Environmental Epidemiology (3) Flashcards
Different types of environmental epidemiologic design
- Descriptive studies
- Ecologic or correlational studies
- Analytic (or etiologic) studies
Descriptive studies
- Not hypothesis driven
2. Cluster studies, surveillance studies
Ecologic or correlational studies
- Association at group level, not individual
Analytic (or etiologic) studies
- Test hypotheses, often informed by previous designs
- RCTs
- Observational studies (cohort, case-cntrl, cross-sectional)
Why can’t we conduct randomized human experiments to investigate health effects of environmental agents?
Ethical reasons so we take an ecological approach to draw comparisons
Concentrated Animal Feeding Operations (CAFOs)
Large-scale factory farms
Problems with spraying of pig fecal waste
- Odor –> eye irritation, respiratory problems
- Hydrogen sulfide –> eye irritation, respiratory problems
- Particulate matter –> eye irritation, wheezing
- Endotoxins –> sore throat, chest tightness, nausea
How can we study health implications of CAFOs if RCTs are unethical?
Option #1: Location contrasts –> ppl living near CAFOs vs. ppl living far away from CAFOs
Option #2: Time period contrasts –> among ppl living near CAFOs, compare illness rates on high vs. low pollution days (case-crossover design)
What is the key assumption in “Location Contrasts” design?
Populations are “exchangeable”
What is exchangeability?
Morbidity in unexposed = morbidity in exposed had they not been exposed
What happens when populations are not exchangeable?
There is confounding
Is there exchangeability in RCTs or observational studies?
- RCTs –> Yes
2. Observational –> No randomization so it is unknowable
In the pig farming example, which design is appropriate for hypotheses about short-term effects?
Case-crossover
Why is the quality of exposure assessments important?
Because it affects the validity of an environmental epi study
What is the case-specular design?
Similar to case-crossover, but the comparison is spatial (rather than temporal)