Article Text

Impact of increasing tobacco taxes on working-age adults: short-term health gain, health equity and cost savings
1. Christine L Cleghorn,
2. Tony Blakely,
4. Frederieke S van der Deen,
5. Nhung Nghiem,
6. Linda J Cobiac,
7. Nick Wilson
1. Burden of Disease Epidemiology, Equity and Cost Effectiveness Programme, University of Otago, Wellington, New Zealand
1. Correspondence to Dr Christine L Cleghorn, Burden of Disease Epidemiology, Equity and Cost Effectiveness Programme, University of Otago, Wellington, Dunedin 9016, New Zealand; cristina.cleghorn{at}otago.ac.nz

## Abstract

Objective The health gains and cost savings from tobacco tax increase peak many decades into the future. Policy-makers may take a shorter-term perspective and be particularly interested in the health of working-age adults (given their role in economic productivity). Therefore, we estimated the impact of tobacco taxes in this population within a 10-year horizon.

Methods As per previous modelling work, we used a multistate life table model with 16 tobacco-related diseases in parallel, parameterised with rich national data by sex, age and ethnicity. The intervention modelled was 10% annual increases in tobacco tax from 2011 to 2020 in the New Zealand population (n=4.4 million in 2011). The perspective was that of the health system, and the discount rate used was 3%.

Results For this 10-year time horizon, the total health gain from the tobacco tax in discounted quality-adjusted life years (QALYs) in the 20–65 year age group (age at QALY accrual) was 180 QALYs or 1.6% of the lifetime QALYs gained in this age group (11 300 QALYs). Nevertheless, for this short time horizon: (1) cost savings in this group amounted to NZ$10.6 million (equivalent to US$7.1 million; 95% uncertainty interval: NZ$6.0 million to NZ$17.7 million); and (2) around two-thirds of the QALY gains for all ages occurred in the 20–65 year age group. Focusing on just the preretirement and postretirement ages, the QALY gains in each of the 60–64 and 65–69 year olds were 11.5% and 10.6%, respectively, of the 268 total QALYs gained for all age groups in 2011–2020.

Conclusions The majority of the health benefit over a 10-year horizon from increasing tobacco taxes is accrued in the working-age population (20–65 years). There remains a need for more work on the associated productivity benefits of such health gains.

• economics
• prevention
• price
• smoking-caused disease

## Introduction

In previous modelling of the New Zealand population,1 tobacco tax increases were estimated to deliver sizeable health gains and health sector cost savings. However, these benefits peaked at around four decades into the future (3% discounting). Since policy-makers are often interested in the impact of interventions in the shorter term (eg, one or two election cycles), research exploring health gains in the shorter term is worthwhile. Furthermore, although a country like New Zealand does have a programme of annually increasing tobacco taxes (at around 10% per annum since 2010 and planned out to the year 2020), such plans are always vulnerable to changes of government and so may need multiple strands of supporting evidence.

Those working in the health sector should also be concerned about how to achieve a healthier population that can either support longer working lives2 or improve quality of life in retirement—without worsening inequities. The decision on when to reduce working hours and retire is multifactorial (eg, lifestyle, cultural norms) but is also influenced by health status.3 4 As governments (and their treasuries) focus increasingly on improving the health of the working-age population and consider lifting the age of entitlement for superannuation/pension payments (typically from age 65 years), more emphasis of studying policy impacts on specific age groups should be considered. Therefore, this study aims to contribute to these discussions by presenting the health and health system cost impacts of tobacco tax increases for the 20–65 year age group and the periretirement age groups of 60–64 and 65–69 years, all using a short (10 year) time horizon.

## Methods

A multistate life table (MSLT) model with 16 tobacco-related diseases that run in parallel was used to model the New Zealand population alive in 2011 (n=4.4 million) out to death or 110 years of age (whichever occurred first).1 5 6 The model is parameterised with rich national data by sex, age and ethnicity (Māori and non-Māori). Proportions of the population simultaneously resided in 16 tobacco-related disease life tables: coronary heart disease, stroke, chronic obstructive pulmonary disease (COPD), lower respiratory tract infection (LRTI) and 12 cancers. The proportion of the population in each disease life table is determined by disease incidence, case fatality and (for cancers and LRTI only) remission. The intervention changes disease incidence, which flows through to impact on mortality and morbidity to change lifetime quality-adjusted life years (QALYs) and associated lifetime health system costs. The intervention modelled was a 10% increase in tobacco tax each year between 2011 and 2020 in the New Zealand population. Further details on the tobacco tax and MSLT simulation methods are presented elsewhere.1 7

The key differences in this paper are: (a) the duration of the tax intervention was ten years (ie, 2011–2020), not to 2030; (b) we present discounted QALYs and costs. Discounting at 3% a year allows QALYs and cost savings occurring in the future to be given less weight than immediate benefits as per standard health economic practices8; (c) we focus on a 10-year time horizon (ie, for years 2011–2020) instead of the rest of the population’s lifespan (which we include for comparison only); and, most importantly, (d) we focus on QALYs and costs at age at which they accrued in the future (20–65 years and specifically 60–64 and 65–69 year olds, the 5-year age groups directly before and after the retirement age).

