Causal association (causal inference)
If we determine that an exposure is associated with a disease, the next step is whether the observed association reflect a causal one. The initial step may consist of clinical observation; the second step is tried to identify correlation from available data. Then we can conduct new studies, ((case-control or cohort studies)), which are specially designed to determined whether there is an association between an exposure and outcome.Association is a statistical relation ship between variables. While causation is a matter of judgment assessment of the meaning of that observed association.
Observation ( Correlation ( Association ( Causation
The presence of correlation not necessarily indicates there is a valid statistical association and the presence of valid statistical association not necessarily indicate there is causal relation.Types of association:
Spurious ((false, factitious)) association (No real association between the factor and out come.
Real ((True)) association, which is either;
Direct ((causal)), e.g. smoking(lung cancer.
Indirect association; the effect is due to hidden factor(s) "confounder", e.g. Coffee drinking
( (
Smoking ( lung cancer.
Or the factor leads to the out come through intermediate steps, Factor ( step I ( step II ( Disease
Types of causal association
Causal association has to fulfill these criteria:
The association observed should be real ((true)) and statistically significant.
Excluding the factors that cause a false association (chance, bias and confounder).
Causal association could be:
Necessary and sufficient. The factor is necessary and sufficient for producing the diseases. E.g., HBV(Hepatitis
Necessary but not sufficient. The factor alone no sufficient to cause the disease, other associated factors are needed.
E.g., T.B bacilli ( Malnutrition ( Low immunity(TB disease
Sufficient but not necessary. The factor can produce the disease, but so other factors.Smpking or HT or hyperlipidemia or DM (IHD
Neither sufficient nor necessary. The factor by it self neither necessary nor sufficient for the development of the disease. E.g., Alcohol (IHD
(Guide lines for judging whether an association is causal.
Temporal relationship: Exposure to the factor must have occurred before the disease developed and also the length of interval between the exposure and the disease. E.g., Asbestos increases the risk of Ca-lung, but Ca occurs after 20 years of exposure. If Ca occurs after 3 years(Asbestos not the cause.
Strength of association: Measured by relative risk and odd's ratio; the stronger the association (the stronger the suggestion about causal association.
Dose- response relation -ship: As the dose of the exposure increase, the risk of the disease also increase( strong evidence of causal association.
Replication of the finding: replication of the finding; the same finding in different studies and in different population.
Biological plausibility: Coherence with current body or biological knowledge. E.g., smoking wasn't considering strongly a cause of Ca lung until researches observed pathological changes in the lung tissue.
Consideration of alternative explanation: The extent to which the investigator have taken other possible explanations into account and the extent to which have ruled out such explanations.
Reversibility: If a factor is a cause of a disease, we would expect reduction in the risk of disease if we decrease the exposure.
Specifity of the association: Specific association between the exposure and the disease.
Consistency with other knowledge: If a relation ship is causal, we would expect the finding to be consistent with other data (biological, chemical .etc.)
Depart. Of Comm. Med. General Epidemiology Lec.5
د.يوسف Al-Kindy College of MedHYPER13PAGE HYPER15
1