TEG horizontal logo 2

Modelling for the smokefree generation policy


This report explains the methodology and data used for the Markov model that we constructed to model the effects of the smokefree generation policy for the command paper Stopping the start: our new plan to create a smokefree generation.

The modelling is for England only and focuses on the 14 to 30 age group, given the primary aim is to further reduce the number of young people taking up smoking (the ‘instigation rate’).

To assess the longer-term impacts on disease incidence, we have modelled the lifetime effects of changes in the instigation rate on disease incidence, mortality and costs, taking into account subsequent smoking behaviours (quitting and relapse).

This analysis focused on:

  • smoking rates
  • the health impacts of smoking
  • costs of smoking (including productivity, healthcare, social care and smoking-related fire costs)

Costs of implementing the policy were not in scope. We will include an analysis of these, as well as a further iteration of this modelling, in the forthcoming impact assessment for this policy.

In developing the model, we have made assumptions based on the best evidence available which influence the results. Also, while a Markov model is a widely used approach for considering smoking behaviour, there is inherent uncertainty in projecting analysis decades into the future. These factors mean that this work should not be considered a precise forecast, but rather an attempt to assess the scale of potential effect. There is further information about the limitations of the model later in this report.

Model structure

The York Health Economics Consortium (YHEC) defines the Markov model as follows:

The Markov model is an analytical framework that is frequently used in decision analysis, and is probably the most common type of model used in economic evaluation of healthcare interventions. Markov models use disease states to represent all possible consequences of an intervention of interest. These are mutually exclusive and exhaustive and so each individual represented in the model can be in one and only one of these disease states at any given time.

Time itself is considered as discrete time periods called ‘cycles’ (typically a certain number of weeks or months), and movements from one disease state to another (in the subsequent time period) are represented as ‘transition probabilities’.

Figure 1: our Markov model structure

Figure 1 is a diagram of our Markov model structure. It divides the population (aged 13 to 89) into 4 states, based on smoking status:

  • non-smokers
  • current smokers
  • former smokers
  • people who are dead

Each cycle of the model is one year, and people can either remain in one of the above states or move to another at each cycle.

People enter the model as non-smokers. If a non-smoker starts smoking, this is known as instigation. Current smokers who quit become former smokers, and if they remain abstinent they eventually move back to being non-smokers (called ‘long-term quitting’ in the model). Former smokers can also relapse. In the model, people die from:

  • smoking-related causes (from current smoking or a history of smoking)
  • other causes, not related to smoking

The model runs from 2023 up to 2100, to assess the long-term impacts on disease incidence, mortality and costs, acknowledging there is greater uncertainty the further into the future the analysis projects.

Transitions between states in the model

Each year, a new set of 13 year olds enters the model as non-smokers. Given the small numbers of under 14s who smoke, after the initial starting population the model assumes no 13 year olds smoke. The number of 13 year olds in England each year is taken from the Office for National Statistics (ONS) 2020-based interim national population projections: year ending June 2022 estimated international migration variant.

We have applied an average of figures over the time period 2026 (the year before the introduction of the policy) to 2050 each year, rather than modelling each year separately. We thought this was reasonable because year-on-year changes are small. So, each year 333,573 males and 316,014 females enter the model at 13 years old.

Each year, people from any state (non-smoker, current smoker or former smokers) can die, although the probability of dying (by age and sex) is different for each state. We have taken baseline mortality rates from the ONS National life tables: UK for 2017 to 2019, to avoid the impacts of COVID-19. We then disaggregated this based on the proportion of smokers who are current, former and ‘never’ smokers and combined with increased relative risks of mortality related to smoking based on data from research on mortality in relation to smoking (Doll and others, 2004). We split data on mortality risks for former smokers by the age of quitting. However, we took an average of these figures for the purposes of this model.

Aside from mortality probabilities, each year non-smokers can either instigate smoking (up to the age of 30) and transition to the ‘current smoker’ state or remain in their existing ‘non-smoker’ state. Current smokers can either quit smoking and transition to the ‘former smoker’ state or remain in their existing state. Former smokers can relapse, returning to being current smokers, remain in their current state, or ‘long-term quit’, which means they move back to being a ‘non-smoker’, as outlined below.

Baseline transition probabilities for instigation (becoming a smoker), quitting (successfully quitting smoking for one year) and relapse (becoming a smoker again after having quit) are taken from the University of Sheffield’s tobacco policy model, using figures available at the time of analysis. More information is available on the Smoking state transition probabilities on the Sheffield tobacco and alcohol policy modelling platform website. University of Sheffield provides data by deprivation quintile, which we converted to an overall figure by calculating a weighted average using the population of smokers in each deprivation decile (from the Office for Health Improvement and Disparities (OHID) Local tobacco control profiles) and assuming each decile had an equal population size.

For the baseline analysis, we have held instigation, quit and relapse rates constant at 2023 values. University of Sheffield projected rates changing over time up to 2030 (at the time of this analysis) but these changes assume some further policy action on smoking. Without this, it is unclear how instigation, quit and relapse rates would change. While smoking overall has been declining in recent years and this trend could be expected to continue, it is plausible that with no action, smoking rates could stall or even rise, as has been seen in Australia and in New York in the USA. So, we assume that instigation, quit and relapse rates remain constant at 2023 values. This results in baseline trends over the coming years that are broadly in line with other estimates from Cancer Research UK’s Smoking prevalence projections for England based on data to 2021 (pdf, 666kb) and University of Sheffield’s projections from 2021, published in the Royal College of Physicians report Smoking and health 2021: a coming of age for tobacco control? The trends then reach a long-run steady state of smoking prevalence that is lower than current levels of smoking (once the starting population has aged out of the model)

Using University of Sheffield’s data instigation, quit and relapse rates were available from the age of 16 (at the time of constructing this model). For our analysis we also calculated instigation rates for 13 to 15 year olds. We did this by taking instigation rates for 14 to 16 year olds from the US SimSmoke model (available to download from the US National Cancer Institute Publication Support and Modeling Resources website), and using these to adjust the Sheffield rates, by assuming the ratio between age groups in the US model applies to our population. For example, SimSmoke suggests 2.4% of 15 year old male non-smokers instigate and 3.1% of 16 year old males instigate. We then divided the Sheffield 16 year old male instigation rate by 2.4 divided by 3.1 to calculate a 15 year old male instigation rate. For 13 year olds, we assumed rates were equal to 14 year olds, as outlined below. We applied long-term quit probabilities (described below) from the age of 24, as they are only relevant for individuals who quit smoking more than 10 years ago.

Each cycle in the model lasts one year. So, transitions between states can only occur between ages. For example, a 17 year old non-smoker who instigates smoking becomes an 18 year old smoker the next year. The model uses rates for a given age to calculate transitions at the end of that year, for example the 17 year old instigation rate is used to calculate those moving to the current smoking state at age 18. When rates are modelled to change over time, the year of the rate used is the year to which it is applied. For example, for a 17 year old becoming a smoker at age 18 in 2030, we used the 2030 instigation rate.

Given the above, when calculating instigation rates for 13 year olds we assumed this would be equal to the rate for 14 year olds. Although we know considerably fewer 13 year olds smoke than 14 year olds, the model applies this number to next year’s 14 year olds and assumes no-one 13 or below smokes.

The ‘former smoker’ state is intended to capture only people who quit smoking less than 10 years ago. A modelling study on risks and mortality (Kontis and others, 2014) shows that 10 years after stopping smoking, the excess risk of cancers and chronic obstructive pulmonary disease (COPD) is less than half that of a smoker, and for cardiovascular diseases is close to zero. Research on long-term smoking relapse (Hawkins and others, 2010) suggests relapse is negligible after 10 years of abstinence. So, the model applies a probability called ‘long-term quit’ to estimate the proportion of people who quit smoking less than 10 years ago (and remain abstinent) who have reached 10 years of abstinence. The model moves these people to the ‘non-smoker’ state, assuming they have the same health risks as ‘never smokers’.

This is a simplification that will underestimate the health consequences of having been a smoker, so will underestimate the effect of the policy to some extent. While the highest relative health risks are among people who quit smoking more recently, analysis of lung cancer, stroke, coronary heart disease (CHD) and COPD incidence data from the 2019 Global Burden of Disease study shows that the main health conditions that can be caused by smoking tend to accrue more in older age. Analysis of the Health Survey for England 2019 data provided by the University of Sheffield shows that most older former smokers quit more than 10 years ago.

