Comments on: How Accurate is Epidemiology? (part 2)

by: Zagreus Ammon

Wed, 19 Sep 2007 21:49:51 +0000

Epidemiology is accurate enough. The interpretation of data leaves much to be desired.

From a clinician, teacher and manager’s perspectives, I wish everyone would please please please realize that every piece of information has its limitations and conclusions drawn by inference cannot be logically assumed to represent complete and unalterable truth.

Ya, take a pill and learn to live with uncertainty.

by: Seth’s blog » Blog Archive » How Accurate is Epidemiology? (part 3)

Tue, 18 Sep 2007 16:29:25 +0000

[…] How Accurate is Epidemiology? (part 2) […]

by: seth

Tue, 18 Sep 2007 03:31:12 +0000

Thanks, Igor.

by: Igor Carron

Mon, 17 Sep 2007 22:09:32 +0000

Seth,

I liked what you wrote but pay attention to that pill:
http://diabetes.webmd.com/news/20070709/selenium-supplements-diabetes-risk

Igor.

by: Tom

Mon, 17 Sep 2007 06:20:30 +0000

Well, much of Taubes’ article is about the kind of problems associated with overstating those cases…the erroneous conclusion that HRT helps, when it actually kills women…the downside of folate supplementation…the panicked switch to transfats after CSPI’s alarums over palm oil in popcorn…and what Taubes explores in his new book, the real result of the bad science that is still believed to support the explosion in carb consumption of the last thirty years.

I wouldn’t argue for censorship, no, but I would support intelligent use of the data, which ain’t happening.

For me, it’s hard to argue with the idea that the double-blind trials that are necessary aren’t being done. The conversion of hypothesis into fact means the “trials” are done on the public at large.

by: seth

Mon, 17 Sep 2007 04:37:29 +0000

The facts (observations) collected by epidemiologists suggest hypotheses; and sometimes those hypotheses are wrong. You may think it is awful that those hypotheses are told to the public; I don’t. Let’s say that an epidemiological study finds a correlation between Behavior X and better health. USA Today publishes this. As a result, many people start doing Behavior X. I don’t see the problem. Sure, they could be wasting their time. But maybe not. Everything has risks, is uncertain. The epidemiological evidence does raise — or should raise — one’s belief that Behavior X causes better health. A little knowledge — a little push in the direction of certainty — is better than nothing.

I’m more worried about poorly-educated science journalists who are overly critical and poorly-educated scientists who are dismissive (e.g., they fail to grasp that correlation raises the plausiblity of causation–I have blogged about this) than “an uneducated population.” It’s the journalists and scientists who have the power.

I think you’re right that scientists sometimes overstate their case. But I don’t see a lot of that. I see much more unwise dismissiveness.

by: Tom

Sun, 16 Sep 2007 18:53:19 +0000

If only 10% of the ideas generated by epidemiology turn out to be correct, well, 10% is more than zero.
__________

Taubes’ points, and I think they’re good ones, are that the hypotheses yielded by epidemiology are confused with facts, and that these questionable hypotheses are immediately implemented as social policy due to a critical storm of:

1. researchers trying to build names and careers

2. climbing reporters, bloggers and a sensationalist press trying to make noise, money and fill space. (USA Today will not be running any articles headlined “Slight correlation of questionable causality found in tiny subset.”

3. opportunistic big Pharma & other entrepreneurs large and small who see trends to milk, and

4. politicians eager to prove their “protecting” Americans from the latest evil.

5. an uneducated population which thinks correlation is causality.