BE.104 Spring
Environetics; Cause and Effect
J. L. Sherley
Outline
I.
II.
Environetics
Cause-Effect Determination
I. ENVIRONETICS
Definition:
Environetics is that discipline that seeks to
detect and quantify cause-effect relationships between noninfectious environmental agents and human health.
Environetics employs concepts from epidemiology,
biostatistics, logic, and good ol’ common sense to make this
determination.
For perspective, consider "The Genetic Paradigm of
Disease"
Figure 1A:
1 2 "All diseases are genetic." Not necessarily "hereditary," but
caused by defects in genes.
Refinement: All diseases are genetic except for those caused
by toxins and infectious agents, which may also have
dependencies on genotype (e.g., susceptibility)
Genetics- Currently this discipline is out of control vs.
environmental mechanisms
Are all diseases due to defects in genes?
Figure 5A:
3 4 Genetic Pedigree Analysis
The gold standard for analysis of genetic disease factors
Discover genetic disease causation (i.e. gene) by:
1) Relating disease patterns to genetic rules ⇒
Genotype classification
a) dominant
b) recessive
c) x-linked
d) age-dependent (e.g. childhood cancers)
2) “Quantitative relating” is called linking,
a statistical method for detecting and quantifying the
significance of associations between genotype and disease
occurrence (e.g., genotyping across 3x109 bp!; mutation
analyses; LOD score)
3) Establishing cause-effect:
strength of association
prospective analysis (genotype before disease)
animal models (some limitations)
5 Genetics Features
Rules
1) gene → product (protein, ribonucleic acid) → function for
health
2) Mendelian segregation- 1884
Confounders of the Analysis
1) Penetrance (“mutation expressivity”)
2) Imprinting- specific subset of penetrance (mechanism
known)
3) Phenocopy - (phenotype for other reasons besides the gene
of origin)
4) Ascertainment (e.g., determining familial relationship)
5) “Environment effects”
6) Statistical variation? Yes, spontaneous mutations (sporadic
cases)
Statistics requirements-
n is small
Key Research Tool- Old - Linkage analysis (stat. associations)
New - Genomics: How genes are expressed
⇒“quantitative gene expression pedigrees”
may be next
Experimental Support- Mouse models-Disease synteny
Transgenic KO’s
Human analysesHuman hereditary disease w/genotyping
Societal benefits-
Advances in clinical medicine
6 Environetics
7 Figure 5B: persons, exposures, geographical space, time
The component that genetics leave out
The Environmental Health Science Study or the Public Health
Study as more commonly known
-the gold standard for detecting environmental health factors
Discover chemical disease etiology (i.e., “toxin”) by:
1) Relating patterns of disease and toxicity to toxicology rules ⇒
toxicology classifications and principles of epidemiology; e.g.,
a) Acute
b) Chronic (and subchronic)
c) Clustered
d) Uniform Distribution
e) Age-dependency
f) Factor-dependency (e.g., workplace in vs outdoors)
2) “Quantitative relating" is Biostatistics, methods for detecting
and quantifying the significance of associations between toxin
exposure and health (e.g., retrospective studies, p-value,
confidence limits, RR, AR, etc.)
3) Establishing cause-effect: Prospective Intervention studies
Humans - accidents
Animal models- limited relevance to
human health
8 Environetics Features
Rules
Maimonides- 1198
Spanish physician & philosopher (Mendel, not until 1884)
“Poisons and Their Antidotes”- book published
“Earliest attempt to describe the field of toxicology”
1) Chemical → target (protein, gene, biochemical) → function
for health
2) Dose ⇒ k(disease phenotype or toxicity); i.e., DoseResponse relationship
Confounders
1) (Resistance/Susceptibility) Sensitivity- general, many
unknown & known causes
2) Barriers - agent-specific
3) “Toxocopy” - health effect for other reasons than suspected
agent (e.g., endogenous causes for asthma)
4) Exposure (were they exposed or not?)
5) Genetic/Familial effects (clusters)
6) Statistical variation and error! Major confounder
7) Human psychology
Statistics requirements- n is large compared to genetic pedigree
analyses
9 Key Research Tool-Epidemiology (associate disease w/ exposure)
On the horizon: Toxicogenomics- toxic
expression profiles
Experimental Support- Rodent- Toxicology Evaluations
Human- Public Health Interventions,
Accidents
Societal benefits- Regulations for Public Health
Populations at Risk are often special groups
(General)
Young
Old
Clinical
Workers
10 The Environetics Method
1) Design a study to look for an association
2) Confirm statistical significance
3) Perform orthogonal statistics-based study
Purpose:
Build scientific confidence & statistical confidence
4) Ideal: Intervention Analysis- Critical for establishing
Cause-Effect Relationship
The intervention is the "laboratory experiment" of a public
health study
5) Attempt to relate to toxicological mechanisms for greater
scientific confidence
11 II. CAUSE-EFFECT RELATIONSHIP
DETERMINATION
What we start with is an association:
agent X α disease or toxicity
(After first establishing a statistically significant
association!)
