A beautiful example of bad science

Topics without replies are pruned every 365 days. Not moderated.

Moderator: Dux

Post Reply
User avatar
Sangoma
Sergeant Commanding
Posts: 6470
Joined: Mon Oct 08, 2007 11:16 pm
Contact:

A beautiful example of bad science

Post by Sangoma » Tue Jul 24, 2018 7:33 am

Low cigarette consumption and risk of coronary heart disease and stroke: meta-analysis of 141 cohort studies in 55 study reports
Conclusions Smoking only about one cigarette per day carries a risk of developing coronary heart disease and stroke much greater than expected: around half that for people who smoke 20 per day. No safe level of smoking exists for cardiovascular disease. Smokers should aim to quit instead of cutting down to significantly reduce their risk of these two common major disorders.
This article is likely to become the basis of "indisputable evidence" and various government policies. Yet the study is seriously flawed: none of the cohort studies included in this meta-analysis studies people who smoke one cigarette a day, and the risk of this pastime is mathematically extrapolated, both in every included study and meta-analysis. This flaw is obvious to anybody who bothers to read more than a summary and has been noted by many professional epidemiologists. The page of rapid responses has been fun to read:

https://www.bmj.com/content/360/bmj.j58 ... -responses

I am not even going to the association-causation argument. I don't smoke, except an occasional cigar once or twice a month and do not advocate it. Yet junk science just doesn't do any good. Every time I read an article of such sort the first thing that comes to mind is the reference to the "indisputable evidence" that climate change is caused by human activity. Specifically the Hockey Stick by Mann. They also used flawed input and also refused to make their database available to other researchers for a long time.
Image

User avatar
Shafpocalypse Now
Lifetime IGer
Posts: 21051
Joined: Fri Feb 04, 2005 11:26 pm

Re: A beautiful example of bad science

Post by Shafpocalypse Now » Tue Jul 24, 2018 1:56 pm

I think we are seeing the death of objective science in our lifetimes.

The Pharma industry has already infiltrated research and influenced results.

User avatar
Sangoma
Sergeant Commanding
Posts: 6470
Joined: Mon Oct 08, 2007 11:16 pm
Contact:

Re: A beautiful example of bad science

Post by Sangoma » Fri Jul 27, 2018 5:08 am

The Pharma is the beast on its own. Quite a few years ago a group of psychiatrists boycotted some psychiatric conference in South Australia because of excessive influence of pharmacological industry. Well done, albeit almost completely unnoticed.

In non-pharmacological research big problem are researchers themselves and so called science journalists. The former are easy to understand: after years of injecting mice with various substances and coming home smelling mice urine you suddenly get the spot on the stage. As the result the statements they make in newspapers are way more assertive than what they dare to say in the actual peer reviewed journal. Journalists, on the other hand, are concerned only with the next article they are going to publish. None of them have the intelligence or attention span long enough to actually get clued up about the subject they are writing about. This is of course true for all kind of journalism.

Nice example - carnitine study. In short, they fed genetically modified mice (these mice are predisposed to cancer) large doses of carnitine. And yes, there was a dose-dependent increase in cancer incidence. The conclusion of the study: carnitine is prevalent in red meat, that's why it is feasible to look into the possible meat-cancer connection. In their interviews with newspapers though they were way less reserved and let the fantasy run amok: "Yeah, meat may very well cause cancer and now we know why". Journalists took up the relay with "Researchers found how meat causes cancer!" I slammed my head against the wall for a while when this was topical, but then stopped.

Still, epidemiological "research" takes the gold. Due to the computerization of various data there are huge databases available for analysis. Proper term is "data dredging". If you run enough tests you will come with various correlations and statistically significant numbers. Now, how many know that in research you have to work out pre-test probability first and only then run this kind of analysis? Hence another well read and quoted but virtually forgotten article by John Ioannidis, Why Most Published Research Findings Are False
Image

User avatar
Fat Cat
Jesus Christ®
Posts: 37511
Joined: Mon Jan 03, 2005 4:54 pm
Location: 鬼ヶ島

Re: A beautiful example of bad science

Post by Fat Cat » Fri Jul 27, 2018 5:18 pm

Shafpocalypse Now wrote:
Tue Jul 24, 2018 1:56 pm
I think we are seeing the death of objective science in our lifetimes.
I think the WWI generation were the ones to hit that mark. That ship sailed a long time ago.
Image
"Prepare your hearts as a fortress, for there will be no other." -Francisco Pizarro González

Post Reply