Risk factors behind the global lung cancer burden: a pan-database exploration

Introduction

Lung cancer is the leading cause of morbidity and mortality worldwide and poses a significant global health challenge. In 2022, an estimated 2.48 million new lung cancer cases were reported, accounting for 12.4% of all cancers globally, and it remains the leading cause of cancer-related deaths. The economic impact is profound, with lung cancer costing an estimated $3.9 trillion in 2017, the highest among all cancers (1-3).

Tobacco smoking has been consistently and strongly associated with lung cancer prevalence globally (4). However, other factors such as air pollution, occupational exposure, and emerging environmental and socioeconomic determinants also contribute to the disease burden (5-9). Recent studies have suggested potential roles for ambient temperature, urbanization, and cooking methods in influencing lung cancer risk, but the evidence remains limited and sometimes inconsistent (5-10).

Globally, patterns of lung cancer incidence and mortality are shifting. While high-income countries have seen a decline in incidence rates due to successful tobacco control measures, low- and middle-income countries are experiencing rising rates, partly driven by increasing tobacco use, environmental exposures, and rapid urbanization (11). These trends underscore the need for a comprehensive understanding of the diverse risk factors that influence lung cancer across different settings.

Despite extensive research, gaps remain in our understanding of how various risk factors interact globally over extended periods. Most studies have focused on specific regions or individual risk factors, limiting the ability to draw broader conclusions about global trends and develop effective, multifaceted prevention strategies.

In this study, we aimed to provide a comprehensive analysis of global lung cancer trends from 1990 to 2021, by examining the associations between lung cancer prevalence and mortality and a wide range of potential risk factors, including smoking, urbanization, ambient temperature, and cooking methods. By integrating data from multiple international sources and applying robust statistical analyses, we sought to elucidate the complex interplay of factors influencing global lung cancer epidemiology. Our findings have the potential to inform public health policies and interventions aimed at reducing the global burden of lung cancer, particularly in the context of ongoing challenges such as climate change and rapid urbanization. We present this article in accordance with the GATHER reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-131/rc).

Methods

Data sources and processing

Data source and inclusion criteria

We conducted a comprehensive analysis of tracheal, bronchial, and lung cancer prevalence and mortality across 201 countries from 1990 to 2021, “lung cancer” was used as a term throughout the full text. Data was sourced from multiple international databases, including the Global Burden of Disease Study (GBD), World Health Organization (WHO), Food and Agriculture Organization of the United Nations (FAO), and others (see Appendix 1 for a complete list and descriptions). These databases provide extensive country-level information on lung cancer outcomes and a wide range of potential risk factors, including socioeconomic indicators, environmental variables, lifestyle habits, and dietary patterns.

The inclusion criteria for data sources were based on the availability of relevant country-level data within the study period and variables of interest. Lung cancer prevalence and mortality data were obtained from GBD, which compiles estimates based on nationally representative surveys, cohort studies, and administrative records. The GBD methodology includes model-based estimations for countries that lack direct data, utilizing region-specific models to extrapolate prevalence and mortality rates. We focused on the total lung cancer prevalence without distinguishing among the cancer subtypes.

Given that not all data sources covered the same time periods or countries, we included all available data for each variable, utilizing the maximum possible temporal and spatial coverage. Analyses of variables with limited coverage were restricted to ensure methodological consistency. In cases where multiple sources provided data for similar variables but differed in measurement units or definition, such as alcohol consumption and smoking rates, we included data from all sources and assigned distinct variable names to differentiate them. We intentionally included multiple smoking-related variables from different global databases to address potential limitations of single-source data.

Detailed descriptions of each variable, including definitions, units, temporal and spatial coverage, and data sources, are provided in Appendix 2.

Our analytical approach treated each variable independently, without the need for the harmonization of units or scales. This method allowed us to utilize all available data comprehensively and explore associations across multiple datasets, thereby enhancing the robustness of our findings.

Data processing

Data processing and analysis were performed using MATLAB R2024a, Python 3.12, and R-4.4.1. All statistical tests were two-sided, and P values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the false discovery rate (FDR).

Calculation of prevalence rate

The GBD provides point prevalence rates, representing the proportion of individuals with lung cancer on December 31, each year. To estimate the period prevalence—reflecting the proportion of individuals who had the disease at any point during the year, we calculated the following:

Period prevalencecurrentyear=Pointprevalencepreviousyear+Incidencecurrentyear[1]

For the initial year (1990), where prior point prevalence data were not available, we approximate period prevalence using a conservative approach. We assumed no disease recovery and calculated period prevalence as:

Period prevalence1990=Pointprevalence1990+Death1990

This approach accounts for new cases and mortality but does not consider remission or recovery due to data limitations. The study period (1990–2021) encompasses significant variations in lung cancer therapies, diagnostic capabilities, and global health events like the coronavirus disease 2019 (COVID-19) pandemic. These factors may introduce potential biases in prevalence and mortality estimates.

