Chapter 14 slides

Report
Conclusion
Epidemiology and what
matters most
Epidemiology matters: a new introduction to methodological foundations
Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Seven steps
1.
Define the population of interest
2.
Conceptualize and create measures of exposures and health indicators
3.
Take a sample of the population
4.
Estimate measures of association between exposures and health
indicators of interest
5.
Rigorously evaluate whether the association observed suggests a causal
association
6.
Assess the evidence for causes working together, i.e., interaction
7.
Assess the extent to which the result matters, is externally valid, to other
populations
Epidemiology Matters – Chapter 1
4
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Comparability and external validity
All epidemiologic studies should be conducted with a
clear intent to improve the health of populations
However no one study can stand alone without an
evidence base, no one study will settle a causal
question, no one study will be the last word on any
issue
Epidemiology Matters – Chapter 14
Comparability and external validity
Comparability: achieving within study sample ensures
causal effect estimate(s) are internally valid
Chapter 10: Randomization, matching, and stratification are
foundational approaches to achieve comparability of study
sample
Epidemiology Matters – Chapter 14
Comparability and external validity
External validity: extent to which our findings are generalizable to a
base population. This requires an understanding of factors that
together are involved in producing a causal estimate
Chapter 7: most causes of disease do not act in isolation, i.e.,
interaction
Chapter 11: assess interaction in data - evident when risk of disease
among exposed to two potential causes > additive effect of each cause
Chapter 12: relation between exposure and health indicator is
externally valid to another population to the extent that interacting
causes with exposure are distributed similarly
Epidemiology Matters – Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Small effects, big implications
Does the causal effect obtained in a study have consequence for
the populations in which burden of disease is greatest?
Are the effect estimates obtained in study translatable to actual
cases of illness and disease potentially prevented by
intervention?
To answer: compare effect estimate magnitude to prevalence of
exposures of interest; small magnitude of effect may translate to
large public health benefits
Epidemiology Matters – Chapter 14
Small effects, big implications
example
Question: intervening to prevent occurrence of disease in
Farrlandia, an overall population risk of 6/100, over 5 years
Two exposures associated with disease:
 Exposure A associated with increased risk ratio of 1.2
disease onset
 Exposure B associated with 5-fold increase disease risk
Which exposure should we invest public health time and money
in preventing?
Answer may depend on the prevalence of these exposures
Epidemiology Matters – Chapter 14
Small effects, big implications
example
Exposure A
Exposure B
Interpretation: Exposure A has prevalence of
Interpretation: Exposure B has prevalence of
80% (800/1000). A risk ratio of 1.2 and 5 year
5%. A risk ratio of 5.0 and 5 year risk of 6%.
risk of 6%. Exposure to A caused 45 cases.
Exposure to B caused 12 cases.
Even though Exposure A has weaker overall effect on disease compared with Exposure
B , it is responsible for almost four times disease more because it is more prevalent in
population
Epidemiology Matters – Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Consequentialist epidemiology
 The ultimate purpose of epidemiology, the quantitative
science of public health, is to understand the causes of
human disease and improve health of the populations
where the burden of disease is greatest
 Health is not distributed equally across populations, a
consequentialist epidemiologists engages in science
beyond local borders
Epidemiology Matters – Chapter 14
Implications
To study under 5 mortality in US
•
Sample the population (Chapter 4)
•
Measure potential causes of interest (Chapter 5)
•
Estimate associations of effect of potential causes
on child mortality (Chapter 6)
•
Assess associations for internal validity (Chapter 8)
•
Assess interaction (Chapter 11)
•
Consider the conditions for external validity across
populations (Chapter 12)
An epidemiology of consequence makes sure to study
child mortality in resource poor versus resource rich
settings
Epidemiology Matters – Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Causal explanation and interventions
Effects of causes are not necessarily equal to the effects of
interventions on those causes
Epidemiologic studies can isolate specific effects of exposures by
creating comparable exposed and unexposed groups
However, exposures cannot be removed in isolation, resulting in
alterations to changing distribution of component causes once
causes are manipulated
This can have unintended consequences including increasing
another adverse outcome
Epidemiology Matters – Chapter 14
1. Seven steps of an epidemiological study
2. Balancing comparability and external validity
3. Small effects, big implications
4. Consequentialist epidemiology implications
5. Causal explanation versus intervention
6. Summary
Epidemiology Matters – Chapter 14
Seven steps
1.
Define the population of interest
2.
Conceptualize and create measures of exposures and health indicators
3.
Take a sample of the population
4.
Estimate measures of association between exposures and health
indicators of interest
5.
Rigorously evaluate whether the association observed suggests a causal
association
6.
Assess the evidence for causes working together, i.e., interaction
7.
Assess the extent to which the result matters, is externally valid, to other
populations
Epidemiology Matters – Chapter 1
19
epidemiologymatters.org
Epidemiology Matters – Chapter 1
20

similar documents