The long-term quit probability is 8.96%, calculated from previous OHID analysis simulating a cohort and using the research by Hawkins and others. It also uses the probabilities of relapse to assess the probability of having remained abstinent for 10 years from a given set of former smokers who quit up to 10 years ago. This analysis assumed a constant number of quitters each year and calculated their relapse and mortality risks each year. Then it calculated at the end of 10 years the probability that a randomly sampled person who had quit in one of the last 10 years, and had remained abstinent (and alive), would be one who had quit over 10 years ago. This is slightly less than 10% given a group of ‘non-relapsers’ will skew more towards more recent quitters but is not significantly less, since relapse becomes progressively less likely with time after quitting.

Starting population

The model starts in 2023. For the first year, a starting population (by age and sex from 13 to 89 years old) is assigned to each state. This is based on:

  • mid-population estimates from 2021 for the English population from the ONS data Estimates of the population for the UK
  • ONS data from Adult smoking habits in the UK: 2022 to determine current and former smokers who are over 18
  • data from University College London published on the Smoking in England website to determine current and former smokers 16 to 18 and under 16
  • NHS Digital data from Smoking, drinking and drug use among young people in England to determine current and former smokers 16 to 18 and under 16

The last 2 data sources are from 2021 and 2022 and we have used these to approximate the 2023 population, which may lead to slight inaccuracies.

Running the model from the components described above, we are able to estimate the numbers of people by smoking status, by age and sex per year as well as the numbers of deaths. This provides a baseline which we can compare an intervention to.

Baseline results

Applying baseline transition probabilities to the starting population gives us results for a baseline scenario of no policy intervention. This shows smoking rates decreasing in the short to medium term, in line with other published estimates from Cancer Research UK’s Smoking prevalence projections for England based on data to 2021 (pdf, 666kb) and University of Sheffield’s projections from 2021, published in the Royal College of Physicians report Smoking and health 2021: a coming of age for tobacco control?

Initial smoking prevalence in 2023 in the model among 14 to 30 year olds is 13.0%. Figure 2 shows the modelled baseline prevalence among 14 to 30 year olds from 2023 to 2050. Without any additional policy measures, baseline prevalence is estimated to decline between 2023 and 2050 from 13.0% to 8.1%.

Figure 2: modelled baseline prevalence in England among 14 to 30 year olds, 2023 to 2050

Impact of the intervention

Different impact scenarios

As we primarily assumed the smokefree generation policy to have an effect on instigation rates, we assume no changes to any other parameters, such as quitting and relapse.

On the impact of the intervention, we constructed scenarios based on available evidence and assumptions. You can see 4 different modelled scenarios below. The scenarios range from pessimistic (less than 10% year on year reduction in the instigation rate) to optimistic (90% year on year reduction in the instigation rate). Each scenario takes into account that, at least in the short term, people under the legal age of sale will still take up smoking, something that already happens today.

We modelled the smokefree generation intervention to start in 2027, with the age of sale first increasing from 18 to 19, and then increasing by one year each year thereafter.

In all scenarios, the model assumes smoking instigation rates reduce year-on-year to reflect ongoing increases in the age of sale (for example in scenario 2, rates reduce 30% in the first year, a further 30% in the second year).

Scenario 1 reflects the Institute of Medicine report Raising the minimum age of legal access to tobacco products in the US in 2015. The report projected raising the age of sale by one year to 19 would reduce rates by 10% for most age groups below the threshold, and 5% for some. This scenario also includes a small ‘rebound effect’, a 5% increase in instigation for the 2 age groups just above the age of sale threshold.

Scenario 2 assumes a 30% reduction in instigation rates per year for people below the age of sale. Reflects a projection from UCL modelling of recommendations for tobacco control in England that raising the age of sale to 21 would reduce prevalence among 18 to 20 year olds by 30% and reduce instigation rates by the same amount.

Scenario 3 assumes a 60% reduction in instigation rates per year for people below the age of sale. Reflects mid-point of scenario 2 and scenario 4.

Scenario 4 assumes a 90% reduction in instigation rates per year for people below the age of sale. Reflects the assumptions used by the New Zealand government for its implementation of a smokefree generation, which assumed a 100% reduction in instigation rates. We have modelled a 90% year on year reduction here rather than assuming smoking instigation will immediately stop.

Reflecting the policy intent, the model assumes the government would introduce the change to the age of sale gradually, one year at a time. So, we modelled changes to instigation rates at or below the current age of sale from the first year. We also modelled changes to instigation rates at ages that subsequently become under the legal age of sale each subsequent year. 

