POMP 3240- Epidemology: Midterm 2

Exposure

any factor that when altered produces or is suspected of producing a change in the frequency or characteristics of the outcome

outcome

response or dependent variable

bias

the systematic error errors that result in an incorrect estimate of the association between exposure and outcome

random error

fluctuations around true value due to chance

systematic error

deviations that disproportioantley affect the data and not due to chance

misclassfication bias

systematic differences in the way that exposure and/or outcome is determined for some participants

selection bias

systematic differences between those who participate in the study and those who do not

confounding bias

the observed association between an exposure and outcome is affected by a third factor

non-response bias

bias occurs from differences in characeteritics between those who choose to participant and those who don’t

detection bias

occurs when the probabilty of detecting the outcome differs by exposure status

loss of follow-up bias

occurs when participants who leave or withdraw from a study are systematically different from those who remain in the study

non-differential misclassfication

magnitude and direction of misclassfication of exposure or outcome is similar across groups

differential misclassfication

magnitude and direction of misclassfication of exposure or outcome is different in the two groups being compared

recall bias

a type of differential misclassification bias that relates to problems recalling the past

confounding bias

the mixing together of the effects of two or more factors

association

identifiable relationship between exposure and outcome

causation

implies that there is a true mechanism that lead from exposure to outcome

how can you minimize recall bias

use shorter time periods in your experiment

how can you minimize misclassification bias

use objective, clear, and explicit guidelines to define measurements among groups

what are the criteria for confounding bias

1. associated with outcome

2. associated with exposure

3. not a consequence of exposure

how can you control confounding bias

1. randomization- only for randomised controlled trials

2. retriction/exclusion- only enroll individuals with one level of CFV

3. matching- evenly distributing potential CFVs between comparison groups

models of causation

theories of how potential causes interact to produce disease at a population level

causal inference

weighing of evidence, judgement, and interpretation to determine if a factor is truly causal

component cause model

illustrates types of causal realtionships using concepts of necssary and sufficent causes

necessary cause

one that precedes the disease must always be present for the disease for the disease to occur so its required for the disease to occur

sufficient cause

one that precedes the disease, and if present, will always produce the disease- so not required for the disease to occur, but could cause it

Bradford Hill’s Criteria

guidelines for causal inference

Name Bradford Hill’s 9 Critieria

1. strength of association

2. consistency

3. specificity

4. temporality

5. dose-reponse

6. biological plausibility

7. coherence

8. experiment

9. analogy

strength of association

the stronger the association between an exposure and disease the more likely it is to be casual

consistency

same results across different populations, different circumstances, different study designs, different researchers

specificity

one exposure causes one disease

temporality

exposure precedes disease

dose-repsonse

greater exposure leads to greater incidence of outcome

biological probalitity

association is biologically credible based on existing scientific knowledge

coherence

association should not seriously conflict with general known facts of the disease

experiment

exposure can be shown to experimentally cause disease

analogy

exposure has been shown to cause disease in other species/organisms/contexts

probability

likelihood that an event (disease) will occur

odds

ratio of two probabilities

what do measures of association tell us

the strength of the relationship between an exposure and an outcome

what does a measure of statistical significance show us

the probability of having an association as great or greater than observed if the null hypothesis was true

what does measure of effect show

the effect of exposure on the outcome as an absolute effect

what advantage does OR have over RR

they are interpreted nearly the same as RR but can be used in studies where RR is not applicable

what types of rates do you need to look at for incidence rate

true rate

when are RR, OR, and IRR not useful

when informing practice like in clinical or public health settings

where are RR,OR, and IRR useful

when investigating causation

excess risk

measures the effect of absolute different when looking beyond the baseline risk

attributable fraction (AFe)

the proportion of disease in exposed group that is due to exposure

what does AFe assume

that exposure is positively associated with disease

population attributable risk

same as risk difference but concered with the increased risk of disease in entire population that is due to the exposure

population attributable fraction

AFe but focused on disease in entire population that is due to exposure

descriptive studies

provide information about occurence of a disease but with no attempt to investigate associations

what do descriptive studies tell us about

the heath of a population and how much the disease is occuring

what are the types of descriptive studies

case reports, case series, surveys

case reports

describe rare of unusual cases, diagnosis, what to expect, signs and symptoms

surveys

collect information from individuals in a population

case series

the usual clinical course of a condition of interest, series of cases of patients with a particular outcome

do descriptive studies use comparison groups

no

can descriptive studies show associations

no

analytical studies

analysis of epidemiological data to establish relationships between risk factors and the occurrence of a disease

observational studies

researchers have no control over the allocation of subjects to the groups being compared

do the researchers expose subjects in observational studies

no the subjects would have been exposed regardless of the study

experimental studies

researchers randomly allocate subjects to the groups being compared

do researchers expose in experimental studies

yes

what are the different types of observational studies

cross-sectional, cohort, case-control

cross-sectional studies

individuals are selected without regard to exposure or disease status

cohort

selection outcome-free participants based on exposure status

case-control studies

select participants based on disease status

type of experimental studies

laboratory based, randomised-controlled trials (RCT)

laboratory based studies

may provide the best evidence of causation

what kind of enviornment are lab-based studies in

artificial

what variables are lab-based studies better at controlling

confoudnry variables or bias

randomised-controlled trials (RTC)

controlled allocation of subjects by natural real-world conditions

why may observational studies be chosen over experimental

exposure of interest is harmful, expensive, complex/hard to control, not practical to administer

advantages of cross-sectional studies

• can determine prevalence of E and O in a population

• can assess multiple exposure-outcome associations

• relatively fast and inexpensive to conduct

• can estimate all measures of association and effect

disadvantages of cross-sectional studies

• bad for rare exposures, outcomes, or short durations

• measures prevalence, not incidence

• cannot establish temporal sequence so inability to make casual inferences

• suscitable to bias

when can you conduct cohort studies

both prospective and retrospectively

if you are conducting a cohort study retrospectively what do you need

excellent historical records

how do you select participants for cohort studies

you can know the exposure status beforehand, you don’t know the exposure status beforhand

how to select participants for cohort studies already knowing exposure status

purposely select study groups from source population based directly on the exposure status

how to select participants for cohort studies without knowing exposure status

• select single group of participants you believe to be heterogenous

• determine exposure status after entire group has enrolled

what are the different types of variables for exposure groups

dichotomous, ordinal, categorical

dichotomous

yes/no

ordinal

levels

categorical

categories