Association between cigarette tax and the global burden of non-communicable diseases: a longitudinal ecological study in 193 countries

Abstract

Background

Tobacco control has been a critical strategy to reduce non-communicable disease (NCD) burden. However, the association between cigarette taxes and NCD burdens and its disparity by income level of countries, age group, sex and cause of NCDs has not been well-investigated.

Methods

This longitudinal ecological study analysed country-level data from 193 countries during 2008–2023, including cigarette taxes, NCD burdens and relevant covariates sourced from the WHO, Global Burden of Disease and World Bank. Fixed-effect panel regression models were used to assess the association incorporating a 2-year lagged exposure to account for the potential delayed effects of tax policies.

Results

Among the included countries, there were 26 (13%) low-income, 52 (27%) lower-middle-income, 53 (27%) upper-middle-income and 62 (32%) high-income countries, respectively. Overall, a 1-percentage-point increase in total tax (the proportion of the average retail price of a 20-pack of cigarettes) was significantly associated with a 16.43 (95% CI −21.63 to −11.24) and 0.63 (95% CI −0.83 to −0.42) reduction in disability-adjusted life years (DALYs) and deaths of NCD per 100 000 people, respectively. Greater associations were observed in high-income countries and upper-middle-income countries but no significant association for low-income countries. Compared with other types of taxes, increase in specific tax (tax charged per quantity) was associated with more reduction of NCD burden (DALY rate: −7.16, 95% CI −11.66 to −2.67). The impact of cigarette tax on NCD outcomes was greater among older adults and males compared with their counterparts. For cause-specific diseases, higher total tax was significantly associated with the greatest reduction in cardiovascular disease, followed by digestive disease and neoplasms.

Conclusions

Higher cigarette tax was associated with lower NCD burdens, particularly in upper-middle-income and high-income countries, with specific excise taxes driving most benefits. Integrated action to increase total tax and optimise the tax structure tailored to countries is needed to accomplish the dual benefits of health and economy globally.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Tobacco control has been recognised as a critical strategy in global efforts to reduce the burden of non-communicable diseases (NCDs).

WHAT THIS STUDY ADDS

• Higher cigarette tax was associated with lower NCD burdens, particularly in upper-middle-income and high-income countries, with specific excise taxes driving most benefits.

• The impact of cigarette taxation on NCD outcomes was greater among males and older adults compared with their counterparts.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The health benefits of raising cigarette tax varied across different subgroups.

• Integrated action to increase total tax and optimise the tax structure tailored to countries is needed to accomplish the dual benefits of health and economy globally.

Introduction

Non-communicable diseases (NCDs) are the leading causes of death globally despite age-standardised death rates declining from 1990 to 2021.¹ In 2021, NCDs caused 43.8 million deaths worldwide, with an age-standardised prevalence of 91 034.0 per 100 000 population. Tobacco use was the third leading risk factor for global disease burden, contributing to 9.9% of NCDs age-standardised disability-adjusted life years (DALYs) and 5.7% of total DALYs.² Although actions by governments and non-governmental organisations focused on tobacco control have contributed to a nearly 35% reduction in the age-standardised rate of global DALYs from 2000 to 2021, persistent and sustained actions are still needed to further reduce the health burden of smoking.^{2 3}

In 2007, WHO recommended monitoring tobacco use; protecting people from secondhand smoke; offering help to quit tobacco use; warning about the dangers of tobacco; enforcing bans on tobacco advertising, promotion and sponsorship and raising taxes on tobacco (MPOWER) measures as a technical package designed to help countries implement the demand-reduction measures of the WHO Framework Convention on Tobacco Control.⁴ At present, over 75% of the world’s population is covered by at least one MPOWER measure at the highest level of achievement.⁵ Among all these measures, raising taxes to increase the price of tobacco products is still the most effective tobacco control measure,⁵ which directly reflects government fiscal policy, whereas cigarette prices may additionally be influenced by tobacco industry pricing strategies and market conditions.⁶ From the perspective of potential causal pathways, higher cigarette taxes could lower exposure to smoke through reducing tobacco consumption, thereby decreasing the incidence of NCDs, and ultimately reducing premature mortality and NCD burden. At the Fourth High-level Meeting of the UN General Assembly, raising cigarette tax was reaffirmed as one of the most cost-effective strategies to curb tobacco use and accelerate progress in NCD prevention and control.⁷ However, the real-world association between cigarette taxes and the global burden of NCDs has not been well-evaluated since the implementation of the MPOWER measures. Several previous studies have shown the higher cigarette tax was associated with decreasing risk of overall child survival.^8–12 Recently, a panel study of 152 countries based on data from WHO indicated the negative association between cigarette tax and premature death.¹³ But the association between cigarette taxes and the measures of NCD burden (including DALYs and mortality) that are more sensitive to tobacco control was still lacking. In addition, tobacco tax rates and their structures vary considerably across countries at different income levels, thus the differential effects should be further explored to better inform policy improvements.^{5 14}