### Input parameters

We used a business-as-usual (BAU) comparator of no tax increases from the starting year of 2011. The BAU parameters include disease-specific incidence, prevalence, case fatality and remission (for cancers and respiratory infection) by sex, age and ethnicity in 2011 (with projected trends to 2026 and then no change). Overall morbidity by sex, age and ethnic groups was specified as ‘prevalent’ years of life lived with disability (YLDs); New Zealand Burden of Disease Study YLDs divided by the population count. Disease-specific morbidity was the total comorbidity-adjusted YLDs for that disease divided by the prevalent population, derived from the Global Burden of Disease Study.9

Individually linked administrative data (national minimum dataset, outpatient data, pharmaceutical dispensing, community laboratory test data, maternity care data, mental health data) for publicly funded health events in 2006–2010 (hospitalisations, inpatient procedures, outpatients, pharmaceuticals, laboratories and expected primary care usage) were used to estimate health system costs (in NZ$2011). Costs of both tobacco-related diseases and other diseases were modelled over the cohort’s lifetime. ### Intervention parameters The effect of the annual 10% tobacco inflation-adjusted excise tax increase (made in annual steps from NZ$14.46 in 2011 to NZ$26.53 per pack in 2020, compared with no tax increases from 2011 in BAU and from NZ$14.46 to NZ$15.36 in 2012 for the 1-year taxation scenario) is captured through changes to BAU smoking prevalence from tax increases acting through price elasticities of demand onto initiation (smoking rates age 20) and net cessation (by sex, age and ethnicity). Non-Māori price elasticities10–12 were scaled up by 20% for Māori, given New Zealand evidence for increased price sensitivity for Māori.13 14 The change in smoking prevalence is then combined with relative risks15–17 for the smoking-incidence rate ratios to generate population impact fractions that alter the incidence rates in the disease life tables. Equations by Hoogenveen et al 18 were used to calculate the decay rate of current-never incidence rate ratios for ex-smokers, ie, the disease relative risks for smokers that have quit up until 20 years postquitting. This accounts for ex-smokers reduction in disease risk as time passes. ### Modelling and analysis The scenarios were simulated 2000 times in Microsoft Excel (with the Ersatz add-in19). Each simulation involved a random draw from the probability density function about each of the input parameters. The difference between the intervention and BAU scenarios is expressed as incremental QALYs and health system costs. The net health system cost was the cost of a new law to legalise a tobacco tax (all occurring in calendar year 2011) and any difference in projected future health system expenditure. ## Results Table 1 shows the health gains (in QALYs) by sex and age that accrue over a 10-year period and over the remaining time horizon for <65 year olds and all ages for a 10% per annum increase in tobacco tax from 2011 to 2020. This tax resulted in an estimated 180 discounted QALYs gained (95% uncertainty interval (UI): 105 to 297) in these 10 years in working aged adults, those 20–65 years of age. This was two-thirds (67.4%) of the gain for all ages over this time period (268 QALYs) and 1.6% of the lifetime QALYs gained in this age group (11 300 QALYs). Table 1 Health gain (in QALYs) from 10% per annum increases in tobacco tax from 2011 to 2020, among the New Zealand adult population alive in 2011 by sex and ethnicity for a differing time horizons, by age at accrual of QALYs in the future (3% discount rate) In the first 10 years, there were an estimated NZ$10.6 million (95% UI: NZ$6.0 million to NZ$17.7 million; equivalent to US$7.1 million) of discounted net cost savings to the health system in those aged 20–65 or 73.1% of the NZ$14.5 million (95% UI: 8.2 million to 24.3 million) cost savings for the total population (after netting out intervention costs of $3.54 million NZ dollars). When comparing the results from this 10-year horizon with the QALYs gained and health system costs saved when the total 2011 population is simulated until death (discounted QALYs 39 100, cost savings$799.0 million), this 10-year horizon is only 0.7% of these QALY gains and 1.8% of the cost savings for all ages. In the scenario of only 1 year of tax increases, 1.7% of these lifetime QALY gains and 4.1% of cost savings now occur in the first 10 years post intervention.

Focusing on just the preretirement and postretirement ages, the QALY gains in each of the 60–64 and 65–69 year olds were 11.5% and 10.6%, respectively, of the 268 total QALYs gained in 2011–2020.

