Thursday, November 10, 2011

Tobacco Control Science as Shoddy as It Gets: The North Carolina Smoking Ban - Heart Attack Report

Believe it or not, the following data - which show a 21% increase in heart attacks among women during the first year following North Carolina's smoking ban - are the actual data behind the report out of the North Carolina Department of Health and Human Services which concludes that the smoking ban led to a 21% decline in heart attacks in the state.

Figure 1. Annual Heart Attack Admissions Among Females -- North Carolina, 2008-2010

These data are taken directly from the report.

Despite what the report calls a 21% reduction in heart attack admissions during the first year of the smoking ban, the number of heart attack admissions among women was higher after the smoking ban than before. The number of heart attack admissions among women, which dropped from 4182 in 2008 to 3521 in 2009, increased to 4273 in 2010, so that it reached its highest level in three years!

It is simply mind-boggling to me how the report, and anti-smoking groups and advocates which have been disseminating the conclusions of the report, can draw the conclusion that the smoking ban reduced heart attacks by 21% when the actual number of heart attacks among women increased by 21% during the first year after the smoking ban was implemented.

Apparently, the model used in the report failed to include a time trend and therefore attributed the lower number of heart attacks overall in 2010 to the smoking ban, rather than to the existing trend of decreasing heart attacks in the state. Properly adjusting for the secular time trend in heart attacks would have negated the finding of a 21% decline in heart attacks associated with the smoking ban and instead would have led to the conclusion that heart attacks were higher than expected in 2010.

To be clear, I am not suggesting that researchers intentionally omitted the time trend to create this finding. However, I do believe there may have been a subconscious bias toward finding and showing an effect of the smoking ban. We all want to be able to demonstrate the positive results of our enacted policies on the public's health. Here, that bias appears to have clouded the interpretation of otherwise very clear data which fail to show any significant decline in heart attacks attributable to the smoking ban during its first year of implementation.

I continue to be boggled by the poor science coming out of the anti-smoking movement in the specific area of the effect of smoking bans on heart attacks. If these were tobacco industry studies purporting to show a negative effect of smoking bans on restaurant sales, but using the exact same methodology, we would be blasting them.

Interestingly, in his note to tobacco control advocates, Dr. Glantz argues that the tobacco industry has been criticizing the smoking ban - heart attack studies and insinuates that the only ones criticizing these studies are tobacco company employees or tobacco "sympathizers." Clearly, that is not true. For one, I have been quite critical of these studies and I am not affiliated in any way with the tobacco industry. Additionally, I am not aware that the tobacco companies have publicly opined either way on these studies.

The rest of the story is that anti-smoking groups have jumped on to a study which concludes that the North Carolina smoking ban decreased heart attacks by 21% in its first year, when in fact heart attacks among women increased by 21% during that year and the rate of decline in heart attacks overall slowed significantly. This is shoddy science and I believe it threatens to undermine the integrity of the tobacco control movement, risking the loss of our ability to effectively advocate for these and other important public health policies.

No comments: