A brief analysis of the PlasticList report

My small contribution to citizen science

As you may recall, over the summer I wrote a piece about determining the safety of microplastics, inspired in part by Nat Friedman’s announced effort to sample a bunch of food for microplastics. Well, 7 months later, Nat’s microplastics effort is out, and it’s interesting. Not perfect, not without problems, but interesting.

I want to talk a bit about some interesting findings from the projects, along with caveats, analysis of the method, and suggestions for future work. But, before I do that, I just want to applaud Nat for this effort. Nat organized a big, expensive effort to do science in the public interest, contributing a fair amount of his own time and money to do something he considered valuable. I think that’s incredibly cool, and, frankly, an inspiration for me.

Ok, enough plaudits. Let’s dig into it.

Brief overview of the report

The PlasticList is a 300 item list of foods and some consumer items. These range from tap water, to grocery store steaks, to fast-food hamburgers, to tampons. Each of these items was obtained from somewhere in the Bay Area, blended up, and analyzed for plasticizers, which are chemicals used in the making of different sorts of plastic.

Most of the plasticizers that these items were analyzed for have some evidence of negative health consequences. A few of the plasticizers do not. The items were not checked for anything else (e.g. actual plastics or heavy metals). Most items had replicates, meaning that multiple of the same item were tested (i.e. multiple Fiji water bottles).

A link to the PlasticList itself.

The methodology, which I’ll get into later, was ok but not amazing. So, assume all of my conclusions are with a low degree of confidence. Anything I state should be assumed to be a call for replication and future investigation, not a definitive claim.

Interesting findings

1. The water, both tap and bottled, is generally good. I, like many of you, drink a lot of water. Water contamination would be bad. Fortunately, the water quality, both tap and non-tap, generally seems pretty good.

There are a few worrying signals, such as one crazy high DEHT signal from the tap water Fort Mason, a neighborhood in San Francisco, but that one didn’t seem to replicate. The only one that I think I’d be confident that is probably bad is the tap water from Hayes Valley, which consistently shows a high DEHT signal.

2. Seafood, steak, and chicken are all bad. Nat’s team tested 3 different varieties of Whole Foods salmon (smoked, farm raised, and wild caught) and one variety of canned tuna. All seafood tested had worryingly high levels of plasticizers, with the canned tuna being especially bad for BPA and the Whole Foods salmon being especially bad for DEHT.

Similarly, all steak and chicken tested, both from the local butcher and from Whole Foods, was bad for plasticizers. Organic steak and chicken was no exception.

3. Vegetables have low levels of plasticizers. While Whole Foods vegetables were the only vegetables tested, all vegetables tested had some level of plasticizers. Lettuce was the worst of the bunch (no pun intended).

4. Almost all beverages that come from a store in a plastic or plastic-y paper cup were bad, whether cold or hot. Metal and glass were ok. Iced coffee, hot coffee, milkshakes, and Gatorade were all bad for various sorts of plasticizers. Flexible plastic cups for cold drinks (i.e. the ones used for bubble tea) seemed to be the worst of the bunch. Metal and glass containers generally got off ok.

5. Fast food is bad, regardless of what you get. Unsurprisingly given the above, all fast food hamburgers, salads (including Sweetgreen), and drinks were bad for plasticizers.

6. Takeout containers don’t make an obvious difference. Indian food at the restaurant had sometimes higher levels of plasticizers than takeout Indian and sometimes lower.

7. Microwaving reduces plasticizers. One very consistent signal is that the plasticizers in prepared foods decreased after microwaving. This happened across foods and across plasticizers.

Tentative conclusions

1. Plasticizers accumulate up the food chain. Water occasionally has plasticizers. Vegetables always have a low level of plasticizers. Meat, including chicken, steak, and fish, always has a high level of plasticizers. This suggests accumulation.

2. Heating often destroys plasticizers. Microwaving seems to destroy plasticizers. Letting takeout food cool down or leaving water bottles in the sun doesn’t seem to reliably increase plasticizers.

