Wednesday 2 September 2009

1+1=11: James Repace thinks of a number (part 1)

The claim that "tornado-level ventilation" is required to make a building "safe" from secondhand smoke was first made by James Repace, a prominent and fanatical anti-smoking activist since the 1970s. In those days, Repace - who has a degree in physics - was working at the electronics division of Washington's Naval Research Laboratory. By the end of the decade he had written his debut study into secondhand smoke, found a job at the Environmental Protection Agency (1) and soon became a well-known and well-paid anti-smoking spokesman and researcher. In 1998 he set up his own company - Repace Associates, Inc - and became a professional 'secondhand smoke consultant'.

His published papers on the subject of tobacco bear all the hallmarks of the theoretical physicist, with the emphasis heavily on the theoretical. In 2005, he made an effort to convince lawmakers that high tech ventilation systems were not the solution to what he called the "mortal hazard" of secondhand smoke and wrote a paper which concluded that the amount of ventilation required to make a room safe from smokers would be the equivalent of a "veritable indoor tornado." For those of us who have followed the rise of junk science in recent years, this quote has become something of a classic of the oeuvre, but the study from which it came is less well known.

Despite being intuitively ridiculous, the tornado claim has been repeated around the world. In addition to appearing six times on Repace's own website, a quick Google search reveals that it is currently being reported as fact by, amongst many others, ASH, Unison, Cancer Research UK, Americans for Nonsmokers Rights, the Clean Air Coalition, Smokefree Ohio, Stop Smoking Manchester, Smokefree Europe, the World Health Organisation and GASP (although the last three groups have settled for a mere 'hurricane').

The article in question was published in the house journal of the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) in 2005 (2). The publication may be obscure but Repace's decision to submit his paper to it was no accident since ASHRAE issue ventilation guidelines to the building industry. The article is unusually dense with technical jargon which, perhaps deliberately, has the effect of distracting and confusing the casual reader. In its closing paragraph Repace finally gets to the point. He makes two crucial assumptions:

1. The annual US death rate due to secondhand smoke in bars is 15 per 100,000 workers exposed.

2. In a nonsmoking bar, the desirable rate of ventilation is 18 air changes per hour*

Both of these statements are highly contentious but it is not necessary to debunk them individually to show that the tornado claim is nonsense and so, for the purposes of this article, we shall hold our nose and go along with them. Repace - in his roundabout and over-complicated way - multiplies the figure of 18 (the air changes) by 6,750 (which we will come to shortly) to show that a bar that allows smoking requires a whopping 121,500 air changes per hour in order to be as 'safe' as a nonsmoking bar. This equates to 33 air changes per second - a 'veritable tornado' indeed!

The underlying premise for all this is, as Repace wrote, that "since risk is inversely proportional to ventilation rate, the ventilation rate would have to be increased by the ratio of the number of estimated deaths." Doubtless there is some truth in the first part of this sentence. Ventilation surely does reduce risk from air-bound contaminants; this is why it is used in workplaces and scientific laboratories where dangerous chemicals are in the air. This was precisely the message that the hospitality industry, ventilation industry and tobacco industry had been trying to get through to policy makers for years. It was, alas, quite the reverse of the message that Repace and the pro-ban campaigners wanted to send and his tornado calculation hinges on the second part of that sentence ("the ventilation rate would have to be increased by the ratio of the number of estimated deaths.")

But why? What is the correlation? Why on earth should it be assumed that the ventilation rate needs to accelerated by the number of supposed deaths? It is rather like saying that speeding kills 1,000 people a year and therefore the speed limit must be reduced by a thousand to make the roads safe. A speed limit of 0.03mph would certainly make the roads safe but no one would propose such a limit because it is insane to take the correlation between speeding and road accidents quite so literally. In the case of Repace's equation, the very fact that the resulting figure amounts to no fewer than 33 air changes per second would have been enough to make most researchers take a step back and return to the proverbial drawing board. But James Repace is not most researchers and he not only stuck by his conclusion but added, presumably with a straight face, that "even greater airflow rates would apply for air cleaning, which inefficiently removes secondhand smoke gases."

The key number in this calculation is 6,750 (see footnotes for the rest**). Without multiplying 18 by such a high figure, Repace could not have arrived at his terrifying conclusion, so where does it come from? As it turns out, the figure of 6,750 is based on Repace's belief that 15 deaths per 100,000 (per year) are attributable to secondhand smoke amongst exposed workers. He then inflated this number by a factor of ten to arrive at 150 deaths per million and then multiplies that by 45 to represent the 45 years of working life and - voila! - 6,750 deaths. Leaving aside the fact that these estimates are themselves based on highly questionable data, the way in which he arrives at the 6,750 figure reveals that it is completely arbitrary.

It is not even an estimate of the absolute number of US deaths in a year, but an estimate of deaths per million per lifetime. And yet, measuring deaths per million is nothing more than an idiosyncratic statistical device (it is conventional to measure deaths per 100,000). By what twisted logic has it become the gold standard for determining ventilation levels? One might just as easily apply the more common usage of deaths per 100,000 (which would slash the figure to 675) or, since we're picking numbers out of the sky, inflate it further by basing it on deaths per ten million (making it 67,500). Where does it end?

Indeed, considering that he is multiplying it by the number of air changes per hour, it would be more reasonable to also use the estimated number of deaths per hour. This, however, is a figure of just 0.017 and it would have forced him to conclude that 0.3 air changes per hour would render a room safe - the very opposite of a 'veritable tornado'. I am tempted to say that using the hourly rate would make more sense but the whole premise is so absurd that sense does not enter into it. At best it would make it more consistent in its absurdity.

