Causation Jay M. Fleisher. Causation Two types of medical research Bench work Epidemiology Bench...

Causation

Jay M. Fleisher

Causation Two types of medical research

Bench work Epidemiology

Bench work usually describes the underlying biology of disease

Epidemiology either tests the results of bench work on human populations or provides input to the biomedical scientist on what we still do not know

What does the term “ Causal” really mean?

Example #1 - HIV and AIDS Epidemiology identifies new disease

caused by defect in immune system Bench science identifies the infectious

agent Epidemiological studies confirm that

agent causes disease in humans Causation is proven

Example #2 - What Causes an MI

Epidemiological studies combined with laboratory study identify risk factors Cigarette smoking Cholesterol Elevated blood pressure Stress Family history Obesity Etc

Which of the above contribute the most risk What are the relationships between risk factors

Therefore: The issue of causation is not as

simple as it first appears Thus, the need for a unifying

concept of causation

A Unifying Model of Causal Relationships

The 2 Components:

Sufficient Cause precedes the disease if the cause is present, the disease always

occurs

Necessary Cause precedes the disease if the cause is absent, the disease cannot

occur

The 4 Models of Causal Relationships

1. Necessary and Sufficient*

Only Factor A Disease

* RARELY OCCUR

Genetic factors Sickle Cell Anemia

2. Necessary but Not Sufficient

Factor A

Disease

+

+

Factor B

Factor C

2. Necessary but Not Sufficient - Example

Initiation

Cancer

+

+

Latent Period

Promoter

3. Sufficient but Not Necessary

DiseaseFactor B

Factor C

Factor A

3. Sufficient but Not Necessary - Example

Leukemia

or

Benzene

Electromagnetic Fields?

Ionizing Radiation

or

4. Neither Sufficient Nor Necessary

Factor A +

+

+

and/or

and/or

Factor B

Factor C Factor D

Factor E Factor F

Disease

4. Neither Sufficient Nor Necessary - Example

Smoking +

+

+

and/or

and/or

Cholesterol

HBP Fam. History

Stress Obesity

MI

Therefore: Concept of Necessary vs. Sufficient

Causes provides a theoretical framework for causation of all disease

How do we actually assess whether a Risk Factor is indeed Causal

Criteria for Assessing Causation

Temporal relationship Exposure precedes the disease

Strength of the Association Measured by the Relative Risk ( either the Rate Ratio or the Odds Ratio)

Dose-response Relationship As the dose of exposure increases the risk of disease also increases Example: Cigarette Smoking and Lung Ca

Replication of the Findings Results replicated in other studies

Biologic plausibility Does the association fit with what we know about the underlying biology Sometimes we know little or nothing about the undelieing biology ( “Black Box” epidemiology

Consistency

Alternative explanation eliminated Cessation effects Specificity of the Association Dose-response

Criteria for Assessing Causation

Biologic plausibility Does the association fit with what we know about the underlying

biology Sometimes we know little or nothing about the underlying biology

( “Black Box” epidemiology) Example – Asbestosis and Lung Ca.. Only have theory of mechanism

Consideration of Alternate Explanations If knowledge exists, rule out or make sure studies took into

account

Cessation of Exposure If exposure is reduced or eliminated Risk will decline

Example Ex-Smokers

Specificity of the Association A specific agent is associated with only 1 disease

OK for infectious agents but falls apart with many Risk Factors for Chronic Illness

Example: Cigarette Smoking associated with several diseases

Criteria for Assessing Causation

Consistency with other knowledge If we have other knowledge regarding a Risk factor then

this comes into play Often we do not

Example: Exposure to Electromagnetic fields is a POSSIBLE risk factor

for Leukemia This finding is new and the only other knowledge we have is

from studies in changes in cells ( in vitro) Again “ Black Box” epidemiology

Criteria for Causation:Smoking and Lung Cancer

Temporal relationship Biologic plausibility Consistency Alternatives Cessation effects Specificity of

association Strength of

Association

Dose-response

Smoking before Ca Yes > 36 studies ? Yes Point of attack 25 x > 25+

cigarettes /day* Yes

*.Estimated that 80% of all Lung cancer due to Cigarette smoking

The usual bit of humor