Per capita QALY gains over the 10-year period were 3.0 times higher for Māori at 0.105 per 1000 people compared with 0.035 for non-Māori in the 20–65 age group. When age standardisation (WHO world population) was applied, this ratio increased to 4.5 times higher. The proportion of the QALYs gained in the 10-year period in 20–65 year olds was also higher in Māori than in non-Māori (see online supplementary figure 1).

## Discussion

### Main findings and interpretation

Understanding the impact of tobacco tax by age helps to highlight the impact on working-age adults and therefore productivity. Approximately two-thirds of all health gains (268 QALYs) in the initial 10-year time horizon from the tobacco tax impact were estimated to occur among those aged 20–65 years (180 QALYs), and so the greatest impact of this intervention is on the working-age population. This is in part a function of the age structure of New Zealand (ie, in 2011, 35%, 27%, 25% and 13% of the population was in the <25, 25–44, 45–64 and 65+ age groups), but this age structure is similar to that in many other high-income countries. Similar results were seen by Holford et al,20 who found that half the premature deaths avoided through tobacco control in the USA occur in under 65 year olds. Approximately 10% of all health gains occur in the 5 years before and again 5 years after the age of entitlement for superannuation/pension payments, indicating the potential for this intervention to positively impact on productivity around the age of retirement.

This paper also illustrates that the predominant benefit of this tobacco tax is a long-term one—0.7% of the lifetime discounted QALY gains in the total population occur in the first 10 years. This is partly a function of the slow reduction in disease incidence but also a function of the intervention itself being annual increases in tax over 10 years. If we increase tax in just the first year, the total QALY gains over the remainder of the population’s lifetime are obviously much less, but 1.7% of these lifetime gains now occur in first 10 years postintervention—2.5 times higher than 0.7%.

The health gains associated with a tobacco tax come mainly from reducing the incidence of COPD, lung cancer, heart disease and stroke.1 These diseases are more prevalent among the Māori population (which is also younger and has high smoking prevalence rates). Hence, it is not surprising that the per capita benefit is greater for Māori aged 20–65 years. A tobacco tax could have additional economic benefits for Māori through a reduction in sick leave and forced early retirement due to ill health in 20–65 year olds (as well as reducing expenditure on tobacco by those who quit or reduce tobacco consumption). It could also have an additional benefit to the whole population from reductions in adverse maternal and child health outcomes.21

### Study strengths and limitations

This modelling work benefited from rich disease and cost data as outlined previously.1 The limitations of this modelling are also outlined previously, but we reiterate the major ones here: there are ongoing improvements in the costing data occurring, meaning the cost data used here will be superseded by more accurate data in future publications22; the price elasticities used were unchanged at higher tobacco prices; the effect of second-hand smoke exposure, smoking denormalisation and any reduction in the number of cigarettes smoked have not been captured in the model; and outcome uncertainty will be underestimated due to model structure assumptions and uncertainty.

### Conclusions

This modelling suggests that the majority of the health benefit over a 10-year horizon from increasing tobacco taxes is accrued in the working-age population (20–65 years). This benefit was disproportionately greater per capita for the indigenous population (Māori). However, compared with lifetime health gains, those occurring in the first 10 years after implementing a tobacco tax are modest. There remains a need for more work on the associated productivity benefits of such health gains.

• Only a small proportion, approximately 0.7%, of the total health gains over the remaining lifetimes of the New Zealand population occur in the first 10 years after the tax interventions start.

• The majority (67.4%) of the health gains in the first 10 years after the start of a programme of 10% per annum increases in tobacco tax accrue in working-age adults, those aged 20–65.

• Despite the small health gain in this initial 10-year time period, the intervention is still cost saving (NZ\$14.5 million) within this 10-year time horizon.

• Health gains in each of the 60–64 (preretirement ages) and 65–69 (postretirement ages) year olds were 11.5% and 10.6%, respectively, of the total health gains for all age groups in 2011–2020.

## Footnotes

• Contributors Conceived and designed the experiments: CLC, TB, GK, FSvdD, NN, LJC and NW. Analysed the data: CLC, TB, GK, FSvdD, NN, LJC and NW. Wrote the first draft of the manuscript: CLC, TB and NW. Contributed to the writing of the manuscript: CLC, TB, GK, FSvdD, NN, LJC and NW. Agree with the manuscript’s results and conclusions: CLC, TB, GK, FSvdD, NN, LJC and NW. All authors have read and confirm that they meet International Committee of Medical Journal Editors criteria for authorship.

• Funding The authors are supported by the BODE3 Programme, which is studying the effectiveness and cost-effectiveness of various preventive interventions and receives funding support from the Health Research Council of New Zealand (project number 10/248).

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.

• Data sharing statement The authors can be contacted for additional data, and these will be provided pending agreement from the agency providing access to epidemiological and costing data (the New Zealand Ministry of Health).

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.