Thursday 22 October 2009

Smoking bans have not reduced heart attacks in Australia

Hospital admissions statistics from Denmark recently confirmed evidence from England, Scotland, Wales and the USA that smoking bans do not significantly reduce rates of acute myocardial infarction (heart attacks). This contradicts several widely reported epidemiological studies which have claimed a dramatic reduction in the heart attack rate as a result of smokefree legislation.

I have recently been looking at hospital admissions data from Australia to see if there is any evidence to support the "smoking ban slashes heart attacks" hypothesis. With the exception of the sparsely populated Northern Territory, the whole of Australia has now adopted comprehensive indoor smoking bans and so, if the hypothesis is correct, we should see signs of rates of acute myocardial infarction declining.

Unlike much of Europe and the US, there is a long-term upward trend in heart attack admissions in Australian hospitals. The reasons for this are not clear. Australia's rising population (of around 1.5% a year) will have been one factor and there will have been changes in methods of diagnosis. Whatever the cause, it is quite obvious from the graph below that smoking bans in Australia have done nothing to change that upward trend.

All the data is shown as financial years (April to March). Australia's states introduced their bans independently between January 2005 and November 2007. The coloured arrows indicate when each ban was implemented.

Black: Western Australia (population: 2.22m)

Yellow: Tasmania (population: 0.5m)

Blue: Queensland (population: 4.18m)

Green: Australian Capital Territory (population: 0.34m)

Red: New South Wales and Victoria (population: 6.89m and 5.40m)

White: South Australia (population: 1.16m)

Although a third of Australia's population was covered by smoking bans by the end of 2006, the two most heavily populated states introduced their bans in July 2007. South Australia was the last to follow suit in November 2007.

There is no hospital admissions data yet available beyond March 2008. The statistics for 2008/09 will complete the picture, but what we can see from the data already available is that there was no decline in the heart attack rate, nor was there even a decline in the rate of increase. Indeed, the rate of increase accelerated slightly in 2007/08, despite this coinciding with bans covering the densely populated states of New South Wales and Victoria.

The Australian data lacks the simplicity of Wales and Scotland, where the authorities helpfully introduced their bans at the start of the financial year. Nevertheless, if the Helena hypothesis is correct, some positive effect on the heart attack rate should be visible, and there should be a cumulative effect as more and more states implemented their bans. There is absolutely no evidence of any such effect. 

Note that the Helena hypothesis specifically claimed a drop in heart attacks of 40% within the first six months of the ban. Although Victoria and New South Wales introduced their bans towards the end of the time-frame, there was still nine months between July 2007 and March 2008 for a drop in heart attacks to occur. Instead, the rate rose from 51,667 to 55,676 between 2006/07 and 2007/08.

Australia's hospital records provide further evidence that the "smoking bans slash heart attacks" hypothesis is incorrect.

This data comes from the Australian Institute of Health and Welfare website (click 'Diseases of the Circulatory System', then 'Ischaemic heart diseases'). There is also a report (Australian hospital statistics 2007-08) which breaks the admissions down by public and private hospitals (page 218).


westcoast2 said...

This is a very nice analysis. The thing that is intersting is the staging of the bans and that clearly none had an affect.

All the work you have done on these 'miracles' is very impressive.

Unknown said...

This is ok, but it would be a lot more compelling if you could disaggregate the 'whole country' AMI admissions into the relative territories and then AMIs within each of the 6 states/territories at the point of the ban intervention, presenting it in 6 separate charts/tables.

Otherwise you allow for the pretty obvious out that the ban's acute effect in one state was swamped in the overall Australia data by a random spike in other areas, a change in classification in other areas, or a really scary movie shown to the rest of the country that generated enough heart attacks to see no drop in the areas where a ban was implemented.

Probably just no available data for this I'm guessing.

Christopher Snowdon said...


Stay tuned. I'll be back tomorrow with the state-by-state data.


Fredrik Eich said...

"Unlike much of Europe and the US, there is a long-term upward trend in heart attack admissions in Australian hospitals. The reasons for this are not clear." This is post hoc proctor hoc. This is because whenever the mighty Englandshire beats them in the rugby world cup and the Ashes, they cry like babies, won't talk about sport, become hypertensive...