Dissecting Hazard Identification
When the Redbook risk assessment paradigm has been applied,
the hazard identification step has typically been a subjective determination
that initiates the more formal process of conducting an exposure assessment and
characterizing the dose-response relationship for a specific cause-effect pairing
should begin. When hazard identification
itself is thought of as a formal process, the iterative nature of risk
assessment becomes apparent. This is
especially true when human epidemiological studies are pivotal in establishing
that there is a causal relationship: Hazard
identification needs a weight of evidence evaluation, and a weight of the
evidence evaluation a dose-response evaluation.
In addition, working with epidemiological data often requires estimation
of the dose that the subjects in the cohorts received, which means the hazard
ID may also require exposure assessment.
In other words, in order to conduct a proper hazard identification, one
must do a risk assessment.
However, it probably won’t be exactly the same risk
assessment as one needed for the decision at hand. For one thing, the exposures will certainly
be different for a cohort that was selected precisely because they are known to
have unusually high exposures. There
also may be differences in other causal influences on a particular disease or
health measure in a cohort that was studied and the population for which a risk
estimate is needed. Nonetheless, one
would generally expect that a dose-response analysis used to establish that
there is causal relationship should resemble the one used to estimate the risk.
Building Weight of the Evidence
So it would seem that the consideration of evidential weight
and the dose-response assessment cannot be treated as entirely separate
processes. For example, Suter and Cormier (2011)
characterized environmental epidemiology as risk assessment in reverse, where
both epidemiology and risk assessment share a common need for weighing evidence
and building the case for a causal relationship. But, the weighing of evidence itself isn’t
really reversed. Yes, one should look
for alternative plausible explanations when conducting a causal assessment, but the
need to characterize uncertainty should make that part of the risk assessment
as well. For example, the fact that an
association may be plausibly explained by both causal and noncausal
relationships may be an important source of model uncertainty. Where epidemiology does, or should, work in
reverse is during study design. Since the
expectation of how the data are going to be used drives what data are to be
collected, knowing how data becomes evidence will determine what a study looks
for. An epidemiology study designed to
support regulatory decision making should look like a risk assessment.
Human Toxicology
But, most environmental epidemiology studies are not
designed to support risk assessment. The
primary reason for this is pretty obvious and deep rooted: In spite of the fact
that it is exactly what Hill (1965) advised them not to do, most environmental
epidemiologist and their statisticians are trained to think that the weighing
of evidence is accomplished with statistical significance test, and just about
any test will do. Find a p that is less
than 0.05, publish the paper in a journal, and then call the press office to
report that an association has been found.
Demonstrating a causal relationship is not on the radar screen. The regulatory agencies need to deal with
that problem (or so I’ve been told): But
they can’t do that very well at all because the studies weren’t designed for
that purpose.
- Ban p values. There is no legitimate scientific argument that cannot be made without a statistical significance test. P values are bad for public health, bad for the environment, and bad for the economy. They are a stupid toy – take them away. A psychology journal recently banned p values; epidemiology journals need to do the same thing. Then just maybe epidemiologists will be motivated to build a case.
- Share data. Besides the p value, another legacy if statistical significance testing is the idea that decisions will revolve around a single experiment. With environmental epidemiology, it almost surely won’t. Putting data will allow data to be pooled and analyzed together. It can be reanalyzed again every time a new study comes on line. Not sharing data hides weak conclusions. Sharing data will allow stronger cases to be developed.
- Theory Matters. Unless it is a very very strong association, just demonstrating an association of some sort isn’t enough. A dose-response relationship has to be at least theoretically plausible. Likely is even better. For environmental studies concerned with the effects of chemicals, toxicology and epidemiology cannot work effectively as two entirely separate disciplines.
- Multivariate analyses. When it is necessary to distinguish the causal influence of multiple variables on a particular health outcome, far greater caution is needed. One bad apple can spoil the whole bunch; if even one variable is modeled incorrectly (i.e. with a mathematical form that isn’t likely or plausible) then all the parameter estimates from a multivariate regression will be off. In many circumstance, to may be preferable to establish the causal relationship for each variable independently, perhaps using different cohorts or studies for each. For example, the studies used to characterize the effect of smoking on lung cancer should be different than the ones used to characterize the additional effect of arsenic on lung cancer.
In fact, maybe epidemiology shouldn’t be a stand-alone
discipline at all. Instead, perhaps it
should be viewed as a form of human study design and data interpretation that
can be used to augment any biological or medical discipline.
References
Hill, Sir Arthur Bradford (1965). The Environment and Disease: Association or
Causation? Proc Royal Soc Med 58:295-300.
Suter, GW and Cormier SM (2011). Why and how to combine evidence in
environmental assessments: Weighing evidence and building cases. Science
of the Total Environment 409:1406–1417.
Official Post Soundtrack
Post Note
Thesis Post #44. Part of the solutions thread, but also an epilogue for the epidemiogical thread
No comments:
Post a Comment