Has it really been a whole year? The longer I blog, the faster time seems to fly. Or maybe it's just because I'm getting older. Whatever the case, you may (or may not) recall that about a year ago I got into a little tussle with a certain Libertarian comic and some smoking cranks over the issue of whether secondhand smoke is a health threat. The discussion escalated a bit, and some serious smoking cranks entered the fray, complete with quote-mining. I ended up discussing a couple of studies that claimed to have found a decrease in hospital admissions for acute coronary syndrome (colloquially known as heart attacks) in the months after the enactment of smoke-free workplace laws, one in Helena, MT in particular. These studies all had a lot of shortcomings, so much so that they made it too easy for those opposed to smoke free workplace laws to dismiss them.
And, so, after a flurry of posts about this topic that ended, ironically enough, exactly one year ago today, I more or less let the topic drop.
Until now.
What should I find in the latest, crispy issue of the New England Journal of Medicine but a large study from Scotland looking at the very same issue that avoids most of the flaws in the previous studies. The study, from the University of Glasgow and the University of Edinburgh, comes to the same conclusion: that the implementation of a smoke free workplace law resulted in a rapid and significant decrease in admissions for acute coronary syndrome, 17% to be precise.
One of the biggest (and most valid) criticisms of the Helena, MT and Pueblo, CO studies before, which are the two studies that I'm familiar with that showed a drop in such admissions after the implementation of smoke free workplace laws was that they were retrospective studies and thus prone to all the numerous possible confounding factors. Another valid criticism is that these studies often didn't cover a large enough geographic area or include enough patients to minimize the effect of having their cachement area include communities that did not have similar smoking bans.
The present Scottish study is the largest and most rigorous study thus far done examining the hypothesis that the enactment of workplace smoking bans, including restaurants and bars, results in a decrease in admissions for acute coronary syndrome (ACS). Scotland has a population of 5.1 million, and there were nine hospitals studied, which made up 63% of the admissions for cardiac disease during the time period studied. Most importantly, the authors prospectively studied hospital admissions for acute coronary syndrome. I can't emphasize how important this is. Prospective studies may be prone to confounders, but the number of potential confounders increases exponentially for a retrospective study to the point where it's impossible to control for them all. Also, the criteria for case ascertainment used to define whether a patient had an admitting diagnosis of acute coronary syndrome were rigorously defined. In addition, smokers and nonsmokers were stratified, and it was verified that nonsmokers' reports of secondhand smoke exposure correlated with serum cotinine levels. (Cotinine is a metabolite of nicotine that is often used to evaluate and quantify exposure to tobacco smoke.)
Investigators then examined hospital admissions for acute coronary syndrome for the ten months before the enactment of a smoke free workplace law at the end of March 2006 and the ten months after. Month-by-month comparisons were made in order to control for seasonal variations in the number of admissions for acute coronary syndrome. Finally, to account for unrelated changes in lifestyle or treatment changes,the historical trend in admissions in Scotland was compared with contemporaneous data from England. The reason for this was to make sure, as much as possible, that a preexisting downward trend in such admissions didn't confound the results.
Once the data were collected and the numbers crunched, the result showed that admissions for acute coronary syndrome decreased by 17% in the period after enactment of the law. Even more striking, there was a decrease each month that tended to be more pronounced as time when on, as demonstrated in this graph:
Patients admitted for acute coronary syndrome were prospectively questioned and blood samples taken from them for cotinine levels, allowing investigators to determine that two-thirds of the decline in admissions was attributable to nonsmokers or former smokers, strongly suggesting that decreased exposure to secondhand smoke was responsible for the decline. Because of its size and rigorous design, this study represents some of the strongest evidence yet supporting the hypothesis that enacting workplace smoking bans results in decreased numbers of admissions for ACS. It also is consistent with the results of the much-maligned Helena and Pueblo studies, which, despite their retrospective nature and many flaws, appear to have been mostly confirmed and supported by the Scottish study. A Medpage Today story summarizes additional results quite well; so I'm going to quote from it, rather than reinventing the wheel and summarizing myself:
Remarkably, Dr. Pell and colleagues found that non-smokers appeared to benefit significantly more than current smokers.
Among the latter, admissions fell 14% (95% CI 12% to 16%) after the ban, whereas among those who had never smoked, they declined 21% (95% CI 18% to 24%).
The decline in admissions was also greater among former smokers (down 19%, 95% CI 17% to 21%).
The researchers found significant differences along age and gender lines.
Women appeared to have benefited from the ban more than men. For example, among those who never smoked, the decline in admissions reached 28% in women (95% CI 23% to 33%), whereas for men the decline was only 13% (95% CI 9% to 17%).
The researchers said this result may reflect the increased risk from smoking to women relative to men. When they reduce exposure to smoke, women may therefore benefit disproportionately as well.
Likewise, older people -- men over 55 and women over 65 -- appeared to benefit more than younger individuals. For non-smoking men 55 or younger, admissions for acute coronary syndrome did not decrease significantly at all.
Dr. Pell and colleagues noted that in England, which did not implement a smoking ban during the same period, hospital admissions for acute coronary syndrome declined by 4% and there was a mean annual decrease of 3% in Scotland during the decade preceding the study.
Another excellent summary of this trial can be found here.
The obvious caveat about this study is, of course, that correlation does not necessarily equal causation. However, when a study is done rigorously enough and with adequate controls for confounding factors, correlation can be highly suggestive of causation, and this is the case here. There is a close temporal correlation between the drop in hospital admissions and a single change in public policy. No other major change that could account for such a dramatic drop occurred during the same time period, and a very similar population on the same island demonstrated no such decrease, which was dramatically more pronounced than the 3-4% per year trend in decreases. Add to that the methodologically rigorous manner in which smoking history, exposure to secondhand smoke, and serum cotinine levels were measured, ant this study is very persuasive evidence suggesting that smoking bans of this sort have a rapid and dramatic effect on public health.
Indeed, by any measure, this is amazing stuff. Even I was skeptical that results this striking could result from such a simple public health measure, given the sometimes difficult to quantify health risks due to secondhand smoke. Moreover, this study demonstrated that, among nonsmokers and former smokers, cotinine levels fell significantly, and the percentage of people who had never smoked who reported no exposure to second-hand smoke increased from 57% to 78%, an increase largely attributable to reduced exposure to smoke in bars, restaurants, clubs, and pubs. There was a component of decreased smoking among current smokers, as well. In the three months leading up to the ban, there was a 67% increase in calls to smoking cessation hotlines. Clearly some people were using the ban as an impetus at least to try to quit smoking, and even if they failed they generally did cut back for a while at least.
Of course, I have no illusions that the resistance to such smoking bans in the workplace will go away after this study. Such resistance is primarily based on ideology rather than science, and I can to some extent understand and partially sympathize with the view that restaurant and bar owners should be free to allow or ban smoking as they please. On the other hand, the libertarian argument that restaurants or bars should be free of such regulation is in marked contrast to the acceptance of routine public health laws. Restaurants and bars are heavily regulated to begin with because of their very nature and how easily slackness in food handling and preparation can endanger public health, and no one would argue against such regulations. Now evidence is now mounting that allowing smoking in public restaurants and bars is similarly a health hazard; it is not unreasonable to consider regulating it as well to prevent health problems attributable to secondhand smoke.
It is also true that there is the tendency to go too far with such bans. For example, smoking bans at outdoor venues have nothing that I can tell to do with protecting public health and lots to do moralizing or "setting an example." Indeed, my campus recently went smoke-free everywhere--including outdoors--and that was almost explicitly part of the rationale. It's one of the dumbest things I've seen; smoking is banned anywhere on university property. However, evidence is accumulating that indoor workplace smoking bans have beneficial effects. The current Scottish study represents the latest and the strongest current evidence suggesting significant public health benefits in communities in which indoor smoking in public places is banned. It represents yet one more nail in the coffin of the argument that secondhand smoke is harmless or that there is not a tradeoff for permitting it. Given that, I expect that it won't take long at all for a a concerted attack on these investigators and their study to be launched, if one hasn't been launched already.
REFERENCE:
Pell, J.P., Haw, S., Cobbe, S., Newby, D.E., Pell, A.C., Fischbacher, C., McConnachie, A., Pringle, S., Murdoch, D., et al. (2008). Smoke-free Legislation and Hospitalizations for Acute Coronary Syndrome. New England Journal of Medicine, 359(5), 482-491.
Wow, that really seems to indicate a health risk of secondhand smoke exposure for public places. I would have expected a small but not very important effect there, and a more relevant effect for home and car exposures when you're living with a smoker. (From my childhood experience, you had to spend a lot of time in smoky restaurants to get the level of exposure you got daily living in a house with two smokers!)
Is there any research on alternatives to smoking bans that decrease this exposure? Supposing you have a house with smokers, or a restaurant/club/whatever permitting smoking, has anyone worked on how to decrease the bad health effects? I mean, you could run a vent fan or a commercial air filter, but does anyone know how much good it does, or how to make sure the ventilation is good enough to have a significant effect? (I assume a lot of this depends on what component of the smoke is causing the problems. If it's particulates, then maybe running a HEPA filter helps. If it's CO, you probably need outside ventilation. Etc.)
Now if they can just stop them eating deepfried Mars bars.
Disgracefully inaccurate blogging!
Its Scottish lilt, (Irish is the one with the brogue!)
So, we stopped breathing large amounts of slightly toxic smoke into our lungs from the workplace and bars and clubs and then we got a bit healthier. Wow...shocking...not.
How pro-smokers can believe that this wouldn't happen borders on, no goes over the borders into fantasy and delusion.
I used to come home stinking of the stuff from the pub/club, it was foul beyond belief and could make breathing difficult just being there.
Interestingly Dr. Michael Siegel first commented on this study here as another example of "science by press release" typical of tobacco control. I believe you when you say you believe the study is well done, but let face it, if it is well done it does not add strength to the junk science studies like Helena's. I would like to get my hands on the actual paper. A good history on passive smoking can be found here and an examination of the evidense against passive smoking is also one site examines most of the spousal smoking studies.
Indeed. As a Scots smoker, I wasn't exactly thrilled when this law was introduced. Given the dramatic nature of these results, I take it all back.
Sigmund - The Scots have Brogues too, we just wear 'em on our feet.
Siegel's full of crap in the link about the Scottish study provided. The study was presented at a conference with other research papers. We may not know how rigorous the peer review for such a conference was (peer review for abstracts submitted to conferences are almost always subject to a lower standard), but it's perfectly legitimate to present abstracts at such conferences and it's also perfectly legitimate to issue press releases about presenting abstracts at such conferences. Indeed, there were a number of studies presented in June at the American Society of Clinical Oncology Meeting in Chicago that hadn't been published yet. They found their way into the press, and for some of them press releases were issued before the meeting. Are you saying this shouldn't be done? Some of the studies were quite compelling. I know, I was there and attended the talks.
Quite frankly, Siegel gives me the willies these days. I don't know what to think of him, except that I've never seen anything on his blog that ever praised (or was at least neutral) to a study showing the health dangers of secondhand smoke. Not once. He's also allowed himself to become too closely associated with some of the most radical pro-smoking cranks I've ever seen, something I've criticized him for before and behavior that I consider to be ill-considered at best. In fact, last fall I summarized my thoughts on Dr. Siegel. A brief perusal of his blog last night shows me that he apparently didn't take my criticism over his frequent use of the denialists' favorite term, "junk science," to heart, despite having shown up in the comments to respond. He's still using the term, and the term still echoes of Michael Fumento or Steve Milloy.
I fully expect he'll try to find nothing but fault with the Scottish study. Based on the link you provided, it appears he already prejudged it 10 months ago. However, I would point out that the published study addresses pretty much all of the major possible shortcomings Dr. Siegel mentioned. I don't see a lot there for him to attack.
I saying it depends with regard to press releases. Tobacco Control (public health) has a tendency to release press releases to studies that may never get published in peer reviewed literature. Examples, include a John's Hopkins study that was released on their web page (link no longer exists) when the Maryland smoking ban was under consideration.
The study went something like this. Students using particle detectors go to local bars within walking distance of the university campus and measure particles in the air in bars where smoking is allowed and outside. They conclude more particles in air inside bars that allow smoking and therefore second hand smoke is bad. There were no controls. They did not mention weather (was it a hot humid smog alert day or nice clear no humidity winter day). They did not control for venues that did not allow smoking, etc. This was clearly a study designed for the press for quick support to get the Maryland smoking ban passed. However it is par for the course for the tobacco control movement.
Just the other day you said someone similar in this thread. You said:
Seems like you are complaining about science by press release in your field no less by a respected researcher. My comments in your blog on the same subject were:
Apparently our opinions on science by press release are similar. As for Siegel's opinions on the current published paper, I cannot comment as he has not written on the subject yet.
I don't consider a press release about a study that is being presented at an academic conference to be "science by press release," because such studies almost always ultimately end up being published. Frequently, the order for reporting studies goes like this: (1) Present as an abstract at a meeting; (2) submit for publication. Usually by the time a study is ready for (1) it is being prepared for (2). Some meetings even require that presenters submit a manuscript at the time of the meeting. As I said, Siegel's full of crap when it comes to that criticism of the Scottish study.
The UPitt issue is different, apples and oranges. Dr. Herberman issued a fear-mongering warning based on a study that he can't seem to get published. Worse, it was a study that even his press release had to admit that the peer-reviewed literature doesn't support. I even pointed out in my post that he should have at least presented it at a meeting before making the warning.
Finally, I don't know what's happened to Dr. Siegel, but he now seems to flirting with crank territory. It's a shame. His eagerness to trash the tobacco control movement over any use of science to advance their agenda has clouded his objectivity, IMHO. He routinely goes far, far beyond just "keeping them honest."
Christopher Snowden, author of the forthcoming book "Velvet Glove Iron Fist - A History of Anti-Smoking" gives the paper a debunking here.
Ok, you've already sent out the bat signal for the loonies who want to kill me with their exhaled smoke. Now if you can just send out the bat signal for the loonies who want to kill me with their car exhaust (the AGM deniers), the loonies who want to kill me with preventable diseases (the anti-vaccination crackpots), the loonies who what to kill me with woo-woo "health care" (lots of 'em including a few so-called "medical schools"), and--... well, actually, that's probably a enough nutcases for one thread. Not quite sure how you'd tie them together though, other than being non-rational. Some heavy chain might work, but being toxic waste, I don't know how you'd then dispose of 'em.
No, on second thoughts, that's way too many. Don't bother. The smokestacks will be obnoxious and deluded enough, if last year's invasion was any guide.
"It is also true that there is the tendency to go too far with such bans. For example, smoking bans at outdoor venues have nothing that I can tell to do with protecting public health and lots to do moralizing or "setting an example." Indeed, my campus recently went smoke-free everywhere--including outdoors--and that was almost explicitly part of the rationale. It's one of the dumbest things I've seen; smoking is banned anywhere on university property."
It may be dumb, but it could be useful. Suppose - which seems quite reasonable - that as a consequence of annoying tobacco control measures such as this, some people stop smoking or don't start. This would indirectly reduce exposure to first hand smoke even if it had no effect on second hand smoke. And it would have nothing to do with moralizing or example-setting, if the effect was simply to discourage smokers from smoking by making it increasingly inconvenient to do so.
Now I don't know whether this would actually work in this way. There must be some studies? But it's silly to overlook the possibility. Every cigarette not smoked is a cancer suppressor gene not threatened.
Now they need to ban alcohol from bars. The lowering of drunk driving, alcoholism, and the many crimes & family troubles caused by it would also be a blessing on society.
Now they need to ban unhealthy foods (Mc Donalds anyone). San Francisco is already doing this by not allowing new fast food resteraunts. The lower rates of obesity, heart disease, etc would also be a blessing on society.
Oh, please, give me a break.
Whether someone eats fatty food or not doesn't affect other people the way secondhand smoke does, and there are already laws against drunk driving. As for that "critique" of the paper, a cursory reading of it showed it to be fairly lame. No time to go into detail now (maybe later).
I agree that outdoor venue bans and all over campus bans, etc. are over the top. But I think it's reasonable to ban from the workplace and, quite frankly, from restaurants. Even if I were still a smoker, I'd be ok with that. While the health benefits to 2nd hand smoke shouldn't be a surprise, it's nice to have it spelled out for those who need it.
Adding to the list of criticisms of the Helena study is the fact that an identical drop in AMI admissions was seen just prior to the study's time frame.
Which raises the questions, Why didn't the authors include that period of time in their analysis? Could it have anything to do with everyone involved with that study were anti-smoking activists?
On waiting to be published vs "publication by press release" or presented at a conference, you may recall the drop in AMIs went from 60% to a later revised 40%.
It was a prospective study. That means that they could not go back and include data from before the study began. Carrying out a prospective study over a fixed time interval prevents them from "cherry-picking" the interval to get data that fits their preferred hypothesis (as you are attempting to do). Note that this was not merely a statistical study of hospital admissions--they were objectively determining smoke exposure by measuring a nicotine metabolite at the time of admission, so it would probably not have been possible for them to go back further in time even if they wanted to do so.
Oops...missed the denialist switcheroo. On rereading, I realize that rather than dealing with the recent study under discussion, James Austin went back to pick on the Helena study some more.
As a former EMT, I have to say that while the two aren't directly comparable, there are certainly public health effects inherent in regular fatty food ingestion that go beyond the individual.
I haven't done any research, but I wouldn't be surprised to find that, after controlling for improvements in patient moving technology, the incidence of lifting injuries among emergency workers correlates with the rise in obesity.
Having been raised by alcoholics (who also smoked), I disagree that allowing alcohol is only harming the drinker. Also, the various bad things to happen to drunks (from assaults to legal fees to very bad decisions) also hurt the family.
Well, you can understand why this study might be greeted with some skepticism. The anti-smokers have a bad history with distorting science. But The Scottish Data looks a lot more solid. I'll have to set my stance on smoking bans from against to leaning for.
17% is huge and way more than I would have expected. This would imply an amazing and frankly unbelievable lethality to second-hand smoke. I"m curious to see how it does in peer review.
Orac, can you explain why they would look for ACS and not just heart attacks?
It may not be for health reasons that smoking is being increasingly banned in outdoor public places such as university campuses and beaches. It may just be that the maintenance crews are just sick and tired of picking up cigarette butts. Smokers, for whatever reason, have dug their own grave on this one. For some strange reason, the probability of a smoker finding a trash can approaches zero the second he steps outdoors. Just look at the ground directly outside any building. Point proven. Providing ash trays and trash cans just doesn't work.
Orac, Dr. Michael Siegel has written his comments about this paper. They can be found here. Perhaps you want to butt heads with him? It would be interesting!!!
Siegel does like the Kuneman/McFadden study which found no effect:
"To really answer this research question, what is needed is a systematic study of multiple locations with smoking bans. I believe that McFadden and Kuneman have conducted such a study, in which they examined heart attack trends in a number of U.S. states with and without smoking bans for which data were available. This study has not yet been published, as not surprisingly, it is probably getting unfair reviews from anti-smoking researchers.
I conducted my own analysis of the data, using McFadden and Kuneman's methods, for all U.S. states with and without smoking bans for which data were available. I found no evidence that there is any substantial, immediate decline in heart attack admissions in states with smoking bans."
http://tobaccoanalysis.blogspot.com/search?q=Kuneman+heart+attack
Do smoking bans cut the smoke exposure of nonsmokers?
http://www.ifs.org.uk/publications.php?publication_id=3523
I should first say I know essentially nothing of the detailed science (aside from the odd scrap of course). And secondly, that I agree with these smoking bans and that they will, in time, have positive health benefits (i.e. don't read my post the wrong way!)
Aside from usual "correlation does not necessarily equal causation" issue, I'm struck the size of the effect over such a limited time. Only one year later, a 17% reduction is seen?
How does this compare with the understanding of the time-course of the effects of smoking on coronary disease? I would have naïvely thought this is longer than a year. Vice versa, what is the expected time course of improvement in coronary disease given you stop smoking, or being exposed to smoke?
I notice that the countries that have banned smoking in public places (Ireland, New Zealand, Norway and Scotland) are all small. I wonder why. Keener to tackle anything that would help their limited health budgets? Easier to sell a message or push a policy on people? Fewer degrees of freedom in who knows who means its easier for people to anecdotally know of the effect?
As a civilian with no scientific training, I can't tell a good study from a bad one. All I know is that down here where the real people live all these "studies" are being used to micromanage my life.
Coffee is bad for me. Wait, it isn't. Now it's good for me. I am sick to death of laws for my "safety." Some recent "studies" have told me that my shower curtain will give me cancer and global warming causes kidney stones. Blah blah blah.
Some "scientists" say a whiff of tobacco smoke will kill you. Others don't. Who do I believe? I can decide for myself what establishment to enter. I can also decide what kind of toilet to buy and which light bulbs to use.
Argue among yourselves all you want. Beat your chests and rend your garments. Distribute your health leaflets like Moonies. Whatever. Just leave me alone.
Sorry Cogerias but it's all percentages. Science can only inform.
Personally I blame the media for jumping on every new study regardless of it's findings and especially if it counters previous knowledge. These things should be taken in context which is hard to do given the information dissemination setup that we have.
Cogerias,
before the smoking ban, I could not enter a single restaurant or bar without being exposed to smoke. The decision was not "do I want Chinese or Italian" or "do I want to hear Jazz or Funk", the decision was "do I stay in and breathe or do I go out and come back with a raging headache and reeking of smoke". The quality of life for *everyone* in Scotland has gotten phenomenally better since the smoking ban.
Catherina,
I assume you are in Scotland, and everyone there shares your view of what good "quality of life" is.
Where I live about 85% of the restaurants were already entirely non-smoking. We have plenty of music clubs, and around half of those were non-smoking. There were plenty of choices for everyone, including waitstaff and bartenders, who are the ones they wanted to "protect." It seemed the only problem was that smoking was still allowed in some places, and some people didn't like it.
If enough patrons make their preferences known to the management, perhaps they will change their policy. Why not collect some signatures to present to the management of your favorite jazz club? Why do you have to have *all* the venues, while patrons who smoke and the owners who wish to cater to them have none?
I cannot be the only non-smoker on the planet who finds this appalling. The smell of smoke doesn't bother me, and some pipe smoke is quite pleasant. Cigars, not so much. I stopped going to my favorite bar after the ban. The crowd has diminished by half, and the stink usually covered by the smoke was unbearable.
I don't like the stink of other people's cologne. There has to be some "science" out there to get it banned. Or I could just walk away.
Richard said "...it's all percentages. Science can only inform."
What does that mean? If this pile of studies says something is harmful, and another pile says it isn't, what should we do? My preference is to ignore the whole load of it, but that's getting harder to do when this crap is used for ban after ban, which in the end, accomplish nothing but political brownie points.
Passing laws banning this and that using this kind of science is way beyond informing. If you want me to know something is dangerous, inform me. I will then CHOOSE whether or not to expose myself to it.
Cogerias, its governments that set the bans, not science.
"I notice that the countries that have banned smoking in public places (Ireland, New Zealand, Norway and Scotland) are all small."
Italy, England and the state of California are not small and have smoking bans.
Heraclides wrote:
Exactly this does not make sense. Only if we believe the public statements Dr. Stanton Glantz (a professor of medicine who like myself is really only a mechanical engine) makes, that as little as 30 minutes exposure to second hand smoke can cause a heart attack would this make sense. He said
Then plausible. I say bullshit. Dr. Glantz is the biggest anti-smoker ever. His science (junk) produced Helena (along with other studies) which was a cherry pick. If you can get a heart attack from just 30 minutes exposure, we would have noticed ages ago. Please note that his papers do not say this, it is his public announcements with regard to his papers that do. Dr. Seigel has his analysis here and to me, make more sense then Orac's analysis.
Heraclides,
Why do the governments set the bans? Because of the scientists pushing for them. Busybodies waving studies bullied restaurants and manufacturers away from saturated fats in the 80s, pushing the use of trans-fats. (Turning McD's fries to crap.) Now new studies and new pressure, not just to change, but to *ban*.
In the 70s, it was butter that was the root of all evil, so better use Crisco. There was science behind that, too.
I just read this morning that granite countertops pose a cancer risk. Good thing I can't afford those! I'm already supposed to be terrified of my killer shower curtain.
"Science" now means the same to me as "clinically proven" in a TV commercial.
"Exactly this does not make sense." That's not what I wrote, Dan. I asked a question, not made a statement. In my post, I said I'm not knowledgeable in this area and then when on to ask as I was surprised at the size of the effect. I wasn't trying to say I thought it unlikely, just unexpected to someone not knowledgeable in the area. It may well make sense, and if so its a good result; its just I had naïvely thought these things were long-terms effects, but the results would suggest there are short-term effects, too. (We can all learn, you know!)
Thinking about it now, its not so silly. Smoke (of any kind) would stress the CVS after all.
