ABSTRACT
Background and Objectives
Smoking is a major modifiable risk factor for upper gastrointestinal cancers (UGIC), particularly in China. This study investigated temporal trends, demographic patterns, and future projections of smoking‐attributable UGIC mortality globally and nationally.
Methods
We extracted data from the Global Burden of Disease 2021. Joinpoint regression was used to assess trends in smoking‐attributable UGIC mortality rates and absolute deaths. Age‐period‐cohort (APC) model was applied to evaluate effects brought by age groups and birth cohort. Decomposition analysis quantified contributions of aging, population growth, and epidemiologic shifts to mortality changes. Bayesian APC (BAPC) model projected mortality trends to 2030.
Results
At the population level, smoking is more strongly associated with esophageal cancer than gastric cancer. Although smoking‐attributable mortality rates of both esophageal and gastric cancers declined globally and in China, the decline was more pronounced for gastric cancer than for esophageal cancer (AAPC: −2.594 vs. −1.229 globally; −2.297 vs. −1.491 in China; both p < 0.001). However, trends in absolute deaths differed by cancer type and geography: smoking‐attributable esophageal cancer deaths increased globally and in China, whereas smoking‐attributable gastric cancer deaths increased in China but remained stable globally. APC analysis revealed the most significant drop among young adults aged 35–44, but mortality trends plateaued or increased in older individuals. In contrast, esophageal cancer mortality is expected to remain elevated through 2030, with an estimated 263,219 deaths globally including 187,567 in China.
Conclusions
Despite declining mortality rates, the absolute number of smoking‐related UGIC deaths remains substantial and is rising in China due to demographic momentum. Greater reductions were seen among young individuals and females. These ecological findings underscore the importance of strengthening tobacco‐control policies to mitigate the population‐level burden attributable to smoking, particularly within high‐risk populations.
Keywords: age‐period‐cohort analysis, esophageal cancer, gastric cancer, risk factors, smoking
1. Introduction
Upper gastrointestinal cancer (UGIC), including esophageal and gastric cancers, continues to be the leading cause of cancer‐related mortality globally [1]. Despite the implementation of various approaches including H. pylori eradication, cancer screening, and advancements in cancer treatment, UGIC still poses a substantial burden on the healthcare system. One of the reasons leading to the persistently high UGIC burden is related to exposure to unhealthy lifestyles such as smoking, alcohol, obesity, and diets [2].
Smoking, being one of the most preventable causes of cancer‐related deaths, has been widely recognized to be an important cause for the development of esophageal and gastric cancers through a variety of mechanisms, including direct effects of carcinogenic substances (e.g., Polycyclic Aromatic Hydrocarbons (PAHs), nitrosamines, and benzopyrene), initiation of oxidative stress, alteration and destruction of the mucosal barrier function of the esophagus and stomach [3, 4, 5]. Although tobacco control policies have been strengthened and smoking cessation services have been provided, smoking prevalence is still high in many countries [6].
Although prior studies have documented the burden of UGIC and their associated risk factors [7], they were often limited to cross‐sectional descriptions. In‐depth exploration of the underlying epidemiological changes, including variations across different periods, birth cohorts, and the factors contributing to alterations in disease burden, remains limited.
The World Health Organization's (WHO) 2030 health goals emphasize the targets of reducing premature mortality from noncommunicable diseases (NCDs), including cancers, and ensuring the implementation of the WHO Framework Convention on Tobacco Control (FCTC) [8]. The WHO report on the global tobacco epidemic in 2023 highlights that Asia, particularly China, has a massive smoking population. The advancement of tobacco control measures in these regions will be pivotal in achieving the global health targets set for 2030 [6]. It faces unique challenges due to the high prevalence of smoking and the growing adoption of diverse smoking products. Understanding the distinctions between global and Chinese trends is essential, as China demonstrates distinct patterns in smoking prevalence, tobacco control policies, and demographic characteristics, which contribute to the global burden of UGIC. This study utilizes the latest Global Burden of Disease (GBD) 2021 data to investigate the temporal trend in the global and Chinese disease burden of UGIC attributed to smoking over the past three decades.
2. Methods
2.1. The Global and Regional Burden of UGIC Attributable to Smoking
The global average and Chinese data on esophageal and gastric cancer burden attributable to smoking were obtained from the Global Burden of Disease Study 2021, which comprehensively assesses disease burden, injuries, and impairments for 369 conditions and 87 risk factors across 204 countries and territories. The disease burden estimation in GBD utilized an integrative Bayesian meta‐regression approach to calculate a generalized negative binomial model [9]. The detailed methodologies used in GBD 2021 were reported elsewhere [1, 10]. To improve cross‐country comparability due to variable data quality, GBD applies standardized modeling tools [spatiotemporal Gaussian process regression (ST‐GPR) for exposures; Meta‐Regression‐Bayesian Regularized Trimmed (MR‐BRT) to derive nonlinear dose–response curves], cross‐walk/bias adjustments to align differing case definitions, and out‐of‐sample predictive‐validity testing [11]. Uncertainty from input data, model selection, and parameters is propagated to final estimates.
We obtained age–sex–location–year specific prevalence of current and former smoking from the GBD 2021 comparative risk assessment. In the GBD comparative risk assessment framework, smoking exposure was defined as current or former use of smoked tobacco products. Electronic cigarettes and vaporizers were not included in this definition. Exposure estimates were synthesized from nationally representative surveys and censuses, and modeled using spatiotemporal Gaussian process regression to improve comparability across age, sex, location, and year [12]. Estimates are age‐standardized to the GBD standard population and reported with 95% uncertainty intervals (UI) propagated from the model. China‐specific data are modeled within this pipeline and contribute to regional estimates. A schematic diagram of the GBD comparative risk assessment framework for smoking is provided in Figure S1. The International Classification of Diseases (10th revision) codes for esophageal cancer were defined according to the GBD study framework, including C15.0–C15.9, and D00.1 (carcinoma in situ of esophagus). For gastric cancer, the codes encompassed C16.0–C16.9 and selected Z codes (Z12.0, Z85.02–Z85.028). Although D00.1 and Z codes are rarely used in conventional cancer registries and may not represent invasive cancers pathologically, their inclusion follows the GBD coding scheme to harmonize heterogeneous data sources and ensure comparability across countries and over time. Detailed information of the included ICD code was shown in Table S1. The age‐standardized mortality rates (ASMR), disability‐adjusted life years rate (DALYs), and population‐attributable fraction (PAF) for smoking were extracted for analysis at national and regional levels and across different age groups. The “early onset” refers to individuals under 40, while “late onset” pertains to those aged 40 and above [13, 14]. An additional sensitivity analysis was conducted to adjust the cut‐off age to 50. Specifically, due to the massive smoking population in China (with a smoker population exceeding 300 million, which constitutes nearly one‐third of the global total), it would justify the calculation of the global and China's average annual percentage changes (AAPC) for UGIC burden using Joinpoint regression analysis [15, 16, 17].
