Sunday, July 25, 2010

Lung cancer, smoking & statistical chicanery

The Lung Cancer Alliance (LCA), as noted in my last post, claims the majority (60%) of new lung cancer cases in the US occur among former smokers, with roughly 21% occurring in current smokers and 18% among those who have never smoked. The information is provided in a neat little chart on their website.

I noted in a previous post that the same site claimed that former smokers (45 million) in the US now outnumbered current smokers (40 million), suggesting this was a recent phenomenon in the US. I really didn't give the numbers much thought. If 35% of lung cancer cases occurred in an estimated 21% of the population (smokers), and only 15% among never smokers (55% to 60% of the population), then it still indicated a strong correlation between smoking and lung cancer.

But, these latest figures from LCA gave me pause. The gap between lung cancer cases among current smokers and never smokers appeared to be narrowing as the number among former smokers increased.

And, although the fact that former smokers outnumber current smokers may be a recent phenomenon in the US, (is that possible?) it certainly isn't in Canada. Data from Health Canada's last report on smoking attributable mortality clearly shows that former smokers far outnumber current smokers in every age bracket.

But, Health Canada data doesn't tell us how many current, former or never smokers die from lung cancer. They speak only of smoking attributable deaths and those ostensibly from other causes.

To be sure, extrapolating data from the US to Canada can be a hazardous proposition, as Health Canada found out a couple of year ago. In 2008, they were forced to revise their smoking attributable deaths downwards from roughly 47,000 to 37,000 because the US data they were using was not appropriate for the Canadian population.

Still, I thought it might be interesting to apply the numbers from the Lung Cancer Alliance to the Canadian mortality data from StatCan to see if a reasonable estimate could be made as to the number of lung cancer deaths in each of the three smoker classifications in Canada.

Not surprisingly, applying the Lung Cancer Alliance estimates to the Canadian mortality data paints an entirely different picture than the one presented by Health Canada's smoking attributable mortality figures, as noted in Chart 1.

It's not that the total smoking attributable deaths varied by a huge amount In fact, using the Lung Cancer Alliance percentages actually increases the estimates given by Health Canada by roughly 4%. Of course, the LCA calculations would be reduced somewhat since a small portion of those current and former smokers would be expected to die whether they smoked or not, reflecting the numbers expected in a never smoking population.

But, what was most perplexing was the ratio between deaths among current and former smokers when using the LCA percentages. In fact, those ratios practically eliminated the strong correlation between smoking and lung cancer. The number of lung cancer deaths among current smokers would be only slightly higher than the number among never smokers, despite similar population sizes. Which, in turn, brings into question the anti-smoker claim that “men who are current smokers are 23 times more likely to die of lung cancer than men who have never smoked.”

And, it also suggests that quitting smoking is not the deterrent to lung cancer the anti-smoker crowd claims.

I know what you're thinking. That's ridiculous. But, before dismissing this possibility as unrealistic balderdash, perhaps we should have a look at the smoking attributable mortality report from Health Canada and what it shows.

Almost 99% of all lung cancer deaths in Canada (2002 data) occur after the age of 45. But, at no time after the age of 45 is the percentage of former smokers less than double the number of current smokers. And the ratio between former smokers and current smokers increases cumulatively as the population ages. You may be a smoker for 5, 10 or 20 years, but you're a former smoker for life.

Looking at the male component in the age group between 45 and 59, we see that former smokers already outnumbered current smokers by roughly 2 to 1, as can be seen in Chart 2. In the age group between 60 and 69, where 28% of lung cancer deaths occur, the ratio between former smokers and current smokers is almost 4 to 1. In the age group between 70 and 79, where 36% of lung cancer deaths occur, the ratio is 7 to 1. And, in the age group 80 and over, where 20% of lung cancer deaths occur, the ratio is a whopping 9 former smokers for every current smoker.

The ratios are only slightly less for the female population, ranging from almost 2 to 1 between the ages of 45 and 59 to a high of roughly 6.5 to 1 in the population 80 and over.

Given these high ratios between former and current smokers, is it really unreasonable that former smokers might account for 50% to 60% (and, possibly more) of lung cancer deaths in Canada? Think about it.

In 2002, roughly 56% of lung cancer deaths were among the population aged 70 or older. In that same population, former smokers outnumbered current smokers by at least 7 to 1 for men, over 4 to 1 for women, and in the over 80 crowd, the ratio was 9 to 1, for men and 6.5 to 1 for women.

Common sense would suggest that the vast majority of those deaths occurred among former smokers, many of whom would have quit decades earlier.

So, just what conclusions can be drawn from this highly speculative foray into the maddening world of smoking attributable mortality.

And, the answer is: absolutely none. I'm neither an epidemiologist nor a statistician. And, this little exercise, as previously noted, is highly speculative. It is also postulated on the belief that smoking related statistics are grossly exaggerated and manipulated to the point where they're totally unreliable, which may introduce an element of bias.

But, it sure as hell raises a lot of questions which deserve answers. And, unfortunately, I don't expect that any will be forthcoming from either Health Canada or the rest of the anti-smoker crowd.

Of course, that's just the opinion of a rapidly aging old rambling man.

1 comment:

Leg-iron said...

Playing with the statistics, it's leditimate to claim from those figures that if you stop smoking, you triple your chances of lung cancer.

It's rubbish of course. There are too many factors and potential causes of lung cancer to make such a statement.

Nonetheless, it is far closer to what the figures show than any of the pronouncements of the antismokers.

I have seen several ex-smokers get lung cancer. Never met a current smoker with it. I've wondered if the abrupt stop that most ex-smokers employ act as a shock to their immune systems, which have been coping with smoke for years.

A highly charged immune system that suddenly finds itself with no target can cause all sorts of mischief.