Monday, November 04, 2013
Here's some of what I will be saying today. The students have read two histories of the controversy over leaded gasoline and its ultimate banning. (One is Kevin Drum's famous Mother Jones article, the other an academic essay by Herbert Needleman, a scientist who was a pioneer in recognizing the harm caused by low levels of exposure to lead.) A major part of this story is of course that a powerful industry for decades dishonestly attacked science and scientists, including Needleman, in the completely amoral pursuit of economic self-interest, at incalculable human cost. This plot has been reproduced endlessly -- tobacco, acid rain, pesticides, global warming, you name it.
But you need to understand not only that this happens, but why it works. The fact is, risk assessment is hard, and inevitably freighted with uncertainty. And even after you manage to narrow down those uncertainties, there are still value conflicts about the implications of our knowledge.
In order to understand an environmental risk, you need to go through four steps:
Hazard identification means that relevant people -- those who can cause the next steps to be taken -- recognize that some chemical or other phenomenon present in the human environment might be dangerous and we ought to study it. The fact is that most of the chemicals commonly used in consumer products or emitted into the air, water and soil have not been studied much, if at all, for their potential harm to humans and other living things. Until Needleman and others came along, this was basically true of lead. We knew it was toxic at high doses, but assumed to be harmless at the levels of exposure people experienced from gasoline and paint.
Dose-response assessment means figuring out what the consequences are of exposure to the chemical at the continuum of plausible levels and patterns. This is very difficult because, obviously, you can't experiment on people. We do experiment on animals, notably mice, but let me give you the news: mice aren't people and there are plenty of well known cases in which exposure harmless to mice are dangerous to people, and vice versa. Epidemiological data is very difficult to interpret, for reasons I have discussed here before. Basically, it's difficult to separate out the exposure of interest from everything else that goes along with it, e.g. exposure to lead from gasoline depended on where you lived and spent your time so maybe the observed effects aren't because of lead at all. A corollary is that you can't easily measure people's exposure. In the case of lead, because it persists in the body, you can measure the concentration in blood, which helps. But there will still be arguments over acute vs. chronic exposure, and yes, there are confounders that go along with blood lead levels as well as imputed exposures. And you have to know what effects to look for. If you aren't measuring the relevant endpoints, you won't see them, obviously.
Exposure assessment means figuring out who is exposed to how much of the stuff and by what routes. This is also very difficult because chemicals travel through the environment along complicated paths that are affected by all sorts of influences -- wind, rain, temperature, the food chain, human behaviors, on and on. And chronicity of exposure matters as much as amount.
Finally, risk characterization means putting it all together: somehow multiplying the dose response by people's actual exposure to say "Ultrafine particle pollution from highways is causing 180,000 heart attacks each year in North America," or something like that. (I just made up the number but it does apparently cause cardiovascular disease and also triggers myocardial infarctions.) However, that's not the only consequence of UFP pollution and you need to somehow translate all of the harms of an exposure into a common currency that can be compared with the benefits. In other words, we aren't going to give up cars, trucks and buses even though we know they cause harms. We need some method of weighing harms against benefits. And of dealing with uncertainties while we do this.
Hence, people argue. That's never going to stop but at least we can develop a common understanding of the available information. Or at least it would be nice if we could.