Smoking Ban Increases Heart Attacks


Anti-smoking activists claim smoking bans in Scotland and Wales led to immediate, sizable reductions in heart attacks. In both cases, Michael Siegel shows on his tobacco policy blog, these assertions are unfounded.

An article in the July 31 New England Journal of Medicine cites a 17 percent decline in hospital admissions for acute coronary syndrome during the first 10 months after the Scottish ban took effect at the end of March 2006. Siegel notes that the comparison was based on data from only nine hospitals. Based on data for all Scottish hospitals, he reports, the post-ban drop in heart attacks looks similar to declines seen in previous years. Between 2005 and 2006, heart attack admissions fell by 4.3 percent, slightly smaller than the 4.6 percent drop from 2003 to 2004. The difference between 2006 and 2007 was 8 percent, smaller than the 10.2 percent difference between 1999 and 2000. The two-year drop from 2005 to 2007 (11.9 percent) was not much bigger than the two-year drop from 1999 to 2001 (10.7 percent). Although "the analysis in this paper assumes that the entire observed change in heart attacks is attributable to the smoking ban," Siegel writes, "one cannot rule out the very plausible alternative hypothesis that the observed decline in heart attacks is explained by random variation in the data and the already existing secular trend of declining heart attacks in Scotland."

In the case of Wales, U.K. Action on Smoking and Health and the British Heart Foundation cited a 13 percent drop in heart attacks as evidence of the smoking ban's immediate positive effect. But as Siegel shows, drawing on data presented by Christopher Snowdon, this decline appears only if you compare the arbitrarily selected months of October through December 2007 to those same months in the previous year. But the Welsh smoking ban took effect in early April 2007. Data for the entire nine-month period between then and the end of the year show virtually no change in heart attack admissions compared to the same months in 2006. The overall drop between 2006 and 2007 was about 1 percent, much smaller than the drops seen from 2004 to 2005 (10.3 percent) and from 2005 to 2006 (6.3 percent). Furthermore, heart attacks rose in the first five months after the ban before falling again in the next four months. Based on the post hoc, ergo propter hoc logic of activists who assume that any decline in heart attacks following a smoking ban must be due to the ban, Snowdon writes, "we might even say that the smoking ban 'caused' a rise in heart attacks in Wales."

As I've said before, many of the jurisdictions that adopt smoking bans inevitably will see declines in heart attacks in the year or two after the bans take effect. If you focus only on those jurisdictions while ignoring random variation and pre-existing trends, it's not hard to create the illusion of an effect, especially if everyone forgets how biologically implausible it is for heart attacks to fall so quickly in response to a smoking ban (whether because of less smoking, less secondhand smoke exposure, or some combination of the two). Given the vast potential for cherry-picking data, the remarkable thing is what a bad job the ban boosters are doing.

More reason coverage of smoking bans and heart attacks here.