Unsurprisingly, lung cancer was most strongly associated with smoking (accounting for 80% of deaths), followed by cancers of the mouth and throat.
It is important to note, however, that these are just estimates based on data taken from previous studies, which may have various limitations. Therefore, we cannot be certain that these figures on the proportion of cancers caused by smoking are 100% accurate – or directly applicable to the UK.
The results still make for sobering reading, with the World Health Organization (WHO) estimating that smoking kills nearly 6 million people a year worldwide, due to cancer and other diseases, such as heart disease.
This study was by researchers from the American Cancer Society in Atlanta; Harvard Medical School in Boston; National Cancer Institute in Bethesda, Maryland; and Fred Hutchison Cancer Research Center in Seattle. The analysis part of this work was funded by the American Cancer Society.
The study was published in the peer-reviewed medical journal JAMA Internal Medicine.
The Mail Online’s reporting of the study was accurate. However, one of the background quotes from the lead author – “e-cigarettes are now the most common form of tobacco use among high school students” – is open to criticism, as e-cigarettes do not contain any tobacco.
What kind of research was this?
This study was titled as a research letter, and the researchers used data from previous studies to estimate the proportion of deaths from 12 different cancers in the US in 2011 that could be attributed to smoking.
The researchers say that the 2014 US Surgeon General’s Report estimated the number of cancer deaths overall and lung cancer deaths specifically that were caused by smoking. However, they missed out the other 11 that are reported to be caused by smoking. Previous data on smoking deaths due to these cancers is said to come from 10 or more years ago. Since then, smoking prevalence has decreased, but risk of cancer among smokers can increase over time. Because of this, the study aimed to look at more up-to-date information on specific cancer deaths in 2011.
The research used data from various previous studies and surveys. Specific methods on how these studies were identified and selected is not reported in this brief publication; therefore, it is not possible to comment on whether all relevant evidence will have been considered.
A systematic review would probably have provided more detailed information. However, these types of reviews can be both expensive and time-consuming, and some research teams just don’t have the resources to carry them out.
What did the research involve?
The researchers obtained information on smoking prevalence from the 2011 National Health Interview Survey. This was based on interviews from a nationally representative sample.
Age- and sex-specific cancer risk for former and current smokers came from cohort studies that had assessed smoking in questionnaires, and then followed the people up looking for risk of cancer and cancer deaths. One data source was the Cancer Prevention Study II, which included people aged 35 to 54 years (covering the follow-up period 1982-88), and the source for other age groups was the Pooled Contemporary Cohort (follow-up period 2000-11) which has pooled data from five cohorts.
Using information from these data sources, the researchers calculated the population attributable fraction (PAF) of smoking for different cancer deaths. The PAF is the proportion of the number of deaths that are caused by smoking, or by what proportion the number of deaths would be reduced if there was no smoking.
What were the basic results?
In 2011, there were 345,962 cancer deaths in adults aged 35 or over with the 12 different cancer sites examined. The researchers estimated that 167,805, or 48.5% (95% confidence interval (CI) 46.2 to 51.2%), of these overall cancer deaths were caused by cigarette smoking. Smoking accounted for 51.5% of cancer deaths in men and 44.5% of cancer deaths in women.
By far the largest proportions of smoking-attributable deaths were cancers of the lungs and airways. 80% of these cancer deaths – broken down as 83% for men and 76% for women – were estimated to be caused by smoking. The second highest proportion was for cancers of the larynx (vocal cords), where smoking accounted for 77% (72% in men and 93% in women).
Smoking accounted for around half of all deaths of the mouth, throat and oesophagus (food pipe), and just under half of bladder cancers.
Smoking was attributable to around a quarter of cervical and liver cancers.
The remaining cancers examined where the PAF of smoking was less than 20% were those of the kidney, pancreas, stomach, bowel, and a type of leukaemia.
How did the researchers interpret the results?
The researchers conclude, “Cigarette smoking continues to cause numerous deaths from multiple cancers, despite half a century of decreasing prevalence. The smoking downturn is likely reflected in the generally lower proportions of deaths caused by smoking in 2011 than in 2000 to 2004”.
Conclusion
This study has used data from previously published cohort studies and national surveys to estimate the proportion of cancer deaths that can be attributed to smoking in men and women. They examined 12 cancers that are already known to be associated with smoking and estimated that smoking may account for half of them overall. The vast majority of cancers of the lungs and airways were estimated to be caused by smoking.
It is important to note that these are only estimates. The study has used data from cohort studies to inform the risk of different cancers in former and current smokers, and those who have never smoked. However, these cohort studies may have various inherent limitations and potential biases in their designs, which cannot be analysed here. For example:
the populations studied may not be representative of everyone
the follow-up period may be too short to capture all newly developed cancers and cancer deaths caused by smoking
they may not have taken into account other confounders (e.g. alcohol intake, diet and physical activity)
there may be inaccuracies around assessments of lifetime smoking habits
they may not have been able to assess the effects of passive smoking from environmental exposure
Specific methods on how the cohorts and national survey were identified and selected are not reported in this brief publication. It is likely that the researchers will have selected the best available and most nationally representative evidence on which to form estimates. However, this cannot be assumed, and it is not possible to comment on whether all relevant evidence will have been considered.
Another point to bear in mind is that the cancers studied were selected as they are known to be linked to be smoking. It is possible that other cancers may be associated with smoking that are currently less well recognised. It is worth again highlighting that these are estimates for the US population – not the UK.
Despite limitations in terms of whether this study provides accurate estimates on the proportion of these cancer deaths that are caused by smoking, it nevertheless reinforces the health message. Smoking is known to have many detrimental effects on health, not only on risk of cancer, but for many other chronic diseases.
As the researchers conclude, “more comprehensive tobacco control, including targeted cessation support” seem to be an important way forward.
Even if you have been a smoker for many years, quitting will still provide a tremendous health benefit. For example, 10 years after you've stopped smoking, your lung cancer risk is half that of someone who has continued to smoke.