Population density-weighted averaging

For variables available in spatial formats (e.g., ambient temperature and air pollution levels), we applied population density-weighted averaging to derive representative country-level values. This method accounts for the population distribution within a country and provides a more accurate estimate of the average exposure experienced by inhabitants. The population-weighted average 𝑇 for a variable 𝑡 (𝑥, 𝑦) was calculated as:

Tcountry=∬countryt(x,y)∗ρ(x,y)dxdy∬countryρ(x,y)dxdy[3]

where 𝜌 (𝑥, 𝑦) represents the population density at location (𝑥, 𝑦).

For nighttime light data from the Defense Meteorological Satellite Program (DMSP)/Operational Linescan System (OLS) and the Visible Infrared Imaging Radiometer Suite (VIIRS), we utilized a harmonized dataset from 1992 to 2018. Light intensity values below a threshold of seven were excluded based on recommendations from prior harmonization studies to eliminate background noise, and the remaining values were transformed using a logarithmic scale to reduce skewness and account for a wide range of values. While population density-weighted averaging provides a more nuanced approach to calculating country-level averages, it is important to acknowledge that this method may not fully capture the complete variability within countries, especially those with significant geographic, socioeconomic, or demographic disparities. Regions with large urban-rural divides or substantial internal migration patterns may experience variations that are not entirely represented by this averaging approach.

Adjustment for confounding factors

To account for potential confounding variables affecting the relationship between risk factors and lung cancer outcomes, we adjusted for each variable using generalized linear models (GLMs). This approach allowed us to isolate the independent association between each risk factor and lung cancer prevalence and mortality. We chose GLMs for several key reasons: flexibility in handling non-Normal distributions, ability to model heterogeneous variance and handling of multiple risk factors. Of note, this approach provides a broad assessment of potential relationships but should be interpreted with caution due to the ecological nature of the analysis.

Statistical analysis

GLMs for effect estimation

To explore both the long-term and short-term impacts of risk factors on lung cancer outcomes, we conducted cumulative and instantaneous effect analyses:

• Cumulative effect analysis: for country, we calculated the mean value of each risk factor over the entire study period (1990–2021), representing long-term exposure. We then assessed the association between these cumulative averages and lung cancer prevalence or mortality in 2021 using GLMs. This study aimed to understand how sustained exposure to risk factors influences recent lung cancer outcomes.
• Instantaneous effect analysis: we analyzed the annual data for both risk factors and lung cancer outcomes across the study period. This approach captures year-to-year variations and assesses how changes in risk factors within a given year are associated with changes in lung cancer outcomes during the same year. This study provides insight into the immediate effects of risk factor fluctuations on lung cancer prevalence and mortality.

The distinction between cumulative and instantaneous effect analyses provides a nuanced understanding of lung cancer risk dynamics. This methodological approach bridges the gap between short-term epidemiological observations and long-term cohort studies.

Canonical correlation analysis (CCA) with feature selection

CCA was utilized to examine the multivariate relationships between sets of risk factors and lung cancer outcomes. CCA identifies pairs of canonical variates—linear combinations of variables within each set that are maximally correlated. This method allows the assessment of how groups of risk factors are collectively related to lung cancer prevalence and mortality.

We employed a genetic algorithm (GA) for feature selection to optimize the variables included in the CCA. The GA iteratively searched for a subset of variables that maximized the canonical correlation between independent (risk factors) and dependent (lung cancer outcomes) variants. The algorithm involves processes of selection, crossover, and mutation over multiple generations until convergence or a predefined number of iterations is reached. The final model included an optimal subset of features, and canonical loadings were examined to evaluate the contribution of each variable to the canonical variates.

Correlation analysis and multiple testing correction

We calculated Pearson’s and Spearman’s correlation coefficients to assess linear and monotonic associations, respectively, between individual risk factors and lung cancer outcomes. For each variable, we retained weaker correlation coefficients to provide a conservative estimate of the association.

rαβpearson=∑i(αi−α¯)(βi−β¯)(n−1)SαSβ[4]

rαβspearman=1−6∑i(R[αi]−R[βi])2n(n2−1)[5]

rαβ={rαβpearson,if|rαβpearson|≤|rαβspearman|rαβspearman,if|rαβpearson|>|rαβspearman|}[6]

In these equations: r represents the correlation coefficient, α and β are the variables being compared, n is the total number of observations, Y¯ represents the mean of the variable, S represents the standard deviation, R[x] denotes the rank of a variable.

Features were filtered based on a significance threshold (adjusted P value P_FDR<0.05) and an absolute correlation coefficient threshold (|r|>0.25 for confounder-adjusted data and |r|>0.3 for unadjusted data) to ensure meaningful associations. The selection of correlation coefficient thresholds (|r|>0.3 for unadjusted data and |r|>0.25 for confounder-adjusted data) was based on a careful balance between statistical significance and biological relevance. These thresholds were chosen to capture substantive associations while minimizing the inclusion of potentially spurious correlations. The Benjamini-Hochberg procedure was applied to adjust for multiple comparisons and control the FDR.

Multidimensional scaling (MDS) for visualization

To visualize the complex relationships between variables, we constructed a dissimilarity matrix using the correlation coefficients, where the dissimilarity d_αβ between two variables α and β was defined as:

dαβ=1−rαβ2[7]

Non-metric MDS is then applied to reduce the high-dimensional dissimilarity matrix to a two-dimensional representation. The MDS algorithm aims to preserve the rank order of dissimilarities while minimizing stress, as defined by Kruskal’s normalized stress-1 criterion:

Y=MDS(dαβ,2)[8]

Ethics statement

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All data used in this study were anonymized and collected at the population level, with no individual patient information. Therefore, patient consent was not required.

Results

Burden of lung cancer, 1990–2021

Global burden of lung cancer from 1990 to 2021

Over the past three decades, the global burden of lung cancer has undergone significant change (Figure 1). The total number of lung cancer cases increased from 1.40 million [95% uncertainty interval (UI): 1.34 to 1.46 million] in 1990 to 3.25 million (95% UI: 2.95 to 3.59 million) in 2021. Despite this substantial rise in absolute numbers, the overall age-standardized prevalence of lung cancer worldwide remains relatively stable, changing from 34.25 per 100,000 population (95% UI: 32.66 to 35.69) in 1990 to 37.28 per 100,000 (95% UI: 33.76 to 40.77) in 2021.

Figure 1 Prevalence and death rate of lung cancer by age (years) and gender across various global regions. (A) Lung cancer prevalence in 2021. (B) Lung cancer death rate in 2021. (C) Global age-standardized lung cancer prevalence for men and women in 1990 and 2021. (D) Global age-standardized lung cancer death rate for men and women in 1990 and 2021. (E) Lung cancer age-standardized prevalence by country in 2021. See Appendix 3 for country codes (ISO 3166-1 alpha-3). TBLCancer, tracheal, bronchial, and lung cancer.

The number of deaths attributed to lung cancer also increased, climbing from 1.08 million (95% UI: 1.02 to 1.14 million) in 1990 to 2.02 million (95% UI: 1.82 to 2.22 million) in 2021. Notably, the global age-standardized death rate decreased from 27.58 per 100,000 (95% UI: 26.09 to 28.99) in 1990 to 23.50/100,000 (95% UI: 21.22 to 25.85) in 2021.

Sex-specific analyses revealed divergent results. The age-standardized prevalence decreased from 53.59 per 100,000 (95% UI: 50.90 to 56.49) in 1990 to 51.50/100,000 (95% UI: 45.62 to 57.98) in 2021. In contrast, women experienced an increase in prevalence from 17.62 per 100,000 (95% UI: 16.47 to 18.66) to 24.95/100,000 (95% UI: 22.28 to 27.52) during the same period. The death rate has decreased for both sexes from 1990 to 2021.

Analysis of lung cancer risk factors

Figure 2 provides a visual representation of the diverse risk factors and their distributions.

Figure 2 Overview of global data representation. (A) Lung cancer 2021 prevalence. (B) Lung cancer prevalence rate of change compared with 1990. (C) Smoking (SEV). (D) Smoking rate of change. (E) Secondhand smoke (SEV). (F) Secondhand smoke rate of change. (G) Alcohol use (SEV). (H) Alcohol use rate of change. (I) Gas cooking (%). (J) Gas cooking rate of change. (K) Urbanization (%). (L) Urbanization rate of change. (M) Temperature (℃). (N) Temperature rate of change. Warmer colors represent higher values. Missing data are shown in white. SEV, summary exposure value.

CCA

Using CCA with variables selected using a GA, we identified smoking, temperature, and urbanization as significant factors associated with lung cancer prevalence. The parsimonious model is shown in Figure 3. Increases in smoking and urbanization rates, along with decreases in temperature, correlated with variations in lung cancer prevalence by 1.76‰, 0.38‰, and 1.21‰, respectively.

Figure 3 Canonical correlation analysis shows a parsimonious model of lung cancer prevalence, using smoking, temperature, and urbanization.

Figure 4 shows β_{interquartile range (IQR)} from the GLMs analysis of lung cancer prevalence, where smoking had the largest effect size (β_{instantaneous}=0.042, IQR: 0.041 to 0.043), followed by tobacco (0.036, IQR: 0.035 to 0.037), cigarette smoking (0.031, IQR: 0.028 to 0.034), tobacco smoking (0.028, IQR: 0.025 to 0.030), secondhand smoke (0.025, IQR: 0.023 to 0.026), alcohol use (0.025, IQR: 0.024 to 0.026) and inequality adjusted Human Development Index (IHDI) (0.029, IQR: 0.027 to 0.031). In contrast, the largest protection effect size was observed for gas cooking (−0.017, IQR: −0.017 to −0.016), maximum temperature (−0.020, IQR: −0.021 to −0.019), universal thermal climate index (−0.020, IQR: −0.021 to −0.019), temperature (−0.019, IQR: −0.020 to −0.018), minimum temperature (−0.018, IQR: −0.019 to −0.016) and potential evapotranspiration (−0.015, IQR: −0.016 to −0.014).

Figure 4 Forest plot of β_IQR (scaled beta coefficients) of risk factors from GLMs analysis of lung cancer prevalence. CI, confidence interval; GLM, generalized linear model; HDI, Human Development Index; IQR, interquartile range.

In terms of the death rate, Figure 5 shows that the top factors were smoking (β_cumulative=0.016, IQR: 0.016 to 0.016), tobacco (0.015, IQR: 0.014 to 0.015), cigarette smoking (0.012, IQR: 0.012 to 0.013), tobacco smoking (0.011, IQR: 0.010 to 0.012) and secondhand smoke (0.011, IQR: 0.011 to 0.012). On the other hand, the top protective factors were maximum temperature (−0.006, IQR: −0.006 to −0.005), universal thermal climate index (−0.005, IQR: −0.006 to −0.005) and temperature (−0.005, IQR: −0.006 to −0.005).

Figure 5 Forest plot of β_IQR (scaled beta coefficients) of risk factors from GLMs analysis of lung cancer death. CI, confidence interval; GLM, generalized linear model; HDI, Human Development Index; IQR, interquartile range.

Social economics and environment

The worldwide increase in lung cancer incidence results from a complex interaction of environmental, climatic, and socio-economic factors. Figure 6 illustrates these complex relationships, highlighting how various determinants collectively influence the global lung cancer epidemiology. Additional detailed correlation matrices and network analyses are presented in Figures S1-S3.

Figure 6 Confounder-adjusted correlations between lung cancer prevalence, death rate, and social economics and environment factors. Color gradient: red (positive correlation), blue (negative correlation). Correlation intensity indicated by color depth. Statistical significance markers included. DALYs, disability-adjusted life years.

After adjusting for various risk factors related to socioeconomic status and environment, our results showed that life expectancy, gross domestic product (GDP) per capita, urbanization rate, medical expenditure, Socio-Demographic Index (SDI), Human Development Index (HDI), and IHDI were significantly positively correlated with the prevalence and mortality of lung cancer, with trends increasing with age. In terms of environmental climate, temperature variables, including the mean temperature, universal thermal climate index, and diurnal temperature range, showed significant negative correlations with lung cancer prevalence and mortality. A similar pattern was observed for potential evapotranspiration. The influence of these socioeconomic and environmental factors on lung cancer is shown in Figure 6.

Figure 7A-7C show the relationships between temperature, urbanization, GDP per capita, and lung cancer prevalence. Specifically, mean temperature was negatively correlated with lung cancer prevalence (r=−0.296, P<0.001), while urbanization rate (r=0.481, P<0.001) and GDP per capita (r=0.584, P<0.001) were positively correlated with lung cancer prevalence.

Figure 7 Relationships between social economics and environmental factors and lung cancer prevalence (confounder-adjusted). (A) Temperature (℃) vs. lung cancer prevalence (r=−0.296, P<0.001). (B) Urbanization (%) vs. lung cancer prevalence (r=0.481, P<0.001). (C) GDP per capita ($) vs. lung cancer prevalence (r=0.584, P<0.001). CI, confidence interval; GDP, gross domestic product.

Lifestyle and diets

Figure 8 presents a correlation coefficient heatmap illustrating the correlation between lung cancer outcomes and lifestyle- and diet-related factors. Comprehensive lifestyle and dietary factor analyses are shown in Figures S4-S6. Analysis of dietary habits revealed positive correlations between alcohol-related factors (alcohol consumption and alcohol use) and smoking-related factors (tobacco use, smoking, tobacco smoking, cigarette smoking, and secondhand smoke exposure) with lung cancer prevalence and death rate. This pattern persisted even after adjusting for confounders.

Figure 8 Confounder-adjusted correlations between lung cancer prevalence, death rate, and lifestyle and diets factors. Heat map showing correlations between lung cancer outcomes and lifestyle/dietary variables. Color intensity represents correlation strength. Adjusted for confounding factors. DALYs, disability-adjusted life years.

Although household solid fuel use and polluted fuel cooking initially showed positive relationships with lung cancer death rates, these associations disappeared after adjusting for confounders. Biomass cooking exhibited a positive correlation with lung cancer prevalence, whereas gas cooking showed a negative correlation. After adjusting for confounding factors, the correlation between dietary factors and lung cancer prevalence remained non-significant. Notably, the negative correlations between calcium, phosphorus, zinc, fats, and vegetable oils with age-specific lung cancer death rates persisted after confounder adjustment, suggesting a potential nuanced relationship between these specific nutrients and lung cancer mortality in different age groups. The lack of a significant correlation between dietary factors and lung cancer prevalence after confounder adjustment does not negate the importance of diet in cancer prevention. Instead, it suggests that dietary influences on lung cancer risk are likely mediated through more complex pathways.

