Biomedical Engineering Reference
In-Depth Information
I said that we scientists often punt on causality. Punting is not a bad thing (Ask the football
coach who decides to go for the first down on fourth and inches and whose team comes up a
half-inch short. He would have likely wished he had asked for a punt!). It is only troublesome
when we use the association argument invariably (the football coach who always punts on fourth
and short might be considered to lack courage). People want to know what our findings mean.
Again, the medical science community may help us deal with the causality challenge. The best
that science can usually do in this regard is to provide enough weight of evidence to support or
reject a suspicion that a substance causes a disease. The medical research and epidemiological
communities use a number of criteria to determine the strength of an argument for causality,
but the first well-articulated criteria were Hill's Causal Criteria.
31
These criteria (Table 3.1) are
a set of guidelines
32
that enhance intuition in bridging observed effects with possible causes,
such as coherence of facts, biological gradient, and the requirement that a cause precede the
effect.
Table
3.1
Hill's
criteria
for
causality
Factors to be considered in determining whether an event is caused by an effect:
Criterion
1:
Strength
of
Association
. For something to cause an effect, it must be associated with that
effect. Strong associations provide more certain evidence of causality than is provided by weak associations.
Common epidemiological metrics are used in association include risk ratio, odds ratio, and standardized
mortality ratio.
Criterion
2:
Consistency
. If a cause is consistently associated with an effect under different studies using
diverse methods of study of assorted populations under varying circumstances by different investigators, the
link to causality is stronger. For example, if the carcinogenic effects of Chemical X is found in mutagenic-
ity studies, mouse and Rhesus monkey experiments, and human epidemiological studies, there is greater
consistency between Chemical X and cancer than if only one of these studies showed the effect.
Criterion
3:
Specificity
. The specificity criterion holds that the cause should lead to only one disease and that
the disease should result from only this single cause. This criterion appears to be based in the germ theory
of microbiology, where a specific strain of bacteria and viruses elicits a specific disease. This is rarely the
case in studying most chronic diseases, since a chemical can be associated with cancers in numerous organs,
and the same chemical may elicit cancer, hormonal, immunological, and neural dysfunctions.
Criterion
4:
Temporality
. Timing of exposure to a causative agent is critical. This criterion requires that
cause must precede the effect. For example, in a retrospective study, the researcher must be certain that the
manifestation of a disease was not already present before the exposure to the chemical. If the disease was
present prior to the exposure, it may not mean that the chemical in question is not a cause, but it does mean
that it is not the sole cause of the disease (see “Specificity” above).
Criterion
5:
Biologic
Gradient
. This is another essential criterion for pharmacological and chemical risks. In
fact, this is known as the “dose-response” step in risk assessment. If the level, intensity, duration, or total
level of pharmaceutical or chemical exposure is increased, a concomitant, progressive increase should occur
in the efficacious or toxic effect.
(
Continued
)
