Smoking linked to raised diabetes risk – including passive smoking

"Passive smoking raises risk of type 2 diabetes," The Guardian reports. A major new analysis of previous studies found a significant association between exposure to tobacco smoke – including secondhand smoke – and type 2 diabetes.

People who had never smoked, but were exposed to secondhand smoke, were at a 22% higher risk of developing type 2 diabetes than people who had never smoked, but had been exposed to secondhand smoke.

The study crunched data on almost 6 million people – an impressive feat – meaning it had lots of statistical power to pick out links accurately. It also having took account of many known contributory risk factors for diabetes, including diet and physical activity. The data for passive smokers came from around 150,000 people. 

The diabetes risk increase varied in line with smoking intensity and length of time a person had quit – suggesting that a direct cause and effect link is possible. A randomised controlled trial would be needed to know for sure; however, it would be unethical to allocate people to something that is known to harm.

It is unclear why smoking would increase diabetes risk. Speculations offered in the paper include the fact that smoking can increase inflammation levels and cause cell damage. Interestingly, a study earlier this week found an association between cannabis smoking and diabetes.

Giving up smoking, if you smoke, is one of the biggest steps you can take to improve your health.  

Where did the story come from?

The study was carried out by researchers from universities based in China, Singapore and the US. It was funded by the Chinese National Thousand Talents Program for Distinguished Young Scholars, US National Institutes of Health, the Chinese National 111 Project, and the Program for Changjiang Scholars and Innovative Research Team in University, from the Chinese Ministry of Education.

The study was published in the peer-reviewed medical journal The Lancet Diabetes and Endocrinology.

Generally, the UK media reported the story accurately, with most headlines focusing on the 22% risk increase attributed to secondhand smoke exposure – otherwise known as passive smoking.

What kind of research was this?

This was a systematic review and meta-analysis of different smoking behaviours and type 2 diabetes.

Smoking remains the biggest cause of self-inflicted death and disease in the world, killing 6 million people each year and causing a higher proportion of a smoker’s life to be lived in poor health than non-smokers.

Many studies have suggested links between different smoking behaviours – active smoking, passive smoking, and being an ex-smoker – with a higher risk of developing type 2 diabetes. This systematic review pooled all the studies it could find on the issue in an effort to better understand the link.

A systematic review and meta-analysis is one of the best ways to summarise the results of many different studies. Pooling results of similar studies creates more reliable and accurate estimates of any links. However, the pooled results are only ever as good as the studies that feed into them. If you put rubbish in, you get rubbish out.

What did the research involve?

The researchers identified 88 prospective studies containing 5,898,795 people, 295,446 of whom developed type 2 diabetes during the study periods. Where possible, they pooled the study findings into summary estimates of how different smoking behaviour was linked to the risk of developing type 2 diabetes.

The team systematically searched electronic databases to identify relevant studies with a prospective design. This means that smoking behaviour was known before people developed type 2 diabetes. This eliminates the risk of reverse causation – where people with diabetes may be more likely to smoke.

Each study was rated for quality, and this took into account whether the studies adjusted for lifestyle variables – such as diet, alcohol intake and physical activity – that could influence diabetes risk independently of smoking behaviour. Studies with substantial loss to follow-up (>50%) were excluded – this is a way of selecting only the more reliable studies.

The main analysis estimated links between current smoking, former smoking and passive smoking, and the risk of developing type 2 diabetes. The study sample was very large, so the researchers were able to analyse the effects of many subgroups. This includes, for example, the effect of smoking intensity, time since someone quit smoking, ethnicity, blood pressure, diet, physical activity, alcohol, and study location, among others.

What were the basic results?

Follow-up times varied between studies and about a third of participants had long-term follow-up, lasting over 10 years.

Current smoking, former smoking and passive smoke exposure in people who had never smoked themselves were all consistently linked with a higher risk of developing type 2 diabetes.

