If smoking causes lung cancer, why has a decline in smoking not led to a decline in lung cancer?

Question

According to the CDC, smoking prevalence among US adults (18+) has decreased quite a bit over the years:

1974: 37.1% of the population
1980: 33.2
1990: 25.5
2000: 23.3
2010: 19.3

According to cancer.gov the rates of new cases of lung cancer in the US haven't consistently gone down over the years:

1975: 52.5/100,000 people
1980: 60.7
1990: 68.1
2000: 64.1
2010: 57.3

If smoking causes lung cancer - according to the CDC "cigarette smoking is linked to about 90% of lung cancers" - shouldn't such a clear decline in the number of smokers have also caused a clear decline in the incidence of lung cancer? How have experts/doctors/medical scientists/etc. addressed these statistics, if at all?

This and many other contradictions are analysed in detail by Vincent Ricardo di Pierri in his book, Rampant Antismoking. Recommended reading. — Pedro Werneck, Dec 04 '14 at 14:48
@PedroWerneck That book seems to be saying that if the theory requires adding some undefined 'lag' between cause and effect, then the theory (that there is cause and effect) becomes unfalsifiable? Perhaps you could post whatever it is you want to say about/from the book as an answer, instead of as a comment. — ChrisW, Dec 04 '14 at 15:28
@chrisw No, thanks. I just want to recommend it for anyone who might be genuinely interested in the subject. — Pedro Werneck, Dec 04 '14 at 15:52
@PedroWerneck Would you be able to identify for me the page number (or chapter title) in which he analyses this contradiction in detail, or an accurately/directly-quoted sentence fragment I can search for to find the place you're referring to in the book? — ChrisW, Dec 04 '14 at 15:54
@PedroWerneck It criticizes (i.e. quotes people who identify possible methodlogical faults/concerns with) epidemiology. Although "many other contradictions are analysed in detail" I didn't notice that it analyzes 'lag', at all, anywhere. — ChrisW, Dec 04 '14 at 17:02
Lol, really? "This is analyzed in this book" "Which part of the book?" "Chapter 2" "I checked and no it doesn't" "Well go read it all then!" Come on dude.. — DaleJohnston, Dec 04 '14 at 23:38
I skimmed Chapter 2 (it's 100 pages): it lists ways in which epidemiology can in theory jump to wrong conclusions (e.g. perhaps only a subset of people are susceptible to getting cancer from smoking; or perhaps smokers tend to be non-exercisers, and it's the lack of exercise that causes lung cancer; or etc.). So, that's interesting as a textbook introduction to ways in which epidemiology might be wrong. I don't think it proved that epidemiology *is* wrong in the case smoking and lung cancer (and I don't know what's in the other chapters). — ChrisW, Dec 05 '14 at 00:05
Consider that in 1975 and 1980 more people died without having been diagnosed with cancer than later years. Throw those two years out and a decline emerges. Our ability to detect cancer has increased over time, and if not accounted for would throw such statistics off. My grandmother was a 2 pack a day smoker for 70 years of her life, and in her late 90's after dementia set in a mass was detected but no further diagnosis was made as family elected for hospice care rather than cancer treatment. No treatment means she isn't included in cancer stats. — Adam Davis, Dec 05 '14 at 00:45

score 114 · Accepted Answer · edited Dec 04 '14 at 12:29

114

The cancer.gov site you referenced says that the median age of diagnosis is 70. That implies that you smoke and then, later in life/time, you may be diagnosed with cancer.

Therefore, for example, Lung cancer incidence statistics from the UK says:

Trends in lung cancer incidence rates reflect past trends in cigarette smoking prevalence. Smoking rates peaked earlier in males than in females, so lung cancer rates in men have been decreasing for some decades, but this decrease is yet to start in women.

Figure 1.2 in your cancer.gov report shows that cancer rates for all people have been decreasing steadily, since 1992, and since 1986 for men.

edited Dec 04 '14 at 12:29

Oddthinking

140,378
46
548
638

answered Dec 04 '14 at 04:36

ChrisW

26,552
5
108
141

Ah that's a great chart. It has much more data than I found. For males it does look like a clear lag on the face of it. However, it doesn't explain the increase in women's rates of lung cancer. Smoking rates have been steadily decreasing for women since 1970, which would mean the lag would have to be 40+ years... but if that's how big the lag is, lung cancer rates for males shouldn't have started going down until 1990 or later. – DaleJohnston Dec 04 '14 at 05:02
4

This `cancer.gov` page says the same thing, i.e. that there's "lag": http://www.cancer.gov/researchandfunding/snapshots/lung – ChrisW Dec 04 '14 at 05:15
Hmm alright, looks like I'll have to read some papers describing the details of the lag. That seems to be the commonly-accepted answer in any case. This one looks promising: http://cancerres.aacrjournals.org/content/44/12_Part_1/5940.full.pdf . – DaleJohnston Dec 04 '14 at 05:27
3

When you research it, you'd be most welcome to write/post an answer of your own to this question. – ChrisW Dec 04 '14 at 05:35
4

It might be a good addition to mention that the preferred CT screening is used *much* more frequently now than it was prior to the 1990s. It may not be that incidences of cancer were rising until the mid-90s, but our *detection* of them was. – Is Begot Dec 04 '14 at 14:24
1

@Geobits Is that so? Instead wouldn't that effect be negligible, because if you develop lung cancer then it's impossible to avoid detecting it (because you die of it) within relatively few years? OTOH life expectancy has increased since 1975, maybe that (longer life) increases (late-in-life) prevalence of cancer. – ChrisW Dec 04 '14 at 14:36
@ChrisW That's a good point (assuming the cause of death is known, which I assume it would be). Life expectancy does seem like a better angle, *especially* given that the median age of diagnosis is 70. I think the main issue with that is that life expectancy seems to have undergone a relatively steady climb, whereas cancer has a sizable peak in the 90s. – Is Begot Dec 04 '14 at 14:42
2

@Geobits Or you die because of heart attack due to long-term lack of oxygen, and you're never diagnosed. Or you die from "flu" which is unrecognized pneumonia that is more common and more severe if you get lungs cancer. Ad nobody cares what it was once you're dead. And so on and so forth. As well, what you mention, we get old enough now to develop various diseases that likely didn't exist before (because we were simply dying too young). There are too many factors here that _increase_ the rate in which many diseases appear. That's why one has to be really careful in statistic epidemiology. – yo' Dec 05 '14 at 11:22
Forgetting to consider this "lag" effect reminds me of the cognitive bias called "[base rate fallacy](http://en.wikipedia.org/wiki/Base_rate_fallacy)" which led to other related myths like the common belief that 1/2 marriages end up in divorces, which [have since been disproved](http://www.nytimes.com/2005/04/19/health/19divo.html?_r=0). – gaborous Dec 06 '14 at 17:01
Perhaps mention that radon gas is the second leading cause of lung cancer. – Andy Dec 13 '14 at 19:06
@Andy [This EPA page](http://www.epa.gov/radon/healthrisks.html) says 160,000 deaths/yer from smoking and 20,000 deaths/year from radon ... and that radon is the leading cause of lung cancer among non-smokers. I don't see how/why to integrate that into an answer to this question, though. The background cancer rate (due to radon instead of due to smoking) is relatively low (20,000 compared to 160,000). – ChrisW Dec 13 '14 at 19:15

If smoking causes lung cancer, why has a decline in smoking not led to a decline in lung cancer?

1 Answers1

Linked