2.2. Age‐Period‐Cohort (APC) Model Analysis
The APC model is an advanced framework used in health and socio‐economic research to provide deeper insights than traditional statistical methods. It analyzes long‐term trends by estimating overall and period‐specific drifts while accounting for the individual and combined effects of age, time, and birth cohorts [18]. In this study, the APC model was applied to examine smoking‐attributable UGIC burden across different age groups, periods, and cohorts. This approach supports disease burden trajectory analysis, detects shifts in disease patterns, disentangles age‐period‐cohort influences, and explores how they affect smoking‐attributable UGIC burden. Additionally, it provides a framework for understanding epidemiologic transitions and identifying key areas for intervention in prevention, policy, and clinical management. Accordingly, this study employed the 2021 GBD dataset to investigate smoking‐attributable UGIC burden over the three‐decade period from 1992 to 2021. The population was categorized into 13 5‐year age groups, ranging from 30 to 34 years (the youngest five‐year age group available in GBD 2021 for this analysis) to 90–94 years. The observation period was also split into 6 5‐year intervals: 1992–1996, 1997–2001, 2002–2006, 2007–2011, 2012–2016, and 2017–2021.
To incorporate the cohort dimension, birth cohorts were established, ranging from 1897 to 1906 through 1982–1991. The APC model then estimated both the general trend across time (net drift) and the within‐group annual change (local drift) in disease burden. Net drift reflects the annual percentage change across all ages and periods, while local drift quantifies age‐specific changes. Statistical significance for net drift was assessed using the Wald chi‐square test. Moreover, within the APC model, the age effect is characterized by age‐specific rates that are aligned with birth cohorts. In contrast, the period and cohort effects are represented as the relative risks of mortality associated with each period or cohort, which are quantified by comparing the age‐specific rates of each period or cohort to those of a reference period or cohort. Notably, the selection of the reference period or cohort is arbitrary and does not affect the interpretation of the results.
2.3. Decomposition Analysis and Correlation With Sociodemographic Index
To achieve a more comprehensive understanding of the factors contributing to the changes in the absolute number of smoking‐attributable UGIC deaths between 1990 and 2021, decomposition analyses were conducted, considering age structure, population size, and epidemiological transitions. We used a stepwise replacement algorithm based on the Das Gupta method, commonly applied in GBD studies. This method decomposes the absolute difference in death counts between two time points into components attributable to: population growth (changes in total population size), population aging (shifts in age distribution), and mortality rate changes (changes in age‐specific mortality rates), all assuming constant other factors. Detailed methods for decomposition analysis were elucidated elsewhere [19].
GBD calculated the sociodemographic index (SDI) to reflect the social and economic factors affecting individuals' health outcomes across countries [10]. The SDI levels were based on the average years of education, the total fertility rate, and lag‐distributed income per capita. The Loess regression demonstrated correlations between average SDI (1990 to 2021) and AAPC of UGIC burden [20]. Summary Exposure Values (SEV) smoking data in GBD 2021 were extracted to demonstrate the regional distribution of smoking exposure [1].
2.4. Smoking‐Attributable UGIC Burden Projection
The future burden of smoking‐attributable UGIC mortality (rates and absolute deaths) from 2022 to 2030 was projected using a Bayesian age–period–cohort (BAPC) model implemented via the BAPC R package [21] with an INLA backend. We modeled age–sex–specific death counts with a Poisson likelihood and log link, incorporating log (population) as an offset. Second‐order random‐walk (RW2) priors were specified for age, period, and cohort effects, with hyper‐priors set to package defaults. Model specifications (RW1 vs. RW2; alternative smoothness parameters) were compared using Watanabe–Akaike Information Criterion (WAIC) and the Deviance Information Criterion (DIC), with RW2 showing the best fit. Predictive validity was assessed through rolling 5‐year hold‐out cross‐validation, examining root mean square error (RMSE) and 95% interval coverage; the final model demonstrated superior performance. Forecasts were generated through 2030, with full posterior uncertainty intervals reported [22].
All statistics were performed using the R program (Version 4.2.3, R core team). A p value of less than 0.05 was considered statistically significant.
3. Results
3.1. Temporal Trend of Smoking‐Attributable UGIC Burden Globally and in China
The global distribution of smoking exposure is shown in Figure S2. Smoking‐attributable UGIC burden (Figure 1A,B) is pronounced in East Asia (particularly in China), Eastern Europe, Russia, Central Asia, and the United States, while it is relatively low in most African and Southern Asian countries. China's portion of the global smoking‐attributable UGIC mortality has been steadily increasing over time (Figure 1C,D). In 2021, China accounted for 67.8% of the smoking‐attributable esophageal cancer deaths and 60.8% of the smoking‐attributable gastric cancer deaths in the world, respectively. Although the contribution of female patients to the mortality burden is notably lower than that of male patients, there is still a noticeable increase over the past 5 years. More detailed data on the regional smoking‐attributable UGIC burden was shown in Table 1.
FIGURE 1.
Global distribution of age‐standardized UGIC burden attributable to smoking in 2021 (A: Esophageal cancer and B: Gastric cancer). Proportion of the smoking‐attributable UGIC deaths in China relative to the total global deaths (C: Esophageal cancer and D: Stomach cancer).
TABLE 1.
The age‐standardized rates and number of deaths for smoking‐attributed esophageal cancer and gastric cancer in 1990 and 2021 for both sexes, different WHO regions, and Sociodemographic Index (SDI) level countries.