Figure 9 illustrates the relationships among alcohol consumption, smoking, secondhand smoke exposure, and lung cancer prevalence. Specifically, alcohol consumption (r=0.606, P<0.001), smoking (r=0.753, P<0.001), and secondhand smoke exposure (r=0.416, P<0.001) were significantly positively correlated with lung cancer prevalence. When examining the influence of cooking methods on lung cancer (Figure 10), we found a positive correlation between biomass cooking and lung cancer prevalence (r=0.361, P<0.001) and a negative correlation between gas cooking and lung cancer prevalence (r=−0.469, P<0.001).

Figure 9 Relationships between lifestyle factors and lung cancer prevalence (confounder-adjusted). (A) Alcohol consumption vs. lung cancer prevalence (r=0.606, P<0.001). (B) Smoking (%) vs. lung cancer prevalence (r=0.753, P<0.001). (C) Secondhand smoke (%) vs. lung cancer prevalence (r=0.416, P<0.001). CI, confidence interval; SEV, summary exposure value.

Figure 10 Relationships between lifestyle and dietary factors and lung cancer prevalence (confounder-adjusted). (A) Gas cooking (%) vs. lung cancer prevalence (r=−0.469, P<0.001). (B) Biomass cooking (%) vs. lung cancer prevalence (r=0.361, P<0.001). CI, confidence interval.

Discussion

This study identified several key factors influencing the global prevalence of lung cancer. Smoking was the most significant risk factor, exhibiting a strong positive association with lung cancer prevalence (r=0.753, P<0.001). Urbanization also showed a moderate positive correlation with lung cancer prevalence (r=0.481, P<0.001). Notably, ambient temperature was negatively associated with lung cancer prevalence (r=−0.296, P<0.001), suggesting that lower temperatures may exacerbate this risk.

The strong link between smoking and lung cancer aligns with the findings of numerous domestic and international studies. A South American study reported that smoking-related lung cancer deaths across all age groups totaled 43,454 cases in 2019 (12). Kuang et al. (1) demonstrated that, globally in 2021, smoking was the leading risk factor for lung cancer disability-adjusted life years (DALYs), accounting for 59.5% (55.4–63.3%), followed by environmental particulate pollution (15.0%, 9.4–21.0%) and asbestos exposure (7.2%, 5.0–9.4%). Our findings reinforce the understanding that smoking is a major risk factor for lung cancer, consistent with numerous epidemiological studies (4,13-15).

Importantly, our analysis provides new insights into the temporal dynamics of the impact of smoking on lung cancer. The cumulative and instantaneous effects of increased smoking per unit were calculated. Smoking showed the strongest correlation with lung cancer prevalence (β_cumulative=0.040, 0.039–0.041) among all the risk factors examined. Moreover, the instantaneous effect of smoking (β_{instantaneous}=0.042, 0.041–0.043) was slightly higher than the cumulative effect, suggesting that recent changes in smoking behavior appeared to correlate with variations in lung cancer prevalence than previously thought. This finding has significant implications for public health interventions, indicating that even short-term reductions in smoking rates could yield substantial benefits for lung cancer prevention. While our population-level analysis provides valuable insights, it is crucial to recognize the limitations of aggregated data. Individual lung cancer risk is not simply linear but accumulates complexly over time, influenced by factors such as smoking initiation age, inhalation patterns, and specific smoking behaviors that vary significantly across different cultural and demographic contexts. These individual-level variations, which may not be fully captured in global datasets, represent an important area for future research.

In exploring the relationship between temperature and lung cancer, we observed that temperature played a significant role, and this effect still existed after adjustment for confounders. Our study observed a significant negative association between ambient temperature and lung cancer prevalence, with revealing a complex relationship where lower ambient temperatures are associated with slightly increased lung cancer risk (approximately 1.9‰ increase per 1 ℃ decrease), potentially due to interrelated factors such as indoor air quality, reduced ventilation, and physiological stress responses in colder environments. During winter months, homes are often sealed tightly for heat conservation, reducing air exchange and potentially leading to the accumulation of indoor pollutants, including tobacco smoke (16). This is particularly concerning in households where indoor smoking occurs, increasing the exposure risks for all occupants, including non-smokers. Physiological responses to cold, such as vasoconstriction and reduced mucociliary clearance in the respiratory tract, may increase susceptibility to carcinogenic particles and hinder the body’s ability to eliminate them (17). Additionally, cold-induced stress responses may alter immune function, potentially influencing tumor surveillance and progression (18,19). Our findings highlight the need for tailored public health strategies for different climatic zones. In colder regions, improving indoor air quality, promoting adequate ventilation, and educating the public regarding the increased risks associated with indoor pollutants are essential. Implementing stricter regulations for indoor smoking and encouraging the use of air purification systems could be particularly beneficial. However, we emphasize that these potential mechanisms require rigorous further research to establish definitive causal relationships. Our findings should be interpreted as an observational association that warrants additional epidemiological and mechanistic studies.