By applying these rates as an input and running the model, we can see the impact of the policy on the differences in numbers of non-smokers, current smokers and former smokers by year, age and sex. We also see differences in mortality.

Life years gained and QALYs from mortality

By looking at differences in the number of people dying when running the policy scenario (with reduced instigation rates) versus the baseline, we can determine the number of smoking-related deaths avoided.

Also, by counting the reduction in the number of people in the ‘dead’ state each year, we can work out the ‘life years gained’. Life years gained is a measure of the total number of years of extra life within the population due to the policy.

We can also estimate the quality of life lost to generate quality-adjusted life years (QALYs) lost due to mortality. QALYs are a measure of (health-related) quality and length of life, where 1 QALY represents 1 year lived in full health (a quality of life score of 1 on a 0 to 1 scale). Research has found that the mean health-related quality of life score (utility value) for the general population was 0.828 (Sullivan and others, 2011). We use this value to estimate the quality of life of the extra years lived by someone who does not take up smoking as a result of the policy, in the absence of any information about their health status. Multiplying this quality of life score by years of life gained gives us total QALYs, which we can multiply by £70,000 as per HM Treasury guidance The Green Book: appraisal and evaluation in central government to represent the monetary value of additional QALYs.

Disease cases

We have estimated the cases avoided of certain health conditions as a result of the smokefree generation policy, specifically:

  • lung cancer
  • COPD
  • CHD
  • stroke

According to Global Burden of Disease data from 2019, these 4 conditions together represent nearly 60% of the disability-adjusted life year (DALY) burden caused by smoking in England. The DALY is a measure of both the mortality and morbidity impacts of a health condition.

We carried out this calculation based on 2 inputs:

  • incidence (number of new cases) data from the Global Burden of Disease 2019 study, for England by age and sex
  • data on the relative risks of developing a disease based on smoking status, from the Royal College of Physicians (RCP) report Hiding in plain sight: treating tobacco dependency in the NHS

The RCP report suggested a relative risk for current smokers of 8.96 for lung cancer. This means current smokers are 8.96 times more likely to develop the condition than non-smokers. For former smokers, the relative risk was 3.85. The table below provides the relative risks we used, noting some of these were disaggregated by males and females and in the case of CHD by age too.

Table 1: relative risks of disease, by smoking status and sex

ConditionMale current smokersFemale current smokersMale former smokersFemale former smokers
Lung cancer8.968.963.853.85
CHD (< 35 year olds)1111
CHD (35 to 64)3.183.931.591.48
CHD (65+)1.961.951.161.37

Given the age-disaggregated risks for CHD implied no increased risk in under 35 year olds, we applied the other risks only to those over 35.


We applied estimates of the cost of smoking to the model outputs, to determine the savings from a reduction in smoking instigation.

We used estimates from Action on Smoking and Health’s (ASH) Ready Reckoner. This cost calculator assesses the annual cost of smoking of:

  • productivity costs (or costs to the economy)
  • healthcare costs to the NHS
  • social care costs to local authorities
  • the cost of smoking-related fires

Below is a summary of the methodology and data used to estimate each component.

Productivity costs

The estimate for the cost of smoking on productivity includes:

  • lost productivity due to smoking-related early deaths (valued at the income lost to people dying prematurely)
  • reduced employment levels for smokers compared to non-smokers
  • reduced earnings for smokers compared to non-smokers

The estimate for the cost of lost productivity due to smoking-related early deaths is based on:

  • the years of potential productivity lost to smoking-attributable early deaths
  • distribution of earnings from employment and self-employment in the UK

The years of potential productivity lost to smoking-attributable early deaths is based on:

  • data on smoking attributable mortality from OHID’s Local tobacco control profiles
  • labour market statistics from ONS’ Data and analysis from Census 2021
  • average remaining years in employment for non-smokers in employment from an analysis of micro data (information at the level of individual respondents) from the Understanding Society (USoc) survey
  • micro data on the distribution of earnings from the Department for Work and Pensions’ Family resources survey

The estimates for the costs of smoking to productivity from reduced employment levels and earnings are based on data from the USoc survey. The data from the USoc survey is used in regressions to estimate the relationship between earnings, employment and smoking status. The analysis attempts to control for other factors that affect people’s earnings and likelihood of being employed, such as age, sex, ethnicity and education.