In this study, we aimed to conduct a longitudinal ecological analysis based on country-level data on 193 countries during 2008 and 2022 to describe the implementation of cigarette tax policies globally, assess the associations of cigarette tax with the age-standardised DALY rate and mortality of NCDs taking into account the delayed effects of the policies. In addition, we also aimed to reveal the various associations between cigarette taxes and NCD burden across countries with different income groups, age groups, sex and specific causes of NCDs to comprehensively evaluate the health effects of cigarette tax policies.

Methods

Study design and data source

This was a longitudinal ecological study using country-level data of cigarette taxes, health outcomes and potential covariates, which were derived from WHO, Global Burden of Diseases (GBD) and the World Bank, respectively. In specific, biennial data on MPOWER measures and cigarette taxes were available from 2008 to 2022 at the country level and were derived from the WHO Global Health Observatory.¹⁵ Data on burden of NCDs were collected from GBD 2023 and were available annually during 2008–2023.¹⁶ Data on all other socioeconomic, health-related and environmental covariates (classification for economies, gross domestic product, rural population, physician density, health expenditure, female primary education completion rate, clean cooking, and national CO₂ emissions) were obtained from the World Bank and were available annually between 2008 and 2022.¹⁷ After excluding countries with missing data on cigarette taxes and NCD burden, 193 countries were included for analysis. This study used anonymised aggregated data from publicly available resources; therefore, ethical approval was not required.

Variable definition

The primary exposure variable was the total cigarette tax, the proportion of the average retail price of a 20-pack of cigarettes, which was analysed as a continuous variable. We also examined this measure as a categorical variable, using 75.0% as the cut-off (<75.0% and ≥75.0%) based on the WHO-recommended threshold for cigarette tax.⁵ In addition, we considered the three separate types of cigarette tax: specific tax (tax charged per quantity, such as per pack, regardless of price), ad valorem tax (tax charged as a percentage of the value of the product) and other taxes (including import duties and value-added tax), all of which were treated as continuous variables. Independent models were fitted for each exposure variable and for each NCD outcome of interest.

Countries were categorised into four income groups according to the World Bank classification for economies based on gross national income per capita¹⁸: low-income countries (≤US$1135), lower middle-income countries (US$1136–US$4495), upper middle-income countries (US$4496–US$13 935) and high-income countries (≥US$13 936). Key health and development variables that might have an effect on cigarette tax and NCD outcomes were selected as covariates based on existing literature.^{8 9 19–21} MPOWER scores, biennial data available in WHO, evaluated national implementation of six core tobacco control policies²²: monitoring use, protecting from smoke, offering cessation support, warning about harms (W consists of two scales, warning about the dangers of tobacco: health warnings and mass media campaigns that were averaged to obtain one singular W score), enforcing advertising bans and raising taxes. Each domain is graded according to the strength of measures based on a five-point scale, with higher scores reflecting closer adherence to WHO recommendations. Monitoring tobacco use and raising taxes on tobacco were excluded due to redundancy.⁸ To simplify the model, the P, O, W, and E scores were summed into one single composite score in this study. Other covariates were annually available in World Bank and used as continuous variables: gross domestic product (gross domestic product parity purchasing power per person), rural population (the proportion of the population living in rural areas defined by national statistical offices), physician density (number of physicians per 1000 people), health expenditure (current health expenditure per capita), female primary education completion rate (the ratio of the number of new female entrants in the last grade of primary education regardless of age and the number of girls at the entrance age for the last grade of primary education), clean cooking (the proportion of the population with access to clean fuels and technologies for cooking) and national CO₂ emissions (carbon dioxide emissions are those stemming from the burning of fossil fuels and the manufacture of cement). Missing values in these covariates were addressed using multiple imputation. Time (calendar year as a categorical variable) was also considered in the analysis.

