Intended for healthcare professionals

Rapid response to:

Papers

Reduced incidence of admissions for myocardial infarction associated with public smoking ban: before and after study

BMJ 2004; 328 doi: https://doi.org/10.1136/bmj.38055.715683.55 (Published 22 April 2004) Cite this as: BMJ 2004;328:977

Rapid Response:

These dramatic findings warrant a more objective view of the study's limitations

This study garnered enormous media attention, coinciding with the
presentation of its initial results to the American College of
Cardiologists last year. Those initial reports claimed heart attacks
quickly fell by nearly 60% as a result of Helena's six months workplace
smoking ban, and this staggering finding seemed to receive far more in the
way of celebratory fanfare than it did sober-minded skepticism.

In this published study, the so-called "Helena Miracle" is a bit more
tempered, but even at 40%, as this drop is now described, it is difficult
to fathom how 40% of all those suffering acute myocardial infarction
actually fell ill due to their exposure to cigarette smoke in restaurants,
bars and other holdout "smoking allowed" workplaces. Even though the
number of hospital admissions appeared to decline dramatically, the
arguments presented attributing this decline to workplace smoking bans are
weak. And many of the arguments describing this finding as consistent in
relation to other related studies appears weak as well.

Others have mentioned important concerns (chiefly the relatively
small numbers involved, the absence of any data regarding exposures to
secondhand smoke, and the absence of any distinction based on the
patients' own smoking behaviors). The authors acknowledged some of the
study's weaknesses as well. However, while it appears great care was
taken to distinguish patients in terms of consistency in the diagnoses and
where they should be assigned (test group or comparison group) in terms of
geography, the authors downplayed the complexities of smoking behavior and
overplayed anticipated impacts of newly enacted smoking bans on that
behavior.

As a result, a finding of 40% decline in AMI admissions resulting
from the workplace smoking ban is as astonishing for what it suggests
about the relationship between smoking bans and smoking behavior as it is
about any relationship between secondhand smoke and AMI risk. Even
without legal smoking bans in place, many workplaces have their own
policies against smoking. A ban would naturally result in an increase in
the number of smoke-free workplaces, but surely a sizeable number of work
places would have been smoke-free already. Would a work place smoking ban
impact residential secondhand smoke exposures within just the first month
or two? How many of those suffering myocardial infarction are exposed to
secondhand smoke in the workplace? How many are actually still working at
all? Wouldn't a disproportionate number be past retirement? Wouldn't the
impact of workplace smoking bans on retired and other nonworking people be
significantly diminished? The decline in heart attack admissions of 40%
is so large that it begs the question: when a smoking ban passes, does
everybody instantly stop smoking around others at home and in cars too,
thus eliminating altogether the 30% expected increased risk of secondhand
smoke exposure? Or is it just more unhealthy to breathe secondhand smoke
when you're in a restaurant or your co-worker's cubicle than when your
spouse lights up at home or in the family car? How could a reportedly
expected risk of 30% from secondhand smoke be completely eliminated within
just a few weeks, even days, simply by implementing a work place smoking
ban?

The authors' assert that the Helena findings were consistent with
other published findings, but the research they cited for review point to
far more modest impacts from smoking bans. It's hard not to conclude that
the authors were overzealous, and it's important that this study be
treated with more objectivity (maybe even incredulity) than has been the
case so far, at least in terms of the media and publicity savvy anti-
smoking activists.

For example, the report claims "Data from California, however, could
be interpreted as supporting our results. Death rates from heart disease
fell faster in California than elsewhere in the United States during the
California tobacco control programme, which, while including a tax
increase and media campaign (including the promotion of smoke-free
environments), focused on creating smoke-free workplaces and public
places."

The study goes on to offer another source, published in the NEJM,
which reportedly "support[s] our results", (Fichtenberg, Glantz.
Association of the California tobacco control program with declines in
cigarette consumption and mortality from heart disease). This study is an
examination of trends in tobacco consumption and heart disease mortality
in California in the years 1989 through 1997, concurrent to California's
aggressive new Tobacco Control Initiative. Unlike this current study,
which leads one to conclude that a workplace smoking ban alone could
account for a 40% drop in myocardial infarction, California targeted
tobacco use not only through smoking bans, but with sharply steeper
tobacco taxes and a powerful anti-smoking media campaign distributed
statewide. Despite its more comprehensive anti-smoking campaign, heart
attack mortality in California didn't decline by anything close to 40%.
During the entire 9 years studied, the decline in the mortality rate was
approximately 27%--yet most of this decline, it was acknowledged, should
be attributable to other factors unrelated to California's tobacco program
- factors which were pushing this mortality rate down nationwide.
Although the study claimed California's rate of decline exceeded the
national decline, this difference was in the area of 5 to 6 percent, and
again, this was over a nine-year period, not a few short months.

