Discover more from Living With Evidence
Science Blinkers and Breakthroughs
What does and doesn't break through our line of sight.
This flower bud breaking open in my garden last summer is Zinnia elegans. Seedlings are popping up all over my garden now, with who knows what spray of colors in store. Such a source of delight! They have elegant Art Deco buds and gorgeous blooms on long stems; they’re prolific and long-lasting. One thing they’re not, though, is woody.
Which is why I was puzzled when I saw zinnia-based biotech hailed as one of the top innovations of 2022. Because of it, went the article, “deforestation is longer needed to produce timber.”
Scientists had created “customizable wood in their lab” from zinnia cells, allowing them “to bio-print wooden pieces of any shape and size. This means that if you need a wooden table, you can directly produce a wooden table from the cells.”
A zinnia timber table? Grown or printed in a lab? Say what?! When I dug out the scientists’ article, the pics showed slimey-looking little bits of stuff in petri dishes. They’re trying to generate plant growth by cellular agriculture instead of horticulture. The zinnia cultures they produced, wrote the authors, “can survive and continue to grow for months after printing.”
At the end of their article, the MIT-based authors suggest – totally hypothetically – that using such techniques “in a woody species” could create material that “would exhibit wood-like attributes and could likewise be grown in a tree-free manner”. MIT’s News Office spun that into a press release titled, “Toward customizable timber, grown in a lab.”
I don’t know enough about the field to know if this was a breakthrough, or even the first the authors’ claim it to be. But it sure is a textbook case of science spin. And if it hadn’t been propelled by a prestigious institution from the Global North like MIT, I doubt it would ever have come to my attention. Even very un-newsworthy work from some countries – and one race and gender – is more likely to reach our line of sight than even the most important work from others.
This set of blinkers also contribute to why some people get credited with discovering something that others had documented well before them. This woman’s work got me thinking about this.
She’s Ume – or Umeko – Tange, and she’s in her 70s in this photo. She was born in southern Japan in 1873. Tange was not only in the first trio of women admitted to a men’s university in Japan in 1913 – from what we know, she was also one of the first two Japanese women to get a science PhD, in 1927. And she got a second PhD in her 60s. Her first PhD was in chemistry from Johns Hopkins University; her second, in agriculture, from Tokyo Imperial University (now the University of Tokyo).
I came across her when I was looking for Japanese women scientists who had pages about them on Japanese, but not English, Wikipedia. While I was digging around, I read somewhere that her lab work was of such a high standard, that her findings always proved reliable. And that her studies contributed to nutritional guidelines and minimum caloric intakes in Japan at the time. But without being able to read Japanese, and with so little background knowledge in her field, I couldn’t establish how reproducible or influential her work was.
I read enough, though, to find her inspiring, and wish her biography was available in English. It was also the first time I got to tag a Wikipedia bio with scientist with a disability. (Tange lost the slight of one eye when she was a child.) You can read more about her in the Wikipedia page I created about her, and in this thread with a bunch of pictures on Mastodon.
In the collection below of other things that caught my eye this week, there’s data on an issue of critical concern in clinical trials, good news on debunking misleading data visualizations, good Mastodon news, and more.
Wishing everyone the best,
Zinnia photos are my own, (CC BY 4.0).
Why can clinical trials have sub-optimal care offered as a comparison for the control group, even though it disadvantages patients and gives a biased impression of how good a treatment is? Talal Hilal and colleagues unpacked this critical issue in a systematic review of leukemia (CLL) trials recruiting in the US. Of the 18 eligible trials they found, 10 did not offer the best standard care to participants in the control group – and 6 of those 10 led to FDA approvals.
Reasons they found for this? Then-standard treatments in the US weren’t approved yet in other countries in international trials, so they weren’t allowed in the US group either. The authors also found that a particular set of clinical practice guidelines (NCCN) was used to justify offering sub-optimal care in control groups. They didn’t check for another reason they said is common – when better treatments become standard after a clinical trial has already been planned.
Winnifred Wijnker and colleagues published results of a trial of 4 ways to try debunking misleading graphs/data visualizations. All helped at least a little: “The most effective one is aimed at correcting the initial image by presenting an accurate alternative graph.”
This study on preprints versus journal versions has a systematically assembled group: 19 preprint/journal pairs from a systematic review of Covid prediction models. Reporting quality of the preprints was lousy. After peer review, it was… trivially less lousy.
And this news is so meta, it feels like a meta-science Möbius strip! There’s now a registry for methodology studies within systematic reviews/meta-analyses (SWARs) as well as within trials (SWATs). For example, someone did a randomized trial of two different methods of screening citations as part of their systematic review. Here’s an introduction, and here’s the registry.
Finally this week, good Mastodon news. The Washington Post is going all-in, linking to journalists’ Mastodon accounts from their stories, enabling verification of accounts, and reportedly starting an instance for the newspaper (I haven’t seen a story on that yet, though). Mozilla – the non-profit group behind the free and open-source browser, Firefox – are launching a Mastodon instance early next year. A trusted brand, with major technical skill and resources, it’s going to be one to watch in 2023.