The data for this study comes from the GBD 2021(https://vizhub.healthdata.org/gbd-results/) study, which provides a comprehensive assessment of population, incidence, and death rates for 371 injuries and diseases and 88 risk factors, forming a disease burden database covering 204 countries and regions (18). This database integrates monitoring and research results from multiple institutions, utilizing a variety of data sources including systematic reviews of literature, survey data, surveillance data, and clinical data from inpatient and outpatient visits, among others. During its construction, significant attention was paid to the quality and completeness of the data. Advanced statistical models were employed to adjust and impute data, thereby enhancing the accuracy and representativeness of the dataset.

Estimates regarding the incidence, prevalence, deaths, and disability adjusted life years (DALYs) of LC, along with their 95% uncertainty intervals (UIs), were sourced from the GBD 2021 database.

Descriptive analyses were conducted on the global LC prevalence, incidence, deaths, and DALYs stratified by gender and age. Additionally, these disease burden indicators were analyzed across income levels and geographic regions. To assess temporal trends from 1990 to 2021, we calculated the estimated annual percentage change (EAPC) for age-standardized incidence, prevalence, deaths, and DALYs. The EAPC, derived from a regression model fitted with natural logarithms [ I n ( rate ) = α + β ( calendar year ) + ϵ ], represents the annual trend magnitude, calculated as 100 * ( exp ( β ) − 1 ) (19).

An upward trend was defined when both the EAPC estimate and the lower bound of its 95% confidence interval (CI) exceeded 0, whereas a downward trend was identified when both the EAPC estimate and the upper 95% CI bound were below 0. This metric quantifies the pace and fundamental characteristics of cancer progression, enabling predictions of future trajectories and evidence-based guidance for public health interventions.

Temporal trends in incidence, prevalence, deaths, and DALYs were analyzed using the Joinpoint Regression software (version 4.7.0.0), which identifies statistically significant inflection points (“joinpoints”) by fitting segmented linear models on a logarithmic scale. Joinpoints represent optimal inflection points where significant trend changes occurred, determined through segmented regression analysis (20). The natural logarithm of the observed rate was fitted to the regression line using the calendar year as the regression variable. The Joinpoint regression model formula is: E [ y / x ] = β 0 + β 1 x + δ 1 ( x − τ 1 ) + + … + δ k ( x − τ k ) + , where y represents age-standardized incidence, prevalence, deaths, or DALY rates; x denotes calendar year; β is the intercept; δ_i (i=1, 2,…, k) indicates regression coefficients for segmented intervals; τ_i (i=1, 2,…, k) marks joinpoint years; and k signifies the number of inflection points.

The model calculates annual percent change (APC), average annual percent change (AAPC), and corresponding 95% CI to characterize trends across identified segments. An upward trend was defined when either APC or AAPC > 0 with the 95% CI excluding 0, while a downward trend was identified when APC or AAPC < 0 with the 95% CI excluding 0. Age-standardized rates were considered temporally stable if the 95% CI included 0.

The age-period-cohort (APC) model based on the Poisson distribution can simultaneously estimate the net effect of age, period, and cohort on the burden of disease (21). The expression of the APC model is usually written as follows: Y = log ( M ) = μ + α × a g e 1 + β × p e r i o d 1 + γ × c o h o r t 1 + ∈ , here M denotes the rate of the corresponding age group, μ denotes the intercept term, and α, β and γ denote the age, period and cohort effects, respectively, and ϵ is the random error. Rate, μ denotes the intercept term, α, β, and γ denote the age, period, and cohort effects, respectively, and ϵ is the random error.

The APC model requires equal time intervals for age, period and cohort. In the GBD 2021 database, the incidence and deaths rates of LC were not recorded for population aged 0–19 years, so the age analysis started from 20 years. At the same time, as the age group 95 years and above did not meet the data format requirements of the APC model, this age group was excluded from the analysis. As a result, the age groups were categorized into 15 groups (starting at age 20, with age groups every 5 years). To match the age classification, the data were further collapsed into 6 periods (1992-1996, 1997-2001, 2002-2006, 2007-2011, 2012-2016, and 2017-2021) and 20 equally spaced birth cohorts (starting in 1902, with one birth cohort every 5 years).