| Esophageal cancer | 1990 | 2021 | ||
|---|---|---|---|---|
| Age‐standardized mortality rate a | No. of death | Age‐standardized mortality rate a | No. of death | |
| Global | ||||
| Both | 3.5 (2.8, 4.1) | 13,6450.4 (109,956.4, 163,494.9) | 2.4 (1.8, 3.0) | 205,462.9 (156,544.4, 257,209.4) |
| Male | 6.9 (5.6, 8.3) | 124,555.5 (99,987.6, 149,983.0) | 4.9 (3.7, 6.1) | 19,3281.1 (147,756.8, 241,824.8) |
| Female | 0.6 (0.4, 0.8) | 11,894.9 (8436.9, 15,434.5) | 0.3 (0.2, 0.4) | 12,181.9 (8538.9, 16,514.6) |
| China | ||||
| Both | 10.4 (8.0, 12.9) | 84,136.2 (65,014.5, 105,498.7) | 6.6 (4.9, 8.4) | 139,240.9 (102,619.0, 182,162.8) |
| Male | 20.8 (16.1, 26.2) | 78,917.7 (60,873.0, 100,292.9) | 13.8 (10.2, 18.1) | 134,429.4 (97,859.9, 176,658.5) |
| Female | 1.4 (0.7, 2.1) | 5218.5 (2651.6, 7611.5) | 0.4 (0.2, 0.7) | 4811.5 (2640.8, 7293.9) |
| WHO Regions | ||||
| African Region | 1.1 (0.9, 1.4) | 2417.8 (1872.5, 2955.6) | 0.8 (0.6, 1.0) | 3884.4 (3003.1, 4914.6) |
| Eastern Mediterranean Region | 1.1 (0.8, 1.3) | 1819.8 (1385.8, 2225.0) | 0.8 (0.6, 1.1) | 3419.1 (2608.4, 4284.8) |
| European Region | 2.1 (1.7, 2.4) | 22,081.0 (18,152.4, 25,701.7) | 1.3 (1.0, 1.5) | 20,943.2 (16,390.3, 25,554.9) |
| Region of the Americas | 2.0 (1.6, 2.4) | 12,070.2 (9769.7, 14,357.9) | 1.2 (0.9, 1.5) | 16,493.2 (12,492.3, 20,736.6) |
| South‐East Asia Region | 0.9 (0.7, 1.1) | 6288.2 (4928.9, 7739.9) | 0.6 (0.5, 0.8) | 10,816.2 (8434.6, 13,300.3) |
| Western Pacific Region | 8.0 (6.3, 9.8) | 90,878.9 (71,135.3, 113,081.0) | 5.1 (3.8, 6.6) | 148,417.5 (110,183.6, 192,623.0) |
| SDI levels b | ||||
| Low SDI | 0.9 (0.7, 1.1) | 1885.3 (1436.6, 2332.9) | 0.6 (0.5, 0.7) | 2892.2 (2196.5, 3617.0) |
| Low‐Middle SDI | 1.1 (0.8, 1.3) | 6223.5 (4933.9, 7593.1) | 0.7 (0.6, 0.9) | 101,74.2 (7978.1, 12,436.3) |
| Middle SDI | 5.4 (4.2, 6.7) | 545,85.5 (43,259.0, 68,159.2) | 3.1 (2.3, 4.1) | 82,572.2 (60,685.4, 108,091.6) |
| High‐Middle SDI | 4.6 (3.7, 5.6) | 46,134.0 (36,653.6, 56,475.3) | 3.7 (2.8, 4.8) | 74,471.4 (55,783.1, 97,500.8) |
| High SDI | 2.5 (2.0, 2.9) | 27,561.9 (22,425.8, 32,120.1) | 1.6 (1.3, 2.0) | 35,277.6 (27,498.4, 42,903.7) |
| Gastric cancer | Age‐standardized mortality rate a | No. of death | Age‐standardized mortality rate a | No. of death |
|---|---|---|---|---|
| Global | ||||
| Both | 2.8 (2.3, 3.4) | 109,817.9 (89,440.6, 131,512.7) | 1.3 (1.0, 1.6) | 107,925.6 (84,603.1, 138,447.8) |
| Male | 5.7 (4.6, 6.8) | 99,506.5 (80,639.1, 120,244.8) | 2.6 (2.0, 3.3) | 100,321.5 (78,097.0, 129,829.8) |
| Female | 0.5 (0.4, 0.6) | 10,311.4 (8230.7, 12,572.8) | 0.2 (0.1, 0.2) | 7604.1 (6002.5, 9441.1) |
| China | ||||
| Both | 6.2 (4.7, 8.0) | 51,502.0 (38,094.0, 65,938.7) | 3.1 (2.3, 4.3) | 65,604.0 (48,080.7, 90,844.3) |
| Male | 12.6 (9.2, 16.4) | 48,596.8 (35,521.0, 62,789.0) | 6.5 (4.7, 9.0) | 63,126.6 (45,790.5, 87,892.7) |
| Female | 0.8 (0.6, 1.0) | 2905.2 (2101.2, 3907.7) | 0.2 (0.2, 0.3) | 2477.4 (1685.0, 3483.3) |
| WHO Regions | ||||
| African Region | 0.4 (0.3, 0.5) | 846.6 (621.1, 1064.5) | 0.2 (0.1, 0.3) | 1120.8 (820.9, 1388.4) |
| Eastern Mediterranean Region | 0.9 (0.7, 1.2) | 1567.3 (1115.8, 1972.7) | 0.7 (0.4, 0.8) | 2672.3 (1668.4, 3307.2) |
| European Region | 2.6 (2.2, 3.0) | 27,868.0 (23,506.0, 32,678.0) | 0.9 (0.7, 1.0) | 14,184.9 (11,496.2, 16,867.5) |
| Region of the Americas | 1.4 (1.1, 1.6) | 8245.0 (6729.0, 9726.8) | 0.5 (0.4, 0.6) | 6752.6 (5402.2, 8283.5) |
| South‐East Asia Region | 0.9 (0.7, 1.1) | 6034.1 (4626.2, 7702.8) | 0.5 (0.4, 0.6) | 8205.3 (6232.7, 10,336.5) |
| Western Pacific Region | 5.7 (4.5, 7.1) | 64,667.5 (50,331.5, 79,994.4) | 2.6 (1.9, 3.5) | 74,464.8 (55,897.1, 101,369.7) |
| SDI levels b | ||||
| Low SDI | 0.6 (0.4, 0.7) | 1287.4 (915.8, 1623.4) | 0.4 (0.2, 0.4) | 1709.9 (1126.7, 2128.6) |
| Low‐Middle SDI | 1.0 (0.8, 1.3) | 5874.8 (4621.0, 7462.7) | 0.6 (0.4, 0.7) | 7783.8 (5977.9, 9585.8) |
| Middle SDI | 3.5 (2.7, 4.4) | 35,854.4 (27,880.6, 45,182.3) | 1.6 (1.2, 2.2) | 43,073.5 (32,394.9, 57,603.4) |
| High‐Middle SDI | 4.0 (3.2, 4.8) | 39,723.0 (32,104.8, 47,502.3) | 1.9 (1.5, 2.5) | 39,046.7 (30,049.7, 51,057.1) |
| High SDI | 2.4 (2.0, 2.9) | 26,991.0 (22,570.4, 31,940.5) | 0.7 (0.6, 0.9) | 16,255.2 (13,204.6, 19,906.9) |
Rate per 100,000.
The Sociodemographic Index (SDI) is categorized into five groups based on the level of development. The boundaries for these categories are: Low SDI: 0 to 0.466; Low‐middle SDI: 0.466 to 0.619; Middle SDI: 0.619 to 0.712; High‐middle SDI: 0.712 to 0.810; High SDI: 0.810 to 1.000; SDI: Sociodemographic index.
Globally, a significant decline in the smoking‐attributable mortality rate of UGIC was observed over the past decades. Notably, the decrease in gastric cancer mortality rate was significantly larger than esophageal cancer (Annual Average Percent Change [AAPC]: −2.594, 95% CI: −2.626 to −2.563 vs. AAPC: −1.229, 95% CI: −1.428 to −1.030, p < 0.001) (Figure 2A). In contrast to the declining mortality rates, the absolute number of smoking‐attributable esophageal cancer deaths increased globally between 1990 and 2021, whereas global smoking‐attributable gastric cancer deaths remained broadly stable or slightly declined (Figure 2B).
FIGURE 2.
Joinpoint regression analysis comparison of all smoking‐attributable UGIC burden and AAPC from 1990 to 2021 at the global level (A and B) and China (C and D). The Population Attributable Fraction (PAF) of deaths due to smoking for esophageal and gastric cancer globally and in China among different age groups (E–G).
In China, the smoking‐attributable mortality rates for both UGICs also showed a significant decline over the years. The reduction in mortality was also significantly more pronounced for gastric cancer than esophageal cancer (AAPC difference: 0.763, p < 0.001) (Figure 2C). The number of deaths from both smoking‐attributed esophageal and gastric cancers still increased significantly between 1990 and 2021, with esophageal cancer demonstrating a more marked increase (Figure 2D). The Disability‐Adjusted Life Years (DALYs) changes followed a similar trend to the mortality rates (Figure S3).
Figure 2E–G shows the population attributable fraction (PAF) of deaths due to smoking for esophageal and gastric cancer. Globally, PAF remained stable, with esophageal cancer consistently higher than gastric cancer. In China, smoking‐attributable esophageal cancer deaths were higher than the global average (Figure 2E). PAF showed an upward trend among individuals aged 40 and above, while it declined among younger individuals (under 40) (Figure 2F,G). For gastric cancer, PAF in younger populations remained stable globally but increased in China over the last decade. When the cut‐off age was adjusted to 50 years, the temporal trend of smoking‐attributable late‐onset UGIC deaths remained generally consistent with that observed using the < 40 years cut‐off. For the early‐onset group, the overall trajectory of PAF changes was also broadly similar. The only noticeable difference was in the most recent 5 years; the trend showed a decline as compared to the relative increase pattern observed under the < 40 years criterion (Figure S4).
3.2. APC Analysis for Smoking‐Attributable UGIC Mortality Rate
Age‐group effects in Figure 3A show the largest declines in smoking‐attributable esophageal cancer mortality rate among those aged 35–39 years globally (−4.80%, 95% CI: −4.25 to −5.35) and in China (−4.63%, 95% CI: −3.53 to −5.72). In contrast, those aged over 80 show an increasing trend (local drift > 0). For gastric cancer (Figure 3B), all age groups experienced declines, with slower rates of decline as age increased, more pronounced than for esophageal cancer. Females generally showed a more rapid decrease in disease burden, especially after age 40. In China, the sex difference in annual percentage change of smoking‐attributable UGIC mortality was greater. Detailed net and local drift results are in Tables S2 and S3.
FIGURE 3.
Age‐period‐cohort analysis of smoking‐attributable upper gastrointestinal cancer mortality by cancer type, sex, and regions. Panels A and B present local drift values, indicating age‐specific annual percentage change in mortality rates. Solid lines represent trends for each sex, and shaded areas show 95% confidence intervals. Panels C and D display period rate ratios relative to the reference period (1992–1996), reflecting temporal shifts adjusted for age and cohort. Panels E and F show birth cohort rate ratios compared to the reference cohort (1942–1951), illustrating generational changes in mortality risks.
Period effect (Figure 3C,D) indicated a consistent downward trend in smoking attributable UGIC mortality rate, with the most pronounced decreases observed among females, especially in China. Compared with individuals in the reference 1992–1996 period, Chinese females showed the relative period risk for smoking attributable esophageal cancer of 0.27 (95% CI: 0.22 to 0.34), which is much lower than the global average (relative period risk: 0.44, 95% CI: 0.42 to 0.47). This difference was not observed in gastric cancer (Table S4).
Cohort effects showed a consistent decrease in smoking‐attributable UGIC mortality rates across birth cohorts (Figure 3E,F). For both UGIC, women born before 1937 had a higher risk than men when exposed to smoking. In contrast, for those born after 1967, sex differences were minimal, and risk ratios declined more gently for both sexes. Chinese women of these cohorts exhibited slightly lower risks than Chinese men, particularly for esophageal cancer.
3.3. Decomposition Analysis and Correlation With Sociodemographic Level
Population growth is the main driver for increased smoking‐attributable deaths for esophageal cancer, contributing 160.7% globally and 131.1% in China. Aging further increases this burden, especially in China (China: 75.1% vs. Global: 41.7%). Epidemiological changes significantly reduce the burden, lowering mortality by 102.4% globally and 106.2% in China. Global gastric cancer deaths attributable to smoking decrease primarily due to epidemiological transitions (−1021.9%), though population growth and aging partially offset this trend. In China, population growth (283.4%) and aging (152.72%) drive a substantial increase in mortality, while epidemiological changes (−936.12%) mitigate part of this burden (Figure 4A and Table S5).
FIGURE 4.
Alterations in the number of smoking‐attributable UGIC deaths based on the population‐level determinants of population growth, aging, and epidemiological alteration between 1990 and 2021 at the global level, China, and by SDI quintile (A). Correlation between the sociodemographic index and disease burden (B) and AAPC of esophageal cancer mortality rate (C) and gastric cancer mortality rate (D) by Loess regression. The black dot represents the cumulative effect of all three components. For individual components, a positive value indicates an increase in the smoking‐attributable UGIC death associated with that component, while a negative value signifies a corresponding decrease in smoking‐attributable UGIC deaths. Changes in death numbers due to epidemiological changes refer to alterations in mortality rates while keeping age structure and population size constant. SDI, sociodemographic index; UGIC, Upper Gastrointestinal Cancer.
Smoking‐attributable esophageal cancer has increased in regions with SDI > 0.4 (r = 0.65, p < 0.01). Gastric cancer burden decreased in high SDI regions but increased in low to middle SDI regions (r = 0.58, p = 0.02). Countries with higher SDI show a faster decline in smoking‐attributable cancer mortality, particularly for gastric cancer compared to esophageal cancer (Figure 4B–D).