Study outcomes

The outcome measures included country-level age-standardised DALY rate and mortality, as well as age-specific (<20 years, 20–54 years and ≥55 years), sex-specific (female and male) and cause-specific (cardiovascular diseases, chronic respiratory diseases, digestive diseases, mental disorders, neoplasms and neurological disorders) DALY rate and mortality from NCDs, which were derived from the GBD 2023 estimates. GBD 2023 applied updated evidence and a unified analytical framework, primarily drawing on three standardised modelling tools: the Cause of Death Ensemble Model, Spatiotemporal Gaussian Process Regression and the Disease Modelling–Bayesian Meta-regression V.2.1. These methods were used to estimate major burden indicators across 204 countries/territories and 660 subnational locations, including incidence, prevalence, mortality, years of life lost, years lived with disability and DALYs. Age-standardisation was performed using the GBD reference population. The estimates were informed by diverse data sources, such as vital registration records, verbal autopsy studies, population censuses, household surveys, disease registries and health service utilisation data. Detailed descriptions of the GBD design and methodology are available in prior publications.^{23 24}

Statistical analysis

We described characteristics of included countries with different income groups in 2008 and 2022 using median (IQR) as appropriate for continuous variables, respectively. To compare the differences of characteristics among different income groups, Kruskal-Wallis tests were conducted for continuous variables and χ² or Fisher’s exact tests were conducted for categorical variables. To visualise trends over time, we plotted the median cigarette tax levels and NCD outcomes (age-standardised DALY rates and mortality) from 2008 to 2023, stratified by income group.

To evaluate the associations between cigarette tax and NCD outcomes, we employed fixed-effect panel regression models, incorporating 2-year lagged tax variables to account for delayed effects of tax policies.^{9 12} Due to the absence of outcomes in 2024 from GBD, we used the outcome data in 2023 instead of the 2024 data. Panel regression models were used to capture changes over time and account for the clustered data (country-level observations) collected repeatedly over time. The fixed-effect specification uses dummy variables to account for time-invariant, country-level potential covariates that are not possible or feasible to be measured. Model selection between fixed-effects and random-effects specifications was guided by Hausman tests.²⁵ Time-fixed effects were included to account for global secular trends. Results were presented as regression coefficients (β) with 95% CIs on the original scale. The overall prediction was obtained by summing country-specific estimates. In addition, panel regression models were fitted separately for each subgroup: age group (<20 years, 20–54 years and ≥55 years), sex (female and male) and cause-specific disease (cardiovascular diseases, chronic respiratory diseases, digestive diseases, mental disorders, neoplasms and neurological disorders). To estimate the potential reduction in NCD deaths attributable to achieving the WHO-recommended tobacco tax level of 75%, we calculated the tax gap for each country below this threshold in 2020 and multiplied it by the overall regression coefficient to derive the predicted mortality rate change. This rate change was then applied to estimate the number of NCD deaths that could have been averted in 2025. As the outcome for 2024 was unavailable, we used the most recent data from 2023 as a proxy. Accordingly, the estimated reduction in NCD deaths reflects the potential impact in 2025.

To enhance policy relevance, we additionally used total cigarette tax as a binary variable based on the WHO-recommended 75%. Random-effects models were used to account for potential clustering of countries within income groups and to allow for both within-country and between-country variation in outcomes, with group <75.0% as the reference. In all these models, potential covariates were adjusted, including MPOWER score, gross domestic product, rural population, physician density, health expenditure, female primary education completion rate, clean cooking and national CO₂ emissions.

To examine the robustness of our analyses, we did three sensitivity analyses. First, the association between 4-year lagged values of total cigarette tax and NCD burdens were evaluated to consider the delayed effect of policy implementation. Second, we repeated main analyses considering additional potential confounders (alcohol use and obesity). Third, we additionally examined the association between total cigarette tax and communicable disease burden as a placebo test.

All analyses were conducted using R V.4.5.1 (The R Foundation for Statistical Computing, Vienna, Austria). Two-tailed tests were used and the p value <0.05 was considered to be statistically significant.

Role of the funding source

The funders of the study had no role in study design, data collection, data analysis, data interpretation or the decision to submit the manuscript for publication.

Patient and public involvement

This study was a secondary analysis of existing global aggregate data (GBD study, WHO and World Bank). Therefore, patients and/or the public were not directly involved in the design or conduct of this specific study. However, the findings of this research on tobacco control and non-communicable diseases are of direct relevance to public health policy and will be disseminated in a manner accessible to the public to inform advocacy and decision-making.

Results

Of the 193 included countries, there were 26 (13%) low-income, 52 (27%) lower middle-income, 53 (27%) upper middle-income and 62 (32%) high-income countries, respectively (table 1). In 2022, only 41 (22.28%) countries met the WHO-recommended threshold for total cigarette tax, of which 26 were high-income countries and only one was a low-income country. High-income countries were more likely to have higher total cigarette tax and other three separate types of taxes in both 2008 and 2022 (p<0.05). Additionally, they tended to have more favourable health and development indicators, including greater physician density, higher health expenditure, lower rural population, more access to clean cooking and higher female primary education completion (all p<0.01).

Table 1. Characteristics of countries according to income group in 2008 and 2022.

	Overall	Income group				P value
		Low-income	Lower middle-income	Upper middle-income	High-income
N	193	26	52	53	62
2008
N	181	24	46	52	59
Tax
Total tax, %	46.52 (30.08–66.67)	32.19 (19.51–53)	35.88 (25.00–49.25)	43.32 (29.72–56.63)	70.1 (44.56–75.77)	<0.01
Specific tax, %	5.60 (0.00–31.40)	0.00 (0.00–0.00)	0 (0–18.18)	4.86 (0–29.84)	24.05 (3.71–39.39)	<0.01
Ad valorem tax, %	6.42 (0.00–28.00)	12.53 (1.69–22.34)	3.15 (0–17.82)	0 (0–25.41)	20 (0–39.85)	0.14
Other tax, %	13.79 (10.62–16.67)	13.18 (9.57–15.57)	13.04 (9.28–16.08)	13.04 (9.61–16.44)	15.97 (13.04–18.03)	0.02
Total tax ≥75%	22 (12.15)	0 (0.00)	3 (6.52)	2 (3.85)	17 (28.8)	<0.01
Covariates
MPOWER score	12 (10–13)	9.5 (8–12)	11.5 (9–13)	12 (10–13)	13 (11–14)	<0.01
Physician density	1.15 (0.25–2.67)	0.05 (0.03–0.18)	0.36 (0.1–0.77)	1.27 (0.63–2.58)	2.92 (1.92–3.58)	<0.01
Rural population, %	43.32 (26.27–63.41)	68.12 (61.82–78.98)	58.11 (44.96–71.96)	38.27 (30.59–51.9)	23.42 (14.16–34.3)	<0.01
CO2 emissions, kiloton	2.48 (0.67–6.68)	0.1 (0.07–0.15)	0.65 (0.34–1.07)	2.76 (1.85–4.57)	8.09 (5.42–11.29)	<0.01
Gross domestic product, per capita	9985.98 (3270.56–21 019.8)	1342.55 (915.88–1631.92)	3253.28 (2581.41–4915.44)	10 962.54 (8202.46–13 741.65)	31 197.22 (20 877.75–42 817.41)	<0.01
Female primary education completion rate, %	94.96 (75.36–100.1)	54.37 (45.31–65.95)	78.69 (60.27–94.89)	98.77 (93.08–102.2)	99.16 (96.06–102.18)	<0.01
Health expenditure, per capita	267.73 (73.74–791.2)	33.52 (23.53–39.99)	70.91 (49.2–133.11)	277.62 (218.75–399.97)	1321.94 (839.28–4391.74)	<0.01
Clean cooking, %	85.40 (24.80–100.00)	1.9 (0.8–3.5)	22.7 (7.53–41.75)	85.4 (60.92–93.5)	100.00 (100.00–100.00)	<0.01
2022
N	184	21	50	53	60
Tax
Total tax, %	55.8 (37.02–74.02)	43.5 (29.75–54.87)	44.5 (30.02–58.2)	50.6 (38.71–67.59)	73.92 (63.53–78.11)	<0.01
Specific tax, %	24.31 (0–39.13)	1.25 (0–22)	15.48 (0–36.83)	25.67 (2.5–37.96)	32.08 (11.8–47.01)	<0.01
Ad valorem tax, %	7.4 (0–26.2)	14.64 (0–36)	0.81 (0–14.31)	0 (0–17.23)	18.17 (0–34)	0.01
Other tax, %	15.25 (11.11–17.36)	14.53 (11.11–15.97)	13.04 (9.09–15.25)	15.25 (13.04–17.36)	16.67 (13.78–18.7)	<0.01
Total tax ≥75%	41 (22.28)	1 (4.76)	5 (10.00)	9 (16.98)	26 (43.33)	<0.01
Covariates
MPOWER score	17 (16–18)	17 (16–19)	17 (16–18)	17 (16–18)	16 (16–18)	<0.01
Physician density	2.16 (0.74–3.46)	1.08 (0.39–3.67)	2.23 (0.87–3.13)	2.38 (0.84–3.77)	2.13 (0.77–3.31)	<0.01
Rural population, %	38.63 (25.85–57.57)	42.33 (23.96–51.88)	41.25 (26.66–64.49)	36.36 (23–50.05)	41.2 (30.78–55.3)	<0.01
CO2 emissions, kiloton	1.96 (0.57–4.36)	1.05 (0.69–2.51)	2.31 (0.56–4.02)	2.09 (0.48–4.77)	1.88 (0.69–4.4)	<0.01
Gross domestic product, per capita	14 076.34 (7674.47–24 608.33)	13 445.25 (8538.2–18 981.66)	14 697.61 (7208.39–24 066.18)	14 853.67 (7915.55–29 827.14)	12 722.36 (7509.21–24 556.16)	<0.01
Female primary education completion rate, %	97.42 (90.2–101.1)	96.93 (90.82–100.5)	97.42 (90.82–100.13)	96.04 (88.38–100.61)	99.22 (92.14–104.06)	<0.01
Health expenditure, per capita	311.93 (92.22–961)	235.73 (67–290)	462.23 (127.19–982)	342.94 (95–1880.44)	277.92 (94.66–872.94)	<0.01
Clean cooking, %	95.2 (46.38–100)	4.3 (1.23–11.35)	40.2 (19.8–71.6)	88.1 (80–95.6)	100.00 (100.00–100.00)	<0.01