Based on trends in the disease burden of LC from 1990 to 2021, we utilized the BAPC model to forecast the incidence, prevalence, deaths, and DALYs of LC from 2022 to 2050 (22). The BAPC model was executed using the INLA package in R software. The model was constructed based on the number of cases per age group, period, and gender, assuming that these follow a Poisson process, with means equal to the product of the population at risk and the estimated incidence rate. The logarithm of the incidence rates was estimated as a linear combination of gender-specific intercepts, overdispersion effects, age effects, period effects, and cohort effects. For age, period, and cohort effects, a second-order random walk prior was adopted, with their sum constrained to zero; log-gamma priors were applied to the precision parameters, where the scale and shape parameters for age, period, and cohort effects were set to 1 and 0.00005 respectively, whereas those for the overdispersion effect were set to 1 and 0.005. An average-zero Gaussian prior distribution was assumed for the overdispersion parameter.

The study analyzed the data using a robust decomposition method that attributed the differences in incidence, prevalence, deaths, and DALYs between the two time points to changes in three independent factors: Aging (A), Population (P), and Epidemiological change (M).

The basic equation is as follows: A = Ma + Iam + Ipa + Ipam , where A denotes the main effect of population ageing, Ma is the effect of age-specific death rates, Iam is the interaction effect of ageing and age-specific rates, Ipa is the interaction effect of ageing and population growth, and Ipam is the joint interaction effect of the three. p = M p + I p m + I p a + I p a m , where P denotes the main effect of population growth, Mp is the effect of population growth, Ipm is the interaction effect of population with age-specific rates, Ipa is the interaction effect of population with ageing, and Ipam is the joint interaction effect of all three. M = M m + I p m + I a m + I p a m , where M denotes the main effect of age-specific rate change, Mm is the effect of age-specific rates, Ipm is the age-specific rate interaction effect with population, Iam is the interaction effect of age-specific rates with ageing, and Ipam is the joint interaction effect of all three.

This method decomposes disease-related data changes into factors caused by A, P, and M. The results of the decomposition include the effect of each factor on the disease. The decomposition results include the absolute and relative contribution of each factor to disease (23).

All statistical evaluations and data representations were conducted utilizing R (version 4.4.2) and JD_GBDR (V2.45.1, Jingding Medical Technology Co., Ltd.). Descriptive analytics were compiled for all principal variables, with findings expressed as means accompanied by 95% UIs or 95% CIs. For the analysis of trends, a p-value threshold of less than 0.05 was deemed statistically significant.

In 2021, there were a total of 200,883 newly identified cases of LC worldwide (95% UI: 186,941–216,097), with 171,788 cases in males (95% UI: 159,470–186,042) and 29,094 cases in females (95% UI: 24,974–33,679). The cumulative number of existing LC cases globally reached 1,103,683 cases (95% UI: 1,033,145–1,186,559), which includes 939,924 cases in males (95% UI: 876,345–1,011,203) and 163,759 cases in females (95% UI: 144,050–186,894). Deaths attributed to LC numbered 117,251 cases (95% UI: 109,354–125,952), including 100,392 male fatalities (95% UI: 93,350–108,829) and 16,858 female fatalities (95% UI: 14,208–19,875). Moreover, LC was responsible for a considerable total of 3,143,308 DALYs (95% UI: 2,922,791–3,383,513), with males accounting for 2,694,178 years (95% UI: 2,491,889–2,926,037) and females for 449,130 years (95% UI: 377,884–540,000). In the year 2021, the worldwide age-standardized incidence of LC was recorded at 2.29 per 100,000 individuals (95% UI: 2.13-2.47), while the age-standardized prevalence stood at 12.56 per 100,000 individuals (95% UI: 11.76-13.49). Additionally, the age-standardized death rate was noted at 1.35 per 100,000 individuals (95% UI: 1.26-1.45), and the age-standardized DALYs were calculated at 35.80 per 100,000 person-years (95% UI: 33.29-38.54). From 1990 to 2021, the EAPC and its confidence intervals for all metrics related to LC were negative, signifying a notable global reduction in the burden of LC ( Table 1 ).

In high SDI regions, the age - age-standardized incidence rate of LC dropped from 3.64 per 100,000 population in 1990 (95% Ul: 3.53 - 3.75) to 2.36 per 100,000 population in 2021 (95% Ul: 2.23 - 2.46), with an EAPCs of - 1.52%. Similarly, the age - age-standardized prevalence, deaths, and DALYs in this region also exhibited a significant downward trend, decreasing by - 1.15%, - 2.59%, and - 2.81% respectively. As the SDI level declined, relevant indicators in the high-middle SDI region and the middle SDI region also showed downward trends to varying degrees. However, compared with the high SDI region, the decline was slightly smaller. It is especially worth noting that in the middle SDI region, although the incidence rate decreased, the prevalence rate increased slightly. This may indicate the existence of treatment effects or extended survival periods. For the low-middle SDI and low SDI regions, although the age - age-standardized incidence rate, deaths, and DALYs of LC in these regions also decreased, the decline was relatively slow. Moreover, the prevalence in these two regions remained almost unchanged or decreased only slightly. This may reflect the challenges of LC treatment in lower SDI regions and areas that need more attention ( Table 1 and Supplementary Figure S1 ).

In 2021, health indicators in different regions around the world presented significant differences. The Central European region ranked first in both the age - standardized incidence rate (4.46 per 100,000 population, 95% UI: 4.07 - 4.87) and the prevalence rate (25.25 per 100,000 population, 95% UI: 23.18 - 27.54). Moreover, it had the highest DALYs (69.93 per 100,000 population, 95% UI: 63.58 - 76.15). Oceania had the lowest incidence rate (0.56 per 100,000 population, 95% UI: 0.44 - 0.74) and prevalence rate (2.35 per 100,000 population, 95% UI:1.84 - 3.01). The Caribbean region recorded the highest age - standardized deaths rate (2.69 per 100,000 population, 95% UI:2.33 - 3.03). In contrast, the high - income Asia Pacific region was at the lowest level not only in terms of the deaths rate (0.31 per 100,000 population, 95% UI:0.27 - 0.35) but also in terms of the DALYs burden (7.10 per 100,000 population, 95% UI: 6.22 - 7.97). According to the time trend, the incidence in 4 regions and the prevalence in 7 regions either have no significant change or show an upward trend. However, all the other regions show a downward trend. Among these regions, Central Asia has the largest decline in the incidence and prevalence, and High - income Asia Pacific has the largest decline in deaths and DALYs ( Table 1 and Supplementary Figure S1 ).

In 2021, there were differences in the age - standardized indicators of LC among 204 countries. Based on the cross - sectional results at that time, the Principality of Monaco had the highest incidence and prevalence, with 10.44 per 100,000 people (95% UI: 7.80 - 14.22) and 69.02 per 100,000 people (95% UI: 52.52 - 93.39) respectively. Montenegro had the highest deaths and DALYs, specifically 4.67 per 100,000 people (95% UI: 3.61 - 6.10) and 127.61 years per 100,000 people (95% UI: 97.86 - 170.11). In contrast, the Republic of Kiribati had the lowest values for all of the above - mentioned indicators ( Figure 1 , Supplementary Table S1 ).

According to the analysis of the change trend of EAPCs from 1990 to 2021, in terms of incidence, 149 countries show a downward trend, while 55 countries show an upward trend. The Republic of Guatemala has the most significant decline in incidence (EAPC = - 3.7, 95% CI: - 3.92 - - 3.49). Regarding prevalence, 124 countries have had a significant decline, and 80 countries have no change or an increase. Among them, the Republic of Uzbekistan has the largest downward trend in prevalence (EAPC = - 3.13, 95% CI: - 3.67 - - 2.59). In terms of deaths, the number of deaths in 168 countries shows a downward trend, and in 36 countries shows an upward trend. The Republic of Korea has the largest downward trend in deaths(EAPC = - 5.65, 95% CI: - 6.01 - - 5.28). Regarding DALYs, 170 countries show a downward trend, and 34 countries show an upward trend. Among them, the Republic of Korea has the largest decline in DALYs (EAPC = - 5.95, 95% CI: - 6.29 - - 5.60)( Supplementary Figure S2 , Supplementary Table S1 ).