3.4. Prediction of Smoking Attributable UGIC Burden to 2030
For smoking‐attributable esophageal cancer, mortality rates are predicted to be stable globally and in China from 2022 to 2030. However, the number of deaths increases globally (213,780.5 in 2022 to 263,218.8 in 2030) and in China (144,880.6 in 2022 to 187,567.9 in 2030). Regarding gastric cancer, mortality rates decline globally (1.25 in 2022 to 1.04 in 2030) and in China (3.26 in 2022 to 2.56 in 2030), while the number of deaths remains stable (Figure 5).
FIGURE 5.
Projections of smoking‐attributable UGIC mortality rates and numbers from to 2030 in China and globally. Panel A shows observed and predicted age‐standardized mortality rates (ASMR) for esophageal cancer. Panel B displays the annual number of deaths due to esophageal cancer. Panel C presents ASMR trends for gastric cancer. Panel D shows the corresponding number of deaths from gastric cancer. UGIC, Upper Gastrointestinal Cancer.
4. Discussion
This study presents an updated and comprehensive analysis of the global and Chinese trends in smoking‐attributable UGIC burden over the past three decades. China accounted for more than 60% of worldwide smoking‐related deaths of esophageal and gastric cancers. At the population level, the smoking‐attributable UGIC burden was comparatively higher for esophageal than for gastric cancer, particularly among males. The smoking‐attributable mortality rates for both cancers have decreased globally and in China. We conducted APC analysis, which demonstrated rapid reductions in mortality rates among younger individuals and in recent periods, particularly among females. In contrast, a plateau was observed in mortality rates among older Chinese males with esophageal cancer. However, due to population growth and aging, the absolute number of smoking‐attributable UGIC deaths in China has continued to increase. High‐SDI countries, despite carrying a significant disease burden, have exhibited a faster decline in disease burden.
The smoking‐attributable burden of gastric cancer declined more rapidly than that of esophageal cancer. This reflects advancements in prevention and treatment strategies for gastric cancer, such as H. pylori eradication, cancer screening programs, and improved cancer therapies, which collectively reduced gastric cancer mortality [23]. Robust evidence indicates that H. pylori eradication lowers overall gastric cancer incidence and mortality [24, 25]. Although this does not directly alter the smoking–gastric cancer relative risk, reducing the baseline gastric cancer burden inevitably decreases the absolute number of smoking‐attributable deaths (calculated as PAF × total deaths). This may be due to the lack of early screening programs, regional variations in risk factors like smoking and alcohol, and slower treatment progress compared to gastric cancer. Incidents have surged in regions with high historical smoking rates, such as parts of Asia and Africa. These disparities contribute to the trends, especially where healthcare has not kept up with rising risks.
Our Joinpoint and APC analysis showed significant reductions in smoking‐attributable mortality rates, especially among younger groups and females. This change may be due to several factors. First, as people age, cumulative tobacco exposure increases, particularly among Chinese males, who have historically high tobacco usage [26]. This leads to a sharp increase in the risk of disease and mortality as individuals grow older. This phenomenon also elucidates the trend of higher mortality rates observed in middle‐aged and elderly populations [27]. In contrast, the mortality rate for birth cohorts after 1980 has declined significantly. This is largely due to effective policy interventions, including China's ratification of the World Health Organization Framework Convention on Tobacco Control (FCTC) in 2005 [16]. Consistent with these measures, national survey data show that adult smoking prevalence in China declined from 26.6% in 2018 to 23.2% in 2024 [28, 29, 30], alongside strengthened implementation of the WHO FCTC and MPOWER measures [31]. However, the youth trends are complicated by the emergence of e‐cigarettes [32], so one should cautiously emphasize overall adult prevalence decline and tobacco‐control progress rather than strong cohort‐specific assertions. In parallel, advancements in medical technology, cancer screening programs, and public health education initiatives have collectively led to declining mortality rates across all age groups. However, given the relatively slower manifestation of health benefits in the elderly population, the reduction in mortality rates is more gradual within this demographic and may even reach a plateau in certain cases. This could be attributed to the lower likelihood of adopting preventive health behaviors during the pre‐disease stage among older adults [33].
Our birth cohort analysis revealed that women born before 1937 exhibited a higher smoking‐attributable UGIC mortality risk than men. This is a counterintuitive finding, given the historically higher smoking prevalence among Chinese men. Several factors may help explain this pattern. First, early‐generation Chinese women may have been heavily exposed to second‐hand smoke in domestic environments, particularly from spouses and family members, which has been shown to elevate cancer risk significantly [34, 35, 36]. Additionally, environmental factors such as poor ventilation, indoor cooking with coal or wood, and exposure to cooking oil fumes, common in mid‐20th‐century households of China, may have compounded carcinogenic exposure, especially among women [37, 38]. Moreover, lower socioeconomic status, nutritional deficiencies, and limited access to healthcare may have increased biological vulnerability [39]. These combined exposures and disadvantages may have rendered early‐generation women more susceptible to the harmful effects of tobacco smoke, despite lower levels of active smoking. Future research should explore these sex‐ and cohort‐specific differences more deeply to better inform gender‐responsive cancer prevention strategies and tobacco control policies.
The absolute number of smoking‐attributable UGIC deaths in China and global esophageal cancer cases has continued to increase. This trend can be attributed to the combined effects of population aging and historical smoking exposure, particularly among high‐risk groups such as older Chinese males. According to the “China Statistical Yearbook 2023,” the population aged 60 and above now exceeds 280 million, accounting for 21.1% of the total population [40]. Although the cancer mortality rate has declined, the rapid growth in the elderly population has resulted in an increase in the incidence of other illnesses and their associated deaths. The majority of these fatalities currently occur among men born between the 1950s and 1970s, who are in the peak risk period for cancer. As a result, despite potential reductions in risk for younger generations, they continue to bear the consequences of “historical smoking exposure.”
