Causality

Causality in Epidemiology

Georgeta D. Vaidean, MD, MPH, PhD

Lecture 4 B PHRM 510

TCOP- Spring 2010

Objectives• To define and describe the concept of cause • To define the concept of risk factor • To define, describe and apply the concept of

Epidemiological Triangle• To define and apply the concept of “web of

causation” • To enumerate the 4 Koch’s postulates and to

apply them in examples, and point its limitation• To enumerate the 9 Hill’s criteria and to apply

them in examples• To explain and illustrate the difference between

association and causation

"Causality. There is no escape from it, we are forever slaves to it. Our only hope, our only peace is to understand it, to understand the 'why‘ “

The Merovingian, The Matrix: Reloaded

What is causality?• The concept of cause/causality continues to be

debated as a philosophical matter in the scientific literature.

• Many definitions:• Science• Philosophy• Law• Psychology• History• Religion• Engineering

• “Cause is the effect concealed, effect is the cause revealed” (Hindu philosophy)

• Medicine is very much centered around the concept of “cause”– Doctor, what caused my diabetes?– Doctor, is this drug going to cause me side effects?

• Epidemiology – What are the causes of CVD epidemic? – Why obese people have more DM than non obese

people?– What are the causes of the H1N1 epidemic?

• Public Health Interventions– What is the best public health measure to take to

cause a reduction in the incidence of Asthma in Manhattan?

Causality: The cause effect relationship

• What do we call “Cause”– Any exposure to anything can be a “cause” – Negative

• 1. “natural”: smoking, obesity, alcohol consumption, viruses, bacteria, “stress”, poverty, genetics, poor diet, accidents, etc.

• 2. determined by us: iatrogenic : caused by medical actions: adverse effects of a drug, post-op complication etc.

– Positive: • 3, “natural “: good diet, exercise, healthy habits, etc• 4. determined by a researcher in a RCT: a drug meant to

ameliorate a condition • What do we call “effect”

– Any consequence, result, or aftermath of a “cause” – Negative: disease, death, complication, accident, epidemics,

depression, any outcome– Positive: longevity, disease in remission, happiness, etc.

What is a Cause?

• Merriam-Webster Dictionary: Something that brings about a result especially a person or thing that is the agent of bringing something about.

• KJ. Rothman (a famous US Epidemiologist): An event, condition, or characteristic without which the disease would not have occurred.

• M. Susser (a famous US Epidemiologist): Something that makes a difference.

What are some of the causes of the following diseases and events?

Think about (with what you already know):

• Influenza:

• Lung Cancer:

• Automobile Fatality:

HISTORICAL DEVELOPMENT OF THEORIES OF CAUSATION

1. Supernatural causes. Divine retribution; imbalance in body humors caused by air, water, land, stars; spontaneous generation

2. Hipoccratic theories3. Miasma: Disease transmitted by miasmas or

clouds clinging to earth’s surface

HISTORICAL DEVELOPMENT OF THEORIES OF CAUSATION

3. Germ Theory of Disease and Henle-Koch postulates for establishing causation in Infectious Diseases

4. Web of Causation- establishing causation in chronic diseases.

Basic Question in Analytic Epidemiology

• Is there a relationship between an exposure and a disease?

ExposureAgent

DiseaseOutcome

Note: we do not call it “cause” because it has to be proved that it is indeed a cause . Declaring some agent as a “cause” require lots of arguments. Keep reading

E D

Basic Question in Analytic Epidemiology-Examples

• Is there a relationship between thimerosal content of vaccines and pediatric schizophrenia?

ExposureAgent

DiseaseOutcome

Note: we do not call it “cause” because it has to be proved that it is indeed a cause . Declaring some agent as a “cause” require lots of arguments. Keep reading

E D

• Diseases/negative outcomes, are rarely caused by one single agent/exposure. – Can you give examples?

• There is only rare that an exposure causes an outcome in 100% of the situations. An example: rabies exposure causes death in most exposed unvaccinated people. Can you give other examples? Otherwise, there are various degrees of variability– Not all smokers get lung cancer– Not all obese people get DM – Not all patients have a particular adverse effect to a drug.

• Most diseases/outcomes have multiple factors concurring to their occurrence. They are called multifactorial diseases. The concurring factors are called “risk factors”.

RISK FACTOR

• A RISK FACTOR is an exposure/agent that increases the chances/probability of getting a particular disease/outcome.