Continuous variables were summarised as median and IQR, and categorical variables were summarised as frequencies and percentages. P values were the results of the difference between groups.

MPOWER, monitor tobacco use; protect people from secondhand smoke; offer help to quit tobacco use; warn about the dangers of tobacco; enforce bans on tobacco advertising, promotion and sponsorship and raise taxes on tobacco.

In general, median total tax tended to increase from 2008 to 2022 across most income groups and types of tax (figure 1). Median total tax of all countries increased from 46.52% (IQR 30.08–66.67) in 2008 to 55.80% (37.02–74.02) in 2022 (online supplemental table S1). The increase of total tax in low-income countries (35.13%) was greater than that in high-income countries (5.44%), but the average total tax of low-income countries in 2022 was still below WHO-recommended threshold of 75%. Median specific tax, ad valorem tax and other taxes also increased from 2008 to 2022. But specific tax was consistently low in low-income countries, and ad valorem tax was relatively lower in lower-middle and upper-middle income countries (online supplemental tables S2–S4).

Figure 1. Trends of cigarette taxes across different income groups during 2008–2020. Results were presented as median.

Median DALY rate and mortality of NCD decreased between 2008 and 2023 across all income groups (online supplemental figure S1). The median DALY rate declined from 18 898.97 per 100 000 in 2008 to 17 590.90 per 100 000 in 2020 in high-income countries. The median mortality decreased from 497.44 to 393.71 per 100 000 in high-income countries. However, the burden of NCDs remained disproportionately high in low-income countries, with DALY rate and mortality in 2020 still nearly 1.44 and 1.81 times higher than those observed in high-income countries. The geographic distribution of the total cigarette tax and burden of NCDs in 2008 and 2022 in the included countries is shown in figure 2.

Figure 2. Geographic distribution of total cigarette tax and NCD burden of the included countries in 2008 and 2020. (A) Total cigarette tax in 2008; (B) total cigarette tax in 2020; (C) NCD DALY rate in 2008; (D) NCD DALY rate in 2020; (E) NCD mortality in 2008; (F) NCD mortality in 2020. DALY, disability-adjusted life year; NCD, noncommunicable disease.

In panel regression analysis, a 1-percentage-point increase in total tax was associated with a 16.43 (95% CI −21.63 to −11.24) reduction in DALY rate from NCDs in all countries, with substantial declines in upper-middle-income (−27.18; 95% CI −37.28 to −17.07), and high-income countries (−7.86; 95% CI −14.28 to −1.44) (table 2). A 1-percentage-point increase in total tax was associated with a 0.63 (95% CI −0.83 to −0.42) reduction in mortality in all countries. Similar negative associations were observed in upper middle-income countries (−1.27; 95% CI −1.67 to −0.86) and high-income countries (−0.24; 95% CI −0.51 to −0.02). Based on our regression models, if the 143 countries in 2022 with total tax <75% raise their total cigarette tax to 75%, an estimated 990 880 (95% CI 660 587 to 1 305 445) NCD deaths in 2025 could be averted. In the models where total tax used as a categorical variable (online supplemental table S5), we found that total tax ≥75% group was associated with dramatic decreases in DALY rate (−294.48; 95% CI −501.63 to −87.32) and mortality (−14.99; 95% CI −23.38 to −6.59) from NCDs compared with <75% group.

Table 2. Associations between cigarette taxes and the burden of NCDs.