When stratified according to SDI, the burden of smoking‐attributable UGIC is more pronounced, but also declines more rapidly, in countries with high SDI. The heavier burden can be attributed to historically high smoking prevalence, an aging population, higher life expectancy, and improved cancer reporting and diagnosis [41]. These regions show a rapid decline in smoking‐related UGIC burden due to strict tobacco control policies (e.g., taxation, smoking bans), cessation programs, and anti‐smoking education [42, 43]. The implementation of these initiatives has significantly reduced smoking prevalence over time. In high SDI countries, improved healthcare access includes screenings and early diagnostic tools for timely lesion and cancer detection, leading to earlier intervention and reduced mortality. Additionally, these countries can invest more resources in healthcare infrastructure, including cancer research, prevention programs, and patient care, which further decreases smoking prevalence and reduces overall UGIC risk [44, 45]. It should be noted that, in certain high‐SDI regions such as Japan [46] and the Republic of Korea [47], nationwide endoscopic screening programs for gastric cancer have been implemented and are reported to be associated with mortality reduction. While our ecological analysis does not directly incorporate screening coverage data, such programs may have contributed to the observed faster declines in smoking‐attributable UGIC burden in these settings. In China, gastric cancer screening is not implemented nationwide but through organized programs targeting high‐risk populations. The Upper Gastrointestinal Cancer Early Detection (UGCED) program (since 2012) covers selected rural high‐risk counties, and the Cancer Screening Program in Urban China (CanSPUC) targets high‐risk urban populations. By 2018, over 2.16 million endoscopies had been performed across 194 sites, with high early‐stage detection, but coverage remains limited to specific regions rather than universal implementation [48].
Our research demonstrates several strengths. First, this study provides an updated comparative assessment of smoking‐attributable UGIC burden in China and globally using the GBD 2021 database. This comparison provides valuable insights into regional and global variations in the impact of smoking as a risk factor. Second, through an APC model and decomposition analysis, we explored the underlying epidemiological drivers (for instance, birth cohort, periods, population growth, etc.) of changes in mortality rates and the number of deaths. Third, we examined the relationship between sociodemographic development and shifts in disease burden, enabling us to elucidate the varying rates of disease burden changes across countries at different levels of development.
Nevertheless, this study has some limitations. First, although the GBD database collects disease data on a global scale, some countries or regions still rely on estimated percentages for their data, particularly those with low sociodemographic levels [49, 50]. Another limitation is the inconsistency of disease data collection methods and accuracy across different countries. Second, the ICD‐10 coding scheme adopted by the GBD study includes certain codes such as D00.1 and Z code, which are rarely used in conventional cancer registries and may not represent invasive cancers pathologically; their inclusion follows the standardized GBD framework to harmonize heterogeneous data sources across countries. This should be considered when interpreting the findings. Third, as an ecological analysis based on modeled, population‐level indicators, we could not directly adjust for other potential confounders at the country level, such as alcohol consumption, dietary factors, or H. pylori prevalence. Although the GBD comparative risk assessment framework derives relative risks from confounder‐adjusted cohort studies whenever possible, residual confounding remains possible in our analyses and should be considered when interpreting the findings. Fourth, it is important to highlight that certain unavailable factors in the GBD 2021 study, such as smoking status (current versus past) and histological category (e.g., esophageal squamous cell carcinoma versus adenocarcinoma), could potentially provide more granular insights. Lastly, the data provided in these databases encompass all forms of tobacco use without explicitly distinguishing between different forms of smoking, including traditional cigarettes, chewing tobacco, or emerging trends such as vaping and electronic cigarettes. Given the increasing popularity of alternative smoking products, such as e‐cigarettes and heat‐not‐burn devices, future studies should aim to monitor the use of these products and evaluate their potential impact on the burden of UGIC. Notably, while some regions promote these products as smoking cessation tools, the lack of long‐term evidence of their safety highlights the critical need for further investigation.
5. Conclusions
The global burden and mortality of UGIC attributed to smoking is declining, with more pronounced changes in gastric cancer than in esophageal cancer. However, China's contribution to the global burden of smoking‐attributable UGIC death is still substantial and increasing due to aging and population growth. These ecological findings support prioritizing comprehensive tobacco‐control strategies in high‐burden settings, including substantial tax increases, strict smoke‐free enforcement, expanded cessation services, and regulation of emerging products such as e‐cigarettes.
Author Contributions
Wai. K. Leung: supervision, writing – review and editing, conceptualization, validation. Yanyan Guo: data curation, resources, writing – review and editing, visualization. Yunhao Li: methodology, formal analysis, software, resources, writing – original draft, investigation, visualization, data curation. Man Ping Wang: writing – review and editing, methodology, data curation. Wen‐Qing Li: writing – review and editing, methodology, data curation.
Funding
The authors have nothing to report.
Ethics Statement
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: International classification of diseases, 10th Revision (ICD‐10) codes for esophageal and gastric cancer.
Table S2: Global, China, and different SDI level regions Net drift results for smoking attributable UGIC burden.
Table S3: Mean percent change per calendar year for different age groups regarding smoking attributable UGIC.
Table S4: Risk ratio for smoking attributable UGIC among different time periods globally and in China.
Table S5: Alterations in smoking attributable upper gastrointestinal cancer death and DALYs (Disability‐Adjusted Life Years) numbers based on the population‐level determinants and causes from 1990 to 2021 (Decomposition analysis results).
Figure S1: Risk factor estimation process in the Global Burden of Disease comparative risk assessment framework. Adapted from Institute for Health Metrics and Evaluation (IHME), Global Burden of Disease Study 2021. Used under the IHME Free‐of‐charge Noncommercial User Agreement.
Figure S2: Global distribution of smoking exposure in 2021 (representing by summary exposure value (SEV)).
Figure S3: Joinpoint regression analysis comparison of smoking‐attributable esophageal cancer and gastric cancer Disability Adjusted Life year (DALY) and AAPC from 1990 to 2021.
Figure S4: The Population Attributable Fraction (PAF) of deaths due to smoking for esophageal and gastric cancer globally and in China for early‐onset (< 50 years old) patients and late‐onset (≥ 50 years old) patients.
Acknowledgments
We appreciate the work of the GBD 2021 Study for their contributions to epidemiological evidence.
Data Availability Statement
The original data analyzed in this study were obtained from https://vizhub.healthdata.org/gbd‐results/.
