Gerstman Chapter 16 1

Epidemiology Kept Simple

Chapter 16: From Association to Causation

Gerstman Chapter 16 2

Cause• Causal inference the

process of deriving cause-and-effect conclusions by reasoning from knowledge and factual evidence

• “Proof” is impossible in empirical sciences.

• However, causal statements can be made strong, or even overwhelming

Gerstman Chapter 16 3

Idea #1: Causal mechanisms essential

Proof of causal mechanisms is essential for effective public health intervention

Consider the case of miasmas and cholera (from Chapter 1)

“For want of knowledge, efforts which have been made to oppose [cholera] have often had contrary effect.” – John Snow

Told ya’

Gerstman Chapter 16 4

Idea #2: Discovery of Preventive Measure May Predate Identification of Definitive Cause

What if we waited until the mechanism was known before employing citrus?

Gerstman Chapter 16 5

§16.2 Surgeon General’s Report on Smoking

• Epi data must be coupled with clinical, pathological, and experimental data

• Epi data must consider multiple variables

• Multiple studies must be considered

• Statistical methods alone cannot establish proof

[Link to Surgeon General’s report]

Gerstman Chapter 16 6

Hill’s Inferential Framework1. Consistency2. Specificity3. Temporality4. Biological gradient5. Plausibility6. Coherence7. Experimentation8. Analogy

* Hill, A. B. (1965). The environment and disease: association or causation? Proceedings of the Royal Society of Medicine, 58, 295-300. full text

A. Bradford Hill(1897–1991)

Gerstman Chapter 16 7

Element 1: Strength

• Stronger associations are less easily explained away by confounding than weak associations

• Ratio measures (e.g., RR, OR) quantify the strength of an association

• Example: An RR of 10 provides stronger evidence than an RR of 2

Gerstman Chapter 16 8

Element 2: Consistency• Consistency ≡ similar conclusions

from diverse methods of study in different populations under a variety of circumstances

• Example: The association between smoking and lung cancer was supported by ecological, cohort, and case-control done by independent investigators on different continents

Gerstman Chapter 16 9

Element 3: Specificity• Specificity ≡ the exposure is

linked to a specific effect or mechanism

• Example: Smoking is not specific for lung cancer (it causes many other ailments, as well)

Aristotle (384 – 322 BCE)

Gerstman Chapter 16 10

Element 4: TemporalityTemporality ≡ exposure precedes disease in

time

Mandatory, but not easy to prove. For example, is the relationship between lead consumption and encephalopathy this?

Gerstman Chapter 16 11

or this?

Gerstman Chapter 16 12

Element 5: Biological GradientIncreases in exposure dose dose-response in risk

Gerstman Chapter 16 13

Element 6: Plausibility• Plausibility ≡

appearing worthy of belief

• The mechanism must be plausible in the face of known biological facts

• However, all that is plausible is not always true

Gerstman Chapter 16 14

Element 7: Coherence

• Coherence ≡ facts stick together to form a coherent whole.

• Example: Epidemiologic, pharmacokinetic, laboratory, clinical, and biological data create a cohesive picture about smoking and lung cancer.

Gerstman Chapter 16 15

Element 8: Experimentation• Experimental evidence

supports observational evidence

• Both in vitro and in vivo experimentation

• Experimentation is not often possible in humans

• Animal models of human disease can help establish causality

Gerstman Chapter 16 16

Element 9: Analogy• Similarities among things

that are otherwise different• Considered a weak form of

evidence• Example: Before the HIV

was discovered, epidemiologists noticed that AIDS and Hepatitis B had analogous risk groups, suggesting similar types of agents and transmission