Healthcare costs

ASH estimates for the healthcare costs of smoking to the NHS is based on the estimate by the Department of Health and Social Care (DHSC) 2017 policy paper Towards a smoke-free generation: a tobacco control plan for England. These estimates are combined with new estimates from Public Health England for hospital admissions attributable to smoking, as outlined in its response to consultation on proposed changes to the calculation of smoking attributable mortality and hospital admissions.

Given the DHSC estimate was from 2015, ASH made further adjustments to account for recent changes in:

  • NHS costs
  • population sizes
  • distribution of ex-smokers

Social care costs

The costs of smoking to social care covers the cost to local authorities of having to provide both care in a person’s home (domiciliary care) and residential care. The cost is estimated based on data on smoking status and receipt of social care services from 2 English datasets, which are the:

  • English Longitudinal Study of Ageing
  • Health Survey for England

The data from these datasets is used in regressions to estimate the relationship between smoking status and the need for social care. The analysis attempts to control for other factors that affect a people’s use of social care, such as age, sex, family composition and health status.

Fire costs

The cost of fires caused by smoking includes the cost of:

  • fatalities
  • injuries
  • property damage
  • responding to fires

The estimates for each component are largely based on data from Home Office Fire statistics data tables and report Economic and social cost of fire.

Calculating unit costs

To calculate a unit cost (the cost for each current or former smoker, except for fires where we only calculate costs for current smokers), we divided the 4 main categories of costs by the number of current and former smokers.

For healthcare, social care and productivity costs, we divided them by the total of all current and former smokers. Our reasoning was that health, social care and employment consequences of smoking can accrue after a person stops smoking. For fires, we divided only by current smokers.

The result of this were average costs by current and former smoker of:

  • £824 for productivity losses per year
  • £109 for healthcare per year
  • £66 for social care per year
  • £56 for smoking-related fires per year (current smokers only)

By applying these figures to the differences in current and former smokers from the model, we can estimate the cost savings due to the intervention.


The tables below show the results under each scenario.

Table 2 shows smoking prevalence among 14 to 30 year olds. Smoking prevalence drops from 13.0% today to 8.1% by 2040 in the baseline, a level it remains at in 2050. Each scenario results in a significant reduction in prevalence, with prevalence close to or below 1% in 3 of the 4 scenarios by the early 2040s.

Table 2: smoking prevalence among 14 to 30 year olds

Scenario 1: < 10% reduction in instigation rate13.0%8.4%5.1%3.1%
Scenario 2: 30% reduction in instigation rate13.0%7.3%1.3%0.0%
Scenario 3: 60% reduction in instigation rate13.0%6.4%0.6%0.0%
Scenario 4: 90% reduction in instigation rate13.0%6.1%0.4%0.0%

Table 3 shows results versus the baseline scenario, illustrating the many thousands of cumulative smoking-related deaths avoided (across the whole model population over time). Given the smokefree generation policy is targeted at younger people, we start to see the full impact on deaths towards the second half of this century. We show the results below up to 2100.

Table 3: cumulative smoking-related deaths avoided

Scenario 1: < 10% reduction in instigation rate12811,46670,205
Scenario 2: 30% reduction in instigation rate35923,925118,447
Scenario 3: 60% reduction in instigation rate53927,399126,829
Scenario 4: 90% reduction in instigation rate63428,688129,593

Similarly, table 4 and table 5 below show results in 2075, just under 50 years after the implementation of the policy. They show the cumulative number of disease cases avoided across the modelled population. They include the:

  • 4 conditions modelled (first table)
  • impact on healthcare costs, social care costs, the cost of smoking-related fires and productivity
  • value of QALYs gained due to a reduction in mortality (second table)

Table 4: cumulative cases of selected smoking-related disease avoided by 2075

ScenarioLung cancerStrokeCHDCOPD
Scenario 1: < 10% reduction in instigation rate2,3311,50715,44528,415
Scenario 2: 30% reduction in instigation rate5,0822,95831,97857,854
Scenario 3: 60% reduction in instigation rate5,9793,29336,29465,142
Scenario 4: 90% reduction in instigation rate6,3283,40937,82867,757

Table 5: cumulative social value gained (total savings, £ billion, against baseline in 2075, undiscounted)

ScenarioHealthcare costsProductivity gainsSocial care costsCosts of smoking-related firesQALY gains [note 1]Totals
Scenario 1: < 10% reduction in instigation rate£7£49£4£2£5£67
Scenario 2: 30% reduction in instigation rate£10£79£6£3£12£111
Scenario 3: 60% reduction in instigation rate£11£83£7£3£14£119
Scenario 4: 90% reduction in instigation rate£11£85£7£3£15£121

Note 1: the monetised value of QALY gains presented here are from mortality only, meaning we have not considered changes to quality of life.