References
- 1. Bray F., Laversanne M., Sung H., et al., “Global Cancer Statistics 2022: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: A Cancer Journal for Clinicians 74 (2024): 229–263. [DOI] [PubMed] [Google Scholar]
- 2. Rawla P. and Barsouk A., “Epidemiology of Gastric Cancer: Global Trends, Risk Factors and Prevention,” Gastroenterology Review 14 (2019): 26–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Trédaniel J., Boffetta P., Buiatti E., Saracci R., and Hirsch A., “Tobacco Smoking and Gastric Cancer: Review and Meta‐Analysis,” International Journal of Cancer 72 (1997): 565–573. [DOI] [PubMed] [Google Scholar]
- 4. Nomura A. M., Wilkens L. R., Henderson B. E., Epplein M., and Kolonel L. N., “The Association of Cigarette Smoking With Gastric Cancer: The Multiethnic Cohort Study,” Cancer Causes & Control 23 (2012): 51–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wang Q.‐L., Xie S.‐H., Li W.‐T., and Lagergren J., “Smoking Cessation and Risk of Esophageal Cancer by Histological Type: Systematic Review and Meta‐Analysis,” JNCI Journal of the National Cancer Institute 109 (2017): djx115. [DOI] [PubMed] [Google Scholar]
- 6. Organization WH , WHO Report on the Global Tobacco Epidemic, 2023: Protect People From Tobacco Smoke (World Health Organization, 2023). [Google Scholar]
- 7. Bai Z., Wang H., Shen C., An J., Yang Z., and Mo X., “The Global, Regional, and National Patterns of Change in the Burden of Nonmalignant Upper Gastrointestinal Diseases From 1990 to 2019 and the Forecast for the Next Decade,” International Journal of Surgery (London, England) 111 (2025): 80–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Meghji J., Mortimer K., Agusti A., et al., “Improving Lung Health in Low‐Income and Middle‐Income Countries: From Challenges to Solutions,” Lancet (London, England) 397 (2021): 928–940. [DOI] [PubMed] [Google Scholar]
- 9. Ward Z. J. and Goldie S. J., “Global Burden of Disease Study 2021 Estimates: Implications for Health Policy and Research,” Lancet 403 (2024): 1958–1959. [DOI] [PubMed] [Google Scholar]
- 10. GBD 2021 Causes of Death Collaborators , “Global Burden of 288 Causes of Death and Life Expectancy Decomposition in 204 Countries and Territories and 811 Subnational Locations, 1990–2021: A Systematic Analysis for the Global Burden of Disease Study 2021,” Lancet 403 (2024): 2100–2132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. GBD 2019 Diseases and Injuries Collaborators , “Global Burden of 369 Diseases and Injuries in 204 Countries and Territories, 1990–2019: A Systematic Analysis for the Global Burden of Disease Study 2019,” Lancet (London, England) 396 (2020): 1204–1222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Reitsma M. B., Flor L. S., Mullany E. C., Gupta V., Hay S. I., and Gakidou E., “Spatial, Temporal, and Demographic Patterns in Prevalence of Smoking Tobacco Use and Initiation Among Young People in 204 Countries and Territories, 1990‐2019,” Lancet Public Health 6 (2021): e472–e481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bleyer A., Barr R., Hayes‐Lattin B., et al., “The Distinctive Biology of Cancer in Adolescents and Young Adults,” Nature Reviews Cancer 8 (2008): 288–298. [DOI] [PubMed] [Google Scholar]
- 14. Alvarez E. M., Force L. M., Xu R., et al., “The Global Burden of Adolescent and Young Adult Cancer in 2019: A Systematic Analysis for the Global Burden of Disease Study 2019,” Lancet Oncology 23 (2022): 27–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kim H. J., Fay M. P., Feuer E. J., and Midthune D. N., “Permutation Tests for Joinpoint Regression With Applications to Cancer Rates,” Statistics in Medicine 19 (2000): 335–351. [DOI] [PubMed] [Google Scholar]
- 16. Chan K. H., Xiao D., Zhou M., Peto R., and Chen Z., “Tobacco control in China,” Lancet Public Health 8 (2023): e1006–e1015. [DOI] [PubMed] [Google Scholar]
- 17. Gao B., Ma S., Zhao W., et al., “Calculated Cancer Risks for Polycyclic Aromatic Hydrocarbon Mixtures in Mainstream Smoke of Cigarettes Sold in China,” Regulatory Toxicology and Pharmacology 142 (2023): 105427. [DOI] [PubMed] [Google Scholar]
- 18. Chernyavskiy P., Little M. P., and Rosenberg P. S., “Correlated Poisson Models for Age‐Period‐Cohort Analysis,” Statistics in Medicine 37 (2018): 405–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Xie Y., Bowe B., Mokdad A. H., et al., “Analysis of the Global Burden of Disease Study Highlights the Global, Regional, and National Trends of Chronic Kidney Disease Epidemiology From 1990 to 2016,” Kidney International 94 (2018): 567–581. [DOI] [PubMed] [Google Scholar]
- 20. Jacoby W. G., “Loess: A Nonparametric, Graphical Tool for Depicting Relationships Between Variables,” Electoral Studies 19 (2000): 577–613. [Google Scholar]
- 21. Knoll M., Furkel J., Debus J., Abdollahi A., Karch A., and Stock C., “An R Package for an Integrated Evaluation of Statistical Approaches to Cancer Incidence Projection,” BMC Medical Research Methodology 20 (2020): 257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Riebler A. and Held L., “Projecting the Future Burden of Cancer: Bayesian Age‐Period‐Cohort Analysis With Integrated Nested Laplace Approximations,” Biometrical Journal 59 (2017): 531–549. [DOI] [PubMed] [Google Scholar]
- 23. Ford A. C., Yuan Y., and Moayyedi P., “ Helicobacter pylori Eradication Therapy to Prevent Gastric Cancer: Systematic Review and Meta‐Analysis,” Gut 69 (2020): 2113–2121. [DOI] [PubMed] [Google Scholar]
- 24. de Oliveira Silva L. G., Lemos F. F. B., and de Melo F. F., “Eradication of Helicobacter Pylori and Its Impact on the Incidence of Gastric Cancer,” AME Clinical Trials Review 3 (2025): 27. [Google Scholar]
- 25. Yan L., Chen Y., Chen F., et al., “Effect of Helicobacter pylori Eradication on Gastric Cancer Prevention: Updated Report From a Randomized Controlled Trial With 26.5 Years of Follow‐Up,” Gastroenterology 163 (2022): 154–162.e3. [DOI] [PubMed] [Google Scholar]
- 26. Fong G. T., Yuan J., Craig L. V., et al., “Achieving the Goals of Healthy China 2030 Depends on Increasing Smoking Cessation in China: Comparative Findings From the ITC Project in China, Japan, and the Republic of Korea,” China CDC Weekly 3 (2021): 463–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Yuan J. M., Ross R. K., Wang X. L., Gao Y. T., Henderson B. E., and Yu M. C., “Morbidity and Mortality in Relation to Cigarette Smoking in Shanghai, China. A Prospective Male Cohort Study,” JAMA 275 (1996): 1646–1650. [PubMed] [Google Scholar]
- 28. Pei D., Popova L., Chowdhury P., Shi J., and Njie G., “Exposure to Anti‐and Pro‐Smoking Messages Among Adults in China: Results From the Global Adult Tobacco Survey, 2018,” PLoS One 19 (2024): e0304028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Organization WH , WHO Global Report on Trends in Prevalence of Tobacco Use 2000–2030 (World Health Organization, 2024). [Google Scholar]