I haven't time to explore Seidel's argument in detail, but two obvious flaws: (a) he tries to imply that the nine hospitals are minority subset of the total, whereas Orac pointed out that they "made up 63% of the admissions for cardiac disease during the time period studied"; (b) his "analysis" doesn't use statistics, but "hand-waves" specific numbers. In particular, there is no discussion of statistical significance, which is odd. Its enough to make you suspicious that this argument is cherry-picked. Wouldn't mind seeing EpiWonk having a crack at this.
Dr. Siegel's article comes across as denialist rationalization. He doesn't want to deal with the big drop in coronary syndrome, which is based upon an objective laboratory test (and so is not vulnerable to possible shifts in diagnosis criteria). Instead, he prefers to talk about heart attack statistics. And while the Scotland study was done in a prospective manner, with the time periods preselected (eliminating the possibility of selecting after-the-fact a particular period in which the change fit what they wanted to see), Siegel goes cherry-picking through the year-by-year data for heart attacks (not coronary syndrome) and argues that there are years when the fluctuation in heart attack admissions is as great as the decrease (in heart attack admissions, not coronary syndrome) seen after the smoking. Of course, from a statistical point of view, this is invalid.
Somewhat ironically, Siegel has titled his article "Apple's to Oranges," because he doesn't want to talk about the Apples--a prospective study using an objective laboratory test, and using a preselected time period. So he is trying to shift the discussion to Oranges--heart attack statistics compared over an interval that he has selected after-the-fact to fit his thesis.
Cogerias, you keep mentioning bans as though there have been wave after wave of laws against particular things recently. All I can think of at the moment are smoking bans in public places and California's recent ban of trans fats in restaurants. So really, what are you complaining about?
And I don't know where you live, but every bar in the Twin Cities allowed smoking before the smoking ban. Nearly every restaurant had a smoking section. And a ban on smoking in restaurants and bars does protect the employees there, in exactly the same was the Clean Indoor Air Act protects employees of every other business. It's not that easy to find a job, and in many cities without smoking bans it's actually not possibly to find a non-smoking bar to work at.
By your logic, we should scrap every worker protection, since people who don't like their job can just quit and find another one.
I didn't look at the blog at all last night; so the conversation has gotten away.
I do have to agree mostly with trrll. A quick reading of Dr. Siegel's critique doesn't impress that much. Indeed, he seems to be harping on the downward trend in ACS admissions before the study period (a downward trend that was acknowledged by the authors to be around 4%). Further he seems to be using the heart attack data to imply that a reasonable counterhypothesis is that ACS admissions would have dropped 17% without the ban. The only reasonable point he makes is that the harping on the 17% figure may have been too much. Had I been writing the paper, I would have said that ACS admissions dropped 13% more than would have been expected based on recent trends and emphasized that number rather than the 17% number.
Asking EpiWonk to take a crack at this is an excellent idea, though; that is, if he's recovered from his migraine that prevented him from taking on the cell phone/cancer story from last week.
That last comment is me, not Cogerias. Sorry for stealing your name, Cogerias.
"Why do the governments set the bans?" Because large numbers of people ask them. They're called the voting public ;-)
It should be noted that what Dr. Glantz claims, that a mere 30 minute exposure to second hand smoke (SHS) can cause a heart attack, are the same effects one would feel if one is eating a hamburger or moving from a warm to cold environment and vise-versa. If Dr. Glantz is to be believe exposure to SHS is worse then smoking. Make sense? I think not. The science does not matter anymore. Just make one outrageous claim after the other and soon enough people will believe it. Dr. Glantz really did cherry pick to come up with his findings. This study is of the same ilk.
I am eternally grateful to the government for implementing the ban.
Governments impose bans because powerful groups like the American Cancer Society pressure lawmakers and scare the public with absurd claims.
http://keepstlouisfree.blogspot.com/search?q=American+Cancer+society+Dr…
While this last heart attack study is better done than Helena, I still find it implausible that small exposures to tobacco smoke in restaurants and bars that still allow smoking could have such a catastrophic effect. Both Dr. Siegel and I find the Kuneman/McFadden heart attack study more believable.
http://tobaccoanalysis.blogspot.com/search?q=Kuneman+heart+attack
In St. Louis, which has no ban, I am working with an air filtration company to get bars and restaurants that want to allow smoking to clear their air with atleast 15 air changes per hour of electronic, HEPA and carbon filtration.
I can't imagine that air cleaned to this extent would still present a health hazard. Plus such cleaning rids the air of many other threats at the same time.
http://keepstlouisfree.blogspot.com/2007/11/st-louis-public-smoking-com…
I also live in Scotland. I second Catherina. It now means that I have a choice of whether and where to go. And the ban on indoor smoking at the various environments where I have worked has made a huge difference to my health and happiness as well, because I don't have a choice about breathing, and previous to these restrictions *I* didn't have any choice about breathing smoke, either. This was of concern to me because I'm at high risk for heart disease and blood clots on several fronts; but my ability to make my own decisions to protect myself had been taken away by the ubiquity of cigarette smoke in pretty much every public environment. Smokers' choice to smoke wherever they wanted existed only at a cost and restriction of freedom for me, and many like me.
Smokers now have to go outside to smoke, and do not have the choice to smoke indoors. Ooh, pardon my lack of sympathy here, but seriously, just cry me a river. And democracies are supposed to be majority rule anyway, right? Depending on whether we are talking about Ireland, England or Scotland, public support for the smoking bans runs between 74-91%.
And business is surviving just fine, too, thanks.
http://www.spiegel.de/international/europe/0,1518,472732,00.html
Powerful? The ACS? You've got to be kidding me. The ACS can't even influence legislators to make sure that routine screening tests are always covered by insurance, much less get smoking bans implemented. Better put on the tinfoil hat and go into hiding. The black helicopters are coming for you. You're starting to sound like the cranks who claim that the AMA is somehow manipulating medicine to its nefarious ends in order to lock out "alternative" medicine.
In other words, you don't like the results; so you find them "implausible." I've commented on the Kuneman "study" before here and here. As for Dr. Siegel and the tobacco control movement, over the last couple of years I think he's gone far beyond "keeping 'em honest" to letting his dislike of the more vocal and ideological aspects of the movement seriously affect his objectivity. He's allowed himself to become far too closely associated with radical pro-smoking advocates who oppose virtually any form of smoking regulation, and whenever a study like the Scottish study comes out, he can be relied upon to take their side.
It's truly disappointing. If Siegel ever reads this (and I suspect he has, given a couple of comments in this thread), I would ask him: Given that you trash virtually every study that concludes secondhand smoke poses a health risk (I've never seen Siegel praise one or even comment on one neutrally), why on earth do you still say that you think secondhand smoke is a health risk? On what evidence do you base your statement? In other words, you keep railing against what you see as "bad science"; so show us some good science instead.
And, no, the Kuneman study does not appear to qualify.
BTW, Mike -- "17% is huge and way more than I would have expected. This would imply an amazing and frankly unbelievable lethality to second-hand smoke."
I think this needs to be taken in context. Sharing a house with one smoker is probably not very good for you, overall -- but nothing like being in an environment where there are more than 10 people smoking at any given moment, and the air so thick with smoke that you can't see the far side of the room. That was what I faced in the breakrooms where I worked, and in any pub or restaurant I went to; and an hour in there had me wheezy and coughing for 5-6 hours afterwards. Personal experience is not a good thing to base science on, but I don't personally find the science surprising in the light of my experience.
Dan writes:
Clearly, the science does not matter to Dan. He doesn't address a single scientific or methodological issue related to the study. His entire argument is that it is somehow of "the same ilk" as his caricature of Dr. Glantz's views. By the way, if anybody wants to find out what Dr. Glantz's actual views are, here is a link. Hint: no, he doesn't claim that SHS is worse than smoking.
"By your logic, we should scrap every worker protection, since people who don't like their job can just quit and find another one."
Worker protection is great, if it actually protects workers from something. I'm not all that fussed about a slight increase in risk over 40 years. That's not quite the same as the danger from coal mining or commercial fishing. I don't really see the Discovery Channel making a shocking new show about the "Deadliest Bar." I live in a city where I have to clean black soot off my window sills once a week. Exactly how "clean" are they making the air at the pub?
If workers are allowed to accept those risks, why not tobacco smoke? Anyone anywhere can just quit their job for any reason, including if they think breathing the air is "dangerous."
I still don't see any reason why smoking should be banned in *all* the bars. If non-smokers have no choice, it's unfair. If smokers have no choice, it's also unfair.
Trans-fats were banned here last year. Smoking the year before. Peanut butter sandwiches were banned at a number of local schools, along with soda machines, cupcakes, and candy. Bans being considered are iPods, several brands of sneakers, processed meats, and hot dogs. Teachers may not wear any personal religious symbols.
Weirdtv is banned, along with various movies, books, and songs over the years.
A ban on blunt cigars and "drug paraphernalia" was recently overturned by the court, as it also banned air freshener and Chore-Boys. A weapons ban was just overturned as well.
The bans being considered are the dumbass foie gras thing, skateboarding in all public places, toy guns, plastic bags, bike riding at certain hours, and teenagers at the mall. Those are just the ones I know about. All of this crap is for "safety," and has done absolutely nothing to make anyone safer. I have no idea who the "voting public" is. I got no vote in any of these bans.
Orac said, "The ACS can't even influence legislators to make sure that routine screening tests are always covered by insurance, much less get smoking bans implemented."
That just means insurance companies are *more* powerful than the ACS. Before the ban was passed here the ACS spent a ton of money on full-page ads in the paper and TV commercials.
A public survey on the city hall website, even with horribly biased questions like, "Even though smoking is BAD, should it be allowed in certain places?"
The response was so strongly in favor of compromise that the survey was taken down after a day, and the results never published.
Plenty of "science" was quoted in those ads, though.
trrll wrote:
To address your comment, that the science does not matter, it does. I was unable to follow your link, because an account was required. If you would like, I could post a review of Dr. Glantz's public statements that relate to his papers. Please note, that I could also critique his work, but will not do so, because others already have, I would just point to where to look. Again to answer your comment, science does matter, but it has to be sound. If 30 minutes of second hand smoke (SHS) exposure does kill, it should be much more deadly for smokers. Don't you think? But health effects for smokers do not occur until 20-40 years. The dose does matter. Even Orac, states that banning smoking outdoors is not justified by the science. Glantz would say otherwise, follow the link in one of my previous posts. Also not I am not claiming that smoking is not bad, it is. What I am claiming is that the health risks associated with SHS are way over blown.
Again, you are simply attacking a straw-man caricature of what Dr. Glantz wrote, rather than dealing with the actual science. He did not claim that a single acute exposure to SHS kills with high probability. He wrote "in healthy young volunteers, just 30 minutes of exposure to secondhand smoke compromised the endothelial function in coronary arteries of nonsmokers in a way that made the endothelial response of nonsmokers indistinguishable from that of habitual smokers." So with respect to one particular measure of coronary artery function, it takes only a small amount smoke to produce a full effect. That does not mean that somebody will drop dead from 30 min exposure to SHS, any more than people drop dead from one cigarette. What it does do is provide physiological evidence that supports the hypothesis that the toxic effect of smoke on the coronary arteries "maxxes out" at a fairly low smoke exposure. There is nothing particularly extraordinary about this--all it says is that something in smoke is quite potent in producing this particular effect. What this means in practice is that it is biologically plausible that long-term exposure to SHS could produce cardiovascular risk that approaches that experienced by a smoker.
Employers are required to take all practicable measures to minimize hazards to workers. In some occupations, there is still considerable risk remaining. Making a bar smoke free does not make it safe to work in a bar--it won't prevent bar fights or robberies, for example. But SHS is an avoidable hazard.
Orac,
You asked me to provide examples of "good" studies about the health effects of secondhand smoke and questioned why I don't write about such studies on my blog. First of all, there are far too many studies to list here - there are literally hundreds. The best summary is presented in the Surgeon General's Report of 2006 and the California EPA Report of 2007, both of which conclude (correctly, in my opinion), that chronic exposure to secondhand smoke significantly increases the risk of lung cancer, heart disease, and a number of other diseases. So I will offer in evidence the entire set of references of the Surgeon General's report and the California EPA report as examples of the "good" science which supports my contention that secondhand smoke is a significant health hazard and that all workers deserve protection from this hazard.
