background image

Association and causation 

Dr. Sijal Fadhil  Farhood AL-joborae 
F.I.C.M.S   community (Baghdad) 
M.Sc. Community (Nahrain) 
M.B.Ch.B (Babylon University) 


background image

introduction 

In previous lectures  we studied the designs of  
epidemiologic studies that are used to determine whether 
an association exists between an exposure and a disease 
,We then addressed different types of risk measurement 
that are used to quantitatively express an excess in risk. 
 If we determine that an exposure is associated with a 
disease. 
The next question is whether the observed association 
reflects a causal relationship???? 


background image

 


background image

TYPES OF ASSOCIATIONS 
 

Real or Spurious Associations 

Interpreting real association: 

A :shows a causal association: we observe an association of 
exposure and disease and the exposure induces development of 
the disease. 
B:shows the same observed association of exposure 
and disease, but they are associated only because they are both 
linked to a third factor, designated here as factor X. This 
association is a result of confounding and is non-causal. 


background image

 


background image

 


background image

TYPES OF CAUSAL 
RELATIONSHIPS
 

A causal pathway can be either 

direct 

or 

indirect 

In 

direct 

causation, a factor directly causes a disease without any 

intermediate step. In 

indirect 

causation, a factor that causes a 

disease, but only through an intermediate step or steps. In 
human 
biology, intermediate steps are virtually always present in any 
causal process. 

 


background image

If a relationship is causal; four types of causal relationships are 
possible:
  

                   (1) necessary and sufficient. 
                   (2) necessary, but not sufficient.  
                   (3) sufficient, but not necessary.  
                   (4) neither sufficient nor necessary


background image

 


background image

background image

Necessary and Sufficient 
A factor is both necessary and sufficient for 
producing the 
disease. Without that factor, the disease never 
develops(the factor is necessary), and in the 
presence of that factor, the disease always develops 
(the factor is sufficient) 


background image

For example, in most infectious diseases, a number of 
people are exposed, some of whom will manifest the 
disease and others who 
will not. Members of households of a person with 
tuberculosis do not uniformly acquire the disease from 
the index case. If the exposure dose is assumed to be the 
same, there are likely differences in immune status, 
genetic susceptibility, or other characteristics 
that determine who develops the disease and who does 
not. A one-to-one relationship of exposure to disease, 
which is a consequence of a necessary and sufficient 
relationship, rarely if ever occurs. 


background image

Necessary, But Not Sufficient 
Each factor is necessary, but not, in itself, sufficient 
to cause the disease Thus, multiple factors are 
required, often in a specific temporal sequence 


background image

carcinogenesis is considered to be a multistage 
process involving both initiation and promotion. 
For cancer to result, a promoter must act after an 
initiator has acted. Action of an initiator or a 
promoter alone will not produce a cancer. 


background image

Sufficient, But Not Necessary 
In this model, the factor alone can produce the disease, but 
so can other factors that are acting alone ,Thus, either 
radiation exposure or benzene exposure can each produce 
leukemia 
without the presence of the other. Even in this situation, 
however, cancer does not develop in everyone who has 
experienced radiation or benzene exposure, so although 
both factors are not needed, other cofactors probably are. 
Thus, the criterion of 

sufficient 

is rarely met by a single 

factor. 


background image

Neither Sufficient Nor Necessary 
In the fourth model, a factor, by itself, is neither 
sufficient nor necessary to produce disease ,This is a 
more complex model, which probably most 
accurately represents the causal relationships that 
operate in most chronic diseases. 


background image

 The Bradford Hill criteria  ((Hill's criteria)) 
for causation
, are a group of guidelines 
that can be useful for providing evidence 
of a causal relationship between a putative 
cause and an effect, established by 
the 

English

 

epidemiologist

 Sir 

Austin 

Bradford Hill

 (1897–1991) in 1965. 


background image

Guidelines for judging whether an observed 

association is causal 


background image

1.

 

Temporal Relationship.

 It is clear 

that if a factor is believed to be the 
cause of a disease, exposure to the 
factor must have occurred before the 
disease developed. 


background image

2-

 

Strength of the Association: 

The strength of 

the association is measured by the relative risk (or 
odds ratio). The stronger the association, the more 
likely it is that the relation is causal. 

3-

 

Dose-Response Relationship:

 As the dose of 

exposure increases, the risk of disease also 
increases. 


background image

4-Replication of the Findings.: 

If the 

relationship is causal, we would expect to 
find it consistently in different studies and 
in different populations. 
5-Biologic Plausibility: Biologic plausibility 
refers to coherence with the current body of 
biologic knowledge. 
 


background image

6-Consideration of Alternate Explanations: 

We have discussed the problem in interpreting an observed 
association in regard to whether a relationship is causal or is the 
result of confounding. In judging whether a reported association 
is causal, the extent to which the investigators have taken 
other possible explanations into account and the extent to 
which they have ruled out such explanations are important 
considerations. 

7-Cessation of Exposure.:

 If a factor is a cause of a 

disease, we would expect the risk of the disease to decline when 
exposure to the factor is reduced 


background image

8-Consistency with Other Knowledge:

 If a 

relationship is causal, we would expect the 
findings to be consistent with other data. 

9-Specificity of the Association:

 An 

association is specific when a certain exposure is 
associated with only one disease; this is the 
weakest of all the guidelines. 


background image

Question: 

** All of the following are important criteria when 
making causal inferences except: 
a. Consistency with existing knowledge 
b. Dose-response relationship 
c. Consistency of association in several studies 
d. Strength of association 
e. Predictive value 


background image



رفعت المحاضرة من قبل: Ahmed monther Aljial
المشاهدات: لقد قام عضو واحد فقط و 114 زائراً بقراءة هذه المحاضرة








تسجيل دخول

أو
عبر الحساب الاعتيادي
الرجاء كتابة البريد الالكتروني بشكل صحيح
الرجاء كتابة كلمة المرور
لست عضواً في موقع محاضراتي؟
اضغط هنا للتسجيل