Age-sex correlation analyses in 2021 showed an overall increasing trend in the incidence, prevalence and DALYs of LC globally, and an overall decreasing trend after the age of 70 years, with males far outnumbering females. However, the difference between males and females initially became progressively larger with age, and this gap gradually narrowed after age 70 years ( Figures 2A, B, D ). In terms of deaths, both males and females show an upward trend, with males rising significantly faster than females; in terms of the number, the number of male deaths is higher than the number of female deaths ( Figure 2C ).

The analysis of sex - time trend reveals that globally, the burden of LC has been decreasing over time for both genders. Nevertheless, in the Middle SDI, Low - middle SDI, and Low SDI regions, the incidence and prevalence have fluctuated slightly or increased. In contrast, both High SDI and High - middle SDI regions have shown a downward trend in disease burden indicators ( Figure 3 ).

Age - time correlation analysis reveals that the disease burden of LC patients across all age groups globally has been on a downward trend over time. Nevertheless, since 2005, among patients of all age groups in the Middle SDI, Low - middle SDI, and Low SDI regions, there has been no difference or an increase in the variation range of the incidence and prevalence. In contrast, a downward trend has been presented in the High SDI and High - middle SDI regions. Regarding deaths and DALYs, patients of all age groups in the five SDI regions have all exhibited a downward trend over time ( Supplementary Figure S3 ).

Joinpoint regression analysis showed that during the period from 1990 to 2021, the incidence, prevalence, deaths and DALYs generally showed a downward trend. For incidence, there were three year - nodes, namely 1995, 2004 and 2014. It showed an upward trend during 1990 - 1995, but it was not statistically significant; since 1995, it has shown a downward trend and was statistically significant. Prevalence also had three year - nodes, which were 1996, 2002 and 2017 respectively. It showed an upward trend from 1990–1996 and a downward trend after 1996. Deaths and DALYs also had three year - nodes respectively, namely (1994, 2007, 2014) and (1994, 2006, 2014), and all showed a downward trend in all time periods ( Supplementary Figure S4 ).

Next, we analyzed the relationship between the disease burden of LC and SDI. In 21 regions, the incidence, prevalence, deaths, and DALYs had a non - linear relationship with SDI. When SDI was less than 0.7, an increase in SDI would result in an increase in the disease burden; when SDI was greater than 0.7 or less than 0.3, an increase in SDI would lead to a decrease in the disease burden. Among these regions, in High - income Asia Pacific, High - income North America, Western Europe, Eastern Europe, and Southern Latin America, the disease burden decreased the most. Most of the other regions showed a downward trend, and only a few regions showed an upward trend ( Figure 4 ). In 2021, the incidence, prevalence, number of deaths, and DALYs of LC in 204 countries also exhibited a non - linear relationship with the SDI. In comparison to countries having a low SDI, those with an SDI greater than 0.7 had a lower disease burden ( Figure 5 ).

The age - effect curve reveals that, in terms of the longitudinal age curve ( Figure 6A ), the incidence of LC rises significantly as age increases, peaking in the 80–90 year old age group. The cross - sectional age curve ( Figure 6B ) further validates this trend, demonstrating a sharp increase in the incidence rate after 60–70 years old and remaining high in the advanced - age segment. Moreover, the long - term and cross - sectional relative risk ratios ( Figure 6C ), along with the local drift ( Figure 6G ) and age - deviation ( Figure 6H ) graphs, also show that the relative risk of LC in the elderly population is significantly higher than that in other age groups.

The period - effect analysis results reveal that the fitted temporal trends ( Figure 6D ) and the period RR ( Figure 6E ) display the LC incidence changes from 1990 to 2021. Overall, the data shows that the LC incidence has a downward trend, which may reflect the influence of factors like the progress of preventive measures and treatment methods or lifestyle changes. Moreover, the period deviation ( Figure 6I ) further verifies that there are differences in the LC incidence across different periods, indicating that environmental factors and social changes have a significant impact on the disease incidence.

Cohort effect analysis indicates that the cohort RR ( Figure 6F ) and cohort deviation ( Figure 6J ) show differences in the risk of LC incidence among different birth cohorts. The data suggest that, among those born between 1900 and 2000, the relative risk of laryngeal cancer is basically stable. However, individuals born in specific years may have different exposure risks. The fitted cohort pattern ( Figure 6K ) further demonstrates the relationship between birth cohorts and the risk of LC incidence. It shows that the incidence in some birth cohorts is significantly higher than that in others, which implies the potential influence of early - life environment and behavioral habits on the occurrence of the disease. These findings are highly significant for understanding the epidemiological characteristics of LC.