As to why I don't highlight these studies on my blog, the purpose of my blog is not to reiterate the things I've been stating and publishing for the past 21 years. It's to tackle new and emerging issues in tobacco control, of which the health effects of chronic exposure to secondhand smoke is not. To me, that's old news and I have little interest in harping on it in my blog. That's what I've been doing for the past 21 years, in scores of published articles, reports, letters, trials, and testimony before public bodies.
With regards to the Scotland study, my main point is that if you look at trends in heart attacks on a national basis (not just in 9 hospitals), you find a much smaller decline associated with the smoking ban (about 4% in the first year, 8% in the second year). However, in the years preceding the smoking ban, there were declines of 6% and 10% observed. Thus, it is clear that a 6-10% reduction in heart attacks from year to year in this data is not at all uncommon. I don't argue that there wasn't a 4-8% decline in heart attacks. I only argue that this decline cannot be attributed to the smoking ban because even larger declines occurred in the past just as part of the random variation and secular trends in this variable.
I think it is very misleading to take the 17% drop in the acute coronary syndrome (which clearly includes heart attacks as its major component as the diagnostic criterion was simply a detectable level of cardiac troponin which occurs in heart attacks as well as unstable angina) as diagnosed in these 9 hospitals under this study protocol and to argue that this drop is much greater than the 10% drop observed in national data. Why? Because the national data is based on a completely different data source and definition and is subject to far less random variation. In fact, if you use the national data to examine the change in heart attacks associated with the smoking ban, you find that it is nowhere near 17%, but it is actually between 4 and 8%, depending on which year you look at. In either case, the decline is less than that observed from 1999 to 2000 and is clearly within the range of changes in heart attacks observed during the baseline period.
Again, I'm not arguing that these data suggest that there was NO effect of the smoking bans - in fact, there almost certainly was SOME effect. I just think that there is no way one can attribute the 17% observed decline to the smoking ban.
I have conducted studies very similar to this myself (for example, analyzing the effects of smoking bans on trends in smoking prevalence). In doing such studies, you need to have a consistent data source over a long period of time (I used a period of over 10 years) to be able to draw credible conclusions.
Finally, you criticize me for this: "He's allowed himself to become far too closely associated with radical pro-smoking advocates who oppose virtually any form of smoking regulation."
I don't understand this criticism at all. I have no "close association" with radical advocates (if anything, the radical advocates I DO associate with are some of my friends at ASH, ANR, and other organizations). I happen to have a lot of the folks who you are presumably referring to as readers on my blog, but so what? It's a free country. Am I to stop anyone from reading my blog if they do not sign a statement declaring that they are not a radical pro-smoking advocate?
It looks like many of the same people read and comment on your blog as well, but I don't claim that you are too closely associated with such people.
Well said Michael. I am one of those that visits this and your blog daily. Also MikeH's (PalMd), PZ Myers, Bad Astronomy, SkepChicks, etc. Of those, I enjoy Orac's the best, even if he gets long winded at times.
To trrll
Agreed, but you do not have to go far within tobacco control to find that they are using these studies to say exactly that to the public. Michael Siegel calls them on that almost daily (exaggeration). Maybe once a month.
No doubt somebody, somewhere, said something unreasonable. But Glantz didn't say it. And your attempt to dismiss the Scotland study by insisting that it is of the same "ilk" as your imaginary caricature of Dr. Glantz's views has noting to do with science.
"Employers are required to take all practicable measures to minimize hazards to workers."
Agreed.
There are several ways to *minimize* risk from smoke in the air. One is to open a window. Another is to install a couple of Smoke-eaters. Contrary to some of the "science" I've seen, air cleaners really do clean the air.
Another is to hire only smokers. No matter how much ETS they breathe, their own smoking will kill them first.
With all the workplace hazards we are free to choose from, what's so special about this one that it needs to be eliminated? Who died and made them OSHA?
Since the cancer risk from ETS is already lower than that for drinking milk, reduction brings the risk to near zero.
That smoking bans are Plan A says to me that "safety" is not the real motive, no matter how many studies are flapped. In the end, I don't give a crap how "dangerous" ETS is. I'm perfectly capable of evaluating a hazard, thanks, so how about I decide for myself, rather than having you do it for me.
trrll said
Ok. I assume you meant nothing to do with science. I was hoping to see Orac's response to Michael Seigel.
So the effect of the ban on a completely different measure of cardiac toxicity gives a somewhat smaller reduction? How specifically does that cast doubt on this result? It seems to me that it reinforces it, but suggests that diagnosis of heart attacks may be a bit less sensitive in detecting the short-term benefit of the ban. Off the top of my head, I can see two plausible reason why this might be the case: the troponin measurement may be more sensitive in picking up the toxic effects of SHS, or the variance could be less (which seems likely, as it is a single, objective laboratory test as opposed to diagnosis of a heart attack by criteria that may differ from hospital to hospital), giving it greater statistical power.
If, in fact, you were going to make an actual scientific argument, rather than a hand-waving, cherry picking one, you would need to make a statistical case that the probability of a chance reduction of that magnitude over this particular preselected interval of time (as opposed to one that you choose after the fact) is not low enough for the effect of the ban to be statistically significant. And it would still be an apples-and-oranges argument, because you are talking about a different measure, over a different patient population.
I thank Dr. Siegel for commenting. Some of the issues we've discussed before in the past. (I also point out that I did not look at my blog at all last night. Believe it or not, I somtimes go 12 or even 24 hours without looking at it. I sat down to check it out this morning, and I saw the above comments.)
Perhaps I should have been more clear about what I meant by "close association." I did not mean that you had personal associations (although your past brief collaboration with Kuneman was highly dubious, as I pointed out before). I also think you probably know that's not what I meant. Still, in retrospect, perhaps I should have said something like "is closely associated with in the blogosphere," which is a more accurate and nuanced way of saying what I meant.
Basically, in your professional realm you may be known as a great anti-smoking epidemiologist and scientist. But people outside your professional realm only know you from your blog for the most part. They have no idea what you do other than in a very vague way. All they know is that you're the tobacco epidemiologist who's always criticizing virtually every study that comes out supporting health risks for secondhand smoke and health benefits from smoking bans. Every one. That makes yours the go-to blog among smoking cranks and opponents of smoking bans for quotes to back them up. It also gives someone who hasn't read anything by you other than your blog the very distinct impression that you are on their side. Your blog certainly gives me that impression, its occasional disclaimer that you really hate tobacco notwithstanding. Certainly your serving up juicy quotes where you claim that the tobacco control movement now distorts science worse than the tobacco companies do are red meat to this crowd and pretty dubious on a strictly factual basis as well. Couple that with your frequent use of the term "junk science," a term that is closely associated with industry shills like Steve Milloy and Michael Fumento (see Chris Mooney's excellent book The Republican War on Science for a detailed discussion of the long and sorry history of that term), and you come across, at least to me, like a smoking crank.
I'm not saying that you are a smoking crank, just that your blog comes across like the blog of a smoking crank to someone like me. My fellow blogger revere over at Effect Measure (an epidemiologist who says he knows you) vouches for you in that you are not a smoking crank, and that's good enough for me. However, in terms of the more radical, ideological part of the tobacco control movement, I think that you've gone far beyond "keeping 'em honest" (a laudable goal) and now seem to have let your gadfly status affect your objectivity. Take my criticism for what you will, but ask yourself: Does your blog really help legitimate, science-based tobacco control efforts? I don't think it does. In fact, I think it probably does the opposite. There are ways to "keep 'em honest" without being so relentlessly one-sided.
Regarding the study at hand, I still think that trrll nailed it pretty well, both in his first comment and his followup a few minutes ago. I don't have all that much to add except this. In your comment you said this:
Oddly absent from your post on your blog about the study is any sort of statement that resembles this strength of "there almost certainly was SOME effect."
In retrospect, I actually agree that the effect is almost certainly not 17% and that that number shouldn't have been emphasized in the paper or in the press releases. As I said before, I would have characterized the results as being a drop that is 13% greater than the long-term trend, and if I were writing an accompanying editorial I'd say that that's probably the upper bound for how large the effect was likely, in fact, to have been. The "true" effect was likely smaller.
Even so, what if the effect of the smoking ban was only 10% Or only 5%? Or only 2-3%, as you seemed to be implying with your use of the 2005-2007 heart attack data (17 - 4.6 - 10.2 = 2.2)? That would still be a benefit. Of course, trrll nailed it when he pointed out that you are actually the one comparing apples and oranges; so my guess is that the "true" effect was significantly higher than 2.2% and probably lower than 13%.
Orac great post thanks for commenting. You to trrll. Now time to spend my saturday cleaning toilets before me wife and child return from Europe after 5 weeks. If not I will be in the literal shit house.
One farther comment, before I start my cleaning duties. Orac, I am impressed. I to do research and actually am fluent in two languages. But when it comes to writing the written word, I suck. You put out well crafted articles almost daily. When I write, I am lucky if I can get out one well crafted paragraph a week (with the help of English majors). Science and math are my strong suites. Writing, spelling, etc. not so much. I read most everything you post (however long winded), but I am impressed I honesty do not know how you do it. The butt kissing stop here, the toilet cleaning begins here.
Off topic (OT). Just delaying the toilet/house hold cleaning thingy. I would have been very interested in your opinion with regard to the PZ Myers cracker thingy. I guess that will not happen? Time to do the toilets.
You probably don't really want to know my opinion.
The reason I never posted on it is because (1) I doubt most of my readers would be interested (I know, I know, that never stopped me from commenting on some issues before, but that only happens when it's something that interests me enough to override that concern); (2) because no one--PZ included--exactly covered themselves with anything resembling glory in this profoundly stupid controversy; and (3) I wasn't in the mood for a troll infestation of a different sort than is the norm around here. Suffice it to say that I lost a lot of respect for PZ over this incident. A lot of respect.
In any case, the time I started to think that maybe I should post something, it was all old news anyway.
Orac,
I'm glad to hear that you don't actually think I am a smoking crank, but that you only called me one. That's gratifying. And thanks, also, for clarifying that you were not accusing me of associating myself with these smoking cranks. I wouldn't want anyone thinking that I communicate regularly with subversive characters like them.
Straw man argument. I did not call you a crank. I simply observed that you came across as one on your blog, and you do--to me. (I can speak for no one else.) At least you didn't refer to the Scottish study as "junk science." I'll give you that much. That was the only indication I got from your critique I that you actually thought it might be OK.
I also note that you didn't bother to answer trrll's and my comments about your critique of the study.
How about this, instead: What specific data would it take for you to believe that smoking bans can significantly reduce admissions for coronary artery disease?
What are you talking about? You communicate with me all the time ;-).
Personally I believe there is no effect. But I am interested in Dr. Siegel's response. Back to cleaning.
Agreed.
It's also one reason why I'm not exactly broken up over PZ's decision not to show up at the ScienceBlogs meetup next weekend. No, his decision had nothing whatsoever to do with Crackergate. It was made months ago. He's going to be on the JREF cruise to the Galapagos. I would have made the same decision, along these lines:
Hmmm. New York City Meetup...Galapagos cruise. Galapagos cruise...New York City Meetup? Hmmmm. I wonder.
Not exactly a straw man argument when you stated: "I don't know what's happened to Dr. Siegel, but he now seems to flirting with crank territory. It's a shame. His eagerness to trash the tobacco control movement over any use of science to advance their agenda has clouded his objectivity."
I'll grant you that "flirting" with crank territory may not be the same as having crossed the line into being a full-blown crank, but it sure sounds like you were accusing me of basically being a crank.
At any rate, that's water under the bridge as you've stated that you don't really think that I am a crank.
"What specific data would it take for you to believe that smoking bans can significantly reduce admissions for coronary artery disease?"
Very simple - a study that uses a consistent data set over a long enough period of time so that a clear baseline trend can be established and which also uses a comparison population so that secular trends can be ruled out in addition to random variation in the data. Ideally, a systematic study of multiple locations that enacted smoking bans would be conducted.
Michael,
I'm still confused as to why you write "just" 9 hospitals as if they were a small portion of the total, yet Orac claims they represent "63% of the admissions for cardiac disease during the time period studied". (Scotland is a small country, so I'm not surprised that nine hospitals represent a large portion of the total admissions.)
Why you think heart attacks would be a better measure of outcome? Surely a wider criteria (e.g. ACS) would represent a larger dataset and be less susceptible to spurious fluctuations and whatnot? (I realise they can include outcomes of other causes, but it seems to me that so do heart attacks.)
I'm also confused as to what you mean by "secular trends"! I'm taking it that secular doesn't mean what most people refer to as "secular stance": its not obvious to me how secular stance would affect someone's susceptibility to smoking-induced effects! (Are you meaning demographic trends?)
