Heart and lungs

Women's smoking death risk 'highest ever'

“Women smokers are five times more likely to be killed by their habit today than in the 1960s,” The Sun has reported, while BBC News reports that “female smoking death risk ‘has soared’”.

These headlines are based on a recent study that assessed the trends in mortality between smokers and non-smokers in the US across several decades.

The authors of the research found an increasing difference in mortality between women smokers and women non-smokers since the 1960s (this means that compared to their risk in the 60s, women who smoke today appear to have even higher risk of dying compared to their non-smoking counterparts).

The increase could be explained by what could be termed the ‘Mad Men’ effect – increasingly from the 60s, women’s smoking habits have been more like men’s in that they are starting at a younger age and smoking more per day. As the authors put it:  “Women who smoke like men, will die like men”, that is they have been dying more commonly of lung cancer, heart disease and stroke.

Risk of death among non-smokers may also be decreasing due to advancements in treating common diseases. So non-smokers may be living longer, while smokers’ lifespans are remaining relatively short.

The good news is that a related study found that quitting at any age dramatically reduced death rates, and quitting before the age of 40 reduced the risk of smoking-related deaths by 90%.

Overall, the results of this study are hardly earth-shattering – smoking is still bad for you and quitting smoking is probably the best thing you can do for your health.

Where did the story come from?

The study was carried out by researchers from the American Cancer Society, University of Queensland in Australia, and other research institutes. The research was funded by the US National Institutes of Health and the American Cancer Society.

The study was published in the peer-reviewed New England Journal of Medicine.

Media headlines accurately report the statistics of this study with the exception of Metro, which featured the story prominently on its front page claiming that the rise in smoking-related deaths is due to the popularity of so-called light cigarette brands. While this could be the case, there is nothing in the current study to support this speculation.

The media also focuses primarily on relative risks, which are difficult to interpret (especially over time) without additional information.

News coverage should not be interpreted to mean that smoking is getting more dangerous, especially when only changes in relative risk are covered in the story. Rather, smoking is as dangerous as ever, but more women are doing it compared to the years prior to the 1960s.

What kind of research was this?

This study analysed data from several different cohort studies to identify trends in the risk of death among smokers compared to non-smokers over the course of several decades. The data was pulled from one study conducted from 1959 to 1965, one from 1982 to 1988, and five studies carried out between 2000 and 2010.

As a comparison of cohort data, this research cannot show that smoking directly caused the deaths among the participants in this study, only that there is an association. That is not to say that a definite link between smoking and death has not been shown, only that the data used in this research did not have the power to look at definite cause-and-effect.

What did the research involve?

The cohort studies included nearly 900,000 men and 1.3 million women from the US. The researchers classified study participants as ‘current smokers’, ‘former smokers’ and ‘never smokers’. They then calculated the death rates for each smoking group, and each time period. They then calculated the risk of dying among current smokers compared to people who never smoked.

Risks were calculated for several outcomes, including overall mortality (death from any cause), and death due to specific smoking-related illnesses (such as lung cancer), and compared across the three time periods (1960s, 1980s and 2000s).

These figures were adjusted for potential confounding factors, including:

  • how much and how long current smokers have been smoking
  • age at quitting among former smokers
  • ethnicity
  • education level

These statistics were calculated for both men and women.

What were the basic results?

When examining trends in all-cause mortality among current, former and never smokers, the researchers found that among both men and women who never smoked all-cause mortality rates were 50% lower among the 2000s cohort compared to the 1960s cohort (among both men and women).

Among women who were classified as current smokers, there was no decrease over time in the all-cause mortality rate:

  • in the 1960s there were 3,225 deaths per 100,000 current smokers
  • in the 1980s there were 2,954 deaths per 100,000 current smokers (not significantly different to the 1960s)
  • in the 2000s there were 3,016 deaths per 100,000 current smokers (not significantly different to the 1980s)

Among women who were classified as never smokers, there was a decrease over time in the all-cause mortality rate:

  • in the 1960s there were 2,884 deaths per 100,000 never smokers 
  • in the 1980s there were 1,741 deaths per 100,000 never smokers (significantly lower than the 1960s)
  • in the 2000s there were 1,248 deaths per 100,000 never smokers (significantly lower than the 1980s)

Over time, this leads to an increasing relative risk (RR) in all-cause mortality in among female smokers compared with female never smokers (after adjusting for confounding factors):

  • 1960s RR 1.35 (95% confidence interval [CI] 1.30 to 1.40)
  • 1980s RR 2.08 (95% CI 2.02 to 2.14)
  • 2000s RR 2.76 (95% CI 2.69 to 2.84)

When deaths due to lung cancer among female current smokers compared with female never smokers are considered, the differences in both absolute and relative risk are more pronounced. The absolute risk of death due to lung cancer increased significantly among currently smoking females over the time periods (in the 1960s there were 30 deaths per 100,000, in the 1980s there were 292 deaths per 100,000, in the 2000s there were 506 deaths per 100,000).

