
geographical boundaries of study
how do you source cases for case-control study
population-based - identify and recruit all incident cases from population using disease registry, vital records, or defined area
hospital based - identify cases where you can find them BUT may not be representative
what should the control group represent
the population of individuals who would have been identified and included as cases had they developed the disease
all exclusions or restrictions must apply to cases and controls
outline population-based controls
usually first choice
drawn from specific geographical area and time
randomly select controls from same area and same time period
use government lists, NHS, electoral role or telephone survey
what are advantages of population controls
exclusion criteria is easy to apply
cases and controls are from the same study base
representative of whole study base
what are disadvantages of using population controls
low participation rates could cause possible bias
inconvenience of finding controls
recall bias - could have been modifications since diagnosis or changes in past habits
may not be motivated to take part
what are neighbourhood controls
selected from residences in same geographical area as cases
attempts to reduce variability of socioeconomic status
what are advantages of neighbourhood controls
selection doesnt need roster of people - avoids problem of using telephone lists
possible risk factors which vary georgraphically may be more balanced between cases and controls
what are disadvanatges of neighbourhood controls
costly
possibly not representative of study base
increased chance of selection of people living alone
difficult to document non-response
possible over-matching if exposure is related to residence
outline hospitalised or disease registry controls
control diseases should be subject to same surveillance and detection procedures as cases
controls come from same source as cases e.g hospital admissions, and disease registries
what is an example of a hospitalised control
study of smoking and MI among women
cases from admission to coronary care units at 152 hospitals
controls drawn from admissions to other wards in same hospital
what are advantages of hospitalised controls
easy to identify controls, readily available in sufficient numbers
more likely to be aware of previous exposures than healthy individuals
more likely to cooperate than healthy individuals - reducing bias due to non-response
what are disadvantages of hospitalised controls
they are ill
disease may have common aetiology or be on the causal pathway
more likely to smoke, more likely to be from lower SES, more likely to be heavy drinkers
-bias
if taken from specialist hospitals there could be differences in SES
what are other categories of controls?
friends - cheap, convenient, but could over-match when related to exposure
relatives - useful if genetic factors are involved, but there is risk of over-matching