	Overall	Low-income countries	Lower middle-income countries	Upper middle-income countries	High-income countries
DALY rate
Total tax	−16.43 (−21.63 to −11.24)	8.18 (−4.96 to 21.31)	−10.83 (−21.93 to 0.27)	−27.18 (−37.28 to −17.07)	−7.86 (−14.28 to −1.44)
Specific tax	−7.16 (−11.66 to −2.67)	16.78 (1.61 to 31.94)	−6.85 (−15.85 to 2.15)	−12.26 (−20.54 to −3.97)	5.39 (−0.22 to 11.00)
Ad valorem tax	−6.71 (−11.61 to −1.80)	0.55 (12.21 to 13.30)	−3.13 (−13.83 to 7.57)	−6.74 (−17.17 to 3.69)	−9.51 (−14.91 to −4.11)
Other tax	2.97 (−5.96 to 11.91)	−20.84 (−48.38 to 6.69)	8.60 (−10.14 to 27.34)	−4.15 (−23.83 to 15.53)	−3.91 (−13.66 to 5.85)
Mortality
Total tax	−0.63 (−0.83 to −0.42)	0.15 (−0.32 to 0.61)	−0.21 (−0.64 to 0.22)	−1.27 (−1.67 to −0.86)	−0.24 (−0.51 to −0.02)
Specific tax	−0.33 (−0.51 to −0.16)	0.47 (−0.07 to 1.00)	−0.14 (−0.48 to 0.21)	−0.65 (−0.99 to −0.32)	0.13 (−0.11 to 0.36)
Ad valorem	−0.17 (−0.36 to 0.02)	0.01 (−0.43 to 0.46)	−0.14 (−0.55 to 0.27)	−0.14 (−0.56 to 0.29)	−0.25 (−0.47 to −0.02)
Other tax	0.06 (−0.29 to 0.41)	−0.92 (−1.89 to 0.04)	0.42 (−0.30 to 1.14)	−0.38 (−1.19 to 0.42)	−0.16 (−0.56 to 0.25)

Results were presented as regression coefficients (β) with 95% CIs on the original scale from the fixed-effects model.

Adjusted for MPOWER score, gross domestic product, rural population, physician density, health expenditure, female primary education completion rate, clean cooking and national CO₂ emissions.

DALY, disability-adjusted life year; MPOWER, monitor tobacco use; protect people from secondhand smoke; offer help to quit tobacco use; warn about the dangers of tobacco; enforce bans on tobacco advertising, promotion and sponsorship and raise taxes on tobacco; NCD, non-communicable disease.

In the models of separate tax types, a 1-percentage-point increase in specific tax was associated with a 7.16 reduction in DALY rate (95% CI −11.66 to −2.67) and a 0.33 reduction in mortality (95% CI −0.51 to −0.16). The significant negative association of specific tax with NCD burden was observed in upper-middle-income countries, but not in other income groups. In addition, ad valorem tax was associated with a 6.71 reduction in DALY rate (95% CI −11.61 to −1.80) overall. Significant association of ad valorem tax with NCD outcome was observed in high-income countries. For other tax, no significant associations were observed.

In age-stratified models, the association between cigarette tax and NCD burden was more pronounced among older adults (figure 3). A 1-percentage-point increase in total tax was associated with a reduction in DALY rate of 2.17 (95% CI −4.28 to −0.06), 9.05 (95% CI −13.35 to −4.75) and 71.82 (95% CI −91.82 to −51.82) among population <20 years, 20–54 years and ≥55 years, respectively. Correspondingly, mortality decreased by 0.20 (95% CI −0.29 to −0.11) in the 20–54 age group and by 3.65 (95% CI −4.71 to −2.60) in those aged ≥55 years for each 1-percentage-point total tax increase. In both males and females, associations between cigarette taxes and NCD outcomes were generally consistent with the overall results. For separate tax types, similar patterns were observed for both specific tax and ad valorem tax. And other tax was not significantly associated with NCD burden. In the models stratified by sex, increase in total tax and specific tax was associated with significant reductions in NCD outcomes in both males and females, and more pronounced associations were observed in males (online supplemental tables S6 and S7). In addition, increase in ad valorem tax was associated with decrease in DALY rate (−9.53; 95% CI −15.48 to −3.58) and mortality (−0.22; 95% CI −0.56 to −0.09) in males, but not in females. No significant associations between other tax and NCD burden were observed in both sexes.

Figure 3. Associations between cigarette taxes and burden of NCDs by age and sex from 2008 to 2020. Results were presented as regression coefficients (β) with 95% CIs on the original scale from fixed-effect model. Adjusted for MPOWER score, gross domestic product, rural population, physician density, health expenditure, female primary education completion rate, clean cooking and national CO₂ emissions. MPOWER, monitor tobacco use; protect people from secondhand smoke; offer help to quit tobacco use; warn about the dangers of tobacco; enforce bans on tobacco advertising, promotion and sponsorship and raise taxes on tobacco; NCD, noncommunicable disease.