The authors referred to two other studies to argue "The fraction of
the population covered by smoking restrictions rapidly increased as a
result of [California's Tobacco Control] campaign" (refer to footnotes 14
and 15, both principal author Pierce), and they return once more to
Fichtenberg's report to assert "there was a parallel reduction in deaths
from heart disease." Again, those reductions in heart disease deaths were
much smaller than 40%. During the period 1990 through 1993, which Pierce,
et al, described as experiencing this "rapid increase" in smoking bans,
heart disease death rates declined between 17 and 18 percent while
nationwide, the rate declined about 13 percent. A case might be made
attributing the difference of 4 or 5 % to California's tobacco
initiatives, a portion of which attributable to their smoking bans, but
these statistics do more to enhance suspicions about the over-hyped Helena
results than they do to squelch them.

Besides relying upon exaggerated implications in findings from other
studies to help explain how AMIs could tumble so substantially, there are
potentially significant problems resulting from the peculiar
"geographical" assignations of the test and comparison cases. This is
potentially significant because the test and comparison groups were
assigned based on these geographical assignations, as opposed to more
selective criteria focusing on the ban's impact on actual secondhand smoke
exposures and/or reductions in active smoking. The authors described
assigning certain patients to the test group, even though they lived
outside the geographical area, as a result of information indicating these
patients had spent time in the "test ban zone" some time prior to their
illness. In other words, a patient might have spent as little as an hour
or two total having lunch or dinner in Helena prior to their illness, and,
if this fact were so indicated in their chart, they'd be assigned to the
Helena test group. This study doesn't describe when these "specially
assigned" patients were admitted to the hospital--whether they were evenly
distributed throughout or whether they may have "clustered" toward pre-ban
or post-ban--but it's important to note that the enactment of smoking bans
have the potential to impact non-local customer or visitor traffic. The
St. Peter's Hopsital may be been "geographically isolated", as the authors
emphasized, but I believe just less than half of this "geographically
isolated" community was subject to the smoking ban. There were numerous
casinos in this area, some of them subject to smoking bans while others
not, and it's reasonable to ask whether or not visitor traffic altered in
response to the smoking ban. If so, this would not only mean that the pre
and post population "pools", but also the AMI patient assignments between
test and comparison group, were variables dependent on the passage of the
smoke ban. In other words, it may be impossible to differentiate how much
of the reduction in AMI admissions was attributable to lessened secondhand
smoke exposures and how much was attributable to a consequent relocation
of the problem to other geographical areas where smoking was still
welcomed.

This study certainly isn't the first of its kind to suffer criticisms
like those others have offered as well as myself, but more and more do
such highly charged issues burst through the walls separating the
political arena from the scientific, infecting the dialogue to such an
extent that the scientific rigor is compromised in furtherance of
political expediency. With encouragement from the study's authors, the
media went wild with a story hinting all you had to do to reduce heart
attack by 60% was to pass a city work place smoking ban. This study
offers a somewhat more subdued promise--a 40% reduction in heart attack--
while at the same time suggesting that this amazing result is plausible
since California, which began passing work place smoking bans a dozen
years earlier, also experienced a reduction in heart attack mortality.
(The authors carefully neglected to mention California's reduction was
more along the lines of 5% as opposed to Helena's 40%, or its initially
touted 60%.)

In anticipation of any suggestion that California's results were
dampened by the "smear out" effect described in this report, let me point
out that California's statewide smoking ban went into effect during the
period used in "support" of the findings observed in Helena, and any
"smearing" would thus come out in the wash. Even after 9 years of gradual
yet steady progression from isolated local bans to statewide smoking bans,
California's overall decline in the heart attack rate was an insignificant
blip when compared to Helena's reported heart attack decline, the majority
of which was traceable to June and July 2002, just one month or two after
the ban went into effect.

The "Helena Miracle". First it's 60%, then 40%, which is supposedly
comparable to California's 5 or 6%? This isn't science, folks. This is a
very strange socio/cultural/political gamesmanship calling itself science
and I hope somebody is getting it all down on paper. Examples like this
are revelatory of the way science intersects with politics, business and
the media today, and examining the distortions which often result from
this intersection would be very enlightening.

Competing interests:
None declared

Competing interests: No competing interests

14 April 2004
Linda N Phillips
retired
Ukiah 95482