- 30. Zeng X., Liu S., Liu Y., and Xiao L., “Smoking Prevalence in Urban and Rural Areas—China, 2024,” China CDC Weekly 7 (2025): 751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Dubray J., Schwartz R., Chaiton M., O'Connor S., and Cohen J. E., “The Effect of MPOWER on Smoking Prevalence,” Tobacco Control 24 (2015): 540–542. [DOI] [PubMed] [Google Scholar]
- 32. Zhang G., Zhan J., and Fu H., “Trends in Smoking Prevalence and Intensity Between 2010 and 2018: Implications for Tobacco Control in China,” International Journal of Environmental Research and Public Health 19 (2022): 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Ko F. W. S., “Smoking: Quit to Win,” Respirology 28 (2023): 681–682. [DOI] [PubMed] [Google Scholar]
- 34. Jiang J., Pang H., Liu B., et al., “Effects of Active, Passive, and Combined Smoking on Cervical Cancer Mortality: A Nationwide Proportional Mortality Study in Chinese Urban Women,” Cancer Causes & Control 26 (2015): 983–991. [DOI] [PubMed] [Google Scholar]
- 35. Yang L., Tong E. K., Mao Z., and Hu T. W., “Exposure to Secondhand Smoke and Associated Factors Among Non‐Smoking Pregnant Women With Smoking Husbands in Sichuan Province, China,” Acta Obstetricia et Gynecologica Scandinavica 89 (2010): 549–557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Su Z., McDonnell D., Abbas J., Shi L., Cai Y., and Yang L., “Secondhand Smoke Exposure of Expectant Mothers in China: Factoring in the Role of Culture in Data Collection,” JMIR Cancer 7 (2021): e24984. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Kim C., Gao Y. T., Xiang Y. B., et al., “Home Kitchen Ventilation, Cooking Fuels, and Lung Cancer Risk in a Prospective Cohort of Never Smoking Women in Shanghai, China,” International Journal of Cancer 136 (2015): 632–638. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Chowdhury S., Pillarisetti A., Oberholzer A., et al., “A Global Review of the State of the Evidence of Household Air Pollution's Contribution to Ambient Fine Particulate Matter and Their Related Health Impacts,” Environment International 173 (2023): 107835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Gu D., Sautter J., Huang C., and Zeng Y., “Health Inputs and Cumulative Health Deficits Among the Older Chinese,” Social Science & Medicine 72 (1982): 806–814. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. China NBoSo , “China Statistical Yearbook 2023,” (2023).
- 41. Levy D. T., Romano E., and Mumford E., “The Relationship of Smoking Cessation to Sociodemographic Characteristics, Smoking Intensity, and Tobacco Control Policies,” Nicotine & Tobacco Research 7 (2005): 387–396. [DOI] [PubMed] [Google Scholar]
- 42. Escobedo L. G., Anda R. F., Smith P. F., Remington P. L., and Mast E. E., “Sociodemographic Characteristics of Cigarette Smoking Initiation in the United States: Implications for Smoking Prevention Policy,” JAMA 264 (1990): 1550–1555. [PubMed] [Google Scholar]
- 43. Vangeli E. and West R., “Sociodemographic Differences in Triggers to Quit Smoking: Findings From a National Survey,” Tobacco Control 17 (2008): 410–415. [DOI] [PubMed] [Google Scholar]
- 44. Myung S.‐K., Seo H. G., Cheong Y.‐S., Park S., Lee W. B., and Fong G. T., “Association of Sociodemographic Factors, Smoking‐Related Beliefs, and Smoking Restrictions With Intention to Quit Smoking in Korean Adults: Findings From the ITC Korea Survey,” Journal of Epidemiology 22 (2012): 21–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Yang T., Li F., Yang X., et al., “Smoking Patterns and Sociodemographic Factors Associated With Tobacco Use Among Chinese Rural Male Residents: A Descriptive Analysis,” BMC Public Health 8 (2008): 1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Yashima K., Shabana M., Kurumi H., Kawaguchi K., and Isomoto H., “Gastric Cancer Screening in Japan: A Narrative Review,” Journal of Clinical Medicine 11 (2022): 11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Jun J. K., Choi K. S., Lee H. Y., et al., “Effectiveness of the Korean National Cancer Screening Program in Reducing Gastric Cancer Mortality,” Gastroenterology 152 (2017): 1319–1328.e7. [DOI] [PubMed] [Google Scholar]
- 48. Fan X., Qin X., Zhang Y., et al., “Screening for Gastric Cancer in China: Advances, Challenges and Visions,” Chinese Journal of Cancer Research 33 (2021): 168–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Susmann H., Alexander M., and Alkema L., “Temporal Models for Demographic and Global Health Outcomes in Multiple Populations: Introducing a New Framework to Review and Standardise Documentation of Model Assumptions and Facilitate Model Comparison,” International Statistical Review 90 (2022): 437–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Guthold R., Johansson E. W., Mathers C. D., and Ross D. A., “Global and Regional Levels and Trends of Child and Adolescent Morbidity From 2000 to 2016: An Analysis of Years Lost due to Disability (YLDs),” BMJ Global Health 6 (2021): e004996. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: International classification of diseases, 10th Revision (ICD‐10) codes for esophageal and gastric cancer.
Table S2: Global, China, and different SDI level regions Net drift results for smoking attributable UGIC burden.
Table S3: Mean percent change per calendar year for different age groups regarding smoking attributable UGIC.
Table S4: Risk ratio for smoking attributable UGIC among different time periods globally and in China.
Table S5: Alterations in smoking attributable upper gastrointestinal cancer death and DALYs (Disability‐Adjusted Life Years) numbers based on the population‐level determinants and causes from 1990 to 2021 (Decomposition analysis results).
Figure S1: Risk factor estimation process in the Global Burden of Disease comparative risk assessment framework. Adapted from Institute for Health Metrics and Evaluation (IHME), Global Burden of Disease Study 2021. Used under the IHME Free‐of‐charge Noncommercial User Agreement.
Figure S2: Global distribution of smoking exposure in 2021 (representing by summary exposure value (SEV)).
Figure S3: Joinpoint regression analysis comparison of smoking‐attributable esophageal cancer and gastric cancer Disability Adjusted Life year (DALY) and AAPC from 1990 to 2021.
Figure S4: The Population Attributable Fraction (PAF) of deaths due to smoking for esophageal and gastric cancer globally and in China for early‐onset (< 50 years old) patients and late‐onset (≥ 50 years old) patients.
Data Availability Statement
The original data analyzed in this study were obtained from https://vizhub.healthdata.org/gbd‐results/.