Monday, October 17, 2005

Helping Mr. Moffitt defend smoking bans

Posted by Craig Westover | 4:29 PM |  

If my good friend Mr. Moffit were on the ball, in answer to my challenge to him, he’d be touting a new study by the California Environmental Protection Agency that anti-smoking ban blogger “Marcus Aurelius” posted on his Clearing the Air blog. Unlike Bob’s American Lung Association blog, those that oppose smoking bans are not afraid to post opposing views or address the questions they raise.

The title of this newly released study by the California EPA is “Proposed Identification of Environmental Tobacco Smoke as a Toxic Air Contaminant.” The report outlines the procedure and the results to date of the process to determine if second hand smoke is a TAC by California Standards. Bob might take note that the report establishes neutral standards for making this determination -- something that he has dismissed as not his responsibility. But that’s another debate.

One can read the report here. I’d like to highlight just one section that reiterates the points I’ve made consistently to Bob that he has failed to address, but first some more pointers for Bob.

Bob consistently calls the use of “relative risk” factors to determine causality between second hand smoke and various diseases “junk science.” I have asked him how, absent relative risk numbers, he determines causality. He has not responded. But here we have a report that concludes that second hand smoke ought to be considered a Toxic Air Contaminant, and what does it base that conclusion on -- relative risk numbers. Relative risk factors are not “junk science,” unless Bob wants to discredit this study, not done by the tobacco industry, that supports his conclusions.

A second point Bob makes is that only studies done by agencies such as the California EPA are unbiased and valid. Studies done by tobacco companies and “so-called” independent researchers are profit-driven and invalid simply because of the source. Well, here we have a study done by the California EPA and, low and behold, the pure science in this report is right in line with the pure science one finds in virtually all second hand smoke studies -- including those done by tobacco companies. The difference between this EPA study and studies done by tobacco companies and independent researchers is not the data arrived at, but the conclusions drawn from the data. And it is here that I disagree with this California study.

Quoting from the data --
The relative risk is a measure of the relation between exposure to a substance and the incidence of a disease. A relative risk of 1.0 indicates no relationship.

For ETS, a relative risk estimate of 1.2-1.7 for heart disease mortality in nonsmokers is supported by the collective evidence; this corresponds to approximately 1,700-5,500 deaths annually in California.

The relative risk estimate of 1.38 associated with low birth weight implies that ETS may impact fetal growth of 1,600 newborns in California.

It is estimated that at least 31,000 children in California experience one or more ETS-related asthma episodes (new onset or exacerbation) each year.

Large impacts are also associated with relative risks for respiratory effects in children such as middle ear infection (RR ≈ 1.62) (about 50,000 children annually), and lower respiratory infection in young children (RR ≈ 1.5 to 2) (18,000 to 36,000 children annually).

ETS exposure is implicated in 21 SIDS deaths per year in California (RR ≈ 3.5).

About 400 to 1,100 lung cancer deaths in California are ETS-related. For nasal sinus cancers, observed relative risks have ranged from 1.7 to 3.0. This is as high as or higher than the relative risks observed for lung cancer.

Finally, for breast cancer, when evaluating younger, primarily premenopausal women at diagnosis, a pooled risk estimate of 1.68 isderived in the meta-analysis, and when restricted to the studies with better exposure assessment, an estimate of 2.20 is obtained (see Table 1). These estimates of association could represent a significant number of cases as this is a relatively common cancer in women.

Adding the mid-point of the ranges for lung cancer deaths and heart disease deaths, and including the SIDS point estimate, one can attribute about 50,000 deaths per year in the U.S. and 4,000 deaths per year in California from ETS-associated disease. This does not include the estimates for other ETS-associated cancer deaths.
What you’ll note in these statistics is that with the exception of the relative risk for SIDS (which is irrelevant to the smoking ban in bars and restaurant debate), all of the relative risk factors hover under 2.0. You’ll also note some extremely large ranges in some categories -- 400 to 1,100 lung cancer deaths in California are ETS-related. No baseline data is given for this figure, but what those numbers mean is that out of some larger population -- all Californians or just adult Californians -- one might expect to see 400 to 1,100 additional lung cancer deaths that are ETS related (which is also undefined). The extrapolation made is that the entire state is exposed to second hand smoke in the same quantity as the subjects in the unidentified studies from which the small relative risk number was derived. That is a pretty big assumption.

Here’s where the California EPA study breaks down. This is also a direct quote from the study, which precedes the above data --
Relative risk estimates associated with some of these endpoints [the data quoted above. ed. note] are small, but because the diseases are common and ETS exposure is frequent and widespread, the overall impact can be quite large.
That conclusion is a misleading interpretation of the statistical significance of relative risk. Low relative risk factors -- certainly those below 2.0, and in the scientific community those below 3.0 to 4.0 -- generally are not taken as statistically significant evidence of causality. Relative risk factors at those levels indicate that other factors are likely contributors to the effect being studied. Attributing causal significance to relative risks below 2.0 gets one to the factious conclusions that drinking coffee and talking on cell phones poses a greater risk for heart disease than second hand smoke.

Does it pass the smell test that if, as the EPA report states, a relative risk of 1.0 indicates no cause relationship that a 1.2 relative risk factor, which the report cites for heart disease mortality in nonsmokers, is statistically significant? If it does, there are a hellevua lot of things we should be banning in the name of public health.

This is the bone of contention that Bob will not address. There is a pretty consistent body of data on the relationship between second hand smoke and various diseases. This research, whether done by tobacco companies, state agencies or health organizations arrives at virtually the same relative risk factors. The problem is that in order to promote a social agenda, people like Bob extend the conclusions of these studies beyond their scientific and statistical limits. Without defining criteria for when a health issue rises to a level of significance that requires government intervention, they advocate government policy that inflicts significant economic harm on a segment of the population not to mention tramples individual property rights and the principle of individual choice.

The challenge still stands, Bob. I’ve again done most of your work for you, but if you want to defend this study by providing baseline data and explaining how, using only statistics provided by this study, my analysis is wrong, I’ll be happy to post it here. Oh yea -- is relative risk data still junk science?