I'm tempted to include a joke or two about the medical effects of the local malts here. Local regions in Scotland are rather loyal to their local malt, but I assume that's not what you mean by "secular trend" either :-)
As an aside, its interesting to speculate that the Christmas silly season has its part to play in that only month after July in which the "after legislation" ACS level approaches those of the "before legislation" is December. Just looking at it by visual inspection, it seems to me that if you took this one month's results out, the before and after distinction would be much more striking.
Oh man can I attest to that. I'm grateful I don't have to clean it, but at my workplace there's a bar, and there's always butts outside, where there are several ashtrays just FEET away! And worse, when I come to work, the floor is dirty with disgusting phlegm that these people spit when they're smoking outside. I wonder if they feel it's OK to spit in the hallway, if they think they can do it inside too.
Cogerias, you've admitted to not even wanting to read the science on this. Well, how do you expect to have a well-informed, coherent opinion on the subject? Even if you're not an expert, the more you read science (valid or not), you'll start seeing which arguments make more sense. I mean, if you're comparing these medical studies with granite countertop cancer, you'll probably need a bit better judgment.
It seems to me that you're just reacting from the gut, because it bothers you that people are banning stuff. All those bans you listed earlier, each had a different cause. Not all of them are comparable to this, and that's obvious. Some of them are even silly, I'll admit but it doesn't mean nothing should be banned.
Something that never comes up around these discussions is this, which I will put in form of a parable (which is true too).
A friend of mine (let's call him, um, "Randy") was at the public city library, reading his email. Then, some older guy sat to Randy's right. After a while the guy leaned to his right, and let out the foulest, most horrific gas ever experienced by Randy, who was no virgin to foul gasses himself. Randy didn't know if to laugh or to pity the guy, he could have some horrible intestine-rotting disease (Randy doesn't know if those actually exist, by the way). So, he just left, baffled, angered and, after a while, just a bit amused.
So, why is this a story that you'd tell your friends? Because most of us have the freaking decency not to fart near other people or in closed spaces (or at least we try). The point is that if smokers had the decency not to smoke near others whom it clearly bothers (and more than a fart, actually) then bans wouldn't be necessary. And you don't even have to mention health risks. If I was a smoker I wouldn't smoke in closed spaces or near people even if it was perfectly legal, and I know smokers who do just that.
Just to be clear, I am not pushing for a fart (no pun intended) ban.
Why do we have to put up with other's people's addictions when they bother us? Many smokers somehow think they have the god-given right to annoy people and we should put up with it, even more than drunkards (they at least don't think straight when being annoying).
And no, many times we can't just walk away, as the employee example makes it clear.
Andyo do hate smoking or smokers? There is a difference.
Why would I hate smokers? Just because they smoke? I just think many of them should be more aware that it's not about them being inconvenienced. We have no right to impose to them to never smoke, but they also have no right to inconvenience others with their addiction and not being complained about. Even more so if it's a health risk, but I'm not touching that.
And if one has an addiction that inconveniences others, and thinks one has the right, nay, the moral high ground, to annoy them regardless, then he's just being a dick. Smokers are the ones who should go jumping around hoops, if anyone has to. I'd be all for not banning smoking if smokers who smoke in closed spaces with no regard for others were widely viewed as dicks. That's a good alternative, instead of most non-smokers being viewed as whiners.
So what you are saying Andyo is that you wouldn't have an objection to pubs that catered to 'Dicks' and pubs that cater to the 'Dickless'
If that is what you are saying, we are in complete agreement.
andyo, I am a business owner, not a pub. But if I were a pub owner I should be able to hire only smokers. I should be able to put a sign on the door say "smoking allowed in this establishment". You find that objectionable do not come in. I would have no problem with that. You have a choice. I do not like it government when takes away my rights as a business owner to cater to the clientèle of my choice. Where is the problem? You not have to visit my business if you did not want to and if you did you would know what to expect.
Should read "You do not" instead "You not". Damn I wish I had Orac'c writing skills.
Prior to the passage of the Scottish legislation, it was claimed that no economic fallout would result from the imposition of the ban. It was said that non-smokers, relieved of the need to breathe in the foul stench of secondhand smoke, would flock to the pubs to replace those smokers who might choose to do their drinking at home.
The question I'm asking, I suppose, is whether those non-smokers who didn't frequent pubs before the ban specifically to avoid contact with secondhand smoke would really experience any meaningful reduction in exposure while frequenting the local pub after the ban became effective?
If there was no significant reduction experienced by those non-smokers, how could anyone draw the conclusion the reduction in hospital admissions was the result of the ban?
Well, Dan, which other arbitrary criteria must your employees meet, which are not necessary for optimum job performance? Could I then hire only non-smokers for my business? I'm not exactly sure how hiring only smokers could not be discriminatory. How about if one of them decided to quit smoking? Would you fire him on the grounds of wanting to be healthier? On the grounds of quitting an addiction?
Matt, I think it's just a matter of people getting used to it. Here in California businesses like clubs and bars are doing just fine, people are already used to it. And anyway, this is tangential to the health issue. If making money is what worries you, it's another argument to be made, but it does not refute the issues raised here.
That is already being done. You want me to provide links?
My main point - which remains unaddressed - is that the investigators FAILED to establish the baseline in the trend of patients admitted to the 9 hospitals over time with elevated cardiac troponin levels. That is the ESSENTIAL information necessary to determine whether the 17% decline was attributable to the smoking ban or not.
Instead, the investigators used a completely different source of data to estimate the baseline trend - this database was NOT based on elevation of cardiac troponin levels.
So far, I haven't heard any adequate defense to this criticism of the study.
The fact that the 9 hospitals observed a 17% decline in acute coronary syndrome, but the national data show a much smaller decline suggests that in all other hospitals in Scotland, the number of ACS admissions actually increased substantially. Obviously, this is not consistent with the conclusion that the smoking ban resulted in a dramatic decline in ACS.
What makes no sense is that the authors fail to present the data on the decline in ACS nationally. That data is readily available -- the fact that they fail to present it raises serious questions about their objectivity.
Finally, the most obvious explanation -- which cannot be ruled out -- for the dramatic reduction in ACS in these 9 hospitals is the use of cardiac troponin itself. Because this assay is much more sensitive, it picks up many more cases of myocardial ischemia, leads to earlier treatment of patients, and reduces the incidence of acute coronary events.
In fact, this may well explain why a dramatic reduction in ACS was observed in these 9 hospitals but apparently not in the rest of the hospitals in Scotland.
This is a common limitation of epidemiologic studies - an intervention effect where the intervention itself causes a change in the outcome variable. I suspect that if the cardiac troponin test were initiated in 2002, there would have been a dramatic decline between 2002 and 2003 in the number of ACS admissions.
The point is - we simply don't know, and the study leaves open this plausible alternative explanation for the study results. Thus, it fails to provide sufficient evidence to support its conclusion.
andyo asks: "Well, Dan, which other arbitrary criteria must your employees meet, which are not necessary for optimum job performance?"
Actually andyo, being able to tolerate second hand smoke is a requirement for working in a place that caters to a smoking clientele. Also remember that the industry is called the hospitality industry, and quite frankly if your niche is smokers, having someone waving their hands about complaining of all those nasty smelly smokers, isn't exactly what I'd call being hospitable. As a result, I wouldn't call it arbitray, and would call it essential for optimum job performance in the stated case.
andyo also asks: "Could I then hire only non-smokers for my business?"
That would depend on your state. Many states have passed laws that prohibit lifestyle discrimination, at the objections of tobacco control. The WHO has set example to encourage other employers to follow suit, by refusing to hire smokers where legal. Bill Godshall of Smokefree PA, is a prominent lobbyist for discrimination directed at smoker, and the opposition of laws to prevent lifestyle discrimination. John Banzhaf, pimple on rump of the world, and founder of ASH is another major promoter of smoker discriminination.
It is very typical for the smoker controllers to point out discrimination when the suggestion is to hire only smoking staff to service smokers, except for a very few will publicly condemn hiring policies to exclude smokers. This is evident when CDC foundation board member Weyers made national headlines for firing off-duty nicotine users at his now smokeR-free company Weyco.
I believe the zealousness of the extreme in tobacco control have undermined the justifications for workplace smoking restrictions, and instead of enabling fair employment opportunity for all, have open the door to Dan's very suggestion, by their apparent motivation to punish those that would dare to smoke in their off duty, and privacy of their own home.
If one wants to find cranks, tobacco control is quickly evolving into a well organized one. Unfortunately healthy skepticism from within is sorely lacking, and Siegel who attempts to provide such insight in his own area of specialty draws Orac's criticism. Obviously there is no prohibitions on living in glass houses.
For the record, I do not believe smoking status should be a qualifier for employment. I further believe employers have a responsibility to provide a safe and healthy working environment for all employees.
Orac writes: "In retrospect, I actually agree that the effect is almost certainly not 17% and that that number shouldn't have been emphasized in the paper or in the press releases. ... The "true" effect was likely smaller."
Why would you cast doubt on the study results despite the fact that you defend the study methodology? Should a prospective study just be considered as a basis for subsequent guesswork?
"Even so, what if the effect of the smoking ban was only 10% Or only 5%? Or only 2-3% ...? That would still be a benefit"
More guesswork? What if the SHS attributable reduction was 0% (still close to your range of error)? Would that still be a benefit and would it justify lying to the public for the greater cause?
Steve Martin noted this a while back.
If it doesn't have to do with job performance, I think that's also wrong.
jpf,
I think there's a difference in hiring smokers and hiring people who can tolerate second-hand smoke. But I concede that I don't really know if I agree that it's a good thing to ask people to tolerate smoking if it's a proven health risk (which is actually the side I'm on). I won't debate this point, I'll leave it to more informed people like our host and others, but I see the need to cater to a smoking clientele. I am perfectly fine with having open spaces or ones with necessary ventilation as others have pointed out. That's the way it's done here in California and I have no complaints.
I also thing what irks people is having to change their routines. If the smoking ban were lifted here in CA probably a lot more people would be pissed about second-hand smoking than other places where there's no ban in the first place (and vice versa).
Tulse,
I swear I'd never heard that before. This library guy though did deserve his own airplane section. I mean, that's what my friend told me.
andyo states: "The point is that if smokers had the decency not to smoke near others whom it clearly bothers (and more than a fart, actually) then bans wouldn't be necessary."
It was my understanding, bans were mandated to protect workers from an unsafe working condition.
Patronage is a voluntary act, and as such those that are "clearly bothered" shouldn't be there.
The fact that smoking is a destructive behavior does not give one the "moral high ground" to sanctimoniously commandeer a proprietors property. The discretion should reside with the property owner, provided it doesn't involve placing employees involuntarily at a substantial risk. This is where smoking bans have entered the picture.
If I do not want people smoking on my property, I should not have to submit to "mob rule" and be forced to have to permit smoking. Likewise, I should not have to submit to "mob rule" and be forced to become the smoking police. I should however have to provide a safe working environment for my employees.
Likewise, using a smoking ban as a tool to coerce smokers into quitting is also a questionable action. Too frequently public health figures attempt to buy public support by giving a rationalization to the public of doing it for the smokers own good. If people need to be coerced into abstinence of a product, then the product should not be manufactured, sold or distributed. Too many young people assume the fact that the government permits cigarettes to be sold, and participates in the collection of revenues from it's sale, that cigarettes usage does not present a clear danger. If people want to smoke cigarettes, they should have to roll their own, and remove some of the convenience that elevated tobacco usage to the height of it's popularity. I believe failure to do so sends a schizophrenic message to young people.
Well said jpf on both posts. I have nothing to add.
Micheal,
Regards "which remains unaddressed": I wasn't trying to avoid your point, I was asking these questions because they underpin other questions I wanted to ask, but I needed to understand what you meant by these first. With that in mind, the lack of answer is frustrating, as is the implication that I'm perhaps "not addressing" something. (Nothing personal, but one of my pet frustrations about blogs is people not answering what is asked.)
The reason I asked about your referring to the "nine" hospitals as if they were a minority subset as to their being close to 2/3rds of the total, is that with such a large fraction of the total sampled, the results couldn't different substantially from the results for the total population unless the remainder (or subsets of the remainder) behaved in a very different fashion. If you don't mind me saying so, you seem to have gotten the reason I asked, and have somewhat shifted your stance (whilst still pointing in the same general direction).
You suggest the difference in the two populations is due to assays leading to a better treatment regime in those in the study. But these people were assessed prospectively. I am to understand that your argument then is that the better treatment to have affected re-admission later in the year by the same individuals?
