From association to causation

From association to causation

*

Objectives of lecture:

• what is the causeconcept of Association and Causation• types of association• does association implies causation• what are the types of causal factorsWhat are the Hill’s criteria for causation. *

- Purpose of Epidemiology? - Basic Question in Analytic Epidemiology

Exposure
Disease

Research Questions/Hypotheses

Is there an association between Exposure (E) & Disease (D)? Hypothesis: Do persons with exposure have higher levels of disease than persons without exposure?

250
200
150
100
50
0
0
1600
1400
1200
1000
800
600
Japan
Denmark
New Zealand
Fed. Repub. Of Germany
France
Canada
Israel
Switzerland
USA
Australia
Yugoslavia
Hong Kong
Romania
Finland
Poland
Spain
Hungary
Norway
UK
Italy
Sweden
Incidence Ratio per 100,000 Women
Per Capita Supply of Fat Calories
Figure 13-4. Correlation between dietary fat intake and breast cancer by country.
Prentice RL, Kakar F, Hursting S, et al: Aspects of the rationale for the Women’s Health Trial. J Natl Cancer Inst 80:802-814, 1988.)


DOES CORRELATION IMPLY CAUSATION?Correlation: it indicates degree of association.Association……Do both invoke causation?

Types of Association: 

A real association (causal) is present if we succeed in reducing the risk of disease by lowering the exposure level. Spurious associations refer to non-causal associations due to chance, bias, failure to control (confounding), etc. e.g; Neonatal deaths at Home vs hosp delivery,

Interpreting Associations - Causal and Non-Causal

Causal
Non-Causal (due to confounding)
Characteristic Under Study
Characteristic Under Study
Disease
Factor X
Disease

Types of real association: A) Negative Association (Inverse Relationship): higher levels of the risk factor are protective against the outcome. e.g. regular physical exercise and ischemic heart diseases B) Positive Association: The magnitudes of both variables appear to move together up or down. e.g. hypertension and ischemic heart diseases

Pyramid of Associations

Causal
Non-causal
Confounded
Spurious / artefact
Chance



Association : the rate of disease in a persons with a specific exposure. statistical dependence between two variables,• • A causal association is one in which a change in thefrequency or quality of the exposure results in a corresponding change in the frequency of the disease or outcome of interest.

What is a Cause?

An event, condition, or characteristic without which the disease would not have occurred. Causative factor:


A good epidemiologic exposure variable should…. Have an impact on health Be measureable Differentiate populations Generate testable hypotheses Help to prevent or control disease

ASSOCIATION VS CAUSATION

To decide whether exposure A causes disease B, we must first find out whether the two variables are associated?

Clinical observations

Available data (Ecological or Cross-sectional Studies)
Case-control studies
Cohort studies
Randomized trials (only used to confirm causation



Surgeon Alton Ochsner observed that almost all lung cancer patients were smokers
Ecologic study of per capita smoking and lung cancer incidence
Case-control study of lung cancer patients versus those without lung cancer 39 cohort studies ?

Types of causal pathway: 1. Direct causation (with out any intermediate step). 2. Indirect causation (a factor causes a disease, but only through an intermediate steps).

- Which one is always present in human biology?

* factor
disease
factor
Step 1
Step 2
disease

Causal Factors in general: Necessary cause: The cause must be present for the outcome to happen. Sufficient cause: If the cause is present the outcome must occur. Rare or Common? Isolation/inclusion

Types of causal factors

1. Necessary and sufficient Without factor, disease does not develop Its presence is sufficient to initiate disease

2. ‘necessary but not sufficient’: causative factor, and enabling factors requiredEach factor is necessary but not sufficient to cause the disease. e.g: initiator + promoter * Factor A
Factor B
Factor C
Disease
Carcinogenesis


3. “Sufficient but not necessary’:

the factor alone can produce the disease, but so can other factors that are acting alone.
* Factor A
OR
Factor B
OR
Factor C
DISEASE

4.Neither sufficient nor necessary

* Factor A
Factor B
Factor C
OR
Factor D
Factor E
OR
Factor F
DISEASE

CAUSAL INFERENCE

Web of Causation
Disease
behaviour
Unknown factors
genes
phenotype
workplace
social organization
microbes
environment

Web of Causation - CHD

CHD
smoking
Unknown factors
gender
genetic susceptibility
inflammation
medications
lipids
physical activity
blood pressure
stress


MAKING CAUSAL INFERENCES
The use of Hill’s causal criteria in making inferences from data.

Hill’s criteria for judging whether an association is causal Temporal relationship Strength of association Dose response relationship Consistency of the findings Biologic plausibility and Coherence with other knowledge
Consideration of alternate explanations Reversibility of exposure Specificity of the association Study design

Temporal Sequence

It means? The length of exposure is also important, Which study designs?

STRENGTH

Is the association strong? Heavy smoking with lung cancer, Heavy smoking with CHD. The stronger the association the more likely it is to be truly causal. WHY?? small associations can be causal?

dose-response relationship relative risk increases with higher exposure dose.Years of exposure? Not necessary. Why? … threshold effect… unascertained exposure level … irrelevant to all exposures.

Consistency of Findings of Effect

Does mean exact replication? Look for consistent findings across different populations in differing circumstances with different study designs Relationships that are demonstrated in multiple studies are more likely to be causal. Why? Smoking with lung cancer. contraceptives and breast cancer.

Plausibility and Coherence (ie., biologic credibility)

• existence of biologic or social mechanistic model to explain the association e.g. Trinitrate → headacheFitness .• does it a provision?

EXAMPLE: Serological marker of hepatitis B infection is associated with elevated rates of liver cancer. Is supported by viral genome in many liver cancers. So… Hepatitis B infection is a true cause of liver cancer


Consideration of alternate explanations - review of literatures for analogue. - depend on depth of knowledge.

* Reversibility (cessation of exposure) : Removal of the exposure leads to decreased risk of outcome. e.g. . stop smoking leads to decreased risk of CHD. HOWEVER, in certain cases, the damage may be irreversible.

Specificity of the Association

Is it a Provision? - H. Pylori? - smoking & lung cancer?

Study design :

All of the study designs are important. The best evidence comes from?. Example: Smoking cessation programs result in lower lung cancer rates.

To put the studies in order according to power:

Ability to " prove causation
Type of study
Strong
Randomized controlled trial
strong
Other intervention study
Moderate
prospective cohort
moderate
Retrospective cohort
Moderate
Case-control
Poor
Cross-sectional study
Poor
Ecological
Very Poor
Parametric study

Judging Causality

Weigh weaknesses in data and other explanations
Weigh quality of science and results of causal models

Framework for the Interpretation of an Epidemiologic StudyI. Is there a valid statistical association? • Due to chance? • Due to bias? • Due to confounding?II. Can this valid statistical association be judged to be one of cause and effect? • Is there a strong association? • Is there consistency with other studies? • Is the time sequence compatible? • Is there biologic credibility to the hypothesis? • Is there evidence of a dose-response relationship?

Why was it relatively easy to determine that smoking was a cause of lung cancer?

Measurable. Used to classify persons accordingly. Lung cancer incidence in smokers is much greater than in nonsmokers.

Why will it be relatively hard to determine if community air pollution is a cause of lung cancer?