• Most diseases have multiple risk factors

E DEEE

EEE

The Epidemiological Triangle (Triad)

Originally developed in order to understand infectious diseases.

Nowadays, it is also used in chronic diseases.

The Epidemiological Triangle (Triad) HOST

Immunity and immunological response

Host behavior

ENVIRON

Physical environment ( hear, cold, moisture, pollution, etc)

Biological environ: flora, fauna

Social environment ( economic, politic, cultural

AGENTS

Living organisms: infectivity, pathogenicity, virulence; may require vectors, carriers

Exogenous & endogenous chemicals

Genetic traits

Psychological factors, stress

Nutritive elements

Physical forces

The Epidemiological Triangle (Triad) With what you know about disease pathology, in general:

Can you use this model to grossly explain the causality of Tuberculosis?

Can you use this model to grossly explain the causality of CHD/heart attack?

How do we establish causality?

• 1. Infectious diseases• 2. chronic diseases

Henle-Koch's postulates (1877,1882)

Koch stated that four postulates should be met before a causal relationship can be accepted between a particular bacterial parasite (or disease agent) and the disease in question. These are:

1. The agent must be shown to be present in every case of the disease by isolation in pure culture.

2. The agent must not be found in healthy or cases of other diseases.

3. The agent can be isolated and once isolated, the agent must produce the disease in healthy subjects (can be experimental animals or humans, by contagion)

4. The agent must be isolated from the experimental disease produced.

Source: http://www.science-art.com

Apply this model to H1N1 infection

• While useful, Koch’s postulates have limitations

-Some bacteria (e.g.leprosy) cannot be "grown in pure culture" in the laboratory.

-Animal models may not always exist

-A harmless bacteria may cause disease if:- it acquired extra virulence factors,- it gains access to deep tissues via trauma, an IV line,- if immunocompromised patient.

-There are subclinical infections, which may not be detected

This model has limitations even in the area of Infectious disease. How can it be applied to non-infectious /chronic diseases? It can not

In chronic diseases, causality is complicated

• We rarely deal with 1 agent. • We deal with “agents” that are multiple. Host-related and

Environmental –related factors are multiple as well. They are all “ risk factors” and furthermore, they interact with each other.

• In trying to explain this multifactorial causation, people came up with various models, expanding the Epidemiological Triangle into more complex models, such as the Web of causation.

• The "web of causation" is not an explanation, it is a metaphor for the idea that causal pathways are complex and interconnected.

Web of Causation

Source: RS Bhopal

Disease

behaviourUnk

nown f

actors

genes

phenotype

workplace

socia

l org

aniza

tion

microbes

environment

Illustrates the interconnectedness of possible causes

Web of Causation –Example CHD

Disease

smokingUnk

nown f

actorsgender

genetic susceptibility

inflamm

ation

med

icatio

ns

lipids

physical activityblood pressure

stress

Causal Pathways• How causes may be linked to each other and to the outcome• Typical for chronic diseases, complex pharmacoepi

situations, pharmacovigilance and other applications.

• The following 4 slides are illustrations/ examples from CVD. You are NOT required to know the components of these pathways. You are only required to understand 3 concepts:– We talk about multiple factors interacting= “web of causation”– We can talk about causality at different levels: Molecular /

Physiological ; Personal / Social ; etc. – There are “direct causes” and “indirect causes”( with intermediate

effects)• Whenever you will ask a “Why” question about any medical

situation, think about how you could draw such a graph for your question of interest. It will help you to think critically, to develop a good decision making habit and to become a good researcher, should you chose to become one.

Source: Pitt edu

* IHD: Ischemic heart Disease = Coronary heart disease

• Epidemiology does not determine the cause of a disease in a given individual

• Instead, it determines the relationship or association between a given exposure and frequency of disease in populations

• In Epi- we ask: Do persons with exposure, E. have more disease, D, than persons without exposure E? In other words: is disease D more frequent among exposed> non exposed?

• We infer causation based upon all information we can gather from various sources.

Problem: We said that a cause is something that “makes a difference”. How

do we know that something makes a difference??

Consider the following statement: If the rooster crows at the break of dawn, then

the rooster caused the sun to rise.

What is the problem with this statement?

Problem: How do we know when something makes a difference?

