The GBD study, initiated in 1990, offers a foundational data source for global health research through its comprehensive scope and systematic methodology. This international collaborative project covers 204 countries and territories, evaluating 371 diseases and injuries alongside 88 risk factors. These locations are further categorized into 21 regions and 7 super-regions (24). In our study, we utilized secondary data from the GBD collaborative platform to assess the burden of TBL cancer attributable to SHS exposure. We analyzed relevant data at global, national, and regional levels for 2021 and projected trends from 2022 to 2046.

The codes for TBL cancer were sourced from the International Classification of Diseases (ICD)-9 and ICD-10 coding books. The ICD-9 codes include 162-162.9, 209.21, V10.1-V10.20, V16.1-V16.2, and V16.4-V16.40, while ICD-10 codes include C33, C34-C34.92, Z12.2, Z80.1-Z80.2, and Z85.1-Z85.20 (25). SHS exposure is defined as the inhalation of tobacco smoke in household, occupational, or public environments (26). Regarding the attribution estimation methods for TBL cancer attributable to SHS exposure, GBD 2021 followed the general framework established by Comparative Risk Assessment (CRA), which can be divided into six key steps: identifying SHS-TBL cancer pairs based on systematic reviews and meta-regressions; estimating relative risks from SHS-TBL cancer risk curves using the meta-regressions, regularized, and trimmed (MR-BRT) method; estimating the age-sex-location-year distribution of ambient SHS exposure levels; setting the theoretical minimum risk exposure level (TMREL) as zero exposure, meaning complete avoidance of SHS exposure; considering different exposure scenarios in household environments, workplaces, and public places, using spatiotemporal modeling methods to estimate exposure distributions in data-sparse regions; and estimating population attributable fractions and attributable burden using formulas that account for risk functions, individual exposure distributions across age-sex-location-years, and TMREL, while quantifying uncertainty intervals of estimates through 1,000 uncertainty draws (8, 25, 27). Notably, the GBD 2021 database excludes data on SHS-attributable TBL cancer burden for populations under 25 years old; hence, our study focuses on those aged 25 and older. Through the GBD Results Tool¹, we extracted annual data on SHS-attributable TBL cancer deaths, disability-adjusted life years (DALYs), age-standardized mortality rate (ASMR), and age-standardized DALY rate (ASDR) from 1992 to 2021, stratified by sex, age, and geographic location. The GBD database includes 21 GBD regions and 204 countries or territories, which are categorized into five SDI quintiles: low, low-middle, middle, high-middle, and high. The SDI is a composite development indicator comprising total fertility rate, average educational attainment, and per capita disposable income. It ranges from 0 (theoretical minimum development) to 1 (theoretical maximum development), reflecting a country’s overall social and economic development. Moreover, this study uses 15 age groups (in 5-year intervals from 25 to 94 years and ≥95 years), providing a multidimensional, detailed framework for evaluating the global burden of TBL cancer attributable to SHS exposure.

To examine the burden of TBL cancer attributable to SHS exposure, we implemented a multi-level statistical analysis. We used the Estimated Annual Percentage Change (EAPC) to quantify long-term trends in the ASMR and ASDR rate from 1992 to 2021. We applied a log-linear regression model: ln ( y ) = α + β x + ε where y represents ASMR or ASDR, and x denotes the year. The EAPC and its 95% CI were calculated as: 100 × ( exp ( β ) − 1 )

Temporal trends were categorized as increasing if the 95% CI of the EAPC estimate exceeded 0, decreasing if the 95% CI was below 0, and stable if the 95% CI included 0.

We evaluated the associations between ASMR/ASDR and the SDI using Pearson correlation and linear regression analyses, with statistical significance determined by two-tailed p-values < 0.05.

The age-period-cohort (APC) model, a cornerstone in contemporary epidemiological research, disentangles epidemiological indicators into three temporal dimensions: age, period, and birth cohort. Age effects describe the impact of biological aging on disease occurrence, period effects capture the impact of specific historical periods on disease incidence, and cohort effects reflect variations in risk factor exposure across different birth cohorts (28). We divided the study population into 15 consecutive 5-year age groups (from 25 to 99 years) during the period from 1992 to 2021, covering 20 consecutive birth cohorts. We analyzed key parameters, using the APC Web Tool provided by the National Cancer Institute. Parameter significance was assessed with Wald chi-square tests.