3. Plasticizers leach from polypropylene and LDPE. The worst offenders for plastic bottles seem to all be polypropylene bottles, such as the Starbucks cups. Paper cups lined with low density polyethylene also seem bad. Polyethylene terephthalate, which is what disposable water bottles are made of, do not seem to leach plasticizers as much.

4. BPA is still a problem, although less so than in the past. People have been talking about the dangers of BPA for decades. Somehow, Boba Guys still has significant amounts of it.

Caveats and compliments about the PlasticList methodology

The PlasticList team clearly put in a lot of time and effort into making this the best project that they could. There’s a lot I like about what they did, and they’ve definitely provided me with some food (and plasticizers) for thought. I’ll put a list of what I liked below.

However, I have two big complaints about this piece, and a bunch of little ones.

PlasticList should have engaged better with the prior literature

As I wrote in my piece 7 months ago, other people have done similar reports to this one, showing that various food items have plasticizers in them. Nat’s team didn’t really build off of that work, and instead tested a bunch of random items.

As a result, a lot of the team’s work can be summarized as “many of the food items and some of the consumer items in the Bay Area have plasticizers in them”, which is exactly what I predicted their work would turn out to be 7 months ago. I would have much preferred them to try to explain, support, or refute prior literature, especially given the $500k (which is a lot!) at their disposal.

Engaging with the prior literature would have also helped them avoid my second major complaint, which is…

PlasticList’s testing should have been hypothesis-driven

Testing 300 items for 18 different plasticizers produces a lot of data. Unfortunately, as any good scientist can tell you, a lot of data means a lot of noise. Some of that noise is interesting, like sudden spikes in plasticizers from one water sample to the next or high levels of plasticizers in Whole Foods salmon.

If the PlasticList had followed up on that, like testing those water samples repeatedly to see if there are any time-related patterns in when plasticizers spike or testing salmon from multiple different grocery stores, we could make strong conclusions from that data. As it is, all I can say is “sometimes Bay Area tap water might have high level of plasticizers” or “Bay Area Whole Foods salmon might have high levels of plasticizers depending on the store or time of year”, which are not strong conclusions.

Assorted complaints

1. They needed more replications per item, especially when the item had very high levels of plasticizers. Some items, like the Chipotle Burrito, were only tested once.

2. They needed more samples per purchased item. As far as I can tell, their replications were done by purchasing another of that item (i.e. another Coke can), but never done twice on the same purchased product. This leaves open the possibility of instrumental error.

3. They needed a better protocol for testing parts of complex items. All items were blended before testing, which makes it impossible to tell where the plasticizers are located in say, a hamburger.

4. Their presentation is biased towards showing higher levels of plasticizers. If you look below, it looks like the items at the bottom have a higher level of DIDA. But that’s inaccurate. All items were below the lower limit of quantitation. It would have been better to just put LLOQ, as is industry standard.

Assorted compliments

1. They are very open about where, how, and when they got their items, making it easy to trace and reassuring me that they were careful with sample handling.

2. They thought a lot about how to make sure the analysis was accurate, including careful selection of the lab, recruitment of expert help, and collecting and shipping samples.

3. Their results are free, open to the public, and easily searchable.

4. They tried to analyze their results in a report. I don’t think it’s a great report, to be honest. It’s difficult to navigate; averages together values I don’t think should be averaged; and doesn’t follow any best practices for writing, presenting, or analyzing scientific data. There’s no way it would ever be published in a journal. But I appreciate the effort.

Suggestions for future work

My suggestions for future work remain the same as they did in my original microplastics piece: either go further upstream, to figure out the source of microplastics, or downstream, to figure out the consequences.

Upstream, we can try to determine how water sourcing, ingredient sourcing, cooking methods, and packaging methods influence the amount of microplastics in foods.

Downstream, we can look at specific populations to see the effects of microplastics. According to PlasticList, Stanford University has high levels of microplastics in their dining hall food. I’d bet there are a fair number of students who eat 3 meals of dining hall food every day at Stanford for 4 years. I’d also bet there are a few students who never eat dining hall food. Doing a survey to see if there are any detectable health differences between these students would be an easy enough place to start.