Association does not imply causation

This is one of the very few things in PHRM 510 that needs to be memorized. You should be able to recite this at any time, day or night.

If tattooing is something that works for you, this might be a good one.

Everyday funny examples• More fatal accidents occur in red cars than blue cars.

Color red kills people• Beds are the most dangerous thing in the world

because more people die in beds than elsewhere• Bread is really, really dangerous (see the joke at the

end of the lecture)

• Yes, we are still talking ‘epidemiology’ here• While these may seem obvious, you w’d be surprised

to learn that many times we can find similar fallacies in published papers.

Association vs. Causation• Association - an identifiable relationship

between an exposure and disease– implies that exposure might cause disease

• Causation - implies that there is a true mechanism that leads from exposure to disease

• Finding an association does not make it causal

• We should not design interventions based upon associations!

Types if Causal Relationships• Necessary and sufficient

– without the factor, disease cannot occur; with the factor, disease always occurs

• Necessary but not sufficient – The factor in and of itself is not enough to cause disease ( e.g.

carcinogens) • Sufficient but not necessary

– The factor alone can cause disease, but so can other factors in its absence e.g. benzene or radiation can cause leukemia without the presence of the other

• Neither sufficient nor necessary – The factors cannot cause disease on its own , nor it is the only

factor that can cause that disease – This is probable the most common situation in chronic diseases.

Comparing rules of evidence Criminal Law

• Criminal present at scene of crime

• Premeditation• Accessories involved in the crime • Severity of crime related to state

of victim• Motivation- there must be gain to

the criminal• No other suspect could have

committed the crime• Proof of guilt must be established

beyond reasonable doubt

Causation • Agent present in the disease• Causal events precede onset of

disease• Cofactors and/or multiple causality

involved • Susceptibility and host response

determine severity • The role of the agent in the

disease must make biological sense

• No other agent could have caused the disease under the circumstances given

• Proof of causation must be established beyond reasonable doubt or role of chance

How do we establish causation ? • In contrast with experimental medicine or experimental

pharmacology, proving causation in epidemiology is much more difficult. Among reasons:– We cannot put people in Petri dishes and alike. The closest to

an experiment as we can get is when we do RCT– We cannot manipulate/control all determinant factors, the same

way we do in a lab with temperature, enzymes, pH, etc. All we can do is to collect data and then do our best to read “patterns” in that data. The closest to controlling anything ( outside a RCT), is what is called “statistical adjustment” ( more later)

– People are more complicated than cells, biopsy tissue, molecules, etc. Either patients in a clinic ( Clin Epi) or communities if people (Public Health) are complicated “study subjects”

• Causation cannot be established with one “experiment” be it even a RCT. Declaring causation needs lots of “evidence” coming from cumulating info from many studies.

How do we establish causality ?

• At individual level: clinical judgment (which management scheme)

• At population level: epidemiological judgment (which intervention)

• When weighing evidence from epidemiological studies, we use “causal criteria” (usually applied to a group of articles).

How do we establish causation?• We saw that in Infectious Diseases (ID) we can

prove causation if the 4 Henle-Koch's postulates are met. This may take at least 4 good experiments

• While useful, these postulate have sometimes limitations, even for ID; and they certainly cannot be applied to chronic diseases

• How do we approach chronic diseases?• To come with something similar to Koch’s

criteria, but applicable to chronic diseases, Austin Hill proposed his criteria, now widely accepted.

Causal "guidelines" suggested by Sir Austin Bradford Hill (1965)

Strength of the associationConsistencySpecificity

TemporalityBiological gradient

PlausibilityCoherenceExperiment

Analogy The most important criteria

1. Strength of the association

In Epidemiology, we can measure how strong is the association between a risk factor and a disease/outcome. No risk factor is a sure cause, but some factors are stronger than others to increase the risk of developing a disease.

(more later)

E.g. Lung cancer is more common among people who smoke, have air pollution,

asbestos or radon exposure, have vit A deficiency; some genes may also be . Smokers are 9 times more likely to develop lung cancer than non-smokers; people with vit A deficiency are 1.3 more likely to develop lung cancer smoking is a stronger RF > Vit A deficiency, and much more likely to be trully “causal”

1. Strength of the association (cont’d)

• Strong associations are more likely to be causal because they are unlikely to be due entirely to study errors ( bias and confounding, more about these later)

• Weak associations may be causal but it is harder to rule out study errors.