Furthermore, if it is such a better diagnostic, wouldn't the diagnostic rates of those outside the survey would be (much) lower?
Andyo wrote: "If making money is what worries you, it's another argument to be made, but it does not refute the issues raised here".
My point was not about making money. It was simply about the amount of exposure sustained by those non-smokers who, prior to the ban, stayed out of pubs specifically to avoid that exposure. If no substantial decrease in exposure could be demonstrated, then any decrease in AMI in those cases could not be attributed to the ban, thus altering the overall numbers.
Furthermore, since there was a substantial decrease in the number of hospital admissions for smokers, wouldn't this support Michael Siegel's contention that a significant proportion of the overall reduction may have been the result of more timely and effective diagnosis of the problem.
The fact that the study failed to consider the two month period (April and May) immediately following the ban and failed to use all the data available does suggest a certain amount of cherry picking of data. There is also the fact that several studies in Helena, Pueblo, France, etc. have been sadly lacking in scientific integrity. The Scottish study does nothing to make those studies any more legitimate.
Until these concerns are addressed, the conclusion that the reduction in hospital admissions is the result of the smoking ban remains a matter of conjecture.
Heraclides stated: "The reason I asked about your referring to the "nine" hospitals as if they were a minority subset as to their being close to 2/3rds of the total, is that with such a large fraction of the total sampled, the results couldn't different substantially from the results for the total population unless the remainder (or subsets of the remainder) behaved in a very different fashion."
I think that is exactly it. These 9 hospitals are not a random sample of the population. They are a specifically biased sample because a special intervention was performed in these hospitals: namely, the use of cardiac troponin to diagnose unstable angina in patients who might otherwise not have been treated as aggressively. We know that troponin is a prognostic tool for high risk of heart attacks. So it is likely that identifying people at high risk and treating them would reduce subsequent hospital admissions. I believe that is a reasonable alternative explanation for the observed decline of 17%. It is also likely the reason why the rest of the country did not observe such a decline.
Wait a minute, here, Dr. Siegel.
In the U.S., the use of cardiac troponin levels is standard of care in diagnosing patients with ACS and unstable angina and has been for several years now. If a patient comes in with symptoms even somewhat suggestive of ACS, he'll almost certainly buy a series of cardiac troponin levels. Pretty much all suspected cases of ACS get it these days; the assay is generally available now except in the smallest of hospitals. It's a more sensitive and specific test, as I recall.
This brings up a question: What makes you think that the other hospitals in Scotland or hospitals in Britain not included in the study don't use troponin assays the same way that the chosen hospitals did? After all, that seems to be the entire crux of your argument, and you haven't shown your speculation to be true. Remember, both nations have a national health service, which tends to standardize care even more than in the U.S. I saw no indication in the paper that a differential in the use of troponin levels to diagnose ACS among hospitals that's large enough to have produced such huge fluctuations exists. You're going to have to explain that one. From the paper:
You say on your blog:
How do you know the diagnosis was "likely" made by other means from 1996 to 2006? That was probably mostly true in 1996-2000 (although I would point out that in the U.S. at least by 2000-2001 troponin levels were routine), but I highly doubt it was true from, say, 2001 to 2006. Indeed, this set of recommendations from the NHS seems to suggest that troponin testing in cases of acute chest pain was already widespread in Scotland and indeed contains a recommendation that "troponin testing should be available in all hospitals in Scotland that receive patients with heart-related chest pain."
The document went on to recommend the standardization of the use of troponin testing, stating:
So tell me: Given that troponin testing was already widespread by 2003, what makes you think that there were differences in its use large enough among hospitals to make such a difference? If you can't give good, concrete reasons to think that there was such a huge difference in the manner in which troponin testing was used in Scotland in the U.K., I find your critique unpersuasive, blowing smoke, so to speak.
Just to make sure I'm not misead: I wasn't stating that as a conclusion (not that anyone can conclude this anyway), but asking you if that's your line, which I take it that it is.
It seems to me that your proposal would only be true on lower RE-admissions occuring later in the year: from Orac's description, the primary sample is collected prior to treatment and measured using the assay. That is, as far as I can make out from the description given, it could only work if the same individuals did not re-present later in the year but they might have otherwise. I'd need to read the paper to confirm how they handled multiple instances of the same person in the data, etc.
It seems to me that this is important as multiple counts of the same individual could also manifest as apparently higher rates in the non-sampled group if not controlled for, too, which would be an alternative explanation rather than what you are presenting. This would have the difference in the examined and non-examined statistcs being sampling individuals once v. possibly several times (with the latter over counting" ACS cases in the non-sampled population if not corrected for the number of individuals v. number of cases). I'd like to the the authors excluded individuals from re-admission to the data, in which case this explanation might also resolve the difference in rates observed.
My time is extremely limited, so I doubt I'll find time to examine the paper closer, at least over the next 3-4 days.
Orac,
You're missing my point. It doesn't matter whether cardiac troponin was being used in the other hospitals or not. Just focus on the 9 hospitals. The point is that implementation of a system that involves the use of troponin testing for every potential cardiac patient in these 9 hospitals could itself be an explanation for the reduction in acute coronary syndrome admissions. It is certainly a plausible alternative explanation. Unless the study can show that the use of troponin testing had no effect on improving diagnosis and treatment of patients, then its conclusion that the 17% decline was attributable to the smoking ban is invalid. I don't see any evidence presented in their paper that improvement in diagnosis and treatment of heart disease (especially the identification of patients at high risk of MI) CANNOT be an explanation for the declining admission numbers they observed in their study.
I will be posting tomorrow morning a detailed commentary explaining my reasoning, so anyone interested in reading it is welcome to come over and have a look.
Actually, you're missing my point. The paper states that a standard definition of ACS was used based on troponin levels. Your point would only be valid if the definition of ACS used was somehow new or nonstandard, something you are only speculating about. My reading of the paper was that it was the standard diagnostic criteria being used; no mention that it changed or was different from before or other hospitals elsewhere in Scotland. You are simply assuming it was different. If it wasn't different, your criticism disappears.
Perhaps you should write a letter to the editor of the NEJM and express your concerns. It would be interesting to see how the authors respond.
Regards Orc's comment: "My reading of the paper was that it was the standard diagnostic criteria being used; no mention that it changed or was different from before or other hospitals elsewhere in Scotland." -- my reply assumed that your claim that there are different diagnostics was true (as I'd hope is clear anyway!) If that's not the case, it'd be nice to have it clarified.
Orac,
Think about it - If what you say is true and the cardiac troponin system and ACS definition had been in place for some time, then the research is really even more suspect. Why on earth would the researchers only examine the 10-month period prior to the smoking ban and the corresponding period in the following year? Why not go back many years and present data on trends in ACS admissions over many years prior to the smoking ban?
By assuming that the troponin system and standardized definition were not in place prior to 10-months prior to the smoking ban, I'm actually giving the authors the great benefit of the doubt. Because if it was in place prior to 10 months before the smoking ban, then it is inexcusable that they would exclude all of that data which would have been critical to establish a baseline trend in ACS admissions.
I've been assuming that the reason they couldn't go back more than 10 months is that this system wasn't in place and there was not a standardized definition based on the detection of troponin. If a standardized system was in place based on this definition, then they are really guilty of conducting crappy research, rather than just a poor conclusion.
Perhaps someone could be so kind as to explain how it is known that some other event has not happened (a new treatment, or drug for those 9 hospitals that was introduced prior to the ban) that has contributed to the reduction seen in both the smoking and non-smoking groups? Or is this just wishful thinking on the authors part that the reduction is due to the passing of legislation?
To me this sounds a bit like what the anti-mercury warriors would cook up. Obviously I'm missing a point here, as it seems to be a given, that this can be attributed solely to the smoking ban.
A key strength of this study is that it is a prospective design. study design prohibits going back before the study began, which would be a retrospective study. Moreover, that weren't merely testing for troponin, they were determining cotinine levels in the same patient population, which certainly is not a routine assay, so this could not have been done historically. But it seems rather unlikely that determining cotinine levels would somehow protect people from presenting again with acute coronary syndrome. Still, contrary to your objection that they have no baseline, they do have 10 months of baseline data preceding the ban, which shows no decrease. So if some change in diagnostic practice caused the decrease, isn't it a remarkable coincidence that the effect only happened to "kick in" at the time of the smoking ban?
So American lawmakers should be told that this new Scottish heart attack study is consistent with no life risk from secondhand smoke exposure in bars and restaurants.
I'll happily confess that if it means a smoke-free workplace (and other venues) I'll support any research - crap or not.
I've always likened smoking to urination (except that the latter is a biological necessity while the former is simply a childhood habit). When someone asks "why shouldn't I be allowed to smoke in a restaurant?" I ask why we shouldn't be allowed to urinate in one? I've never received a sensible reply.
Other analogies include playing your own loud music and having sex - neither of which is usually acceptable in the average restaurant - but why?
trrll suggests that the baseline was limited to 10 months because of the cotinine assays, and therefore we can't establish a longer and stronger baseline. It seems to me, that since it's safe to say the cotinine assays didn't influence the diagnosis, the data prior would be valid so long as the other assays were documented and available as standard practice.
So what exactly was the cotinine level of the people who weren't hospitalized? This would be the group from where the reduction comes from, and not the group that was hospitalized, correct? It seems to me that it's nothing but pure speculation that the reduction is due totally to the ban, as there is no link between the ACS "spared" population and their cotinine level. The best you can do is only assume, and not very scientific.
What purpose did measuring cotinine levels accomplish if we didn't measure cotinine levels in those not being hospitalized, except to prevent the study from going back further and taking a larger look at the downward trend?
I think Siegel makes a valid point in stating the study goes out of it's way to fail to establish a reasonable baseline.
One clue that there are other factors may be involved is the 14% reduction in ACS cases from the smoking subset. This also just happens to be the one group that would still be subjected to significant levels of smoke pollution, and the least sensitive to SHS. I believe you will be hard pressed to document an addition 14% quit rate for the study period to justify this reduction, but the data should be available if not now, soon. Anybody want to take their credibility out on this limb?
Damn. This discussion has got interesting. Will have to check back tomorrow. I see Orac and Dr. Siegel talking. That is what I was hoping for. Unfortunately I do not time right now to read it all. Hopefully, tomorrow. Goo Dose (that problably not how it is spelled,)to both Orac and Dr. Siegel for an interesting exchange on the topic of science of second had smoke.
Orac has called James Enstrom's big ETS cohort study that found no link between secondhand smoke exposure and lung cancer or heart disease "crap".
Yet I think Enstrom has pretty ably defended his work against the accusations of Stanton Glantz and the ACS:
http://www.epi-perspectives.com/content/4/1/11
Two other studies using American Cancer Society data have come to the same negative result as did Enstrom and Kabat.
http://legacy.library.ucsf.edu/tid/jvh19a00/pdf
http://legacy.library.ucsf.edu/tid/sqg22d00/pdf
Actually, reading all this, I urge Dr. Siegel to write a letter to the NEJM outlining his criticism. Blogs don't matter very much when it comes to this discussion; the peer-reviewed scientific literature does.
Nice straw man. Not even Dr. Siegel has said that. His argument, as I understand it, is that there may be an alternative hypothesis that explains the results as well, not that there wasn't an effect due to the smoking ban; indeed, he said there "almost certainly" was some effect due to the ban; he's just saying he doesn't think the authors can conclude it was 17% from their data. From the man himself:
His question is thus what is the "true" effect or whether the effect can be entirely explained by something else. He didn't say that this study is consistent with no life risk from SHS, only that it might be.
Do learn to read.
Orac is right about my argument - based on the known effects of secondhand smoke on the cardiovascular system, there is little question that a smoking ban is going to have some beneficial effect - over time - in terms of reducing coronary events. My argument here is purely a scientific one: I don't believe this particular study demonstrates that the observed 17% decline was due to the smoking ban.
trrll commented - correctly I believe - that because the study involved the measurement of cotinine, the full study results could only be presented for the 10 months prior to the smoking ban. However, that didn't preclude that authors from presenting data on trends in acute coronary syndrome at these hospitals in previous years. The question is: why didn't they? Anyone who truly wants to determine whether a drop in events observed over a year is a real effect of the smoking ban or not would go back as many years as possible to establish a solid baseline before drawing any conclusion. Why didn't the authors do that?