This analysis used a model to help understand (among uncertainty) the extent of some of the likely consequences of the smokefree generation policy. In developing the model, we made assumptions and simplifications, so it has limitations.

Potential underestimation

Some elements of the model likely underestimate the impacts. For example:

  • we assumed that former smokers who quit 10 or more years ago have the same risk profile as non-smokers and the model only applies per person risk and cost figures based on former smokers in general to those who quit more recently
  • the model assumed the policy only impacted on instigation rates rather than any further effects like people smoking less
  • the model calculated health outcomes only in terms of mortality and the onset of some smoking-related diseases – this includes QALY calculations that refer only to mortality effects, so do not include the considerable quality of life impacts of smoking-related morbidity

So, as well as other diseases, the analysis does not include other health consequences of smoking, including 2 areas where outcomes are particularly poor for younger people:

1. Smoking during pregnancy, which is a major cause of:

  • stillbirths (Flenady and others, 2011)
  • low birth weight (Selvaratnam and others, 2023)
  • impairment of childhood lung development (McEvoy and Spindel, 2017)

Local tobacco control profiles shows that the prevalence of smoking in pregnancy is high for the under 18 age range, at 31.8%, and the 18 to 19 age range at 31.2%.

2. Passive smoking, which can cause all the harms of smoking, although at lower levels. Children exposed to parental and household smoking are more likely to become regular smokers. Smoking, drinking and drug use among young people in England shows that in 2021, 52% of pupils reported being exposed to secondhand smoke in a home or in a car.

As well as these limitations, QALY calculations refer only to mortality effects, so do not include the considerable quality of life impacts of smoking-related morbidity.

Potential overestimation

On the other hand, the model may overestimate effects in some areas. It does not consider the effects of smoking policies on vaping, so does not including potential detrimental health effects of increased vaping (although neither does it consider the effects of vaping-related government policy). It also relies on ASH estimates on the cost of smoking. These estimates are the best available that we know, but they may potentially overstate the effect of smoking on employment and earnings as well as the effect on social care. They also do not include all quantifiable costs of smoking, which would offset this to some extent.

So, for example, their productivity loss regression analysis controls for age, ethnicity and education, but does not control for all aspects of deprivation, which is correlated with higher smoking rates. It is possible that some factors related to deprivation may result in both reduced earnings and higher smoking rates, but those reduced earnings are not due to smoking.

Also, we applied societal costs of smoking per person to the whole modelled population of current smokers and former smokers (who quit up to 10 years ago). So, we modelled these to accrue earlier in life than when they might occur in reality, given these costs predominantly arise in older age.

Costs of living longer

The model does not include the costs incurred in remaining alive longer. This is standard practice for health economic analysis. And in line with the National Institute for Health and Care Excellence (NICE) guidance NICE health technology evaluations: the manual, we have not included costs unrelated to the conditions of interest. However, it is true that there will be additional costs for people who live longer, even excluding government payments like pensions that represent a transfer between parties and do not constitute a societal cost. We have not estimated the extent of these costs here. People who live longer will also contribute to society, and this is not captured beyond direct productivity impacts either.  

Limitations in the structure of the Markov model

There are limitations, too, in the structure of a Markov model. Markov models only measure changes each cycle, and only look at the aggregate numbers of people in each state. It is not possible to measure an individual and their history in a Markov model. For example, it is not possible to apply a relative risk of disease function to people who stop smoking based on years since quitting.

Other limitations

Other more minor limitations exist, such as the model not including smokers under 14 or over 90, nor the effects of population growth or migration.

Other uncertainties

More generally, there is inherent uncertainty in the analysis, including uncertainty:

  • over the impact of the policy
  • over the baseline trends in smoking
  • in forecasting far into the future

It is not possible to overcome these points without further research. So, this analysis should be considered an attempt to assess the scale of potential effect, rather than provide a precisely accurate estimate.


Source: Department of Health & Social Care

Kongsikan | Share


Bincang | Discuss