2. Consistency/Replication

• Relationships that are demonstrated in multiple studies are more likely to be causal

• Look for consistent findings– across different populations– in differing circumstances– In different samples– with different study designs– With different investigators

• Consistency could either mean:– Exact replication (as in lab sciences, impossible in epidemiological studies)– Replication under similar circumstances (possible)

• Meta-analysis ( combining similar studies together) is a good method for testing consistency.

2. Consistency (cont’d)

Example: Smoking has been associated with lung cancer in at least 29 retrospective and 7 prospective studies.

Note: Sometimes there are good reasons why study results differ. For example, one study may have looked at low level exposures while another looked at high level exposures.

3. Specificity

• A single exposure should cause a single disease.

• This is a hold-over from the concepts of causation that were developed for infectious diseases. There are many exceptions to this.

3. Specificity (cont’d)

Example: Smoking is associated with lung cancer as well as many other diseases. In addition, lung cancer results from smoking as well as other exposures.

• When present, specificity does provide evidence of causality, but its absence does not preclude causation.

4. Temporality

• Time order ( directionality): The causal factor must precede the disease in time.

• A sine qua non for causality

• Can be established by prospective studies only ( cohorts, RCTs)

Example

• Low serum cholesterol has been linked to increased risk of colon cancer in various studies. But, is a low serum cholesterol a cause of colon cancer, or does an early phase of colon cancer cause low cholesterol levels?

5. Dose-Response/Biological Gradient

• A “dose-response” relationship between exposure and disease. Persons who have increasingly higher exposure levels have increasingly higher risks of disease.

Example: Lung cancer death rates rise with the number of cigarettes smoked.

• Some exposures might not have a "dose-response" effect but rather a "threshold effect" below which these are no adverse outcomes.

6. Biological Plausibility• Biological explanations exists to explain the association.

Example: Cigarettes contain many carcinogenic substances, and the mechanism of action of these carcinogens is known, from various studies

• But: many epidemiologic studies have identified cause-effect relationships before biological mechanisms were identified. For example, the carcinogenic substances in cigarette smoke were discovered after the initial epidemiologic studies linking smoking to cancer.

7. Coherence

• A cause-and-effect interpretation for an association is clearest when it does not conflict with what is known about the variables under study and when there are no plausible competing theories or rival hypotheses. In other words, the association must be coherent with other knowledge.

• Similar with biological/theoretical plausibility

Biological Coherence • Reserpine ( an early anti-hypertensive

drug) was thought to be a cause of breast cancer, based on some studies in the early 1970’s. But, there was no other supporting biological information, or any truly plausible biological mechanism. Subsequent larger studies failed to support this association.

8. Experiment

• Investigator-initiated intervention that modifies the exposure through prevention, treatment, or removal should result in less disease.

Examples: Smoking cessation programs result in lower lung cancer rates.

• Provides strong evidence for causation ( this is one reason why we put so much weight on the evidence coming from RCTs)

9. Analogy

• Sometimes a commonly accepted phenomenon in one area can be applied to another area.

• Example: Effects of Thalidomide and Rubella on the fetus provide analogy for effects of similar substances on the fetus.

Hill concludes…“Here then are nine different viewpoints from all of which

we should study association before we cry causation.... None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. What they can do, with greater or lesser strength, is to help us 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 was ..modest. Temporality IS a sine qua non for causality.

Bradford Hill's “guidelines" are useful guides for:

•Remembering distinctions between association and causation in epidemiologic research

•Critically reading epidemiologic studies

•Designing epidemiologic studies and interpreting the results of your own study.

•Remembering that no single study is sufficient for causal inference•Remembering that causal inference is not a simple process

•consider weight of evidence•requires judgment and interpretation

Judging Causality

Weigh weaknessesin data and other

explanations

Weigh qualityof science and

results of causalmodels

Source: RS Bhopal

Prevailing Wisdom in Epidemiology

• Most judgments of cause and effect are tentative, and are open to change with new evidence

Pyramid of Associations

Causal

Non-causal

Confounded

Spurious / artefact

Chance

Source: RS Bhopal

BMJ. 2001; 322(7280): 226–231.

The Dangers of BreadSource: The Internet

• A recent Cincinnati Enquirer headline read, "Smell of baked bread may be health hazard." The article went on to describe the dangers of the smell of baking bread. The main danger, apparently, is that the organic components of this aroma may break down ozone (I'm not making this stuff up).

