KHPM202 2022학년도 2학기

역학개론
Introduction to Epidemiology

2. Causality
Homework assignment

• Please read a paper titled “Epidemiology and the public

health movement” by M. Terris (Journal of Chronic Diseases
1986, 39(12):953-960 or Journal of Public Health Policy, 1987,
8(3):315-329) and submit a summary report on the paper.
‒ It should be written either in Korean or English.
‒ Please upload an MS Word file of no more than 3 pages.
• Due date: September 26, 2022

2
What is cause

Definition of “cause”
• Any event, act, or condition
preceding disease or illness without
which disease would not have
occurred or would have occurred at
later time

Disease results from the cumulative

effects of multiple causes acting
together
(causal interaction)

Kenneth J. Rothman

3
What is cause

Causes of hip fracture

ü poor weather
ü uncleared path for
pedestrians
ü the choice of footgear
ü the lack of a handrail
ü …

Even..
ü person’s weight
ü equilibrium disturbance
ü having a hip

4
Causal models

• Conceptual model that can serve as a starting point in

discussions of causation
• Causal model should address
‒ multifactorial causation
‒ interdependence of effects
‒ direct and indirect effects
‒ levels of causation
• Types of causal models
‒ epidemiologic triad(or triangle)
‒ web of causation
‒ sufficient-component cause model
‒ potential-outcome or counterfactual model
‒ graphical causal model (causal diagram)

5
Epidemiologic triad

• The triad consists of an external agent, a susceptible host,

and an environment that brings the host and agent together

Agent ü An infectious microorganism or

pathogen
ü Ex) virus, bacterium, parasite,
or other microbes

Host
ü The human who can get the disease
ü Ex) age, sex, hygiene, genetic
composition, nutritional and
immunologic status
Environment
ü extrinsic factors that affect the agent
and the opportunity for exposure
ü Ex) geology and climate, vectors,
socioeconomic factors
6
Epidemiologic triad

• In this model, disease results from the interaction between

the agent and the susceptible host in an environment that
supports transmission of the agent from a source to that host.
H A
A H

E E
Agent becomes more pathogenic Host becomes more susceptible

H A

A H

E E
Environmental change Environmental change
facilitates agent spread alters host susceptibility 7
Web of causation

Macro-level
(indirect, distal cause)

Individual-level
(intermediate-level cause)

Micro-level
(direct, proximal cause)

8
Sufficient-component cause model

“Necessary”
Single component cause component cause
One causal mechanism

9
Sufficient-component cause model

• Sufficient cause
‒ a minimal set of conditions and events that are sufficient for the
outcome to occur (that makes disease inevitable)
• Component cause
‒ each component of a sufficient cause
• Necessary cause
‒ a component cause that appears in every sufficient cause
• Complementary cause
‒ for each component cause in a sufficient cause, the set of the other
component causes in that sufficient cause

10
Example 1. COVID-19

SARS-CoV-2

Susceptibility

• Necessary cause: SARS-CoV-2

• Complementary cause: Susceptibility

11
Example 2. Yellow Shank

• A trait in chickens called yellow shank

occurs when a specific genetic strain of
chickens is fed yellow corn. Yellow corn
‒ What would the farmer think if he started Genetic
feeding yellow corn to a susceptible flock? strain
‒ What would the farmer think if he added
susceptible chickens to a flock being fed
yellow corn?

Is yellow shank
environmental or genetic?

12
Strength of cause

• With respect to an individual case of disease,

‒ Every component cause that played a role was necessary to the
occurrence of that case
(i.e., there is no such thing as a strong cause or a weak cause)
‒ There is only a distinction between factors that were causes and
factors that were not causes
• With respect to the total burden of cases occurring in a
population,
‒ We can define a strong cause to be a component cause that plays a
causal role in a large proportion of cases
‒ The strength of a cause defined in this way necessarily depends on
the prevalence of other causal factors that produce disease

13
Strength of cause

Exposure
Population A Population B
Smoking Radon

Radon Smoking 1 1 900 100

1 0 100 900
0 1 900 100
Others
0 0 100 900

• on the population level,

strength of a cause ≠ biology of causation

14
Is catalyst a cause?

After 10 years
A
Disease
U

Is B acting as a catalyst or
as a cause of disease?

After 2 years
B A
Disease

15
Scientific inference

• Induction
‒ The method of induction starts with observations on nature.