The following results were found:

  • Current smokers were 27% more likely to develop type 2 diabetes than current non-smokers (relative risk [RR] 1.37, 95% confidence interval [CI] 1.33 to 1.42) based on 84 studies, totalling 5,853,952 people)
  • Former smokers were 14% more likely to develop type 2 diabetes than those who had never smoked (1.14 95% CI 1.10 to 1.18), based on 47 studies with 2,930,391 people)
  • Those who had never smoked, but had been exposed to passive smoke, were 22% more likely to develop type 2 diabetes than those who had never smoked (RR 1.22, 95% CI 1.10 to 1.35, based on seven studies with 156,439 people)

The risk of diabetes increased in proportion to the amount smoked, adding weight to a possible causal link. Compared with those who had never smoked, the relative risks were 21% higher (1.21, 95% CI 1.10 to 1.33) for light smokers, 34% higher (1.34, 95% CI 1.27 to 1.41) for moderate smokers, and 57% higher (1.57% 95% CI 1.47 to 1.66) for heavy smokers.

The risk also started to decrease in proportion to the time since a person kicked the habit – another signal that the link may be causal. Compared with those who had never smoked, new quitters (less than five years since they quit smoking) were at a 54% raised risk of type 2 diabetes (RR 1.54, 95% CI 1.36 to 1.74), 18% for middle-term quitters (5-9 years, RR 1.18 95% CI 1.07 to 1.29) and 11% for long-term quitters (10 years or more, RR 1.11, 95% CI 1.02 to 1.20). These results came from 10 studies with 1,086,608 participants.

Based on the assumption that the association between smoking and diabetes risk was 100% causal – that is, that all the diabetes risk increase was due to smoking – they estimated that 11.7% of type 2 diabetes cases in men and 2.4% in women were attributable to active smoking. This amounts to around 28 million cases worldwide.

How did the researchers interpret the results?

The study group concluded that: "Active and passive smoking are associated with significantly increased risks of type 2 diabetes. The risk of diabetes is increased in new quitters, but decreases substantially as the time since quitting increases. If the association between smoking and risk of type 2 diabetes is causal, public health efforts to reduce smoking could have a substantial effect on the worldwide burden of type 2 diabetes".


This large, robust systematic review and meta-analysis of prospective studies shows a consistent and dose-responsive link between smoking and a higher risk of developing type 2 diabetes. This is suggestive of a causal link. This included exposure to secondhand smoke through passive smoking – a link that grabbed the media’s attention.

The study crunched data on almost 6 million people, meaning it had lots of statistical power to pick out links, having taken account of many known confounders.

The results were consistent and the increases in diabetes risk linked to smoking varied in line with smoking intensity and length of time a person had quit. While prospective studies cannot prove cause and effect, these findings do hint at one. A randomised control trial would be needed to know for sure, but is not feasible, as it would be unethical to allocate people to smoke, due to its known health effects.

A comment article published alongside the Lancet study says "smokers tend to have lower average educational attainment, worse diets, lower physical activity levels, and greater alcohol consumption than non-smokers". This signals that smokers are generally less healthy than non-smokers. This underlying unhealthiness could account for some of the diabetes risk increase – an example of residual confounding. How much of the risk increase is due to this underlying unhealthiness and how much is due to smoking is not easy to define.

The article also reminded us: "we cannot say definitively, based on the existing evidence, that smoking directly increases diabetes risk". 

Though the link appears clear for passive smoke exposure, it is also worth noting that self-reported passive smoke exposure could have covered various intensities of smoke exposure. This result was based on seven studies – three from the US, two from Europe, one from Korea and one from Japan. The specific questioning to establish passive smoking status is not reported. For example, some people could have meant they have been extensively exposed to smoke in their homes throughout their lives, while others could just have been referring to being exposed to passive smoke in public places occasionally. Therefore, although the link seems clear, the 22% increased risk estimate may be imprecise and could not easily be applied to particular individuals with passive smoke exposure.

Overall, while there is not a conclusive proof that passive smoking can increase diabetes risk, the harms of exposure to smoke, such as increased cancer risk, are well established.

NHS Attribution