Hazard Assessment and Contaminated Sites - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 541

themselves to cause the disease. A “sufficient” cause is one which “inevitably pro-
duces or initiates a disease” and a “necessary” cause is one for which “a disease
cannot develop in its absence” (Beaglehole et al. 1993 ). In the biological sciences
there is often a constellation of components acting in concert for a cause to create
an effect, and many of the components of a “sufficient cause” may be unknown
(Rothman and Greenland 1997 ). At the low levels of exposure commonly encoun-
tered in the environment and where there may be a range of contributory factors
present, it may be difficult or inappropriate to assign this nomenclature to a contam-
inant even though the contaminant is accepted as causing a specific effect with high
exposures.
As with other scientific disciplines, epidemiology has attempted to define a set
of causal criteria to help distinguish causal from non-causal associations. In the first
place other explanations for a potentially causal association must be excluded (such
as chance, selection or measurement bias, or confounding. Particularly rigorous
scrutiny should be given to studies giving a positive but not statistically significant
result. Figure 12.2 illustrates this process.
If alternative explanations such as bias and confounding can be excluded, it
is then useful to systematically apply guidelines for assessing causation from
Beaglehole et al. ( 1993 ) as shown in Table 12.2 . The concepts in these guidelines
derive from work by Hill ( 1965 ) and others. However, as Rothman and Greenland

OBSERVED ASSOCIATION

Could it be due to
selection or
measurement bias?

NO

Could it be due
to confounding?

NO

Could it be a
result of chance?

PROBABLY NOT

Could it be causal?

Fig. 12.2 Assessing the
relationship between a
possible cause and an
outcome when an association
is observed (Beaglehole et al.
1993 , p. 75)

APPLY GUIDELINES
AND MAKE JUDGEMENT