All swans we have seen have been white

All swans are white

‒ The sharpest criticism came from the skeptical philosopher David

Hume, who pointed out that induction had no logical force.

16
Scientific inference

• Refutationism
‒ Popper’s philosophy, known as refutationism, held that statements
about nature can be “corroborated” by evidence, but corroboration
does not amount to a logical proof.
‒ On the other hand, Popper also asserted that statements about
nature can be refuted by deductive logic.

“All swans are white”

REFUTE
We observed a black swan

“All swans are white or black”

?
In this philosophy, what we call scientific knowledge is a body of currently
unrefuted hypotheses that appear to explain existing observations
17
Scientific inference

• Refutationism in epidemiological research

“A and B are not associated” (null hypothesis)

REFUTE
We observed that A and B are associated

“A and B are associated”

18
Counterfactual (potential-outcome) concept

• How to measure causal effects?

‒ The comparison of people with themselves, followed through time
simultaneously in both an exposed and an unexposed state
‒ Such a comparison envisions the impossible, because it requires each
person to exist in two incarnations
‒ Because this situation is impossible, it is called “COUNTERFACTUAL”

actual me
outcome
(exposed)
vs.
counterfactual me
(unexposed) potential-outcome

*all other factors remaining the same

19
Causal inference in Epidemiology

20
Causal inference in Epidemiology

• “Causal inference” is the process of deriving cause-and effect

conclusions by reasoning from knowledge and factual evidence
• “Proof” is impossible in empirical sciences

21
Koch’s postulate

22
Hill’s criteria (considerations)

1. Strength
2. Consistency
3. Specificity
4. Temporality
5. Biologic gradient
6. Plausibility
7. Coherence
8. Experimental evidence
9. Analogy

23
1. Strength

• The stronger the association, the more likely it is that the

relation is causal (less likely to be confounded)
• Usually measured by relative risk or odds ratio

• However,
‒ (By causal pie model) Strength depends on the prevalence of other
causes
‒ No general rule for how large an association needs to be met
‒ There are many weak associations that are generally agreed to
reflect causal effect

24
2. Consistency

• If the relationship is causal, we would expect to find it

consistently in different studies and in different populations
• Replication of findings is particularly important in
epidemiology

• However,
‒ The absence of consistency does not imply the absence of causal
effect
‒ Consistency cannot rule out bias entirely

25
3. Specificity

• An association is specific when a certain exposure is

associated with only one disease

• However,
‒ A cause can have many effects (e.g., smoking)
‒ This is the weakest of all the guidelines

26
4. Temporality

• A cause must precede it’s effects

• A necessary condition

• However,
‒ Not easy to prove
(potential reverse causation)

27
5. Biologic gradient

• As the dose of exposure increases, the risk of disease also

increases (monotonic dose-response relationship)

• However,
‒ There may be a threshold effect
or U-shaped dose-response effect

Oh, H., Kwak, S. Y., Jo, G., Lee, J., Park, D., Lee, D. H., ... & Shin, M. J. (2021).
Adiposity and mortality in Korean adults: a population-based prospective
cohort study. The American Journal of Clinical Nutrition, 113(1), 142-153.
28
6. Plausibility

• A causal explanation is plausible if it appears reasonable or

realistic within the context in which the hypothesized cause
and its effect occur
‒ biology, pathology, toxicology, …

• However,
‒ Plausibility can change as the context evolves

29
7. Coherence

• The hypothesized causal relation of exposure to disease does

not conflict with current understanding of the disease
process
• Similar with consistency and plausibility

30
8. Experimental evidence

• Clinical trials, animal experiments, or experiments on tissue

• However,
‒ Not always available
‒ Uncertainty in extrapolation from animals to humans

31
9. Analogy

• Drawing inference about the association between a given

exposure and disease based on what is known about other
exposure-disease relations

• However,
‒ Absence of analogies may reflect only lack of imagination or
experience, not falsity of the hypothesis
‒ The more apt the analogy, the less specific

32
Summary

• Aside from temporality, there is no necessary and sufficient

criterion for determining whether an observed association is
causal

“What [the nine viewpoints] can do, with greater or less

strength, is to help us to make up our minds on the
fundamental question—is there any other way of explaining
the set of facts before us, is there any other answer equally,
or more, likely than cause and effect?” (Hill, 1965)

33
Modifications of the Guidelines
for Causal Inferences

34
Modifications of the Guidelines
for Causal Inferences

35
Thank you

36