In cause-specific models, higher total tax was significantly associated with lower burdens of cardiovascular disease, digestive disease and neoplasms (online supplemental figure S2). A 1-percentage-point increase in total tax was associated with reductions in DALY rate of 11.01 (95% CI −14.08 to −7.95) for cardiovascular disease, 1.55 (95% CI −2.16 to −0.94) for digestive disease and 1.85 (95% CI −3.37 to −0.33) for neoplasms. Corresponding reductions in mortality were 0.52 (95% CI −0.67 to −0.38), 0.04 (95% CI −0.06 to −0.02) and 0.06 (−0.11 to −0.01), respectively. No significant associations were observed for mental disorders and neurological disorders. Similarly, the patterns were consistent across sexes and aligned with the overall results, with stronger associations observed in males than in females (online supplemental figures S3 and S4).

In the sensitivity analyses, the results were not materially changed when using 4-year lagged values of total tax (online supplemental table S8) or considering other potential confounders (online supplemental table S9). No significant association was observed between total cigarette tax and communicable disease burden among all population (online supplemental table S10).

Discussion

Main findings

In our panel regression analysis of 193 countries, we found that higher cigarette tax was significantly associated with reductions in DALY rate and mortality of NCD, with more pronounced associations observed in high-income countries and upper-middle-income countries. It is estimated that 990 880 (95% CI 66 0587 to 1 305 445) NCD deaths could be averted in 2025 if all countries raise total tax to 75%. For separate tax types, increases in specific taxes were associated with more reduction of NCD burden. The impact of cigarette tax on NCD outcomes was greater among males and older adults compared with their counterparts. The increase in cigarette tax was associated with the strongest reduction in CVD burden, followed by digestive diseases and neoplasms.

Comparison with other studies

By using data from 193 countries during 2008–2023, we provided more comprehensive quantitative assessments of the association between cigarette tax and the burden of NCDs and its variations across country-level income groups, age groups, sex and specific causes of disease. Our results showed that higher total tax was significantly associated with lower DALY rate and mortality of NCDs, which are consistent with most prior studies.^{13 26} Raising cigarette tax could directly inhibit consumption via price mechanisms, thereby reducing tobacco use rates, especially for those vulnerable populations sensitive to tobacco prices.²⁷ In addition, increased cigarette tax revenues provide additional financial resources for health systems, further amplifying the health benefits of tobacco control. Given this, the fiscal policy of raising cigarette tax seems to be one of the most cost-effective ‘Best Buy’ interventions.²⁸

Except for the overall association of cigarette tax with the burden of NCDs, our study added to the existing literature in several ways. First, disparities in association between total tax and the burden of NCDs were first reported in countries with different income levels. The association was more pronounced in high-income and upper-middle-income countries, while no significant association was observed in low-income countries. In high-income countries, higher cigarette tax (with a median over 70%) was widely implemented as early as 2008. Although subsequent increases were limited, the cumulative health effects of long-term policies have become fully evident. The significant association observed in upper middle-income countries partly reflects the progressive increase in total cigarette tax from 2008 to 2020 within these countries. In contrast, no significant association was observed in low-income countries, which might be explained by the following two reasons. On one hand, rapid population and economic growth in some low-income countries may have offset the disease-control effects of price increases in cigarettes,²⁷ thereby weakening the impact of tax policies on disease burden. For example, cigarettes became increasingly more affordable from 1999 to 2019 in ASEAN countries (Cambodia, Indonesia, Lao PDR, Thailand and Vietnam) because cigarette tax growth failed to keep pace with income growth.²⁹ On the other hand, cigarette tax in low-income countries predominantly relied on ad valorem taxes, but specific tax rates were generally low. Ad valorem tax regimes provide tobacco companies with greater flexibility to manipulate prices, allowing consumers to partially mitigate the impact of tax hikes and price increases by switching to lower-priced brands. This substitution behaviour is particularly pronounced under ad valorem systems, leading to significant price differentials between brands and undermining the tax’s effectiveness in reducing tobacco consumption. Previous studies have found that when ad valorem taxes constituted a higher proportion of total excise taxes, they were often associated with increased tobacco consumption levels.³⁰

Second, we found specific tax seemed to have greater health effects than ad valorem tax and other taxes. Specific taxes could significantly raise prices across all brands, narrowing the price gap between premium and budget brands.³⁰ This reduces consumers’ incentive to switch to cheaper brands to evade taxes, thereby effectively decreasing tobacco consumption and promoting NCD prevention and control. This could explain the significant association between specific tax and NCD burden in upper middle-income countries. However, in high-income countries, increasing ad valorem taxes was associated with a significant reduction in disease burden. This might benefit from their robust tax enforcement and structural adjustment capabilities, allowing the elasticity advantages of ad valorem taxes to play a role in tobacco control.