I have given them the benefit of the doubt by assuming that the reason they only went back 10 months is that they had no choice - the ordering of a troponin test on every cardiac patient was only in place in all 9 hospitals during the study period. Orac suggested that I'm probably wrong in that assumption, which seems reasonable. But remember, if that's true, then there is no rationale for excluding the prior data from the analysis.
trrll asked: "So if some change in diagnostic practice caused the decrease, isn't it a remarkable coincidence that the effect only happened to "kick in" at the time of the smoking ban?"
The problem, trrll, is that we don't know when the effect kicked in. Did it kick in during the ten months prior to the smoking ban, or had the effect already kicked in prior to that? There's no way of knowing that without examining trends prior to the 10-month period before the smoking ban.
Orac is right about my argument - based on the known effects of secondhand smoke on the cardiovascular system, there is little question that a smoking ban is going to have some beneficial effect - over time - in terms of reducing coronary events. My argument here is purely a scientific one: I don't believe this particular study demonstrates that the observed 17% decline was due to the smoking ban. The study really has nothing to do with whether there is a lifetime risk from secondhand smoke. We already know that there is.
trrll commented - correctly I believe - that because the study involved the measurement of cotinine, the full study results could only be presented for the 10 months prior to the smoking ban. However, that didn't preclude that authors from presenting data on trends in acute coronary syndrome at these hospitals in previous years. The question is: why didn't they? Anyone who truly wants to determine whether a drop in events observed over a year is a real effect of the smoking ban or not would go back as many years as possible to establish a solid baseline before drawing any conclusion. Why didn't the authors do that?
I have given them the benefit of the doubt by assuming that the reason they only went back 10 months is that they had no choice - the ordering of a troponin test on every cardiac patient was only in place in all 9 hospitals during the study period. Orac suggested that I'm probably wrong in that assumption, which seems reasonable. But remember, if that's true, then there is no rationale for excluding the prior data from the analysis.
trrll asked: "So if some change in diagnostic practice caused the decrease, isn't it a remarkable coincidence that the effect only happened to "kick in" at the time of the smoking ban?"
The problem, trrll, is that we don't know when the effect kicked in. Did it kick in during the ten months prior to the smoking ban, or had the effect already kicked in prior to that? There's no way of knowing that without examining trends prior to the 10-month period before the smoking ban.
Orac is right about my argument - based on the known effects of secondhand smoke on the cardiovascular system, there is little question that a smoking ban is going to have some beneficial effect - over time - in terms of reducing coronary events. My argument here is purely a scientific one: I don't believe this particular study demonstrates that the observed 17% decline was due to the smoking ban. The study really has nothing to do with whether there is a lifetime risk from secondhand smoke. We already know that there is.
trrll commented - correctly I believe - that because the study involved the measurement of cotinine, the full study results could only be presented for the 10 months prior to the smoking ban. However, that didn't preclude that authors from presenting data on trends in acute coronary syndrome at these hospitals in previous years. The question is: why didn't they? Anyone who truly wants to determine whether a drop in events observed over a year is a real effect of the smoking ban or not would go back as many years as possible to establish a solid baseline before drawing any conclusion. Why didn't the authors do that?
I have given them the benefit of the doubt by assuming that the reason they only went back 10 months is that they had no choice - the ordering of a troponin test on every cardiac patient was only in place in all 9 hospitals during the study period. Orac suggested that I'm probably wrong in that assumption, which seems reasonable. But remember, if that's true, then there is no rationale for excluding the prior data from the analysis.
trrll asked: "So if some change in diagnostic practice caused the decrease, isn't it a remarkable coincidence that the effect only happened to "kick in" at the time of the smoking ban?"
The problem, trrll, is that we don't know when the effect kicked in. Did it kick in during the ten months prior to the smoking ban, or had the effect already kicked in prior to that? There's no way of knowing that without examining trends prior to the 10-month period before the smoking ban.
Orac - Thanks for letting me comment here and be a part of this intriguing discussion.
Matt,
yes, I do go out more and I do enjoy going out more because I don't have to put up with second hand smoke.
Data on coronary heart disease / acute myocardial infarction in Scotland for the years 1997 - 2006 is available here.
Re: Harping on the investigators not using earlier data.
Dr. Siegel,
Once again, as has been pointed out many times, this was a prospective study. There was no time to get the study together as a prospective study and cover more than 10 months before the ban.
The reason is that the Scottish smoking ban was introduced into Parliament on December 17, 2004 and finally passed on June 30, 2005. It took effect March 26, 2006. That's less than 10 months from when the bill containing the start date was passed. Tell us: How would one do a prospective study that covers more than 10 months on either side of the ban? It can't be done and still be a prospective study. Going to a retrospective design, as you well know, would introduce all sorts of other confounders that would be very hard to control for.
Also, if you keep in mind that the ban wasn't passed until June 2005, it makes perfect sense to look at the period from June 2005 to March 2006 because that's all the time there was between final passage of the law and its implementation. To keep it even, one then looks at a ten month period after passage of the law that includes the same months of the year, to account as best as possible for seasonal variations in ACS admissions. There's nothing sinister or dubious about it. It's the best prospective design that could be produced in such a short period of time.
Most likely all that happened is that the investigators started planning the study when the bill was introduced (or maybe slightly before if it was thought that the bill had a high chance of passing) and finalized the time periods to be studied prospectively once the start date of the legislation was known. Once those time periods were set, they couldn't go back and look at longer time periods. It's really that simple. Again, there's nothing tricky or sinister there. The odd ten month period is nothing more than a consequence of doing a prospective study and not having more time between the passage of the ban and the date it took effect. I really don't see why you keep harping on this. It's incredibly obvious. Indeed, your constant asking "Why didn't they look at a longer and earlier time periods?" is starting to strike me as a bit disingenuous.
Perhaps you could educate me, though, given that I'm not an epidemiologist. How could the investigators have designed a prospective trial that covered more than the ten months that they covered?
Until my parents finally gave up the ciggie butts, I hated going to their house because the smoke residue was in everything: Furniture, wall paint, upholstery, you name it. Their car reeked of it. I remember steam-cleaning a living-room furniture set they'd given us and watching as gobs and gobs of brown tarry crap filled the bucket.
RE: PZ --
Actually, considering that he was responding to the news that the life of Webster Cook, a student at a state-run (and therefore legally secular) university (University of Central Florida) is in the process of being ruined because he took a communion cracker out of a Catholic church facility that is illegally operating on university grounds (remember the signal importance the Founders placed on separation of church and state?), I can understand exactly why he did what he did.
Webster Cook is probably going to get kicked out of his school, at great expense to himself and his family. He will have to travel far and wide to find a school that will take him, because the people responsible for hounding him out of UCF will attack any school that dares consider him. His career path is shot and he will be lucky to be employed as a janitor. All because he took a wafer out of a church that was illegally set up on school grounds.
Yeah, I can see why that ticked PZ off.
I know the background story quite well. My comment stands. The way he decided to stand up for Webster Cook is what I have a serious problem with and what lowered my opinion of PZ a lot.
If it completely "kicked in prior to the 10 month period before the smoking ban" then there shouldn't have been any difference between the 10 month period before the ban and the 10 month period after, which is contrary to the result. If it was somehow in the process of "kicking in" then there should have been some indication of a decline in the 10 month period before the ban, yet there is not. So you are arguing that some change in diagnosis or treatment (for which you have no actual evidence) fortuitously "kicked in" simultaneously with the implementation of the smoking ban.
By measuring cotinine levels, the investigators had an independent, objective check upon subjects' self-characterization as smokers or nonsmokers. They also had a way of determining whether the smoking ban actually resulted in an appreciably decreased exposure to smoke.
Exactly. Not only was it a prospective study limited by the time frame between passage of the law and its start date, but all of Dr. Siegel's complaints are starting to seem more and more like obfuscating to me. I really hate to say it, but his harping on this same point over and over again when it's obvious to me why the study must have been designed the way it is makes me wonder about him.
But, hey, I'm just a surgeon, scientist, and a blogger. I'm not an epidemiologist. That's why I urge him to write a letter to the NEJM and outline his problems with the study. It would be really interesting to see the authors' response.
trrll remarks "They also had a way [cotinine testing] of determining whether the smoking ban actually resulted in an appreciably decreased exposure to smoke."
But ONLY for those having a heart attack or cardiac event.
Thanks, but you haven't addressed how this applies to demonstrating that those NOT having a heart attack were any more or less exposed. This is the hypothesis isn't it? Less smoke = less heart attacks?
Since cotinine testing doesn't alter the diagnosis. And since we are looking at a discrete point in time, without a means to linking the two periods otherwise, we can use historic information to determine if there is a change in the trend. While we wouldn't have a detailed breakdown by authenticated group status, we would gain the insight of the general trend.
By failing to consider this, the public might be mislead into thinking this is the result of the smoking ban, and not part of the prevailing general trend.
Clearly the bulk of such improvements are consistent with previous years, and are independent of the ban. A simple before/after comparison failing to consider such abandons good science and delves into the realms of activism and propaganda.
Based on my casual observations of the data presented, it appears there was an improvement above the general trend between the two time periods. However, it appears the author is attempting to overstate the case. Instead of making a strong case, taking such liberties draws only suspicion, and violates the publics trust.
By looking at the previous trends, and the smoking group which would have minimally been effected, I'd say the difference between these two years is in the 3% - 8% range, how much of that can be attributed to the smoking ban? Good question.
Let me see if I can put this simpler.
Hypothetically speaking.
I know the general trend for cardiac disease has show steady improvement over the last 10 years.
I design a prospective study to look at events. I can reasonably expect the trend to continue.
Since this is a prospective study, we avoid looking back at the previous trend, and simply compare this year to the next.
I publish my results and attribute them to some event.
Unless the trend reverses, it will appear to confirm that the event had some impact, regardless of outcome.
Would this be sound science, and what exactly have we really learned?
So what is your hypothesis? That the law somehow resulted in reduced exposure to smoke only for people destined to have a cardiac event?
We have 10 months of data to look for a trend prior to the law going into force. So if there is a steep ongoing trend toward a reduction in coronary syndrome, it should be evident in that time frame. If the trend is too shallow to be evident over 10 months, then it can hardly explain the magnitude of the decrease observed.
As cited in the paper, the 10-year trend in Scotland prior to the beginning of the study was 3% per year. The reduction in England for the same period of time as the study was 4%. That hardly seems to account for the "bulk" of a 17% decrease. So what is your hypothesis? That the trend in the population served by these particular hospitals was actually much steeper than for the country as a whole, yet somehow this failed to show up in the 10 months of data prior to implementation of the law?
Even Dr. Siegel seems to think that the data indicate a decrease due to the law, he just thinks that the study may somewhat overestimate the magnitude of the effect.
After reading Senator Thomas Bliley's testimony about the Oversight Committee's investigation into the EPA, and especially the antics of James Repace, I have to ask if there is any appointed watchdog who makes sure Dr. Pell and her colleagues do all they say they are doing? I know Dr. Siegel has opined that he wished the Big Tobacco watchdog was still around to keep the antismoking movement honest.
http://www.pipes.org/Articles/Bliley.html
I nailed Repace myself for inflating the likely death toll at Pennsylvania casinos. But what else goes on that nobody catches?
http://keepstlouisfree.blogspot.com/search?q=Repace+Pennsylvania+Casinos
Apparently in the absence of a big study by an antismoking activist such as Glantz or Pell, the heart attack rate fails to drop.
http://www.velvetgloveironfist.com/index.php?page_id=59
http://www.acsh.org/factsfears/newsid.990/news_detail.asp
trrll -- This whole paper seems to be clear as mud but...
You say "As cited in the paper, the 10-year trend in Scotland prior to the beginning of the study was 3% per year. The reduction in England for the same period of time as the study was 4%."
Am I correct that this is based on national stats about AMI? If so, what is your explanation for the 17% decrease found in this study, versus the 8% or so decrease found based on continuation of the conventional method(s) of diagnosis of AMI -- as reported in the continuation of the national stats for Scotland? Is it that AMI is different from ACS? If so, then the 10 year baseline trend is irrelevent?
How can we know that if the tropinine measurement had been consistently used in the previous year (in the 9 hospitals)we would not have seen a large drop in that year as well? If we can't know that -- would it not be better to simply stick with the conventional (national) stats which as you point out DO have a 10 year baseline?
I, of course, meant "troponin"
My hypothesis is you can't derive the exposure level of the general population from only measuring those that had heart attacks, as this is not a randomized sample representative of improved outcome population.
By the demographics we know the largest improvements were those least likely to frequent areas covered by the ban.
Going to more detailed demographics from all Scotland we learn that the greatest improvements in AMI's and CHD Incidents were the +75 crowd. See link above for the ISDSCOTLAND 10 year datasets.
