Hearts in Helena

For those who can't get enough debunking of the Helena, Montana, smoking ban study, here are some problems I didn't have space for in my column:

The heart attack data do not really match the story the researchers are telling. Starting at the beginning of 2002, the pattern was:

January: 8
February: 6
March: 2
April: 8
May: 6
June: 2
July: 1
August: 3
September: 7
October: 5
November: 4
December: 6

The ban took effect in early June and was suspended in early December. Notice that the 8, 6, 2 sequence appears twice in a row; the tail end of the second sequence happens to fall in June, at the very beginning of the ban. There's no reason to assume that the low number seen that month had anything to do with the ordinance.

Also, it's implausible that the ban would have its greatest impact early on, when compliance and enforcement were spottiest. The local paper was still running stories about the need to crack down on noncompliant establishments in late September. And why would heart attacks rise and go down again in the middle of the period when the ban was in effect?

Comments

[Lefty]

They could pick up the phone and talk to the relatives.

[Dr. Kevin Felles]

This really isn't a scientific sample; Helena, Montana is not a large city. I'd echo what Chuck said above--during the winter, Helena receives huge amounts of snow. What's to say that some people didn't get heart attacks from shovelling snow? The sample is so small that it can't possibly take that into account, nor can it be analyzed for the potential of foreign visitors.

Dr. Kevin Felles
Assistant Professor of Applied Statistics
Boise State University

[The Mountain Goat]

There is credible scientific theory that heart attacks are frequently preceded by chest colds. That would put less heart attacks in the summer months, building up again in the indoor season, which in Montana would be coincide with this data.

[Chuck]

Another factor might well be related to seasonal physical activity and population fluctuations, namely due to snow skiing. I don't know whether Helena has many ski lodges nearby, but there probably are some. And with a relatively small population base, it doesn't take too many visitors to increase the seasonal population by a large percentage. The original article didn't say that they were only counting residents, so probably visitors are included as well. I wonder how many otherwise-sedentary, middle aged professionals hit the slopes there in winter, for some vigorous exercise that they are not in shape for?

These data are so sparse that any sort of statistical analysis is pretty much useless. The best one could do is some sort of nonparametric test--perhaps a runs test to determine if there is any seasonal variation. But my hunch, on a first reading, is that one would have to dig REALLY hard to mine any statistically significant ANYTHING out of this dataset other than a possible seasonal fluctuation, which could have any number of explanations.

[Brian]

In line with Chuck's comment...you'd also really need several years' worth of data to establish a seasonal variation (or lack thereof). This data is almost certainly available to the researchers. That it is not addressed is pretty telling.

[Jacob]

As I noted in my column, the study did include a comparison with previous years (1998 through 2001). It's clear that the heart attack numbers were unusually low in July, August, and November compared to those months in the previous four years. The question is whether the smoking ban had anything to do with it.

[JBurns]

I am NOT a number, damn it!

I'm 89, smoke, and I'm still around.

(This lumping of humans into statistical boxes is purely Prussian.)

[Frenk]

Hey, JBurns! As a resident of Westfalen (heart of Prussian Germany), I would like to defend this place against accusations of fascism. In fact, most rules are taken quite lightly around here. And we can smoke nearly anywhere--including my university office. Compare that to the "freedom" of the US!

[JBurns]

I know that, Frenk. I was referring to the old, classical "Prussia" -- the one before WW-II. (I'm sure you know what THAT was like.)