Saturday, 14 April 2012

Not normally glamorous but still deadly?


A continuing bid to promote the cause of plain packaging gives Cancer Research UK cause to publish lung cancer figures. The graph published in the CRUK report is not new (appearing also here): it gives data to 2007.

The Telegraph quotes CRUK director Jean King:
“It’s vital that the UK closes one of the last remaining loopholes that portrays smoking as something glamorous and normal, rather than the lethal product it truly is."
Glamorous and normal? What does she want?

As a non-scientist the graph gives me immediate problems because the trajectories for male and female lung cancers do not even begin to resemble each other. The Telegraph tells us that the incubation period is 25 to 30 years. The male smoking rate appears to follow the same trajectory as the male lung cancer rate with about 18 years' difference between the two. The female rise from 1975 does reflect the rise in the female smoking rate from 1950 to 1975, but fails to start falling 25 years after female smoking peaked.

Another pair of graphs here (blue for men, pink for women, split into age groups) shows lung cancer for men at 1960 at less than half the 1975 level, but for women the contrast is less stark with the rate in women in 1960 closer to the 1975 level. The same clear difference shows: while men's lung cancer declines, the women's rate continues to climb. There are no smoking rates plotted against the earlier figures before 1975, but the huge disparity between the two correlations prompts questions about other factors that might affect lung cancers.

As far as lung cancer is concerned, smoking is the only suspect in the frame. Huge amounts of evidence will not help to eradicate the problem of lung cancer if they have the wrong suspect. Cancer Research doesn't seem to want to know who the other suspects might be, because its tobacco research agenda has its own specific goals.


Anonymous said...

As you rightly say, these trends do not appear to make much sense. Cutting off the lung cancer incidence in their graph at 1975 is a classic example of how to lie with statistics. Clearly the figures are available, as you pointed out, in your external link (figures 2.4 and 2.5) and so could have been included.


Fredrik Eich said...

That graph has always been susect. Soon I plan to create my own version of it. The IRAC now has an online tool for creating charts which can be accessed from the "SOURCE" link that can be found below the charts here

Fredrik Eich said...

Ok, embeded hyperlinks not allowed.
Here it is agin

Anonymous said...

Mortality Trends for Selected Smoking-Related Cancers and Breast Cancer -- United States, 1950-1990
TABLE 1. Age-adjusted death rates * for selected smoking-related cancers, by sex and
race + -- United States, selected years, 1950-1990(per 100,000)
Total - Lung

1950 = 13/100,000 (Incidence Rate = 13/100,000)

1860 = 21

1970 = 32.

1980 = 42.

1990 = 50.3

American data; but, interesting trends.
Gary K.