Lecture 5 - Observational Studies II Flashcards
In cohort studies _________ is assessed at baseline
exposure
In cohort studies, participants are ______ at baseline
disease-free
What are two other terms for prospective
longitudinal, concurrent
What are two other terms for retrospective
historical, nonconcurrent
for cohort studies, _______ is measured before ____
exposure is measured before outcome
When are prospective studies used?
time btw exposure and disease is short, LTFU is low, adequate funding
When are retrospective studies used?
long follow up time required, temporality preserved
when are cohort studies used
for rare exposures, multiple exposures for same disease, long follow up time btw exposure and disease, RCT is unethical, medium-long term follow up of drug performance
What are the cons of cohort
high drop out rate, residual confounding, long wait for rare diseases, lil evidence to justify long study examining exposure-disease relationship, data quality might be poor, historical effect
Define historical effect
advance in medicine may render old data obsolete for answering today’s questions
Cohort participants are selected on
exposure
Case participants are selected on
outcome
Define selection bias
participants LTFU (drop out associated with outcome or unique exposure pattern)
what are three examples of information biase?
observational/misclassfication bias, ascertainment bias, analytic bias
define information bias
quality of info is diff btw groups
define observational bias
quality and extent of info is diff btw exposed and unexposed
define misclassification bias
quality and extent of info is diff btw exposed and unexposed
define ascertainment bias
disease assessor knows exposure status
define analytic bias
belief of investiggators influence data analysis and interpretation
define relative risk
ratio of risk of disease in exposed to risk of disease in unexposed
interpret relative risk
the risk of outcome in exposed is ___% higher/lower than risk of outcome in unexposed
RR = 1
null value, risk in exposed = risk in unexposed
RR>1
positive association
RR<1
negative association
When can OR approx RR?
when disease is rare
When does ability of OR to estimate RR become worse
as disease prevalence increases
Why cant we do RR for case control
bc we fix # of cases and controls
What measure indicates the potential for prevention if exposure is eliminated?
attributable risk
Equation for attributable risk?
incidence in exposed group - incidence in non-exposed group
Attributable risk as proportion?
AR/ incidence in exposed group
Incidence in total pop?
incidence in exposedproportion of popEXPOSED + incidence in unexposedproportion of popUNEXPOSED
Population attributable risk?
incidence in total pop - incidence in unexposed group
Population attributable risk %?
PAR/ incidence in total population
Equation for incidence density?
of cases/# of PY
Equation for incidence density ratio?
ID exposed/ ID unexposed
Interpretation of ID ratio?
incidence of disease in exposed group is ___xx the incidence of disease in unexposed group
What is a cross sectional study
identify exposure and disease status at same time and there is no follow up
Cross sectional studies identify what
prevalent cases of disease in exposed and unexposed
What are the pros of cross sectional studies
quick, cheap, useful for hypothesis generation, suggest need for CC or cohort
What are the cons of cross sectional studies?
reverse causality bias, prevalent may not be representative of all cases
Why doesnt RR apply to cross sectional studies?
bc subjects arent at risk or follows up
What is causality bias
cant be sure exposure proceeded disease
What is prevalence ratio
prevalence of disease in exposed/prevalence of disease in uexposed
What is the interpretation of prevalence ratio
prevalence of disease in exposed is ____% higher/lower than prevalence of disease in unexposed
What observational study does prevalence ratio apply to
cross sectional study