Furthermore, we employed the Bayesian age-period-cohort (BAPC) model with integrated nested Laplace approximations (using R packages BAPC and INLA) to project ASMR for TBL cancer attributable to SHS exposure from 2022 to 2046 (29). The ‘BAPC ()’ and ‘INLA ()’ functions were employed to fit the age-period-cohort model and estimate posterior distributions of model parameters, respectively. Data were structured into 5-year age intervals (specific age groups were categorized as under 5 years, 5–9 years, 10–14 years, etc., up to 95 years and above, totaling 20 age groups), with a time span from 1990 to 2021, incrementing annually. Population data and disease rates for each age group and period were prepared in matrix format, with rows representing age groups and columns representing periods. These data were input into the ‘BAPC ()’ function, with age and time dimensions specified accordingly. Standard population data were sourced from the GBD Study 2021, available at [IHME GBD 2021 Demographics] (Global Burden of Disease Study 2021 (GBD 2021) Demographics 1950-2021|GHDx). This dataset provides age-specific population weights from 1950 to 2021, which were used for standardized analysis in this study. The 95% uncertainty intervals (UIs) for predicted rates were calculated using posterior distributions derived from the BAPC model. These data were obtained through the ‘INLA’ package, which provides posterior marginal distributions for each parameter. The uncertainty intervals were calculated as the 2.5th and 97.5th percentiles of the posterior distributions (29). Compared to other forecasting methods, this approach not only provides superior prediction coverage and accuracy, but also offers important scientific evidence for future public health policy development (30). All statistical analyses were conducted in R software (version 4.4.1), ensuring methodological consistency and reproducibility.

The global burden of TBL cancer attributable to SHS exposure has shown significant changes from 1992 to 2021. The number of global deaths increased from 59,998.3 (95% Uncertainty Interval [UI]: 7,076.9 to 113,306.0) in 1992 to 97,910.8 (95% UI: 11,955.2 to 184,912.9) in 2021. Despite the rise in absolute numbers, the ASMR decreased from 1.44 (95% UI: 0.17 to 2.73) to 1.14 (95% UI: 0.14 to 2.15) per 100,000 population, with an EAPC of −0.92 (95% CI: −0.98 to −0.87). DALYs increased from 1,654,275.7 (95% UI: 195,097.3 to 3,113,584.6) to 2,355,866.0 (95% UI: 290,210.9 to 4,442,996.3), while the ASDR decreased from 38.09 (95% UI: 4.49 to 71.66) to 26.93 (95% UI: 3.32 to 50.83) per 100,000 population, with an EAPC of −1.30 (95% CI: −1.35 to −1.25) (Table 1).

Significant disparities in the burden and trends of TBL cancer were evident across SDI regions in 2021. High-middle SDI regions experienced the highest burden with 39,124.1 deaths (95% UI: 4,613.2 to 73,341.1), followed by middle SDI regions with 35,511.4 deaths (95% UI: 4,496.6 to 67,456.6). In contrast, low SDI regions had the lowest burden with 769.7 deaths (95% UI: 93.9 to 1,556.2) (Table 1). High SDI regions demonstrated the most significant improvements in both the ASMR (EAPC: −2.64 [95% CI: −2.75 to −2.54]) and ASDR (EAPC: −3.01 [95% CI: −3.12 to −2.90]), with deaths decreasing to 16,945.3 (95% UI: 2,216.0 to 32,742.6) and DALYs declining to 380,284.8 (95% UI: 49,959.2 to 738,239.8) (Table 1). Conversely, both high-middle and middle SDI regions showed increases in absolute deaths and DALYs, despite moderate decreases in age-standardized rates.

Among GBD regions in 2021, East Asia, Western Europe, and High-income North America bore substantial burdens. East Asia faced the highest burden with 59,196.3 deaths (95% UI: 7,267.3 to 111,539.5) and 1,387,974.7 DALYs (95% UI: 172,610.9 to 2,594,495.8), showing minimal improvement in ASMR (EAPC: 0.06 [95% CI: −0.07 to 0.18]) (Table 1). Western Europe ranked second with 6,226.9 deaths (95% UI: 775.5 to 12,307.0) and 150,798.7 DALYs (95% UI: 19,175.0 to 295,048.8), followed by High-income North America, which demonstrated the largest improvements in both ASMR (EAPC: −3.87 [95% CI: −4.01 to −3.73]) and ASDR (EAPC: −4.22 [95% CI: −4.36 to −4.09]). Western Europe also showed notable improvements (EAPC in ASMR: −2.64 [95% CI: −2.73 to −2.54]; EAPC in ASDR: −2.85 [95% CI: −2.97 to −2.73]), while Oceania notably exhibited an upward trend in ASMR (EAPC: 0.29 [95% CI: 0.21 to 0.38]) (Table 1).

