The data are what the data are

The data are what the data are

I once made the mistake on Facebook of challenging the data in a statistic provided that bolstered a position in a very political hot topic. As hard as I tried, it seemed some could not get past the notion that I was actually arguing for one side or the other of the political issue when in fact I was simply criticizing the data. Dare I try again?

Well hopefully THIS isn't too terribly controversial. In the interest of full disclosure (though most of you know already) I am not a big fan of heavy handed public health science in the political arena and additionally I enjoy a cigar and pipe about 3-5 times a year. I'll start off by saying I am not going to enter into an argument about the dangers of secondhand smoke, nor the extent to which the government can or should legislate such bans. But I will posit this:

I do not believe there is any meta-analysis that could possibly convince me that this is accurate: "Secondhand smoke causes close to 50,000 deaths per year." I don't think anyone doubts that secondhand smoke is bad for you, but when people start tossing out "body counts" I start having doubts. It ends up being more emotion than real science. So I followed the rabbit down the hole.

CNN says the data came from the American Lung Association's webpage. I went there and looked and looked and looked. They cite the source as being the California Environmental Protection Agency. A reputable source? I'll let you decide. Still seeking the data. The CEPA cites that its source is an Executive Summary intended to set public policy in California. Hmmm. I kept searching...a Fact Sheet...a report...and finally a good deal of details HERE.

In all honesty I do not have time to plow through this in its entirety...but a couple of points to ponder: Did they pick and choose studies? In my experience with nearly every human behavior/health issue there is conflicting evidence. Proving behavior as a cause of death is exceptionally difficult...unless of course it involves an unopened parachute, a very very long skid mark leading to a tree, or a bullet hole. All of us know people who have smoked all their lives and never got lung cancer and some of us no doubt know people who got lung cancer and never smoked a day in their lives - and no exposure the second hand smoke. The point being: smoking isn't the only thing that causes lung cancer. It may be a prominent risk factor, but we have to ask how this understanding of lung cancer plays into the body count? And remember we are talking about a solid number...not increased risk. Insurance companies are probably a good indicator here...they know well that it is a serious risk factor, but that's different than a politically charged body count.

Heart Disease? They cite a number of studies that offers "estimates of the number of CHD deaths in nonsmokers that could be attributed to ETS exposure" and the keyword here is "COULD." That's different from a "FACT sheet" data point...that sounds more like a "Maybe sheet" data point. Heart Disease is a very complex disease and from all I've read the science leaves many questions unanswered. Given all the evidence out there I think it's fair to conclude that MANY factors may or may not play a role in any one person developing heart disease. I will have to look at the listed papers to see how exactly they decided that secondhand smoke COULD be attributed to the disease. But I have a really sneaky suspicion that my concerns will not be assuaged.

I have colleagues who regularly design human subject studies and they have backed me on this point saying in essence that designing studies that prove causality in human behavior/disease is indeed very difficult. And in addition the numbers are spurious.

Anyway...I'll reiterate I'm not suggesting smoking is safe or that second hand smoke is not bad for you. So please don't engage that route. I just don't believe that a body count can be accurately proposed and only serves to win political fights as opposed to providing real scientific insights.

Comments

Popular Posts