• I was horrified. When are we going to do something about bread- induced global warming? Sure, we attack tobacco companies, but when is the government going to go after Big Bread?

• Well, I've done a little research, and what I've discovered should make anyone think twice....

The Dangers of Bread-cont’ed

• More than 98 percent of convicted felons are bread eaters. • Fully HALF of all children who grow up in bread-consuming

households score below average on standardized tests. • In the 18th century, when virtually all bread was baked in the home,

the average life expectancy was less than 50 years; infant mortality rates were unacceptably high; many women died in childbirth; and diseases such as typhoid, yellow fever and influenza ravaged whole nations.

• More than 90 percent of violent crimes are committed within 24 hours of eating bread.

• Bread is made from a substance called "dough." It has been proven that as little as one pound of dough can be used to suffocate a mouse. The average American eats more bread than that in one month!

The Dangers of Bread-cont’ed

• Primitive tribal societies that have no bread exhibit a low occurrence of cancer, Alzheimer's, Parkinson's disease and osteoporosis.

• Bread has been proven to be addictive. Subjects deprived of bread and given only water to eat begged for bread after only two days.

• Bread is often a "gateway" food item, leading the user to "harder" items such as butter, jelly, peanut butter and even cold cuts.

• Bread has been proven to absorb water. Since the human body is more than 90 percent water, it follows that eating bread could lead to your body being taken over by this absorptive food product, turning you into a soggy, gooey bread-pudding person.

• Newborn babies can choke on bread. • Bread is baked at temperatures as high as 400 degrees Fahrenheit!

That kind of heat can kill an adult in less than one minute.

• Most American bread eaters are utterly unable to distinguish between significant scientific fact and meaningless statistical babbling.

• In light of these frightening statistics, we propose the following bread restrictions:

• No sale of bread to minors. • No advertising of bread within 1000 feet of a school. • A 300 percent federal tax on all bread to pay for all the societal ills

we might associate with bread. • No animal or human images, nor any primary colors (which may

appeal to children) may be used to promote bread usage. • A $4.2 zillion fine on the three biggest bread manufacturers. • Please send this e-mail on to everyone you know who cares about

this crucial issue.

The Dangers of Bread-cont’ed

Supplemental Slides for Optional Readings

• The following slides are optional, for those with a special interest

• I highly encourage everyone to retrieve the papers and read, for your own intellectual delight

• NO questions will be asked from the material in the following slides, at any exam, quiz, or assignment in PHRM 510. No bonus points either.

The concept of “Sick individuals and sick

populations”• An exceptionally illuminating paper, a

landmark for understanding subtleties in in Epidemiology work and research

• Sick individuals and sick populations. Geoffrey Rose, 1985 – Reprint: Rose G.Int J Epidemiol. 2001

Jun;30(3):427-32; PMID: 11416056

Causal propositions in clinical research and practice.

– “The word cause is an abstract noun and, like beauty, will have different meanings in different contexts”[ PMID: 1607903]

• “Philosophers agree that causal propositions cannot be proved, and find flaws or practical limitations in all philosophies of causal inference. Hence, the role of logic, belief, and observation in evaluating causal propositions is not settled. Causal inference in epidemiology is better viewed as an exercise in measurement of an effect rather than as a criterion-guided process for deciding whether an effect is present or not.”

– Rothman KJ, Greenland S. PMID: 16030331

• “Knowledge and sound methods are both vital for understanding diseases, but so are the investigator's vision, intuition, and imagination. Historical example shows that significant research is done not by taking the world as one finds it, but by viewing it differently, by the disciplined use of creativity and imagination, and a willingness to challenge accepted categories and models of disease.”

• [Mawson AR., 2002 .On not taking the world as you find it-epidemiology in its place. J Clin Epidemiol. 2002 Jan;55(1):1-4. Review. PMID: 11781115]

Counterfactual models

Exposed

Non-Exposed

Yes

Yes

No

No

EFFECT, OUTCOME“CAUSE”, Exposure

amount of the effect which would have been observed, if the

same population would not have been exposed to that cause,

all other conditions remaining identical.

Causal criteria and counterfactuals; nothing more (or less) than scientific common sense.

Phillips CV, Goodman KJ.Emerg Themes Epidemiol. 2006 ;3:5.PMID: 16725053