In 2021, significant disparities in the disease burden of TBL cancer attributable to SHS exposure were observed across national regions. The highest ASMR were recorded in Montenegro (3.45, 95% UI: 0.38 to 7.06), China (2.80, 95% UI: 0.35 to 5.27), and North Macedonia (2.55, 95% UI: 0.29 to 5.31) (Supplementary Table S1; Figure 1A). A similar distribution pattern was seen in the ASDR rates, with Montenegro (84.34, 95% UI: 9.36 to 170.96), China (63.32, 95% UI: 7.95 to 117.85), and North Macedonia (63.26, 95% UI: 7.03 to 132.65) reporting the highest burdens (Supplementary Table S1; Figure 1B). Conversely, the lowest burdens were predominantly observed in African regions, with Nigeria (0.02, 95% UI: 0.00 to 0.04), Kenya (0.04, 95% UI: 0.01 to 0.09), and Burundi (0.05, 95% UI: 0.01 to 0.11) reporting the lowest rates (Supplementary Table S1; Figure 1A). Between 1992 and 2021, most countries showed declining trends in both ASMR and ASDR rates, mainly in Western Europe and North America, such as the United Kingdom, Ireland, Denmark, Switzerland, the United States, Canada and Mexico (Supplementary Table S3; Supplementary Figure S2). In contrast, a few countries showed an upward trend mainly in Africa, such as Egypt, Lesotho, Guinea-Bissau, Mali, Mozambique, Angola, Georgia (Supplementary Table S3; Supplementary Figure S2). Developed nations typically demonstrated sustained improvements, whereas certain developing regions faced deteriorating conditions. Notably, some African countries consistently reported the lowest disease burdens, exemplified by Nigeria and Burundi. This pattern likely reflects limitations in disease surveillance and healthcare system capacities rather than genuinely lower disease burdens.

The disease burden of TBL cancer attributable to SHS exposure also shows significant differences between genders. Across both sexes, males consistently experienced a higher burden, with 56,848.1 deaths (95% UI: 6,655.0 to 109,070.9) and an ASMR of 1.44 (95% UI: 0.17 to 2.78) in 2021, compared to 41,062.7 deaths (95% UI: 5,535.9 to 78,059.2) and an ASMR of 0.89 (95% UI: 0.12 to 1.69) in females. Likewise, DALYs for males rose to 1,359,556.6 (95% UI: 158,640 to 2,596,453.4) with an ASDR of 32.88 (95% UI: 3.85 to 62.82), while DALYs for females reached 996,309.4 (95% UI: 135,988.2 to 1,907,079) with an ASDR of 21.81 (95% UI: 2.98 to 41.75) (Table 1). However, males showed greater improvement in age-standardized rates compared to females (EAPC in ASMR: −1.11 [95% CI: −1.19 to −1.03]; EAPC in ASDR: −1.51 [95% CI: −1.58 to −1.44]) (Table 1). Notably, TBL cancer deaths attributable to SHS exposure for both sexes followed a similar age pattern, increasing with age before gradually decreasing, predominantly concentrated in the 65–74 age group (Figure 2A). The burden of DALYs showed similar age and sex distribution patterns but peaked earlier at ages 65–69, with the primary burden concentrated in the 55–69 age group (Figure 2B). Notably, the disease burden was consistently higher in males than in females, a pattern that persisted across both developed and developing countries.

A national-level analysis in 2021 revealed a significant positive correlation between the ASMR of TBL cancer attributable to SHS exposure and the SDI (R = 0.441, p < 0.01) (Figure 3A). This relationship exhibited a distinct non-linear pattern: when SDI < 0.75, ASMR gradually increased with rising SDI; however, when SDI > 0.75, ASMR began to decline, forming an inverted U-shaped distribution. Significant geographical disparities were evident in the regional distribution. Several countries deviated notably from expected ASMR levels. Montenegro, China, North Macedonia, and Greenland exhibited substantially higher ASMR than anticipated based on their SDI. At the same time, many African countries — including Nigeria, Malawi, Kenya, and Burundi — demonstrated significantly lower ASMR compared to nations with similar SDI levels (Figure 3A). Notably, several high-SDI countries such as Switzerland, Sweden, Finland, and Singapore maintained relatively low ASMR levels (Figure 3A). The ASDR displayed distribution patterns highly consistent with ASMR, showing a significant positive correlation with SDI (R = 0.446, p < 0.01) and exhibiting the same inverted U-shaped trend (Figure 3B). This consistency not only reinforces the reliability of the findings but also underscores the complex relationship between the TBL cancer burden attributable to SHS exposure and socioeconomic development levels.

