Observ. Study Designs Flashcards
define association
a statistical measure of dependence/relationship between two events, characteristics, or variables
define causation or causal inference
Observations made in a research study provide evidence for
making causal inferences
what are Bradford Hill’s Criteria for Causation
strength of association
consistency
specificity
temporality
biologic agent
biologic plausibility
coherence
experimental evidence
analogy
what are causes of causal inference?
true/causal effect
chance
confounding
effect modification
bias
what is the hierarchy for causal inference? (most likely to have causal inference -> least likely) and outline the the variables that make the hierarchy not absoulte
Randomized clinical trial -> cohort -> case-control -> cross-sectional -> case series
A poorly documented cohort study may not provide as high of quality of knowledge than a well-designed case-control study, so quality of study is important
what do risk measures quantify?
probability of a given outcome in groups of people
as an example, we want to answer the question of: *Is Azithromycin use associated with increased risk of death due to cardiovascular causes?
we must first identify the exposure of interest and the outcome of interest. What would E,D,Ē,Ď be/what do they mean?
E= exposure -> person who has used azithromycin and did not experience CV death
D= disease/outcome -> Person who has used azithromycin and did experience CV death
Ē= without exposure -> Person who has not used azithromycin and did not experience CV death
Ď= without disease/outcome -> Person who has not used azithromycin and did experience CV death
outline the method of a case series
studies “case” patients only (no control group) who have the disease or outcome of interest. also for patients who all took the same treatment and the study monitors the outcome
T/F case series cannot measure association or causal inference
true
what are the steps involved with conducting a cross-sectional study
- identify population of interest
- define point/period in time
- identify who is exposed and who is disease
- calculate burden of disease
what is a main disadvantage of cross-sectional studies?
no time component: collects info for a single point or short duration of time
what are the steps involved with conducting a case-control study?
- Within the population of interest
- Sample disease and non-diseased persons
- Retrospectively identify exposure status
what are the steps involved with conducting a cohort study?
- Beginning with persons without disease,
but at risk - Identify exposed and unexposed persons
- Follow through time to observe incidence
of disease - Measure association by comparing
incidence in exposed and unexposed
there are two types of cohort studies: prospective and retrospective. what are the differences?
prospective studies recruit patients in present time and “follow” the moving forward in time whereas retrospective starts in some point in the past and measure data using an existing database or registry
what is prevalence and how do we calculate it?
a measure of the burden of disease
equals= # diseased during specified time / # in population during specified time
what is incidence and how do we calculate it?
a measure of risk
#New persons with diseaseduring given time period / Total # at risk persons during given time period
*review slide 12-18 of Measures of Burden, Risk, and Association
*REVIEW SLIDE 22 TO CALCULATE RELATIVE RISK AND RISK DIFFERENCE
how do we interpret the value we get for relative risk?
RR > 1: E is a risk factor for D
RR = 1: E is not associated with D
RR < 1: E is protective of D
how do we interpret the value we get for risk difference?
RD > 0: E is risk factor for D
RD = 0: E is not associated with D
RD < 0: E is protective of D
both odds and risk can be identified for in which study design?
cohort studies
what are the methods used for a cohort study?
aims to sample a group of people who share a defining characteristic (such as smoking) through a period of time to determine if the similar exposure (smoking) has any correlation to the disease (lung cancer)
what are methods used for a case series study?
what form of bias are they selectively prone to and why?
study that tracks multiple subjects with a known exposure, such as patients who have received a similar treatment, or examines their medical records for exposure and outcome.
they are vulnerable to selection bias because they often use subjects from isolated environments (such as the hospital)
what are methods used for a case-control study?
study that compares two groups of people: those with the disease or condition under study (cases) and a very similar group of people who do not have the disease or condition (controls). used to determine causality
what are the methods used for a cross-sectional study?
what form of bias are they selectively prone to?
a type of research design in which you collect data from many different individuals at a single point in time.
recall bias