Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/5/84; site terak.UUCP Path: utzoo!decvax!decwrl!Glacier!oliveb!hplabs!hao!noao!terak!doug From: doug@terak.UUCP (Doug Pardee) Newsgroups: net.med Subject: Re: Risk factors Message-ID: <822@terak.UUCP> Date: Fri, 25-Oct-85 14:49:21 EST Article-I.D.: terak.822 Posted: Fri Oct 25 14:49:21 1985 Date-Received: Tue, 29-Oct-85 15:11:41 EST References: <1939@aecom.UUCP> <796@terak.UUCP> <1955@aecom.UUCP> Distribution: na Organization: Calcomp Display Products Division, Scottsdale, AZ, USA Lines: 36 [It's hard to argue with someone who knows what he's talking about] > For instance, in american males, heart disease is related to smoking both by > the number of years smoked and number of packs (it can be expressed as pack- > years.) So quitting smoking does help. But, but, but... The conclusion does not *necessarily* follow. I do accept the conclusion, but I don't believe that it has been properly proven. It is the method of "proof" that I'm questioning. As a counter-theory, try this: The kind of person who smokes is often a "Type A" person. The kind of person who has heart disease is often a "Type A" person. If we express the current theory as Type A --> smoking --> heart disease then we can express an alternate theory as Type A --> other factor --> heart disease Type A --> smoking Under this alternate theory, people who quit being Type A quit smoking *and* reduce the risk of heart disease. This gives the effect which has been observed in population studies. But if a person quits smoking in order to avoid heart disease *when he otherwise would continue* and remains a Type A person, there would be no effect. It is *very* difficult to set up a properly controlled experiment when the variable to be introduced is human behavior. It is equally difficult to draw reliable conclusions from population studies when the independent variable is human behavior. Human behavior is too often not (maybe never is?) an "independent" variable. -- Doug Pardee -- CalComp -- {calcom1,savax,seismo,decvax,ihnp4}!terak!doug