Third, we identified differential associations between cigarette tax and NCD burden across population subgroups and cause-specific diseases. There was a stronger association between cigarette tax and NCD burden among older adults. This finding may be related to baseline risk levels and cumulative exposure effects within the population: as individuals age, the risk of developing NCDs due to long-term smoking significantly increases. Consequently, tax-induced reductions in smoking could produce more substantial health benefits in this subgroup. In addition, we observed that associations were generally stronger among males than females, consistent with the reality that male smoking rates are significantly higher than female in most countries. Tax policies reduce smoking by increasing prices, thus producing stronger associations on the disease burden among males. In addition, we observed higher cigarette tax was associated with the greatest reduction in the burden of cardiovascular diseases, followed by digestive diseases and neoplasms. This is primarily because CVD accounts for the largest proportion of NCDs, and smoking is a major risk factor for CVD.

Implications

Our study has some potential public health implications. First, increasing cigarette tax appears to be one of the essential strategies for reducing the burden of NCDs globally. However, the implementation of tobacco control measures has been uneven across countries, particularly regarding the enforcement of cigarette taxes.¹⁴ Although WHO anticipated that 56 countries were likely to have achieved at least a 30% relative reduction in tobacco use by 2025 if current efforts were maintained or accelerated, 40 countries had not taken any of the MPOWER measures by 2024, leaving nearly 2 billion people experiencing the harms of tobacco.⁵ NCDs continue to account for a substantial proportion of the global disease burden, and their burden is rising rapidly in many low- and middle-income countries undergoing epidemiological transition, which highlights the need for effective tobacco control policies. As our findings suggest, if all countries could raise total cigarette tax to 75.0% in 2023, 990 880 deaths from NCDs could have been averted in 2025. These findings support global calls to strengthen tobacco taxation in line with national circumstances.⁷ One of the main implementation barriers identified by WHO FCTC Parties is the interference of the tobacco industry,¹⁴ highlighting the need for multi-departmental collaboration. second, tax structures should be further optimised by enhancing the proportion of specific taxes to strengthen policy effectiveness, particularly in low-income countries, while avoiding excessive reliance on ad valorem taxes. Thirdly, the health benefits of raising cigarette tax varied across different age groups. For younger people, strategies that shift social norms may complement taxation, such as leveraging role models to diminish societal acceptance of smoking,³¹ whereas cessation services and pharmacological support remain essential for high-risk smoking groups. Fourthly, cigarette tax has demonstrated a significant association with reducing the burden of cardiovascular, digestive diseases and neoplasms, which remain the major causes of global disease burden in the following decades. Thus, it is necessary to prevent and control these diseases through increasing tobacco tax. Of course, NCDs are influenced by multiple factors including diet, alcohol consumption, physical activity and environmental conditions. A comprehensive and multisectoral approach is needed to address behavioural risks, social determinants and health system capacity alongside policies.²⁸

Limitations

This study has some limitations. First, even with fixed-effect models and additional covariates, residual and unmeasured confounding should be considered in future studies. Second, although we considered 2-year delayed effect of cigarette tax, policy measures typically require more time to produce health benefits, so the association between cigarette tax and NCD burden in this study might be conservative. Third, only data on cigarette tobacco taxes were available and analysed in this study. Emerging products like e-cigarettes, snus and waterpipe tobacco, which are increasingly popular among young people, raise concerns about reversing advances in tobacco control.^{14 32} The impacts of these products on the disease burden need further studies. Fourth, we focused on the impact of cigarette tax on NCD burden. Future research could extend this work by incorporating additional strategies outlined in the WHO MPOWER package, including smoke-free policies, graphic health warnings, comprehensive bans on tobacco advertising, promotion and sponsorship and cessation support services. Fifth, the observational nature of this ecological panel study limits causal inference.

Conclusions

Higher cigarette tax was significantly associated with burden of NCDs, with more pronounced associations observed in upper-middle-income and high-income countries. Specific taxes appeared to be the primary tax component driving these health benefits. Integrated action to increase total tax and to optimise the tax structure tailored to specific country is one of effective strategies for reducing burden of NCDs globally.

Data availability statement

Data are available in a public, open access repository.