Study methodologies: Cross sectional
The Canary database curators determine, for each included study, the
type of study methodology employed by the researchers (using
this classification protocol).
The possible categories are:
Fox has outlined criteria for objectively evaluating the relationship
between an environmental hazard and an observed health effect in an
observational study of animals (Fox 1991). These include probability, time
order, strength of association, specificity, and consistency on
replication, predictive performance, and coherence. The choice of study
design can have a major effect on the ability of a study to fulfill such
Our preliminary review of the animal sentinel literature has found that
some potentially useful study designs, such as case-control and cohort, are
under-utilized in animal sentinel research.
In a cross-sectional study, both outcomes and exposures are assessed on
the individual level, but at a moment in time, without either forward or
backward timing. Cross-sectional studies may be almost as simple and quick
to perform as aggregate studies. In an elegant example of this, investigators
used paired samples of eggs to correlate dioxin exposure levels to brain
development in wild blue herons (Henshel, Martin et al. 1995). Many of
the infectious disease studies in the Canary database measure antibodies
to infection, and were therefore classified as cross-sectional since the
test determined, on the individual level, both the occurrence of exposure
to an infectious agent as well as evidence of effect (infection).
Since information on both exposure and effect is available for each subject,
cross-sectional data can be analyzed either in terms of outcome or exposure.
When they compare groups on the basis of exposure, they resemble a cohort
analysis. When they sample groups for analysis based on outcome, the analysis
is similar to that of a case-control study. Analyzing on the basis of outcome
involves taking a group of affected individuals affected by a particular
health outcome, and comparing them with a group that does not have the outcome.
Levels of contaminants or evidence of infectious exposure can then be compared
between groups. Conversely, analyzing a cross-sectional study in terms of
exposure involves comparing the degree of health effects in a group that
has a certain exposure, such as a certain elevated level of contaminants,
with another group that has a lower level of contaminants.
In human epidemiology, the decision whether to sample and analyze based on
outcome or exposure is generally made on the basis of whichever is rarer
(Kramer and Boivin 1987). If an outcome is fairly rare, such as a limb
deformity in a frog population, it makes sense to attempt to sample on the
basis of outcome, since a sufficient number of cases could be assembled
more economically and compared with a number of unaffected individuals than
if an entire population had to be sampled. An advantage of this approach
is that for a particular outcome, a number of different etiologies can be
explored. If on the other hand, an exposure is fairly rare, it makes sense
to concentrate on the subjects that have that exposure (as in a cohort
design), and look for health effects in this population compared to an
equivalent number of individuals that do not have the exposure.
Animal sentinel studies often sample on the basis of exposure rather than
outcome. For toxin studies, this often involves comparing individuals in a
"polluted area" to individuals in a "reference" (less polluted) area and
measuring differences in outcomes between the two groups. For example, a
study compared lung histopathology in ring-billed gulls living near a
point-source of air pollution with that of gulls living in a geographically
removed area with better air quality (Yauk, Smits et al. 2001). Similarly,
infectious disease studies often compare rates of infection in animals
living in one geographic area to those in another area to determine risk
factors for infection, as when raccoons in two distinct areas were compared
for rates of leptospirosis (Mitchell, Hungeford et al. 1999).
This assumes that there is little diversity of exposure among individuals
in the "polluted site", and that it is necessary to choose "unexposed"
individuals from a separate population. In studies of animal populations,
where natural selection may be at work even in populations separated by
small distances, such an approach runs the risk of "selection bias", i.e.
making comparisons between two groups of individuals that are not really
For example, in their study of fish exposed to sewage and other water
pollutants, Karels et al found differences in the population structure
between fish who lived near sewage outflow compared to those individuals
living in a geographically removed location (Karels, Markkula et al. 2001).
Such selection pressures can cause genetic shifts, perhaps toward greater
adaptation, for a population exposed to hazards, and make comparisons to
another population less valid. For example, deer mice captured from areas
of high air pollution level have been found to be more resistant to the
effects of ozone compared with deer mice from less polluted areas, indicating
that a process of genetic adaptation on a population level had taken place
(Dickerson, Hooper et al. 1994). Similarly, it is possible that different
animal populations could have different levels of immune function that
could affect infection rates.
A key weakness of both ecologic and cross-sectional studies of toxins and
health outcomes is that they cannot determine facts about "time order" of
exposure and effect, in other words whether one preceded the other. This
is less of a problem for infectious disease studies where both the outcome
and the exposure are for the same infectious disease. Another disadvantage
of cross-sectional studies is that if a toxin or infection has a fatal effect
on animals, a cross-sectional study will tend to under-select individuals
who have been exposed. This is known as "late-look bias." Alternatively,
the cross-sectional study may over represent exposures that have sub-lethal
effects (Neyman bias). This can be a problem in studies of asymptomatic animal
reservoirs of infection. It is possible that the only survivors in an infected
population are those individuals who experienced a mild infection, thereby
again introducing selection bias into the study. In this way, a cross-sectional
study could miss a transgenerational effect that initially caused widespread
mortality of more susceptible individuals.