Environmental and Occupational Epidemiology Flashcards
Is there an interest in causation?
Yes; if yes, how important
Types of association?
- no association
- artificial association: die to chance or bias
- indirect association: confounding
- direct association: causal
Observational studies?
-cohort (prospective and retrospective)
-case control
cross-sectional
analytical observational studies do what?
observe events as they happen (or did happen) w/ no active role in events
- measure the exposures, outcomes, & potential confounders that might effect the relationship
- statistical tests
Another name for cohort studies?
- incidence studies
- longitudinal studies
- prospective studies
Types of errors?
- random error (chance)
- systematic error
Random errors?
more likely to cancel out as the measurements increase
Confounding variable?
a variable that distorts the association between the exposure and outcome
-either strengthen or weaken the association; Type I/II error
What makes a confounding variable?
- must be independently associated w/ the outcome
- must be associated w/ the exposure
- must not be in a causal pathway between exposure and disease
Alcohol leads to (arrow) lung cancer. What’s the confounding variable?
Smoking
-correlated w/ alcohol and a risk factor for lung cancer even if you don’t drink alcohol
Controlling confounding factors? (at design stage)
- restricting
- randomization
- matching
Controlling confounding factors? (at analysis stage)
- stratification
- multivariable adjustment
Randomization?
an attempt to evenly distribute potential (unknown) con founders in the intervention and control groups in a randomized controlled trial
Restriction?
prevents confounding, but may be reduce the study size reducing statistical power
Ex: Concern that maternal smoking may be a confounder in associations between childhood lead exposure and reduced IQ. Restrict study population to kids w/ non smoking moms
Multivariable adjustment?
can adjust (control) for the effects of many variables simultaneously
- linear regression
- logistic regression
Selection bias?
selective differences between the comparison groups that distorts the relationship between exposure and outcome
-Ex: healthy worker effect; only picking ppl who are considered healthy to work or be in the study
Information bias?
measurement bias; Investigators who know exposure status may be more or less likely to ascertain the outcome (diagnostic bias)
-to fix this, blinding of investigator and subjects
Controlling bias?
selection bias and measurement bias must be controlled at the design stage
How to limit selection bias?
Representative sampling from the same population
How to limit measurement bias?
blinding
Strengths of cohort studies?
- good when exposure is relatively rare
- can look at multiple outcomes from a single exposure
- can minimize measurement bias
Weaknesses of cohort studies?
- inefficient for rare outcomes
- long time to complete
- expensive
- risky following people at risk
Weaknesses of cohort studies?
- inefficient for rare outcomes
- long time to complete
- losses to follow up
- expensive
- potential ethical issues. may be a risk
Case control studies?
a type of observational study commonly used to look at factors associated with diseases or outcomes
-one group that has the disease (cases) are tested while another group w/o the disease (controls) are given the same exposure
What is the measure of association for case control?
odds ratio (AD/BC)
Strengths of case control studies?
- useful for diseases that are rare, have long induction periods, like cancer, and to explore a wide range of exposures
- cheap
- quick
- safe
nested case control studies?
select cases and controls for the study
-the study is “nested” w/in the cohort
Recall bias?
cases tend to closely scrutinize their past history looking for explanations for their illness
-the disease status of subjects affects their likelihood of reporting the exposure.
Ex: a patient with cancer may be more likely to recall being a smoker
How to avoid recall bias?
- avoid using records of exposures prior to disease
- choose new incident cases (for better memory,
- choose appropriate control group (those w/ a different disease not related to the exposure)
- blind study subjects
- verify exposure from another source (spouse, friend, etc.)