A significant but more modest increase in absolute risk was seen among female never smokers over the same time period (in the 1960s there were 18 deaths per 100,000, in the 1980s there were 28 deaths per 100,000, in the 2000s there were 22 deaths per 100,00). This substantially higher rate of lung cancer among current smokers compared to a relatively steady mortality rate among never smokers leads to the trend of increasing relative risks reported in the media:

  • 1960s RR 2.73 (95% CI 2.07 to 3.61)
  • 1980s RR 12.65 (95% CI 11.15 to 14.34)
  • 2000s RR 25.66 (95% CI 23.17 to 28.40)

Among men, the patterns were slightly different, with all-cause mortality rates decreasing among both current and never smokers. However, rates dropped more substantially in never smokers, leading to a similar increase in relative risks seen in women smokers compared with women who were never smokers. The significant increases in both absolute and relative risk of death due to lung cancer peaked during the 1980s among male smokers, and are not significantly different among today’s smokers to the previous generation.

How did the researchers interpret the results?

The researchers conclude the risk of death (in both absolute and relative terms) is increasing among female smokers, and the rates seen today are nearly identical to those seen among males.


This large-scale research adds to the already prodigious amount of evidence on the risks that come from smoking. This research estimates the risk of death among smokers, and examines trends in this risk over time and by gender. There are difficulties, however, in teasing apart just what the figures mean.

Multiple risk factors contribute to risk of death

Several factors influence the relative risk of death between smokers and non-smokers, including background rates of mortality from smoking-related illnesses. An increase in relative risks of death from smoking-related illnesses could be attributable to certain smoking behaviours (such as increased smoking among females, or changes in types of cigarettes smoked) that increase the risk among smokers. However, in the case of all-cause mortality, this relative increase appears to be due to a reduction in overall mortality among non-smokers, as opposed to an increase in overall mortality among smokers.

In the case of lung cancer, however, this relative increase appears to be due to a steep increase in the absolute risk of lung cancer death among female smokers, compared to a more modest increase in female non-smokers.

Difficulty directly comparing the data

In addition to difficulties in interpreting the figures presented in the media, there are some limitations to the research methods that should be considered. For instance, each of the cohort studies assessed smoking status at different time points. While the 2000-10 studies updated smoking status information throughout the course of the study, the older studies collected information on smoking status at the start of the research period only. This may lead to a misclassification of the participants, as smoking status could have changed over the course of the study, with current smokers quitting, former smokers relapsing, or never smokers starting.

Will this apply in the UK?

It is also important to remember that these figures came from a US population, and that these estimates are essentially differences between observed disease rates and those that would be expected given general population rates. As disease and mortality differ between countries and populations, the differences in absolute and relative risk of dying will differ as well.

The limitations of the study and difficulty in interpreting the data should not be taken to mean that smoking isn’t as bad for you as the headlines are making it out to be. Indeed, most media coverage also referenced another study on the associations between smoking and mortality (also published in the New England Journal of Medicine this week), which concluded that quitting before the age of 40 “reduces the risk of death associated with continued smoking by about 90%”.

Put another way “because the absolute risks of continuing to smoke are large, the absolute benefits of cessation will also be large”.

Smoking doesn't just kill you

Finally, it is also important to remember that the outcome assessed in this research was restricted to mortality. There are, however, other factors to consider, especially surrounding living with chronic illnesses such as cardiovascular disease, chronic obstructive pulmonary disease and the effects of stroke and other smoking-related illnesses. So even if smoking does not kill you it can cause your quality of life to suffer.

Overall, while the statistics of this study are complicated and reporting of the results is somewhat complex and varied, the take-home message is neither complicated nor different than that given for many years:

Analysis by *NHS Choices

. Follow Behind the Headlines on Twitter*.

NHS Attribution