Just a quick one. Perhaps this will be taken the wrong way, but how often do you actually read the Lancet Don?

The Lancet is not a serious scientific journal. It is a newspaper for doctors.

It was the Lancet that started the MMR scare, and the Lancet that published the wild estimates of “excess deaths” in Iraq. I wouldn’t believe anything I read in it, any more than the Daily Mail.

Clearly the market has resulted in such an even distribution of wealth elsewhere it would be the perfect model to apply to scientific knowledge.

In my copy of ‘Reliable Knowledge’ John Ziman proposes that the goal of science ‘is a consensus of rational opinion over the widest possible field’, p3 L4. Furthermore ‘imperfections of communication or of critical analysis reduce the reliability of science in every field’, p4 L 29

I could go on quoting but to my mind science consists only of the messages communicated and recorded between scientists. It is the scientific archive that is key, and by implication access to it.

I strongly feel that many of the woes so often expressed here regarding popular scientific illiteracy would be consigned to history if only we got ourselves out of those ivory towers and opened the libraries to all comers.

This may also result in an significant improvement in the quality and relevence of the messages we store in the archives. No more me-too drugs on the front page, or indeed hundred thousand dollar a year treatments in a world where in the last thirty years we’ve invented under a score of new tropical medicines.

How many Daily Mail, or even Guardian readers are going to go and read this paper, even if it’s freely available, in order to make their own mind up? I would posit that an insignificant number will pursue the matter any further than reading the pieces published in the papers, or watching the news.

Thus any interpretation that journalists make embeds itself in the public’s mind as “true”, and is never challenged by the original paper itself.

Equally, if other scientists cannot read the paper before the media publish their opinions, then they can’t respond in a timely manner. Nor can other journalists who weren’t privy to the early release write anything that may offer a different view of the paper

Generally, I hate embargoes and exclusives, and not just for scientific papers either. I think it encourages lazy and irresponsible journalism – if you’ve been given an ‘exclusive’, are you really going to do a rigorous fact check? Are you going to quote contradictory evidence or opinion? Methinks, probably not.

arxiv.org also contains a lot more than just preprints; whole books, and irrelevant (but interesting) v.v.v.niche stuff that no journal would ever want to bother with.

- 800 more psycho cases [(c) 2007 Alasdair Philips] per year. Oh noes!!1!!1! Well it’s hardly an epidemic is it. Especially combined with:

- That claim in the Indie and elsewhere about incidence of schizophrenia actually staying flat. [Are schizophrenia and psychosis medically the same thing? If not then someone's talking at cross purposes.]

- Also: paranoia being closely associated with schizophrenia and psychosis, I can’t help wondering if knowing that the blue meanies could lock you up for enjoying a smoke might be a contributory factor in borderline cases.

Andrew.

PS superburger (50) — OA journals don’t charge if you’re from a third world country. Are there non-OA journals where you pay to publish? What fool would do that?

However, I’m not convinced by their argument that reverse causation was unlikely given putative genetic, neurodevelopmental or neurobiological factors in these disorders and the long prodromal period.

Note that the effect on schizophrenia incidence is even less convincing than they make out since the odds ratio specifically for psychotic illness is nearer 2.0 than 1.4.

http://news.independent.co.uk/health/article2809179.ece

It seems then, that the only increase has been in the number of diagnoses implicating cannabis as a causal factor. This suggests that we may be witnessing merely a diagnostic fashion trend, to which medicine is no less prone than any other field of human endeavour.

I would very much like to see a comparison with the risks intrinsic to other, unpersecuted pursuits, like leisure motorcycling or mountaineering or eating sashimi.

Given that cannabis is commonly available and used, it is certainly also possible that people feeling psychotic (perhaps for the first time, as teenagers when they are also trying cannabis out for size) might use it to feel better (and it might work), or that it eventually makes their symptoms worse, or both.

Nothing in the paper, unfortunately, can answer the fundamental question about causality (as they point out – this would be very difficult). Knowing you have the relevant symptoms and reporting them might come later than their actual onset, for instance, which makes it hard to assess whether or not cannabis comes first. Dose-response is a bit of a red-herring as well for the same reason(i.e. worse psychosis might make you want to smoke more dope).

This obviously doesn’t mean that their conclusions are wrong – just that they might be.

A small technical point re. the 1.4 estimate: meta-analyses are often a bit dodgy (adding apples to oranges etc.). You use I2 (measure of heterogeneity) to measure the extent to which your estimates are measuring the same thing. Low I2 is good. Some of the pooled estimates in this paper have very high I2 (>50% is considered in some quarters to render pooled results meaningless). Haven’t had time to read the paper carefully enough to see if the 1.4 increased odds comes from from pooled figures with high I2 or not. Hope not, but it might be worth checking.

(I thought the same thing too)

I just downloaded the “Lancet story” pdf from the Guardian, and it has only the accompanying editorial. Have they pulled the research paper as a result of your comments?

i believe they controlled for your option 3, or ‘reverse causation’ in the analysis, so it is much more likely to be a combination of 1 and 2.

i’m no expert on meta-analysis, so i’m happy to be corrected, but i think the relative risk of 1.4 seems fairly generated from the data they have, and they are very cautious in their conclusions. which obviously results in “1 spliff and you’re mad” if you write for the mail.

re: Physics/Chemistry publication cultures. As a chemist in a physics department I’ve had the opportunity to experience both cultures of publication and there really are equally good arguments for both, so I think it is just down to consensus and tradition.