Decomposition analysis shows that, globally and in the five SDI regions, the contribution patterns of population growth, aging, and epidemiological changes to the burden of LC are largely consistent. This is also similar among different gender groups. Overall, population growth and aging are the main factors leading to the increased burden of LC, while epidemiological changes help to reduce this burden. Specifically, globally, population growth and aging have increased the incidence rates caused by LC by 81.6% and 84.12% respectively; in contrast, epidemiological changes have reduced the incidence rate by 65.71%. The impact of population growth is particularly significant in high SDI regions (235.22%) and middle-high SDI regions (161.2%). Similarly, aging has also significantly increased the incidence in these regions, with an increase of 113.41% in middle-high SDI regions and 111.11% in high SDI regions. Epidemiological changes have greatly reduced the disease burden in all regions, with the largest decreases in middle-high SDI (- 174.61%) and high SDI (- 246.32%) regions ( Figure 7A and Supplementary Table S2 ). Decomposition analysis of the deaths and DALYs also shows a similar trend ( Figures 7C, D and Supplementary Table S2 ). However, in terms of prevalence, the changes in SDI regional epidemiology have increased the disease burden of LC (7.36%). Male (10.27%) also show the same trend. In the high - middle SDI regions, the changes in epidemiology have increased the LC disease burden of female (6.33%) ( Figure 7B , Supplementary Table S2 ).

Predictive analysis indicates that by 2050, the global age - standardized incidence, prevalence, number of deaths, and DALYs of LC will respectively decrease to 1.75 per 100,000 people (95% CI: 1.46 - 2.05), 13.67 per 100,000 people (95% CI: 2.5 - 24.84), 0.85 per 100,000 people (95% CI: 0.72 - 0.98), and 24.67 per 100,000 people (95% CI: 19.40 - 29.95) ( Figures 8A, C, E, G ). This implies that globally in 2050, there will be around 167,468 new LC cases (95% CI: 139,090 - 195,846), the number of patients will reach 1,300,968 (95% CI: 241,648 - 2,360,288), the number of deaths will be 81,213 (95% CI: 68,758 - 93,668), and the total number of DALYs will be 2,356,583 years (95% CI: 1,852,666 - 2,860,501) ( Figures 8B, D, F, H ). After 2021, in the short term, the incidence and prevalence will first rise and then promptly decline over time, while the deaths and DALYs generally present a downward trend ( Supplementary Tables S3, S4 ). In addition, through predicting the changing trends of the global LC disease burden in different age groups, it was found that generally, its incidence, deaths and DALYs showed a gradually decreasing trend. Regarding prevalence, for the population aged 20 and above, it increased from 2030 - 2035, then decreased rapidly and tended to be stable, with a relatively small change range ( Supplementary Figures S5–8 ).

The purpose of risk factor analysis is to evaluate the impact of certain behavioral, environmental or metabolic factors on the overall burden of diseases. This burden is usually measured by the number of deaths caused by diseases or DALYs induced, rather than simply the incidence or prevalence of diseases. In 2021, there were nine different causes contributing to LC globally. Among them, Behavioral risks and Tobacco (Smoking) had the greatest impacts. The proportions of deaths caused by them were 69.58% and 66.23% respectively, and the proportions of DALYs were 66.87% and 65.28% respectively. Five SDI regions showed a similar distribution( Supplementary Figure S9 , Supplementary Table S5 ). Among all age groups, behavioral risk factors (such as smoking and high alcohol consumption) account for a significant proportion of mortality factors. Among them, smoking is the main risk factor for death in all age groups, especially in the 60–69 age group, where the proportion of smoking exceeds 70%. High alcohol consumption also shows a relatively high proportion of risk factors in some age groups, reaching 14% in the 45–59 age group. Relatively speaking, the proportion of risk factors such as occupational exposure to carcinogens (such as asbestos and sulfuric acid) is lower. Overall, as age increases, the proportion of smoking as a risk factor gradually rises, while the influence of other factors is relatively stable or only has small fluctuations. The contribution of DALYs is similar to that of mortality risk factors ( Figures 9 , 10 , Supplementary Table S6 ).