Analysing Smoking Harms – New statistics using up to date evidence

The challenge of measuring smoking related diseases and deaths can be complex. Craig Collins, Service Manager, Health Harms Intelligence, explains the reason behind the change in measuring smoking harms. 

Smoking kills. It is uniquely lethal, increasing the risk of developing certain cancers, respiratory diseases, circulatory diseases, mental health and other conditions.  

However, measuring the harms of smoking is complex. This is because smoking is linked to a wide range of diseases and conditions, many of which can also be caused by other factors unrelated to smoking. As a result, it’s rarely possible to attribute any single death entirely to smoking, making it challenging to quantify its true impact on health and mortality. 

How do we measure smoking harm 

The solution is to estimate what fraction of all hospital admissions and deaths due to conditions where smoking is a known risk can be attributed to smoking. These are called Smoking Attributable Fractions (SAFs). 

Our SAFs analysis looks at the admissions or deaths for the conditions where smoking is a known cause. It combines an evidence-based risk factor (or relative risk) for each of these conditions with the smoking prevalence rate, taken from the Scottish Health Survey for a given year. This is then used to calculate the estimated annual number of admissions or deaths due to these conditions that can be attributed to smoking.  

Keeping up to date with new evidence  

New research is constantly being added to the evidence base in public health. It’s crucial that official statistics move with the times and reflect new evidence available. The new smoking related deaths statistics that we are publishing today are based on the latest evidence on smoking harms. 

Our previous method of calculating SAFs was based on a key report by the US Surgeon General in 2004 which used evidence from 1982 to 1988. Our updated methodology is based on a 2018 Royal College of Physicians (RCP) report, which incorporates more recent research and analyses on the effects of smoking.  

Updating our published statistics  

The SAFs published today produced under the new methodology are not comparable with figures produced using the previous approach. Earlier estimates, which were published on the ScotPHO website, should not be compared to SAF estimates released today. Under the new methodology, annual estimations of smoking attributable hospital admissions and deaths are lower than those produced under the old method by approximately 16% and 17% respectively. 

This should not be considered a suggestion that smoking is somehow less harmful to individual health but instead reflects a range of improvements in public health and clinical treatment. With the updated methodology, we have improved the accuracy of the estimate of deaths and admissions that are attributable to smoking.  

Smoking is still as harmful as it has ever been, but the context has changed since 2004. For example, the ban on smoking in enclosed public places has reduced exposure to second-hand smoke by 96% over the past 20 years. People who smoke are also smoking less on average, with daily cigarette use falling from around 15 a day in 2003 to 11 in 2025. 

At the same time, some smoking-related cancers are being found earlier, and treatments have improved. Taken together, these changes at a population level may help explain shifts in how we estimate the risks of smoking. 

Consistency across the UK 

The new method is the same as that used by England and Wales, meaning Scottish figures are now comparable with other UK nations. This allows us to make comparisons across the different countries of the UK and get a broader and more accurate picture of smoking related harms.  

Further information 

Read the report on Smoking Attributable Fractions here. Further information on PHS’ work in this area can be found on the PHS website. 

For those looking for support to quit smoking Quit Your Way is an advice and support service for anyone trying to quit smoking.