Our study also revealed a positive correlation between urbanization rates and lung cancer prevalence, adding a new dimension to our understanding of the relationship between urban environments and cancer risk. This aligns with a study that highlights the multifaceted impact of urbanization on health outcomes (20). Urban areas often experience rapid economic development, leading to lifestyle and behavioral changes such as increased sedentary behavior, higher stress levels, and altered dietary patterns compared to rural areas (21). Research by Crippa et al. (22) and Xue et al. (23) has shown how these urban lifestyle factors contribute to an increased risk of various cancers, including lung cancer. Urbanization is also associated with shifts in occupational patterns, exposing workers to different hazards compared with rural agricultural work (24). Some urban occupations may involve exposure to carcinogens that increase lung cancer risk, although further research is required to establish direct links.

The implications of our findings are significant for public health policies, particularly in rapidly urbanizing regions. We found that the instantaneous effect of urbanization (β_{instantaneous}=0.22, 0.21–0.23) was more pronounced than the cumulative effect (β_cumulative=0.18, 0.17–0.19), suggesting that rapid urban growth may pose immediate public health risks related to lung cancer. This underscores the urgent need to implement prompt and targeted interventions in rapidly urbanizing areas. As the United Nations projects that 68% of the global population will live in urban areas by 2050 (25), agile and responsive public health policies are essential to address the immediate challenges posed by urbanization to lung cancer risk.

Cooking methods have emerged as another important factor that influences the prevalence of lung cancer. We observed that gas cooking was associated with a lower prevalence of lung cancer than biomass cooking, and this association persisted after adjusting for confounding factors. Recent epidemiological research provides substantial support for these observations. A population-based case-control study by Liang et al. (26) in Rural China confirmed the increased lung cancer risk associated with long-term biomass fuel use, particularly among never-smokers. Bruce et al. (27) conducted a study showing the ORs for lung cancer risk with biomass for cooking was 1.17 (95% CI: 1.01 to 1.37). These findings add to the growing evidence that household air pollution from cooking fuels plays a significant role in lung cancer etiology (28). The negative correlation between gas cooking and lung cancer prevalence (r=−0.469, P<0.001) aligns with research indicating that cleaner cooking fuels reduce the risk of respiratory disease. For example, Raspanti et al. (29) found that switching from biomass to gas stoves significantly reduces indoor air pollution and improves respiratory health. Conversely, the positive correlation between biomass cooking and lung cancer prevalence (r=0.361, P<0.001) was consistent with the extensive literature on the health risks of solid fuel use. The International Agency for Research on Cancer has classified household air pollution from biomass fuel combustion as a group 2A carcinogen (30) and a meta-analysis by Kurmi et al. (31) reported an increased risk of lung cancer among biomass fuel users.

The mechanisms underlying the differential impact of gas and biomass cooking on lung cancer risk are likely multifaceted. Biomass fuels produce high levels of particulate matter, polycyclic aromatic hydrocarbons, and other carcinogenic compounds when burned in traditional stoves (32). These pollutants can cause chronic inflammation and oxidative stress in the respiratory tract, potentially leading to DNA damage and carcinogenesis (33). Gas cooking, which is not entirely emission-free, generally produces lower levels of harmful pollutant than biomass burning (34). A cleaner combustion process may reduce exposure to carcinogens, potentially explaining the lower lung cancer risk observed. However, factors such as ventilation, cooking practices, and exposure duration also influence the relationship between cooking fuel and cancer risk. Yu et al. (35) demonstrated that proper ventilation significantly reduces the health risks associated with both gas and biomass cooking. Our findings have important implications for public health policies, particularly in low- and middle-income countries where biomass fuels are prevalent. Promoting the transition to cleaner cooking fuels such as gas or electricity could effectively reduce lung cancer risk (36). However, such interventions should be tailored to local contexts and should include education on proper stove use and ventilation.

This study has several strengths. Our comprehensive analysis spanning 1990 to 2021 provides a unique long-term perspective on global lung cancer trends, which is rarely seen in existing literature. By considering traditional risk factors, such as smoking, alongside environmental and social factors, such as temperature, urbanization, and cooking methods, we offer a holistic assessment of lung cancer risks. Temporal dynamics analysis, which compares cumulative and instantaneous effects, provides valuable insights into how recent changes in risk factors might have immediate impacts on lung cancer risk. This differentiation allows for a better understanding of which interventions may yield immediate results versus those requiring sustained, long-term efforts, bridging the gap between short-term epidemiological and long-term cohort studies.

