Last week I had the privilege to attend a conference on risk science in New Orleans, and hold a roundtable together with neuroscientist Alison Bernstein from SciMoms. We were kindly invited to talk about our approach to risk communication by Center for Food Integrity.
This was awesome! Not only did I and Alison finally get to meet face to face, we also got to present what we had learned about messages that work on social media (see our series: Risk In Perspective), as well as ones that don’t. In between many interesting sessions, on everything from food fears to radiological risk to supplements and climate change mitigation, Alison also introduced me to fresh beignets, fried green tomatoes and other New Orleans essentials.
For our presentation, one of the things we talked about was the varied success of our infographics for the piece Measures of Toxicity. In the article, we underlined the distinction that acute toxicity info (like LD50) is mostly relevant to accidents, murders and suicides, whereas what the public is often most concerned about are the gradual chronic effects. These are more appropriately reflected by the chronic intake limits, which are set in a way to try hard to avoid such effects.
We talked about what these metrics can and can’t tell us. We stressed that they themselves tell us nothing about the risks to consumer, because you have to consider actual exposures relative to these limits to assess risk.
We made tables of toxicity measures to go with the article, keeping in mind that social media heavily favours visual posts. With the infographics we wanted to make the point that any substance can cause great harm if the dose is high enough, and similarly, will have levels at which we expect there to be virtually no impact.
Now, always when you post an infographic, you know there will be a lot of people who will not read the accompanying article, so we have to try to make them cogent and clear enough to stand on their own. These tables evoked a lot of discussion, which might be their strongest point, and many appreciated and shared them. But it is also questionable whether they actually work well on their own, and people’s reactions to them taught me several things, some of which I want to discuss here.
First, people love their coffee! They will not take kindly to anything suggesting coffee could have risks. (No, the table values do not suggest we should give up coffee – or that we should take up a habit of drinking lead or hydrogen cyanide instead 😉 )
Secondly, how the regulatory intake limits (used in the chronic table) relate to consumer risk is not very clear to many people. When we think of the word ‘acceptable’ in everyday sense, it sounds rather like a mediocre test score, not a value hundred or thousand times below levels where minor harm might occur.
Considering that these limits are referred to in health-related news as well, this is a potentially a very large point of misunderstanding. Hearing there is an ‘acceptable’ limit for a substance, people quickly assume that crossing that limit must land us in dangerous territory.
That it takes a thousandfold level of caffeine to achieve the effects many of us do want to have, and that such doses are not actually threatening to most people’s health, is a good demonstration of how conservative these ‘acceptable limits’ can be. Their purpose is to define a level where particularly at-risk populations (such as children, pregnant women) are very unlikely to experience any unwanted effects, even if the intake is constant throughout life.
The power of visuals
Having realized how elusive the nature of these acceptable limits can be, I was struck by a very simple graph I saw on a slide during an excellent presentation at the risk conference on intuitive toxicology and irrational food fears by Dr. Angela Bearth. Her research has reached heartening conclusions such as in her paper on Informing Consumers About the Scientific Risk Assessment of Food Additives:
it appears that informing consumers about complex food safety topics, such as the scientific risk assessment of food additives, is possible, and using a carefully developed information video is a successful strategy for informing consumers.
… and that video (in German) included a graph on finding the No Adverse Effects Level (NOAEL) of aspartame. Next to the NOAEL was a bar for the acceptable limit – hundred times lower. So simple, so elegant. Why hadn’t I thought about it? A simple visual approach would make the information more intuitive. Being a bit short on video-making skills, I decided to go at it with an infographic.
While setting the limit, factors like the speed of breakdown (the potential for accumulation), what foods it is found on, and how much of those foods people tend to eat, are taken into account. For a detailed account of how the safety margins are determined, please see Alison’s piece where she uses coffee and glyphosate limits as the examples, or the EPA’s page on how pesticide tolerance limits are set, and similar guides for the OECD countries. Note that depending on region and class of substances, the names of the limits vary (common ones include Acceptable Daily Intake, Reference Dose, and Tolerable Daily Intake) – more about that in our piece Measures of Toxicity.
Regulatory limits vs reality?
In addition to the limit level, I decided to look at some real world data on what kind of trace levels are actually found. I chose pesticides residues, because they are a prime suspect for inflated sense of risk: they tick the boxes of being unfamiliar, manmade, and imposed, so we are biased to view them with exaggerated caution. As a result, many consumers worry about pesticide residues out of proportion.
One way to look at this are to consider various food-stuffs and their typical residue levels – as Steve Savage does here, looking at the annual USDA pesticide residue database. Detailed information is also available on the Californian pesticide testing regimen, which samples thousands of food items every year. They tend to find that about 40 % of foods tested have no detectable residues at all, and that in the remaining portion the residues tend to fall much below 10 % of the limit. (The limit which is already 100-1000 times below a level where we still would expect no effects.)
Added: Similarly for Europe, as analysed by EFSA, yearly tests find about half of the samples to be free of residues, and more than 96 % to fall below the Maximum Residue Limits (MRLs), which are often set at even lower levels than the Acceptable Daily Intake (ADI), taking into consideration that farmers are often able to use the pesticides at levels where residues remain far lower.
A few percent of cases in Europe, as well as those in California, do have either levels above the limit, or an unapproved kind of pesticide residue present. Both regulators note that majority of the time, this does not indicate a health concern – with rare exceptions, like in 2013, when cactus pads imported to California from Mexico were found with potentially harmful levels of organophosphates. (They were subsequently removed from sales and imports monitored.)
Looking at how much above the acceptable limit we need to go before we even expect effects, this lack of health concern is not all that surprising. I wanted to try to make this information more intuitive as well, and I think the most relevant approach would be to look at our actual average exposures – how much do we end up taking in through our diet?
I know of at least two studies that have made these kinds of estimates, a Danish one looking at cumulative pesticide exposures from earlier this year, and one from US from 2011. These studies looked at 25 and ten most commonly detected residues, respectively. Neither found consumer exposures to exceed the acceptable limit. The scientist in me did not feel comfortable without allowing for at least some uncertainty, however, so I chose to go with “more than 99.9 % of the time our exposures are below the acceptable limit.” This infographic is much more cluttered compared to the first one… but it’s a start.
Both studies found that the highest exposures were around 2 % of the limit, and that majority were even below 0.5 % – in fact, the US study found that in almost half of the cases, we are exposed to less than 0.001 % of the acceptable limits. Even with a combined assessment, assuming that these residues could together amount to more risk, the Danish paper finds that consumers cumulatively reach about 16 % of the acceptable intake limit.
UPDATE: I couldn’t stop thinking about wanting to improve the residue-graphic, and decided to try a different approach, here:
Using areas, it is easier to keep all the different levels visible (the visual comparison of areas instead of heights may underestimate the values somewhat for the viewer, but I think here it is useful nevertheless, because the differences are still huge, but it brings the levels to a form that is more easily viewable at a glance.
I am still digesting the insights and extra materials I accumulated from the risk conference, and I hope that they can make me better at communicating risk. When I think of how much anxiety goes into many areas that pose very tiny risks, it really makes me wish that I could find a way to make the information tangible in a way that would take away needless fear and help empower the public.
I want parents, grandparents, and all worried or health-conscious consumers to know that they are in control of their well-being: there are no invisible threats to eating plenty of fruits and vegetables, which, along with regular exercise, is the single most effective secret to living long and healthy lives.
If you would like to have a discussion in the comments below, please take note of my Commenting policy. In a nutshell:
- Be respectful.
- Back up your claims with evidence.