The age effect revealed significant heterogeneity across different SDI regions. The high-middle SDI region exhibited the highest death rates across all age groups, with a linear increase from ages 25–29 to 85–89, followed by a gradual decline (Figure 4A). This may be due to the fact that these two regions experienced decades of high smoking rates and SHS exposure, and the implementation of relevant tobacco control policies lagged behind that of High SDI regions. The middle SDI region and the global average displayed similar inverted V-shaped distributions, peaking at ages 85–89 before decreasing (Figure 4A). The high SDI region showed a distinct pattern: death rates increased gradually from ages 25–29 to 65–69 (second only to the high-middle SDI region), then stabilized and declined below the global average (Figure 4A). Notably, in the middle SDI region, the death rate surpassed both the high SDI region and the global average at ages 65–69, rising to the second-highest level. In contrast, the low-middle and low SDI regions showed gradual increases in death rates from ages 25–29 to 75–79, followed by relative stability, with the disease burden consistently remaining below the global level (Figure 4A).

Sex-stratified analyses revealed notable differences, despite similar overall mortality patterns between males and females. In the high-middle SDI region, middle SDI region, and globally, male death rates peaked at ages 90–94 before declining rapidly (Supplementary Figure S5A). In contrast, female death rates peaked at 85–89, followed by an initial decline and then a secondary increase (Supplementary Figure S5B). Particularly in the high SDI region, male death rates remained relatively stable after ages 65–69, consistently staying below the global average (Supplementary Figure S5A). In comparison, female death rates began to fall below the global level from ages 55–59, displaying this trend approximately 10 years earlier than males (Supplementary Figure S5B). This may be due to female having less cumulative effects from SHS exposure than males.

Period effect analysis revealed an overall declining trend from 1992 to 2021. The high SDI region demonstrated the most substantial improvement, with the period rate ratio (RR) decreasing steadily from its peak in 1992–1996 to the lowest level in the most recent period (Figure 4C). Although other SDI regions also exhibited declining trends, their period RR values exceeded the global average after 2002–2006, particularly in the low-middle SDI region (Figure 4C). This region showed a progressive increase, rising from the lowest level in 1992–1996 to the highest level after 2002–2006 (Figure 4C). This may be related to the industrialization process and economic development. Notably, before 2002–2006, the high SDI region had the highest period RR among males (Supplementary Figure S5C), while the middle SDI region exhibited the highest period RR among females (Supplementary Figure S5D). After 2002–2006, period RR values in all SDI regions—except the high SDI region—exceeded the global average among females (Supplementary Figure S5D).

Cohort effect analysis showed that in the high SDI region, the cohort RR increased slightly between the 1897–1901 and 1917–1921 birth cohorts before declining, then decreased consistently from the highest to the lowest level after the 1942–1946 cohort (Figure 4E). This could be related to the development of medical and the improvement of people’s health awareness. The middle SDI and high-middle SDI regions, along with the global average, followed a similar trend, with brief increases from the 1897–1901 to 1932–1936 cohorts before gradually declining (Figure 4E). However, these regions maintained cohort RR values above the global average after 1942–1946 (Figure 4E). In contrast, the low-middle and low SDI regions showed minimal improvement in cohort RR, eventually surpassing other regions and the global average after 1942–1946 (Figure 4E). Sex disparities were most evident before the 1942–1946 cohort. Among males, the middle SDI region recorded the lowest cohort RR (Supplementary Figure S5E), while among females, the high-middle SDI region had the lowest (Supplementary Figure S5F). Importantly, the burden of DALYs from TBL cancer attributable to SHS exposure exhibited similar patterns to death rates across age, period, and cohort effects (Supplementary Figure S6).

The ASMR for TBL cancer attributable to SHS exposure is projected to decline gradually from 2022 to 2046, with notable differences between sexes. In 2022, the overall ASMR is estimated at 1.12 (95% UI: 1.10 to 1.15), with significantly higher ASMR in men than in women (Figure 5). By 2046, the overall ASMR is expected to further decline to 0.90 (95% UI: 0.24 to 1.56), with the male ASMR projected to decrease to 1.03 (95% UI: 0.19 to 1.87) and the female ASMR is expected to decline to 0.85 (95% UI: 0.15 to 1.55) (Figure 5).