I am not disputing that it's likely the ban made some positive difference, however, I believe the data has been highly overstated.
As for the bulk. Using national statistics the long term average is more in line with 2006 showing some improvement, and this holds true looking across several aspects. AMI's and the wider diagnosis group.
And the question remains why is there such disparity from national statistics to the 9 hospitals that are suppose to represent the bulk?
I would expect the data to be much closer. However there does seem to be something unusual in the older age groups, which is contradictory with expectations concerning the bans.
The proper thing to do would be to derive a trending from the same hospitals going back 10 years, because without it, the study is not of much value.
No. From the paper:
"The Scottish Morbidity Record 01 routinely collects information on all hospital admissions, including the date and type of admission and disease codes.15 We used these data to determine trends in admissions for acute coronary syndrome (International Classification of Diseases, 10th revision [ICD-10], code I21) in the preceding 10 years. "
Yes, ACS is a broader diagnosis than AMI, and might reasonably be imagined to be a more sensitive indicator of cardiac problems. Since the 10 year trend is for ACS, not AMI, it is relevant.
The same methodology was used in both 10-month periods. You seem to be grasping at straws here.
Calling an assertion a hypothesis does not make it one. If you want to make the case that a reduction in the cotinine levels of people being admitted for ACS does not reflect lower exposure to smoke exposure among the population vulnerable to ACS (which is the population that we are interested in), you need to offer at least some kind of reasonable hypothesis as to why only those admitted would show lower cotinine. "This is not a randomized sample" is only a rational objection if you can provide some rational reason why the lack of randomization would affect the study's conclusions.
I haven't seen national statistics for ACS over the same period. Have you? Can you provide them?
I would tend to expect the older age groups to be more likely to have underlying heart problems that would increase their vulnerability to a toxic insult, and thus to experience the greatest benefit from the ban. What would you expect?
I suspect that jpf would expect that the 75+ crowd would be less likely to be hangin' in smoky bars.
Ya know trrll -- I have no need to grasp at anything because I have no dog in this fight. And I find the grasping at straws comment not necessary. What I was trying to do was to explore the reasons for, and the implications of, switching from a previous standard to a different system of measurement (which Siegel argues may have had the effect of an intervention as well) in mid-stream, so to speak.
"We used these data to determine trends in admissions for acute coronary syndrome (International Classification of Diseases, 10th revision [ICD-10], code I21) in the preceding 10 years. "
ICD 10 I21 and I22 seem to be AMI... No?
"The same methodology was used in both 10-month periods"
I understand that, I was referring to the baseline trend vs. the study time-frame.
I think Dunc provided the link to the stats way upthread (@11:31)
clear as mud...
The following url will take you to the ISDSCOTLAND.ORG data source.
http://www.isdscotland.org/isd/information-and-statistics.jsp?pContentI…;
The closest datasource to that of the study appears to be the table AC1.
Selecting Diagnosis Angina; Type of Admission ALL; Health Board Scotland; and age group all
You will find the standardized Rate per 100,000 population for Apr 1 1997 through Mar 2007. This will be all hospital discharges with a diagnosis I20, I249
Next you can select Acute Myocardial Infarction for IDC-10 codes I21,I22
From this you can derive a year over change for which I get the following results Beginning with the ending year 1999:
-0.60% -8.55% -2.56% -2.11% -6.14% -5.82% -8.13% -5.04% -9.80%
As you can see, the year following the ban, the reduction from the year over is -9.8% The year before the ban was -8.13% and the year ending at the start of the ban was -5.04%. This is fall ALL Scotland.
I see a slight difference after the ban, but I don't find a 17% improvement that I can attribute to the ban.
Now please explain to me how measuring for cotinine changes these findings, and should invalidate the knowledge we gain by looking at the longer term trends.
And the next question is it appears the author used only I21-22 for her 3% while ignoring I20, and comparing that against a 17% which appears to be I20-I22.
The average change for the years prior to the ban were -2.9% AMI, -6.84% Angina and -4.87% combined.
For those that don't have excel, here is the standardized rates:
Diagnosis1998199920002001200220032004200520062007
I21-I22324.7319.9283.2283.0289.3285.3277.0268.1254.9236.3
I20358.0358.7337.4321.7302.7270.3246.3212.6201.6175.4
I20-I22 682.67 678.59 620.59 604.69 591.96 555.61 523.25 480.72 456.47 411.72
And the difference by year:
I21-I22-1.48%-11.47%-0.07%2.23%-1.37%-2.92%-3.20%-4.95%-7.27%
I200.20%-5.94%-4.65%-5.92%-10.70%-8.89%-13.68%-5.16%-13.00%
I20-I22-0.60%-8.55%-2.56%-2.11%-6.14%-5.82%-8.13%-5.04%-9.80%
I'm sorry if they are hard to read.
I believe you missed the point trrll. It is my contention that the cotinine measurements are only good for determining the accuracy of the self reporting of the cardiac cases in terms of their smoking status. Knowing that those that had a heart attack have a higher or lower cotinine result is inconsequential to the diagnosis. It also does not reflect the status of those not having a heart attack.
The flaw I see with this study is not the fact they did or didn't measure cotinine levels, but rather failed to consider the history and comparing only one year to the next, rather then consider the long term trend. As you have pointed out, this appears to be because of the desire to measure cotinine levels which wouldn't be possible for prior years.
While the defenders of this study seem to promote a prospective study as being positive. It appears to me to be a severe handicap, as a whole new picture develops once you look at the trend. One has to ask is the loss of the historical trend more important then losing a non-critical piece of information on the older data. In other words, the lack of cotinine doesn't change the diagnosis on the older data, AND having it only for the most recent data doesn't invalidate the findings in terms of which groups showed the most improvement.
GDF hit on it, and I could go into further discussion about why I believe the cotinine levels of the group at risk that wasn't hospitalized levels may be higher after the ban, but it's not important and only distracts from the point that not looking back at historical data is the flaw, and the reason for not looking back (lack of cotinine data) is inconsequential, and paints an incomplete picture because of the loss of trend information.
I'd be interested to know how big a drop would you expect in the short term from enacting a smoking ban in terms of reduction over the prevailing trend? I believe Orac used the word amazing to describe the results. I think miraculous is more appropriate.
The first study of this nature was released to the press as being a 60% drop in AMI's.
However the AHA claim that by the end of the first year, a smokers risk only improves by 50% of not having a heart attack. I would expect the effect on exposed non-smokers to be realistically even less. So for a community that enacted a ban, where everyone smoked, and they quit, the best based on the AHA risk factors would be is a 50% improvement. Somewhere before it got published someone must have realized this, and scaled it back a little to 40%. If looking for "Woo", you'd be in paydirt.
Here is a copy of the press release which was issued prior to publication. http://www.eurekalert.org/pub_releases/2003-04/uoc--sps032603.php and http://pub.ucsf.edu/today/cache/news/200304012.html
While I support workplace smoking bans, I do not believe the definition of "Woo" should depend on which side of an issue you stand on. The rules should be the same, and studies like these can undermine the publics trust.
I emailed the lead author to see if they just used admissions for I21, or if they were including unstable angina, which would be under I20. The were in fact using only I21 as stated, so the comparison trends listed in the paper are probably better characterized as AMI rather than ACS.
Thanks trrll for making that effort -- I hope you now see the point we were trying to make. To compare this 17% drop with the previous drops is indeed a case of apples and oranges. Especially when you have possible differences in standard procedures at these 9 study centers. (The troponin measurement and whatever else may have influenced the centers that were involved in the study). OTOH, to simply do pre-post at those centers and leave out the trend data leaves a quite a a hole in the analysis.
It's just a very tricky bit of work -- when comparison of the national stats for the smoking ban period for the entire country with the existing national trend data would have been a quite simple analysis. It certainly leads one to at least consider if politics or ideology is driving this bus.
But most importantly, as jpf points out, it's vital for all of us who want to call ourselves scientists to be able to recognize and acknowledge flaws (and dare we say biases?) in studies -- whatever side of a political issue they may support.
And then -- we have what appears to be the Wales anti-miracle... But it's turned into a miracle nonetheless. Nothing much to say about that. Just FYI.
http://www.reason.com/blog/show/127939.html
It also provides evidence that the law has been effective in reducing smoke exposure for at least some segments of the population. One does not need a randomized sample for this, although one would if one wanted to quantitate the average reduction in smoke exposure over the entire population.
You are missing a key strength of a prospective study, which is that there is no way that the investigators could be influenced by the data to cherry-pick a particular time interval or location where the evidence favors their hypothesis. For this reason, scientists give considerably greater weight to prospective studies.
So are you trying to argue that exposure to secondhand smoke is protective?
That depends upon your hypothesis as to how the toxic effect works, and I cannot see any basis for forming an expectation one way or another. One plausible toxicity model is that there is a highly potent, but rapidly-eliminated, substituent of smoke that produces an acute elevation in the risk of cardiac ischemia. In this model, the cardiac effect of the toxin "maxxes out" at a low level of smoke exposure, so smoking a cigarette is no worse (for this particular effect) than being in the same room as a smoker for the same duration of time. In this model, the critical variable for determining risk is not whether one is directly smoking or not, but simply the total amount of time one spends exposed to smoke. So even a smoker would be expected to benefit from the ban even if their cigarette consumption is unchanged, because the total amount of time spent exposed to smoke would be less (unless they are such a heavy smoker that they smoke continuously).
I'm a little disappointed trrll, that you didn't bother to extend an estimate of what you though one should expect as results from a smoking ban.
One key that should have lead you to the conclusion that there is something wrong with this study is only 43% of the never smokers reported being exposed prior to the ban, so at best the never smoker category could show due to the ban is a 43% improvement, and this is if smoke exposure caused every one of those heart attacks. We know 22% are still reporting exposures, so that means the MOST it could drop would be 21% percent, and that is if every heart attack in never smokers was caused by smoke exposure. And what was the drop? 21% for all never smokers? So you want us to believe that every heart attack in the exposed never smokers was caused by exposure to second hand smoke?
Sorry, I have a more plausable senario; Other factors dominate the drop observed one year over the next, as we have seen from prior trend.
I'll respond to your previous post later this weekend when I have a bit more time.
I've noticed over the years that one of the real giveaways that somebody is trying to twist data to support a preconceived notion the practice of taking an obviously error-prone value as an exact number and trying to do precise calculations with it.
The first question that comes to mind is, "Just how accurate is people's self-reporting of smoke exposure?" To what extent is this number determined by actual exposure, and to what extent is it a measure of individual awareness and recollection of smoke exposure? One of the strengths of this study is that there is an independent measure of smoke exposure from blood cotinine levels. And it is notable that the correlation coefficient with self-reports, while statistically significant, is not terribly high, about 0.33, telling us that people really are not all that good at estimating their own exposure to smoke. So taking self-reports as an exact measure of how much smoke exposure people are experiencing is ridiculous.
Perhaps you've put a litle too much faith into the cotinine assays, and failed to consider their limitations. Perhaps it never occurred to you that one reason for the poor correlation between cotinine and reported self values is likely due to the fact that only 17% and 13% (before/after) of never smokers reported a weekly exposure of >5 hours. Considering cotinine has a fairly short half-life of 20 hours doesn't make for a good infrequent or occasional exposure monitor. Clearly these people do not believe they experience much exposure what-so-ever.
Even if we discount self-reported and only consider cotinine assays, the mean cotinine level decreased by 21% and that corresponds to a 21% reduction in never smokers which again implies that of cotinine values went to zero, so would all the heart attacks. Clearly the percentage reduction in heart attacks seems inflated for the measured amount of reduction.
The true strength of this study if it actually had any would have been confirmed had the results been close to the NHS data which demonstrated a comparable result. However it appears to some type of anomaly between the national average, and these nine hospitals. And this is reflected in AMI's, Angina, and Deaths, as there was not a major deviation in these indicators outside of the prevailing downtrend following the ban, except in this study.
Even looking emperically at the best case senarios the risks factors for exposure appear to eclipes the risk factors for smoking.
This was a study in wishful thinking.
Who needs to cherry pick in a major downtrend? When the preceeding years for I20-I22 is -6% -6% -8% -5% and the year following the ban is -10%, if you only compare one year to the next and ignore the rest it looks like a 10% decline caused by the ban, but clearly the average for the preceeding years was 6%. This is a major flaw. Going back even further, we see another large decline followed by a one or two years of smaller decline. Perhaps there is a small contribution, but you're living in a fantasy world believing there was a real 17% decline attributable to this law.