What ‘Science By Press Conference’ looks like

Exhibit A: 1989 and Pons and Fleischmann announce cold fusion — an “inexhaustible source of energy” — at a press briefing in Utah, before they had applied for patents or published their technology. Too bad they were just plain wrong. I bet they feel embarrassed now.

(Pic: scientificamerican.com)
(Pic: scientificamerican.com)

Exhibit B: 2002 and five-word headlines circle the globe as Clonaid announce their human cloning activities. Twelve years on, and we’re still waiting for any evidence whatsoever of said clone.

(Pic: ipscell.com)
(Pic: ipscell.com)

Exhibit C: The WHO issue a press release concerning a yet-to-be-published paper on mobile phones and brain cancer. The world media reports a definite causal link, even though the yet-to-be-published-paper was (a) unseen by everyone in the world media, (b) focused on only a small subset of possible cancers, and (c) was merely a document where a group of boffins “discussed and evaluated” the available research literature, rather than a new study bringing new data to bear on the issue.


* * *

And finally, apropos of nothing at all, and arising just randomly from miasmic boredom of a Bank Holiday, here is today’s Irish Indo front page…

(Pic: Twitter)
(Pic: Twitter)

Hmmm. Stream of consciousness, eh?

Here are some observations on that story featured in the main headline.

1. The study mentioned is not yet published. Yes, it has been accepted for publication, but it is not yet available to readers — or peer scientists — to review and comment upon in detail. The journal’s webpage has an ‘Early View’ section where not-yet-published papers are posted, but this one isn’t there yet. So all we have is the information that has been released to the media.

2. The study is in fact a meta-analysis of other, already published, studies. It is a statistical study which seeks to combine findings from other studies together. In this regard, it combines 25 previous papers all of which have been published before. To date, few if any experts who have read those papers have been confident enough about the findings to declare any kind of definitive increased risk for autism resulting from Caesarian section.

3. The authors of this paper seem at pains to emphasise that they have not established a causal link. It’s a shame that this emphasis doesn’t make it  into the large-font headlines, but then I suppose there is only so much space on a page. After all, if the headline-writers sought to convey the equivocal nature of the research conclusion, we might never have learned of Laura Whitmore’s divided football loyalties.

4. Of course, notwithstanding the equivocation, the authors seem to have found time to expand upon a range of gut-flora- and brain-oxygen-related hypotheses to explain the finding of causality that they have definitely not made and are definitely unwilling to stand over. (Perhaps their difficulty is in having too much paper to fill, what with only having done a meta-analysis.)

5. The most important point made by the authors appears in the Irish Times. It is as follows:

However, the authors point out the underlying reasons for a Caesarean may lead to the increased risk of autism, not the Caesarean per se.

In other words, it’s not the C-section that’s the problem, it’s the underlying medical condition that made it necessary to have a C-section. There are several naturally occurring neurodevelopmental disorders that both complicate childbirth and are themselves linked with autism diagnoses. Saying — or, rather, implying — that C-sections cause autism is a bit like saying that being referred for a psychological assessment causes autism. And there is little a statistical meta-analysis can do to remove that confound.

Although, as none of us has actually seen the meta-analysis, all we are left to go on is the Press Conference.

Still, since when has Science-by-Press-Conference been a problematic method for announcing theories on autism causality?

A guy talking about a thing (Pic: bmj.com)
A guy talking about a thing (Pic: bmj.com)


Share this:

Leave a Reply