However, there are many limitations in this study. As an ecological study based on population-level data, we cannot establish individual-level causal relationships, potentially masking important individual variations. Confounding is most precisely assessed at the individual level, and our GLM-based method may not completely remove all sources of bias. While population-level analyses provide valuable insights into broad epidemiological trends, they are inherently limited by the ecological fallacy, which means that associations observed at the population level may not necessarily hold true for individuals. An important limitation of our study is the use of country-level data, which may obscure significant internal variations. Countries can have substantial heterogeneity in cultural groups, socioeconomic conditions, and environmental exposures. And data quality and comparability issues across different countries and regions may have affected the accuracy of our findings. Additionally, the lack of differentiation between lung cancer subtypes and the absence of genetic susceptibility data limits our comprehensive understanding of lung cancer risks and the COVID-19 pandemic, which occurred within our study period, presents unique challenges for epidemiological analysis.

Conclusions

Our comprehensive global analysis reveals that lung cancer risk extends far beyond traditional understanding. While tobacco remains the primary risk factor, emerging environmental and socioeconomic determinants play crucial roles in disease development. Urbanization, ambient temperature, and cooking methods are not peripheral concerns but integral components of lung cancer epidemiology. For policymakers and public health professionals, these findings demand a holistic approach to prevention. In rapidly developing regions, interventions must consider how urban transformations, climate variations, and household practices intersect with cancer risk. The transition from biomass to cleaner cooking technologies, strategic urban planning, and targeted health education can substantially mitigate lung cancer prevalence. While our study provides global insights into lung cancer risk factors, we acknowledge that public health interventions must be carefully tailored to local contexts. Smaller countries, in particular, may need to develop strategies that are specifically adapted to their unique cultural, economic, and political landscapes.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the GATHER reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-131/rc