We want to know:
Does agent X cause the disease or toxicity?
Often, the standard applied is “Preponderance of evidence”
(the civil case); instead of “beyond a reasonable doubt” (the
criminal case)
Why do we care to establish C-E (cause effect) relationship
or to rule them out?
12 1) To intervene: control permits development of sound
and effective public health policy
2) To predict future disease/toxicity occurrences
3) To prevent wasting resources on erroneous
conclusions
4) To avoid litigation
5) To assign responsibility
13 Ideally,
To define an absolute C-E Relationship, two conditions
must be satisfied.
(Absolute means that an agent is solely responsible for an
observed disease or toxicity. This is the most
straightforward case to evaluate. Cases of multiple
responsible agents are addressed in the following
discussion).
If we say, “A is the cause of B”, then
1) If A, then B
I.e., A is sufficient for B
2) If not A, then not B
I.e., A is necessary for B
“Necessary and sufficient”
Therefore, if we alter A, we will also alter B in a
predictable fashion.
This is the Ideal C-E Relationship for effective
intervention
14 This a “Single Factor Mechanism”
(We often face multi-factor systems)
*When these are true, there is a C-E relationship; BUT, a
C-E relationship could also exist in their absence, too!
The necessary-and-sufficient test is poor for
sensitivity, but has great specificity. I.e., it
may miss responsible factors if multiple
factors can cause the disease or toxicity.
However, it gives few false positives! If an
agent meets this test, you will never
conclude it is C-E related when it is not.
15 Common Confounders & Errors when "A alone causes B"
is thought to be true.
1)
Carcinogen Fatty Foods ⇒ ↑Absorption
A (+ X)
Liver Cancer
B
A necessary, but not sufficient
Test needed to discover error:
Evaluate A vs B in different people ⇒
sometimes B, sometimes not B.
Only changing A, will mislead!
2) Carcinogen
A
Dust
Liver Cancer
B
Asthma Attack
&
HBV
(X
)B
Exercise
(“Riding bike down dusty road”)
A is sufficient, but not necessary
Test needed to discover error:
{Remove A, B remains; ↑A, will mislead!}
16 3) (Bacteria)
(X)
Fever
Sepsis
A + B
Highly correlated,
but A neither necessary, nor sufficient for B
Test needed to discover error:
Remove A, B remains; ↑A, B unchanged; fails both tests
4) A:B Chance Association!
E.g. same location, same time, etc.
Not necessary, not sufficient
Test needed to discover error:
{Remove A, B remains; ↑A, B unchanged}
5) Artifactual- Errors in method
• Statistics
• Measures, instruments
• Fraud, competency
17 Two Common Questions for C-E Analysis
1) Most common: Does A cause B?
Watch out for bias! Suspected
2) What X causes B?
Unknown!
18 Summary of the Environetics Approach
1) Rule out genetic (look for familial relations)
2) Establish statistically significant relationships
a) Retrospective typical
b) Small “pilot” study
3) Apply scientific method
a) Observe = statistical analyses
b) Develop hypothesis
c) Test hypothesis
i. statistical predictions- NEW ONES
Avoid just conducting a larger study
(e.g., first before-after exposure study; next length of
exposure study)
ii. Intervention (dose effects)
Necessary-sufficient
“Koch's Postulates”
19 Henle-Koch’s Postulates for Microbes versus Environetics
I.
Find agent in all cases of disease
II.
Isolate from diseased individuals and grow in pure culture
III. Cause disease is susceptible host
IV. Re-isolate the agent from infected hosts with disease
4) Prospective Public Health Study (also fits in 3c above)
With possibility of varying A “experimentally”
20 Consider a classic Public Health Study applying the
Environetics Approach to an example of a beneficial
chemical in the environment
1) 1930’s
Recognized that fluoride in water supply was associated
with increased prevalence of mottled enamel on teeth
because of effects on dentin deposition
2) A dentist formed the impression that people with mottled
teeth had fewer cavities:
NaF : ∆Health?
3) The U. S. Public Health Service (PHS) conducted
surveys with children 12-14 yrs old.
Fig 1-1 ⇒ F is associated with positive ∆Health (statistical
analysis not shown!)
4) Intervention:
Conduct study with similar communities with low natural
fluoride
Fluorinate some & not others
Evaluate ∆Health at a later time (i.e., prospectively)
Table 1-1
If NaF, ∆Health?
Not NaF, Not ∆Health
Sufficient (Newburgh)
Necessary (Kingston)
21 "And the rest is Environetics history…"
22
