Last week the Oregonian ran an op/ed about lottery revenue that turned out to really be primarily about the Oregon smoking ban. I missed it at the time, but it’s worth rebutting now. The piece is by Ty Gluckman, director of the coronary care unit at Providence St. Vincent Heart and Vascular Institute and a board member for the American Heart Association’s Portland chapter. In short, he’s someone who ought to know better than to write this:
In the midst of the messy process of health care reform, it’s refreshing to see one particular health policy, smoke-free workplace laws, having a documented impact. But despite two recent studies clearly linking the laws to a decrease in heart attacks, the laws continue to be a scapegoat for declines in video lottery sales — with no evidence of this link.
In the first study, published in Circulation: Journal of the American Heart Association, researchers James Lightwood and Stanton Glantz looked at 10 cities in the U.S., Italy and Scotland that have adopted laws restricting smoking in public areas. Their conclusion was astounding: The rate of hospitalization for acute heart attacks was reduced by 17 percent after one smoke-free year.
Even better, the bans continued to reduce heart attack rates over time, with about a 36 percent drop three years after the laws took effect. These results are consistent with the fact that nonsmokers exposed to second-hand smoke at home or at work have a 25 to 30 percent increased risk of developing heart disease. In fact, Lightwood believes that the new findings suggest that the increased risk may be even higher. […]
The two studies released this month should give comfort to our state legislators that the smoke-free workplace law will help make Oregon a healthier place to live. In fact, according to the studies, it’s highly likely that Oregon’s heart attack rates are already dropping as we near the law’s one-year anniversary.
If we reduce the number of acute heart attacks by 17 percent, there will be at least 1,100 fewer hospital admissions in Oregon in just one year. At a cost of more than $35,500 per admission, the savings will be substantial: $40 million.
Before getting in to what’s wrong with this reasoning it’s worth taking a look at Oregon heart attack rates. These are conveniently listed on the state DHS site for the years 2000-2006:
2000 — 36
2001 — 36
2002 — 37
2003 — 36
2004 — 33
2005 — 32
2006 — 31
There are two things to notice here. One is that, as is the case in many regions, there’s a secular downward trend in the rate of heart attack admissions. It’s dropping for reasons other than the imposition of smoking bans. Secondly, there can be significant fluctuations from year-to-year with no clear cause. From 2003 to 2004 there was a decrease of more than 8%. I’m sure there were several laws that took effect on January 1, 2004, but it would be silly to credit them with the drop in heart attack rates just because they occurred contemporaneously.
It’s also worth looking at the county-by-county break down of heart attack rates. Multnomah County, home to Portland’s large population, is very stable, with rates never going up or down by more than two points each year. Baker County (population 16,000+), the first on the list, varies much more wildly:
2000 — 39
2001 — 53
2002 — 31
2003 — 38
2004 — 23
2005 — 20
2006 — 16
Heart attack rates in Baker County jumped nearly 36% from 2000 to 2001. Just as miraculously, they then dropped more than 40% the next year. Does this mean anything? No, of course not. These are just the kinds of fluctuations you get when you examine a small sample. Yet an unscrupulous researcher could take the years 2001 to 2006 and make it look like a change in policy in 2002 brought about dramatic health results. Clearly, if you want to get an accurate view of the effect of a smoking ban on heart attacks, you’d want to look at studies with large sample sizes. Fortunately this data is easily available and there are dozens of states and countries than have implemented comprehensive smoking bans. (Incidentally, the city of Baker passed a partial smoking ban in February of 2000, just prior to the dramatic rise is heart attacks. For some reason anti-smoking researchers don’t include Baker in their studies.)
In the studies cited by Gluckman, where do the authors look? The first, by Lightwood and Glantz, includes such giant metropolises as Helena, MT and Pueblo, CO. Not only that, but because the authors found two studies each for Pueblo and Piedmont, Italy, they double-counted the same data and treated them as independent observations. (For a critique of one of the Piedmont studies, see here.)
The Helena and Pueblo studies are notorious examples of junk science, akin to taking the wild fluctuations in Baker County as significant findings (see my earlier post on the Pueblo study here). No impartial researcher would give them such weight when there is so much other data from larger samples available.
The second study from the IOM isn’t much better. I won’t get into it here, but suffice to say that it too ignores data that show no dramatic declines in heart attack rates following a smoking ban. See, for example, the United States, New Zealand, Australia, and the United Kingdom.
It’s possible that smoking bans do have some small effect on heart attack rates but it’s clear that the “heart miracles” claimed by Glantz and other researchers are illusions of the data created by small sample sizes. It will be important to keep this in mind as Oregon’s data for 2009 come in. Will there be a drop in heart attack rates, maybe even a drop comparable to that seen statewide in 2004? Will there be a few places the size of Baker County that experience not a mere drop by a full-blown miracle? Both outcomes are possible, but it would be foolish to attribute them solely to the smoking ban.
As Gluckman himself writes, the conclusions in the papers he cites are “astounding.” I’ve come to expect that reporters will take such claims at face value, but I hope for better from professional cardiologists like Dr. Gluckman. Those astounding claims should have aroused some skepticism in him. I’m sure he’s an intelligent man and perhaps he just didn’t look into the studies as thoroughly as he should have. As so often happens, confirmation bias can lead smart people to believe outlandish things.