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-131/prf

Funding: This work was supported by the Open Project of the State Key Laboratory of Respiratory Disease (SKLRD-OP-202402), Zhongnanshan Medical Foundation of Guangdong Province (ZNSXS-20220019), State Key Laboratory of Respiratory Disease (SKLRD-OP-202402), and Guangzhou Municipal Science and Technology Bureau (SL2024A04J00706).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-131/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Kuang Z, Wang J, Liu K, et al. Global, regional, and national burden of tracheal, bronchus, and lung cancer and its risk factors from 1990 to 2021: findings from the global burden of disease study 2021. EClinicalMedicine 2024;75:102804. [Crossref] [PubMed]
2. 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 Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
3. Chen S, Cao Z, Prettner K, et al. Estimates and Projections of the Global Economic Cost of 29 Cancers in 204 Countries and Territories From 2020 to 2050. JAMA Oncol 2023;9:465-72. [Crossref] [PubMed]
4. Islami F, Torre LA, Jemal A. Global trends of lung cancer mortality and smoking prevalence. Transl Lung Cancer Res 2015;4:327-38. [Crossref] [PubMed]
5. Hirst AM, Frame FM, Arya M, et al. Low temperature plasmas as emerging cancer therapeutics: the state of play and thoughts for the future. Tumour Biol 2016;37:7021-31. [Crossref] [PubMed]
6. Theodorescu D. Cancer cryotherapy: evolution and biology. Rev Urol 2004;6:S9-S19.
7. Di W, Yu J, Zong D, et al. Effects of ambient temperature, relative humidity and absolute humidity on risk of nasopharyngeal carcinoma in China. Int J Cancer 2024;155:646-53. [Crossref] [PubMed]
8. Repasky EA, Evans SS, Dewhirst MW. Temperature matters! And why it should matter to tumor immunologists. Cancer Immunol Res 2013;1:210-6. [Crossref] [PubMed]
9. Povinelli BJ, Kokolus KM, Eng JW, et al. Standard sub-thermoneutral caging temperature influences radiosensitivity of hematopoietic stem and progenitor cells. PLoS One 2015;10:e0120078. [Crossref] [PubMed]
10. Analitis A, Katsouyanni K, Biggeri A, et al. Effects of cold weather on mortality: results from 15 European cities within the PHEWE project. Am J Epidemiol 2008;168:1397-408. [Crossref] [PubMed]
11. Torre LA, Siegel RL, Ward EM, et al. Global Cancer Incidence and Mortality Rates and Trends--An Update. Cancer Epidemiol Biomarkers Prev 2016;25:16-27. [Crossref] [PubMed]
12. Salerno PRVO, Palma Dallan LA, Rodrigues Pereira GT, et al. Trends in tracheal, bronchial and lung cancer attributed to smoking in South America: Global Burden of Disease analysis 1990-2019. Rev Panam Salud Publica 2024;48:e30. [Crossref] [PubMed]
13. Centers for Disease Control and Prevention (US), National Center for Chronic Disease Prevention and Health Promotion (US), Office on Smoking and Health (US). How Tobacco Smoke Causes Disease: The Biology and Behavioral Basis for Smoking-Attributable Disease: A Report of the Surgeon General. Atlanta (GA): Centers for Disease Control and Prevention (US); 2010. Available online: https://www.ncbi.nlm.nih.gov/books/NBK53017/
14. Ribassin-Majed L, Hill C. Trends in tobacco-attributable mortality in France. Eur J Public Health 2015;25:824-8. [Crossref] [PubMed]
15. Zheng W, McLerran DF, Rolland BA, et al. Burden of total and cause-specific mortality related to tobacco smoking among adults aged ≥ 45 years in Asia: a pooled analysis of 21 cohorts. PLoS Med 2014;11:e1001631. [Crossref] [PubMed]
16. Institute of Medicine (US) Committee on the Assessment of Asthma and Indoor Air. Clearing the Air: Asthma and Indoor Air Exposures. Washington (DC): National Academies Press (US); 2000.
17. Eccles R. An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 2002;122:183-91. [Crossref] [PubMed]
18. Brenner IK, Castellani JW, Gabaree C, et al. Immune changes in humans during cold exposure: effects of prior heating and exercise. J Appl Physiol (1985) 1999;87:699-710. [Crossref] [PubMed]
19. Eklund LM, Sköndal Å, Tufvesson E, et al. Cold air exposure at - 15 °C induces more airway symptoms and epithelial stress during heavy exercise than rest without aggravated airway constriction. Eur J Appl Physiol 2022;122:2533-2544. [Crossref] [PubMed]
20. Gong P, Liang S, Carlton EJ, et al. Urbanisation and health in China. Lancet 2012;379:843-52. [Crossref] [PubMed]
21. Popkin BM. Urbanization, Lifestyle Changes and the Nutrition Transition. World Development 1999;27:1905-16.
22. Crippa A, Larsson SC, Discacciati A, et al. Red and processed meat consumption and risk of bladder cancer: a dose-response meta-analysis of epidemiological studies. Eur J Nutr 2018;57:689-701. [Crossref] [PubMed]
23. Xue XJ, Gao Q, Qiao JH, et al. Red and processed meat consumption and the risk of lung cancer: a dose-response meta-analysis of 33 published studies. Int J Clin Exp Med 2014;7:1542-53.
24. Blair A, Marrett L, Beane Freeman L. Occupational cancer in developed countries. Environ Health 2011;10:S9. [Crossref] [PubMed]
25. United Nations Department of Economic and Social Affairs. World Urbanization Prospects: The 2018 Revision. ; 2019.
26. Liang D, Wang J, Li D, et al. Lung Cancer in Never-Smokers: A Multicenter Case-Control Study in North China. Front Oncol 2019;9:1354. [Crossref] [PubMed]
27. Bruce N, Dherani M, Liu R, et al. Does household use of biomass fuel cause lung cancer? A systematic review and evaluation of the evidence for the GBD 2010 study. Thorax 2015;70:433-41. [Crossref] [PubMed]
28. Gordon SB, Bruce NG, Grigg J, et al. Respiratory risks from household air pollution in low and middle income countries. Lancet Respir Med 2014;2:823-60. [Crossref] [PubMed]
29. Raspanti GA, Hashibe M, Siwakoti B, et al. Household air pollution and lung cancer risk among never-smokers in Nepal. Environ Res 2016;147:141-5. [Crossref] [PubMed]
30. Guercio V, Doutsi A, Exley KS. A systematic review on solid fuel combustion exposure and respiratory health in adults in Europe, USA, Canada, Australia and New Zealand. Int J Hyg Environ Health 2022;241:113926. [Crossref] [PubMed]
31. Kurmi OP, Arya PH, Lam KB, et al. Lung cancer risk and solid fuel smoke exposure: a systematic review and meta-analysis. Eur Respir J 2012;40:1228-37. [Crossref] [PubMed]
32. Naeher LP, Brauer M, Lipsett M, et al. Woodsmoke health effects: a review. Inhal Toxicol 2007;19:67-106. [Crossref] [PubMed]
33. Sood A. Indoor fuel exposure and the lung in both developing and developed countries: an update. Clin Chest Med 2012;33:649-65. [Crossref] [PubMed]
34. Zhao Y, Hu M, Slanina S, et al. Chemical compositions of fine particulate organic matter emitted from Chinese cooking. Environ Sci Technol 2007;41:99-105. [Crossref] [PubMed]
35. Yu K, Qiu G, Chan KH, et al. Association of Solid Fuel Use With Risk of Cardiovascular and All-Cause Mortality in Rural China. JAMA 2018;319:1351-61. [Crossref] [PubMed]
36. Quansah R, Semple S, Ochieng CA, et al. Effectiveness of interventions to reduce household air pollution and/or improve health in homes using solid fuel in low-and-middle income countries: A systematic review and meta-analysis. Environ Int 2017;103:73-90. [Crossref] [PubMed]