posted 07-01-2004 12:03 AM            

I'll start the ball rolling towards deconstruction with this: Per a 1999 ALA-sponsored study (Ahijevych et al) the blood cotinine levels in less than pack-a-day smokers ranged from 182 -249 ng/ml. The RR for smokers v nonsmokers for heart disease is generally reckoned at 1.7. Yet this study attributes 1.6 to ostensibly "exposed" nonsmokers with blood cotinine levels of 8- 14 ng/ml.

Any other (de) constructive criticisms welcome.

posted 07-01-2004 09:35 AM            

quote:

---

Originally posted by Cantiloper:
I didn't analyze it in depth, but I did read the full text without any glaring errors/problems springing up at me.

That's not to say there aren't any... just none that jumped out. However, as those of you who've read Antibrains may remember, Antismoking studies tend to be like the "one-ahead" mind-reader charletons: sometimes they look REALLY good, and seem like bedrock stone solid proof to a believer; but to a skeptic who knows that there's just GOTTA be a fake somewhere in there it's usually just a matter of finding the "trick" that was used.

Once in a while though there IS no trick: a study will come up with a statistically significant effect and will also have been both well-designed to correct for confounders and probably honestly done (I believe there are SOME cases of outright fraud in these studies, but I don't think they're real frequent.) This may be such a study.

BUT... even if it is, remember that a significant finding does not necessarily mean a true finding. Again, check Appendix A in Antibrains (or just look online at
http://www.nycclash.com/philly.html

and read the Abstract to the WHO study. The World Health Organization found a strongly significant effect showing that children of smokers got 22% **LESS** lung cancer later in life than matched children of non-smokers. Was that finding reflective of the truth? Maybe. But will you ever find an Anti who believes it? Doubtful. In the same way, a single well-designed study showing a significant relation between secondary smoke and heart disease is not the be-all and end-all of what the truth is.

By the way, the "significance" of the figures in this study is a bit on the shaky side. While it's true that any confidence interval that does not include the value 1.0 is significant, some of the intervals found in this study are VERY close to 1.0 (The lower bounds on the three main measures were 1.01, 1.03 and 1.08. Given that the upper bounds were over 2.0 in each case it seems that there was a lot of variability that could easily have nudged those lower bounds below statistical significance with a bit of luck/chance involved.) The closeness of the figures to nonsignificance is very similar to what we saw in the "Great Helena Heart Miracle" study... although that study never actually examined secondary smoke.

- Michael

Michael J. McFadden
Author of "Dissecting Antismokers' Brains" http://www.Antibrains.com

---

Welcome abboard Michal,
Firstly let me state this, and please don'tread this as a flame, but this is a serious mistake that most people don't realize. Any attributable risk factor increases of less than 1 are NOT statistically significant, i.e. a 22% reduction in lung cancrer equates to going from a 1 to a 0.78 risk factor. This is not statistically significant. When the media claims that there is signifcant increase of 66% in COPD this changes the risk factor from 1 to 1.66, again not significant statistically. To be statistically significant the risk factor would have to increase from 1 to AT LEAST 2 and this would still be viewed with some skepticism from serious scientists until a biological pathway was proven.

There are some instances when even very high risk factors are viewed with some doubt. Now this is not necesarily my oppinion but there are some who view the link with smoking and lung cancer as tenous even though there is a RR of like 33. The problem is the shear complexity of the system (the human body) and the fact there is no way to prove that those who contracted lung cancer wouldn't have gotten it even if they hadn't smoked. This is because there has been no proven biological pathway.

On a side note for those of you who are curious the merger between RJR and B&W is going to go thru and I got an offer to continue work with the new company (hehe acronyn RAT Co.). I will be working in the in vivo group so I will get some whole new perspectives doing rat inhalation and skin painting testing. I am very excited as I have been mainly concentrating on the analytical side of the industry till now.

posted 07-01-2004 03:44 PM            

While it's true that relative risks of less than 2 (or .5 in the case of a "protective" factor) should be treated as relatively insignificant in most epidemiological studies due to lack of control and unknown confounders, this concept is distinct from "statistical significance."

If my understanding is correct, statistical significance usually means that there is a 95% chance that the "real" result does not include the value of 1.0. purely on the basis of statistical chance. It says nothing about confounders or how well a study was constructed. It's usually seen as merely a minimum standard to be met before bothering to look more closely at the more complicated questions of what might have produced a study's result.

In the case of the WHO study I believe the 95% confidence interval for matched sets of children exposed and not exposed to secondary smoke extended from .58 to .96. Since .96 is below 1.0 the results were statistically significant. Not "proof" of any sort of protective effect, any more than the new study on CHD and secondary smoke is proof of a deletrious effect, but a sign that an effect is worth looking at more closely.

Michael

Michael J. McFadden
Author of "Dissecting Antismokers' Brains" http://www.Antibrains.com

posted 07-01-2004 03:46 PM            

1. WHO tried to bury it! They spent all this time and money on a study to prove the secondhand smoke is the Great Satan, and when they didn't get the "right" result they tried to bury the report. When they were caught, of course, they reversed position and claimed no intention of burying the study. Riiiiiight.

posted 07-02-2004 03:37 PM            

First of all, the most relevant subgroup for examination is NOT "All Men," but rather "Excluding Former Smokers" since the effects of past smoking and exposure to hundreds or thousands of times the concentration of smoke compared to nonsmokers is a variable far too important to throw into an examination of such low level exposures. Once we concentrate on TRUE nonsmokers the evidence is not nearly as strong as the news media is making it out to be.

Take a look at the nine "Hazard Ratios" for true nonsmokers exposed to secondary smoke in Table 2. Of those nine, only ONE passes the test of statistical significance and it's not even the category of heaviest exposure.

Then take a look at Table 3. As the so-called "passive smokers" are exposed to more and more years of secondary smoke, their heart disease risk comes closer and closer to showing no increase at all over those with "no exposure" (i.e. <.7ng/ml) This is strongly counterintuitive of what we would expect if secondary smoke was actually promoting heart disease.

Finally, spend some time looking at Table 1. Although tables 2 and 3 "adjust" for the wildly varying variables noted in Table 1, those variables ARE different enough across the different exposure groupings that such adjustment and the manner in which it is made becomes a very important issue. Note for example the percentages of heavy alcohol intake across the none, light, medium, and heavy exposure groups: 2%, 4%, 8%, and 16%. The effect of such intake, even after "adjustments" are made, could still overwhelm any actual effect secondary smoke exposure might have.

Look at the FEV(1) <Forced Expiratory Volume> values and at first it looks like there is a very clear effect from secondary smoke exposure: the numbers go from 357, to 355, to 346, to 329. However once one looks at the value for Active Smokers, those whose exposure as measured by cotinine is 28 times as intense as even the most heavily exposed nonsmokers and we find that their FEV(1) is *identical* to that of the nonsmokers exposed to about 3% of their "dose" of smoking.

Take a look at the percentages showing "Evidence of CHD" : 23%, 25%, 23%, 28%, and then back down to 25% for active smokers.

Overall this study is light years ahead of the sort of propaganda piece exemplified by the "Great Helena Heart Miracle" study, but despite the efforts of the authors to correct for a very respectable array of variables, I think the overall results are far from convincing.

All that being said, I think it's important to recognize that there have been a fairly good number of studies done to this point that indicate at least mild positive association between CHD and secondary smoke exposure. While the association does not seem to be logical in terms of the size of the claimed effects versus the intensity of exposure, the existence of such an association should at least form the basis for further investigation as to what the true cause *might* be. Of course I may myself be misled here: the perception that studies generally tend to support some association may be more of a media fed fantasy than a scientific reality.

In any event, neither this study nor any other study before it is relevant to the questions facing bars, restaurants, and other workplaces today that are willing to install filtration and ventilation equipment to remove 99% of smoke components from the air. **NO** study has been done that would indicate that air cleaned to such an extent poses **ANY** reasonable hazard to people. Indeed, if the right equipment is installed and the air in a smoking bar is made cleaner than the air in a nonsmoking bar without such equipment we may see studies down the line that will urge nonsmokers to hang out in smoking bars for their health!

posted 07-04-2004 02:16 AM            

quote:

---

Poor Science

Anthony S Ward, retired

Sir,
re: Passive Smoking report 30/06/04:

I am very concerned that a respectable scientific journal could publish such a study. Further, I am shocked that the press have further exaggerated the reports 'findings'. This at a time when a major political decision on the subject is imminent. Very few doctors, politicians and members of the public will examine the report in detail and fewer will understand it and so will rely on press reports.

My scientific/statistical objections to the study could fill several pages, but I will just point out two:
1. The risk factors in the conclusion are taken from table 2 (part 1). These figures include ex-smokers (potentially ex heavy smokers of 40+ years). CHD risk is known to be increased for this group and has nothing to do with passive smoking. Table 2 (part2) gives the figures excluding ex-smokers and hey presto, 2 of the 3 results are no longer statistically significant. The remaining one still is (just) but only after the third adjustment, which at +0.45 is huge. Such adjustments are little more than best guesses and a result that only occurs through adjustment is highly suspect.
The authors comment that the figures after excluding ex-smokers look similar has no place in a serious paper. In conclusion - the study shows no statistically significant evidence for a link between CHD and passive smoking (let alone a causal one).

2. Page 4 states that ex-smokers had 'no specific incentive to provide inaccurate information' (i.e. that they had taken up smoking again). Except that if their insurer found out, they could invalidate their life assurance. (Non-smoking discounts were introduced by all major insurers during 1981/2).

What happened to peer review and editorial scrutiny ?

---