One can argue over whether a room really needs 18 air changes per hour to be 'safe' from environmental contaminants but Repace provides no evidence that this rate of ventilation needs to be increased at all where smoking is permitted, let alone by a factor of nearly seven thousand. What is plainly obvious is that the number Repace multiplies it by is plucked out of thin air. None of the numbers he uses have any relevance to one another except in his own imagination and they are applied to a premise that is, in any case, bereft of rationale. His whole study is based on a series of basic logical errors which, when exposed, bring the rest crashing down.

Postscript

In 1997, the Bellagio hotel and casino opened in Las Vegas with the kind of high-tech ventilation system that Repace claimed was hopelessly ineffective. The Bellagio ventilated its casino several times time every hour with 100% outdoor air. Smoking was permitted throughout, except for the high stakes poker room. In 1999 and again in 2005, air quality assessments found that the air in the casino was as clean or cleaner than that measured outdoors. Respirable suspended particulates (RSP) were found at the barely measurable levels of 12 to 58 micrograms per cubic metre (a microgram is a millionth of a gram) and less than half of the RSP was made up of tobacco smoke. This is far lower than the 400-600 micrograms found in some unventilated bars and is a fraction of OSHA's limit of 5000 micrograms(4)(5). Carbon monoxide levels ranged from 0.8 to 2 parts per million (ppm) - far lower than OSHA's limit of 25ppm. Nitrogen dioxide levels ranged from 0.23ppm to 0.46ppm - well below the permissible 3ppm limit - and nicotine levels were found to be between 3.4 and 9.2 micrograms - a fraction of OSHA's 500 microgram limit (6)(7).

Neither ASHRAE nor OSHA have ever endorsed Repace's study.


* Repace comes up with the figure of 18 air changes an hour by extrapolating from an ASHRAE guideline which stated that this was the amount of ventilation required in a smoky bar. Why Repace then decided he needed to inflate it further is anybody's guess. For the record, the typical ventilation in a bar is between 1 and 2 air changes an hour. In the home, it is even less; ASHRAE recommend private homes receive just 0.35 air changes per hour (3).


**It is worth studying how Repace arrives at the final figure as it provides further evidence of his numerical illiteracy. To do this we must bear two facts in mind:

1. The recommended number of hourly air changes (ACH) is equivalent to 60% of the recommended number of cubic feet per minute per person (cfm/occ) eg. 30 cpm/occ = 18 ACH.

2. There are two ASHRAE guidelines, one for smoky rooms and one for smoke-free rooms. The former requires 30 cfm/occ, the latter 9 cpm/occ.

Repace starts off with the 30 cfm/occ level that applies to smoky rooms and multiplies it by 6,750 = 202,500 cfm/occ. Had he followed this thinking through and worked out the number of air changes, he would have arrived at a figure of 121,500 (which is 60% of 202,500). Instead, he completely changes tack. "However," he writes, "the default ventilation rate for a smoke-free bar under [ASHRAE's] Standard 62.1-2004 is 9 cfm/occ (equivalent to 5.4 ACH)." This is quite true and he proceeds to work out what the 'safe' level would be using the 9 cfm/occ level.

There is a certain logic to using the standard for a smoke-free room since the aim of Repace's study was to find what ventilation is needed to make a room free of smoke. What happens next, however, is very strange.

Instead of multiplying 6,750 by 9 cpm/occ, he takes the figure of 202,500 which he arrived at before and divides it by 9. Why he does this is anybody's guess. This is just plain wrong - he should have divided by 30 and multiplied by 9 - but here, as in much of the rest of the paper, one searches in vain for any semblance of logic. This gives him 22,500, which he then multiplies by the leftover figure of 5.4 for no particular reason and arrives at...121,500 - the exact same number he would have got had he finished off his sums using the 30 cpm/occ figure (above).

A remarkable coincidence? Not a bit of it. Look again at where the crucial figures of 9 and 5.4 came from. 9 ccpm/occ is the ASHRAE indoor air guideline and 5.4 (60% of 9) is the recommended number of air changes pertaining to 9 cfm/occ. Now think of a number, any number. Let's say, for instance, that ASHRAE required a whopping 1,000 cpm/occ indoors and, therefore, 600 air changes per hour. Now do what Repace did. First take the figure of 202,500, divide it by 1,000 ( =202.5) and multiply it by 600. What do you get? That's right: a tornado level of 121,500 air changes per hour. No matter what number you start with, you will always end up with 121,500 because the equation will only ever give you 60% of 202,500. And so, whether ASHRAE requires one cubic foot of fresh air or a million, Repace's equation will always tell you that you need a tornado.

It's like a card trick in which the sleight of hand is executed before you start paying close attention and you end up with the card the magician wants you to have. In this case, the sleight of hand is in the first calculation (6,750 x 30 cfm/occ = 202,500) which looks like a red herring and which Repace appears to disown with a quick "however..." But he has not disowned it and he slips the 202,500 back in as if it were a relevant figure while appearing to be using the more relevant 9 cfm/occ figure. He then employs a final flurry of disorientating calculations, dividing it by 9 and then multiplying by 5.4. The effect is to give us 60% of the first number, which, remember, never belonged in the equation to begin with. It is a very roundabout way of getting there, but this boils down to multiplying 6,750 x 18 = 121,500.

(1) http://www.repace.com/Repace-CV.pdf

(2) http://www.repace.com/pdf/iaqashrae.pdf

(3) http://www.ieqcorp.com/ventilation.htm

(4) http://www.ornl.gov/info/press_releases/get_press_release.cfm?ReleaseNumber=mr20000203-00

(5) http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1637107

(6) www.americangaming.org/assets/files/studies/final_iaq_white_paper_7-7-06.pdf

(7) One nonsmoking visitor to The Bellagio said it has "an air filtration system that makes cigarette smoke almost undetectable, which was a big plus for us and a definate (sic) advantage over other casinos."

No comments: