Gastric cancer incidence trends in China: 1989-2018 registry data, 2022 Bayesian age-period-cohort estimates, ecological correlates

Abstract

BACKGROUND

Gastric cancer is a major global public health challenge. Although previous studies have described incidence trends in limited populations, the relationship between long-term incidence trends and their driving factors remains unclear.

AIM

To investigate the temporal trends in gastric cancer incidence in China over three decades, identify key influencing factors, and assess changes in urban-rural disparities to inform prevention strategies.

METHODS

This time-series correlation study used National Cancer Center data (1989-2018) with Bayesian age-period-cohort (BAPC)-model-projected incidence estimates for 2022. Joinpoint regression calculated annual percent change (APC) and average APC (AAPC) by gender and urban/rural residence. Bivariate correlation analysis examined associations between incidence and economic, environmental, and lifestyle factors.

RESULTS

Gastric cancer incidence declined significantly across all groups: Rural males [AAPC: -3.47%; 95% confidence interval (CI): -4.19 to -2.74; P < 0.001]; urban males (AAPC, -1.51%; 95%CI: -2.68 to -0.33; P = 0.012); rural females (AAPC: -3.18%; 95%CI: -3.86 to -2.50; P < 0.001), and urban females (AAPC: -1.73%; 95%CI: -2.91 to -0.53; P = 0.005). Rural rates exceeded urban rates throughout, but the gap narrowed substantially by 2022. Strongest correlates of declining incidence included economic income (R², urban: 0.514; rural: 0.914); fresh food consumption (R², urban: 0.507; rural: 0.914); population migration (R², urban: 0.471; rural: 0.906); fresh fruit consumption (R², urban: 0.561; rural: 0.863); and industrial wastewater treatment (chemical oxygen demand R², urban: 0.253; rural: 0.766).

CONCLUSION

Thirty-year declining gastric cancer incidence in China showed ecological temporal correlations with economic development, environmental improvement, and dietary improvement. Rural areas demonstrated steeper declines but persistent disparities, highlighting priorities for targeted prevention.

Key Words: Gastric neoplasms; Incidence; Risk factors; Trends; China

Core Tip: Gastric cancer incidence in China has declined continuously over the past three decades, with the decrease in rural areas being significantly greater than in urban-areas and the urban-rural gap narrowing markedly. Ecological temporal correlations were observed between higher economic income, greater fresh-fruit-and-vegetable consumption, improved industrial-wastewater treatment and declining incidence; these are exploratory findings and causal relationships cannot be established. However, the persistently high incidence in rural areas indicates the need for targeted prevention interventions.

INTRODUCTION

Gastric cancer imposes a major socioeconomic burden and was the fifth most common cancer and fifth leading cause of cancer mortality in 2022[1]. It is a major public health challenge in China. In 2022, China had an estimated 358700 new cases, ranking it fifth in incidence, and 260400 deaths, ranking it third in mortality, with higher rates in males and rural areas[2].

Globally, gastric cancer incidence and mortality rates are declining, but some countries show increasing trends[3]. Japan once had the highest gastric cancer incidence and mortality rates globally[4], but Wang et al[5] reported that the mortality rate for individuals aged > 40 years decreased by 60% between 1965 and 1995. In contrast, Ebrahimi et al[6] and Ebrahimi et al[7] found that in Babol, Iran, the age-standardized gastric cancer mortality rate rose from 20.1 per 100000 in 2013 to 20.5 per 100000 in 2021, with an annual percentage change (APC) of 2.95% from 2015 to 2021. Lin et al[3] revealed that from 2003 to 2012, most countries, including China had downward gastric cancer trends: Average APC (AAPC) was -3.0% in males and -3.1% in females. In contrast, in Canada, gastric cancer incidence increased in females (AAPC 1.2%), and in Thailand, gastric cancer mortality increased in males (AAPC 3.5%) and females (AAPC 4.7)[3]. In regions of China with a high risk of gastric cancer, Niu et al[8] found that from 2010 to 2019, the AAPC of gastric cancer incidence in older adults was -2.59%, and the AAPC of mortality was -2.55%, suggesting stable incidence and mortality rates. Based on gastric cancer incidence data from 1989 to 2018 collected and published by China’s National Cancer Center (NCC), earlier annual reports showed that age-standardized gastric cancer incidence had already declined markedly[9-13]; subsequent annual reports indicated that this downward trend continued[14-18]; and more recent annual reports supported a sustained long-term decline in gastric cancer incidence[19-23]. Earlier trend analyses and projections based on Chinese registry data reached the same conclusion[24].

Some stuides[8,25,26] have reported the trends and possible causes of gastric cancer incidence and mortality for some Chinese populations and time periods. However, few studies have utilized statistical methods to analyze the relationship between time trends in incidence rates and influencing factors. Therefore, it is essential to conduct relevant research to provide a scientific basis for the prevention and treatment of gastric cancer in China.

MATERIALS AND METHODS

Research characteristics and methods

This is an ecological time-series correlation study, which cannot prove causal relationships between exposures and outcomes. This study investigated gastric cancer incidence trends and influencing factors in mainland China from 1989 to 2022 via time-series correlation analysis. The research subjects were the total population aged 0-85⁺ years in China’s 31 provinces, categorized into four subgroups by urban/rural and male/female. Over the study period, the population aged 0-85⁺ years increased from 1.13 billion to 1.41 billion, with the urban population proportion rising from 26% to 64%, while the male population remained around 51%. Demographic data were taken from the China Statistical Yearbook[27].

The study, spanning > 30 years across China, provided sufficient data points to identify long-term trends and factor impacts, ensuring reliable and valid analysis. Data on age-standardized incidence rate by Chinese standard population of gastric cancer were obtained from the China Cancer Registry Annual Report, collected by the NCC from 1989 to 2018[9-24]. The 2022 incidence data were obtained from the official cancer statistics report published by the National Cancer Center of China, which estimated 2022 age-specific incidence rates using a Bayesian age-period-cohort (BAPC) model fitted to data from 106 cancer registries during 2010-2018, with bias correction using 2018 data from 700 registries[2]. All primary trend analyses were anchored on the observed 1989-2018 dataset. The 2022 projected value was not included in any bivariate correlation calculations. These sources are authoritative and rigorously quality controlled. The incidence rate was calculated as new cases divided by the population at risk, standardized to the Chinese standard population aged 0-85⁺ years to ensure comparability. We collected raw data on influencing factors from various sources, statistically standardized part of the data, and checked data integrity to avoid missing or abnormal values.

Search strategy and selection criteria

China’s gastric cancer age-standardized mortality rate by Chinese standard population comes from multiple data sources: (1) Three retrospective mortality sampling surveys in 1973-1975[28], 1990-1992[29], and 2004-2005[30]; (2) Cancer mortality data (1987-2000) from the World Health Organization[4]; (3) Cancer mortality data (2006-2010) published by the NCC of China[31-35]; (4) Cancer mortality data (2011–2014) from subsequent national cancer registry reports[36-39]; and (5) The 2022 model-based estimates from Han et al[2]. Due to varying statistical methods, data from different sources show significant discrepancies. In contrast, gastric cancer incidence data are sourced from the NCC of China, ensuring a unified data source and minimal discrepancies[2,9-24]. Furthermore, incidence rates better reflect the relationship between influencing factors and gastric cancer. Therefore, our primary focus was on analyzing the relationship between influencing factors and gastric cancer incidence rates.

According to the third edition of the Cancer Report published by the United States in 2018[40], the risk factors for gastric cancer included smoking, Helicobacter pylori (H. pylori) and Epstein-Barr virus (EBV) infections, exposure to high-temperature and dusty environments, daily alcohol consumption > 45 g, consumption of high-salt preserved foods, and consumption of processed meats. The protective factors included consumption of fresh vegetables and fruits, especially citrus fruits. Other potential factors included economic factors, environmental pollution, and urbanization. The potential socioeconomic and dietary factors included (Figure 1A): High-salt preserved foods (vegetables, fish, meat, etc.); daily per capita consumption of and processed meats (sausages, bacon, ham, etc.); annual per capita consumption of fresh vegetables and fresh fruits; annual per capita disposable income; annual ownership of refrigerators per 100 households. The potential lifestyle and environmental factors included (Figure 1B): H. pylori infection rate; proportion of smokers; average daily alcohol consumption per capita; proportion of urban and rural populations; annual chemical oxygen demand (COD) and ammonia nitrogen emissions from industrial wastewater and municipal domestic sewage.

Figure 1 Trends in potential influencing factors of gastric cancer in China from 1989 to 2018. A: Socioeconomic and dietary factors. High-salt pickled food[50]; processed meat[50]; annual per capita consumption of fresh vegetables[27]; annual per capita consumption of fresh fruits[27]; annual per capita disposable income[27] (standardized to the 1978 price level using the consumer price index); annual ownership of refrigerators per 100 households[27]; B: Lifestyle and environmental factors. Helicobacter pylori infection rate (derived from systematic literature reviews[41,42], with urban-rural calibration based on period-specific ratios; see Supplementary Table 1 for details); proportion of smokers[50]; average daily alcohol consumption per capita[49] (standardized using the urban-rural consumption expenditure ratio); proportions of urban and rural populations[27]; annual chemical oxygen demand emissions from industrial wastewater and municipal domestic sewage[52]; annual ammonia nitrogen emissions from industrial wastewater and municipal domestic sewage[52]. All trends are presented as observed; no interpolation or smoothing was applied.

The H. pylori infection rate data were collected through a systematic literature review and meta-analysis covering six time windows (1989-1994, 1995-1999, 2000-2004, 2005-2009, 2010-2014, and 2015-2018)[41,42]. These data were standardized for urban-rural differences based on additional regional studies, including earlier urban-rural comparative surveys[43-45] and later regional prevalence surveys[46-48]. The H. pylori infection rates by time window were derived from the systematic review and meta-analysis by Li et al[41] and the surveillance system study by Wang et al[42]. Urban-rural stratification was performed based on infection rates reported in the literature for each specific time period, rather than applying a fixed ratio[43-48]. The full literature search strategy, inclusion criteria, extracted data, urban-rural calibration method, and standardization protocol are provided in Supplementary Table 1 to support reproducibility. The proportion of smokers, derived from the China Health and Nutrition Survey (CHNS), was calculated using the number of smokers in the surveyed population in 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011, and 2015[48]. Data on average daily alcohol consumption per capita, sourced from the Food and Agriculture Organization of the United Nations, was collected for 30 time points from 1989 to 2018 and standardized using the urban-rural consumption expenditure ratio from the China Statistical Yearbook[27,49]. Data on annual per capita consumption of fresh vegetables and fruits; annual per capita disposable income; annual ownership of refrigerators per 100 households; and the proportions of urban and rural populations, sourced from the China Statistical Yearbook, were collected for 30 time points from 1989 to 2018[27]. The annual per capita disposable income was standardized to the 1978 price level using the consumer price index from the China Statistical Yearbook. The daily per capita consumption of high-salt preserved foods and processed meats, sourced from CHNS and China’s Health Statistics Yearbook, was collected for 1992, 1997, 2000, 2002, 2004, 2006, 2009, 2011, 2012, and 2017[50,51]. COD and ammonia nitrogen emissions from industrial wastewater and municipal domestic sewage, sourced from China’s Environmental Statistics Yearbook, were collected for 27 time points from 1992 to 2018 for industrial wastewater variables and 21 time points from 1998 to 2018 for municipal domestic sewage variables[52]. Annual analytical data, data source descriptions, units, data conversion procedures, missing-data handling, time-point alignment, all Joinpoint parameters, and statistical results are detailed in Supplementary Table 2, Supplementary material 2, and Supplementary material Date.

Statistical analysis

Joinpoint regression analysis was used to examine gastric cancer incidence trends across urban/rural areas and genders, as it can identify significant changes in trends over time, which is crucial for understanding the dynamics of disease incidence. We calculated APC and AAPC to provide a comprehensive overview of incidence changes throughout the study period. Correlation analysis, using bivariate statistical methods, assesses relationships between factors and incidence. We acknowledge that simple bivariate correlation cannot account for secular trends, temporal autocorrelation, plausible latency periods, confounding effects, and multipletesting bias. In the absence of further timeseries modelling, all bivariate correlation outputs in this manuscript are treated exclusively as exploratory descriptive temporal associations, and no causal or independenteffect inferences are drawn. The Shapiro-Wilk test was used to assess normality due to the sample size being < 50. For data meeting bivariate normal distribution assumptions, the Pearson correlation coefficient (r) was calculated to assess linearity. For non-normal data, coefficient r was calculated to assess monotonic associations. Two-tailed P < 0.05 was considered statistically significant. For significant factors, the coefficient of determination (R²) was calculated to quantify relationship strength, and factors were ranked by their R² values. All analyses were conducted using Joinpoint Trend Analysis Software (version 5.4.0) and SPSS (version 22.0).

Ethical approval and consent to participate

All data were sourced from public databases and were anonymized. Given the retrospective nature of the study and the use of aggregated, de-identified population-level data, ethical approval was not required.

RESULTS

Trends in gastric cancer incidence in China

Incidence data for 1989-2018 are observed registry-based records; values for 2022 are BAPC-modelprojected estimates without realworld registry verification. Rural males exhibited a declining trend from 1989 to 2000 [APC, -3.03%, 95% confidence interval (CI), -3.89% to -2.16%; P < 0.001] and from 2010 to 2022 (APC, -6.45%, 95%CI: -7.98% to -4.9%; P < 0.001), with an AAPC of -3.47% (95%CI: -4.19% to -2.74%; P < 0.001) (Figure 2). For urban males, a downward trend was observed from 1989 to 2007 (APC, -2.73%, 95%CI: -3.21% to -2.25%; P < 0.001) and from 2010 to 2022 (APC, -4.13%, 95%CI: -4.91% to -3.33%; P < 0.001), but an upward trend occurred from 2007 to 2010 (APC, 18.18%, 95%CI: 3.69% to 34.7%; P = 0.015), resulting in an AAPC of -1.51% (95%CI: -2.68% to -0.33%; P = 0.012). Rural females experienced a decline from 1989 to 2010 (APC, -1.89%, 95%CI: -2.29% to -1.49%; P < 0.001) and from 2010 to 2022 (APC, -5.41%, 95%CI: -7.18% to -3.61%;P < 0.001), with an AAPC of -3.18% (95%CI: -3.86% to -2.5%; P < 0.001). Urban females showed a decreasing trend from 1989 to 2007 (APC, -2.82%, 95%CI: -3.29% to -2.35%; P < 0.001) and from 2010 to 2022 (APC, -3.52%, 95%CI: -4.33% to -2.7%; P < 0.001), with an AAPC of -1.73% (95%CI: -2.91% to -0.53%; P = 0.005). The incidence in rural areas was higher than in urban areas, and it was higher in males than in females. However, the disparity between urban and rural areas was gradually narrowing according to observed 19892018 data; modelprojected estimates suggested near equalization by 2022.

Figure 2 Age-standardized incidence rate of gastric cancer in China, 1989-2022, by sex and urban/rural residence. Symbols: Filled symbols (blue circles, orange diamonds, black triangles, green squares) represent observed age-standardized incidence rates (per 100000, Chinese standard population) from the National Cancer Center of China (1989-2018). Open symbols (circles, diamonds, triangles, squares) represent projected values for 2022 estimated using a Bayesian age-period-cohort model fitted to data from 106 cancer registries during 2010-2018, with bias correction using 2018 data from 700 registries[2]. Solid lines represent Joinpoint regression fitted segments; joinpoints (trend change points) are indicated by line intersections. Rural male (blue, 2 joinpoints): Average annual percent change (AAPC) -3.47%^a [95% confidence interval (CI): -4.19% to -2.74%]; annual percent change (APC) -3.03%^a (1989-2000), -0.25% (2000-2010), -6.45%^a (2010-2022). Urban male (orange, 2 joinpoints): AAPC -1.51%^a (95%CI: -2.68% to -0.33%); APC -2.73%^a (1989-2007), 18.18%^a (2007-2010), -4.13%^a (2010-2022). Rural female (black, 1 joinpoint): AAPC -3.18%^a (95%CI: -3.86% to -2.5%); APC -1.89%^a (1989-2010), -5.41%^a (2010-2022). Urban female (green, 2 joinpoints): AAPC -1.73%^a (95%CI: -2.91% to -0.53%); APC -2.82%^a (1989-2007), 13.09%^a (2007-2010), -3.52%^a (2010-2022). ^aP < 0.05. Source: National Cancer Center of China[2,9-24]. The increases in urban males and urban females during 2007-2010 likely reflect data artifacts from the 2004 expansion of cancer registries and changes in diagnostic practices, rather than true epidemiological increases. Interpretation of incidence trends is primarily based on the filled-symbol observed data (1989-2018). Open-symbol projected values should be interpreted cautiously.

Exploratory ecological temporal correlates of gastric-cancer incidence in China

Bivariate correlation analyses were performed exclusively using 1989-2019 observed incidence data; the 2022 projected estimate was not included in correlation calculations. Each correlation analysis was conducted on different numbers of available time-points according to variable data availability (Table 1). All correlation results are for exploratory descriptive purposes only. The proportion of smokers, and annual COD and ammonia nitrogen emissions from municipal domestic sewage were not correlated with the incidence of gastric cancer in urban and rural areas (Table 1). Average daily alcohol consumption per capita, annual per capita consumption of fresh vegetables, daily per capita consumption of high-salt preserved foods, processed meats and annual ammonia nitrogen emissions from industrial wastewater were not correlated with the incidence of gastric cancer in urban areas. The following were negatively correlated with the incidence of gastric cancer in urban and rural areas: Annual per capita consumption of fresh fruits (urban, -0.749, 95%CI: -0.878 to -0.507;P < 0.001; rural, -0.929, 95%CI: -0.973 to -0.810; P < 0.001); annual per capita disposable income (urban, -0.717, 95%CI: -0.889 to -0.422; P < 0.001; rural, -0.956, 95%CI: -0.980 to -0.883; P < 0.001); and annual ownership of refrigerators per 100 households (urban, -0.712, 95%CI: -0.901 to -0.406; P < 0.001; rural, -0.956, 95%CI: -0.980 to -0.884; P < 0.001). In urban areas, only the proportion of urban population was negatively correlated with the incidence of gastric cancer (-0.686, 95%CI: -0.831 to -0.486; P < 0.001). In rural areas, only daily alcohol consumption per capita was negatively correlated with the incidence of gastric cancer (-0.373, 95%CI: -0.652 to -0.033; P = 0.042). In urban and rural areas, annual COD emissions from industrial wastewater (urban, 0.503, 95%CI: 0.256-0.703; P = 0.008; rural, 0.875, 95%CI: 0.765-0.951; P < 0.001) was positively correlated with the incidence of gastric cancer. The following factors were positively correlated with the incidence of gastric cancer in rural areas: H. pylori infection rate (0.993, 95%CI: 0.976-1.0; P < 0.001); daily per capita consumption of high-salt preserved foods (0.847, 95%CI: 0.690-0.962; P = 0.002) and processed meats (0.745, 95%CI: 0.114-1.0; P = 0.013); annual per capita consumption of fresh vegetables (0.866, 95%CI: 0.675-0.934; P < 0.001), annual ammonia nitrogen emissions from industrial wastewater (0.811, 95%CI: 0.570-0.929; P < 0.001) and the proportion of rural population (0.952, 95%CI: 0.904-0.986; P < 0.001).

Table 1 Correlation coefficient (r) between various influencing factors and age standardized incidence rate by Chinese standard population of gastric cancer in China.

Influencing factors	Year	n	Urban r (95%CI)	Correlation method	Rural r (95%CI)	Correlation method
Helicobacter pylori infection	1989-2018	6	0.533 (-0.457 to 0.996)	Pearson	0.993a (0.976 to 1.000)	Pearson
Smoking	1991-2015	9	0.167 (-0.603 to 0.878)	Spearman	0.500 (-0.555 to 0.948)	Spearman
Drinking	1989-2018	30	-0.047 (-0.396 to 0.272)	Spearman	-0.373a (-0.652 to -0.033)	Spearman
High salt pickled food	1992-2017	10	0.435 (-0.200 to 0.908)	Pearson	0.847a (0.690 to 0.962)	Pearson
Processed meat	1992-2017	10	0.455 (-0.283 to 0.914)	Spearman	0.745a (0.114 to 1.000)	Spearman
Fresh vegetables	1989-2018	30	0.338 (-0.087 to 0.703)	Spearman	0.866a (0.675 to 0.934)	Spearman
Fresh fruits	1989-2018	30	-0.749a (-0.878 to -0.507)	Spearman	-0.929a (-0.973 to -0.810)	Spearman
Disposable income	1989-2018	30	-0.717a (-0.889 to -0.422)	Spearman	-0.956a (-0.980 to -0.883)	Spearman
Refrigerator ownership	1989-2018	30	-0.712a (-0.901 to -0.406)	Spearman	-0.956a (-0.980 to -0.884)	Spearman
COD in industrial wastewater	1992-2018	27	0.503a (0.256 to 0.703)	Pearson	0.875a (0.765 to 0.951)	Pearson
Ammonia nitrogen in industrial wastewater	1992-2018	27	0.393 (-0.005 to 0.714)	Spearman	0.811a (0.570 to 0.929)	Spearman
COD in municipal domestic sewage	1998-2018	21	-0.074 (-0.534 to 0.387)	Spearman	-0.204 (-0.729 to 0.272)	Spearman
Ammonia nitrogen in municipal domestic sewage	1998-2018	21	0.234 (-0.113 to 0.538)	Pearson	-0.163 (-0.768 to 0.291)	Pearson
Urban population ratio	1989-2018	30	-0.686a (-0.831 to -0.486)	Pearson
Rural population ratio	1989-2018	30			0.952a (0.904 to 0.986)	Pearson

^aP < 0.05.

All correlations are calculated using observed incidence data from 1989-2018; Bayesian age-period-cohort projected 2022 estimate was excluded. For each variable, the table below reports effective sample size (n), covered study years, and correlation method (Pearson/Spearman). All correlation outputs are exploratory descriptive temporal associations only and cannot indicate causality. COD: Chemical oxygen demand.

Methods for identifying confounding factors

According to the Cancer Report published in the United States in 2018[40], the risk of gastric cancer increases when daily alcohol consumption exceeds 45 g. Although alcohol consumption in rural areas is on an upward trend, the average daily intake is < 30 g. Therefore, alcohol consumption, which was negatively correlated with the incidence of gastric cancer in rural areas, was identified as a confounding factor. Fresh vegetable consumption is considered a protective factor against gastric cancer, reducing its risk. However, our data indicated a positive correlation between fresh vegetable consumption and the incidence of gastric cancer in rural areas. This seems contradictory but can be explained by Engel’s Law in economics. The law states that as household income rises, the share of spending on food gradually declines. As rural economic levels improve and fridges become more common, households allocate more income to nonfood items. Some stuides[53-55] proposed that this leads to a decrease in the consumption of low-value foods such as vegetables and grains, and an increase in high-value foods such as meat, dairy, eggs, seafood, and fruit. Although fresh vegetables are a protective factor against gastric cancer, their changes in consumption in this study mainly reflected dietary shifts due to rising living standards. These shifts, along with other lifestyle changes, such as reduced intake of pickled foods and better storage conditions, have a combined effect on gastric cancer incidence. Therefore, fresh vegetable consumption is classified as a confounding factor in this context.

Descriptive ranking of temporal associations with the incidence of gastric cancer in China

After eliminating the confounding factors, we calculated the coefficient of determination (R²) for the remaining data and ranked them (Table 2). These R² values reflect unadjusted bivariate associations and are presented for descriptive ranking only; they do not imply causal strength or independence from confounding by shared time trends. In rural areas, H. pylori infection rate showed the highest R² value (R² = 0.986). Factors with relatively higher R² values in urban areas included increased consumption of fresh fruit (R² 0.561), increased income (R² 0.514), greater consumption of fresh produce (R² 0.507), growing urban population (R² 0.471), and effective treatment of industrial wastewater (R² COD 0.253, ammonia nitrogen 0.154). In rural areas, factors with relatively higher R² values included eradication of H. pylori infection (R² 0.986), increased income (R² 0.914), greater consumption of fresh produce (R² 0.914), decline in rural population (R² 0.906), increased consumption of fresh fruit (R² 0.863), effective treatment of industrial wastewater (R², COD 0.766, ammonia nitrogen 0.658), and reduced intake of high-salt pickled foods (R² 0.717) and processed meats (R² 0.555).

Table 2 Coefficient of determination (R²) for statistically significant influencing factors.

Urban	R2	Rural	R2
Fresh fruits	0.561	Helicobacter pylori infection	0.986
Disposable income	0.514	Disposable income	0.914
Refrigerator ownership	0.507	Refrigerator ownership	0.914
Urban population ratio	0.471	Rural population ratio	0.906
COD in industrial wastewater	0.253	Fresh fruits	0.863
		COD in industrial wastewater	0.766
		High salt pickled food	0.717
		Ammonia nitrogen in industrial wastewater	0.658
		Processed meat	0.555

R² values are derived from bivariate correlations on observed 1989-2018 dataset; projected 2022 data was excluded. These R² represent unadjusted ecological temporal associations and do not indicate causal strength. COD: Chemical oxygen demand.

DISCUSSION

Our study shows a downward trend in gastric cancer incidence in China and highlights potential factors affecting urban-rural incidence rates. It emphasizes the significance of these findings for understanding gastric cancer trends and prevention in China. We aim to offer a scientific basis for gastric cancer prevention and treatment, easing the disease burden.

Lin et al[3] revealed that gastric cancer incidence and mortality rates are declining, but some countries show increasing trends. Some stuides[6,7] reported that in Babol, Iran, the age-standardized gastric cancer mortality rate rose from 20.1 per 100000 in 2013 to 20.5 in 2021, with an APC of 2.95% from 2015 to 2021. A global study on gastric cancer incidence from 2003 to 2012 found that most countries, including China (AAPC, males -3.0%, females -3.1%), had downward trends, while Canada saw an increase in females (AAPC 1.2%)[3]. A study on high-risk gastric cancer regions in China from 2010 to 2019 indicated a stable incidence rate[8]. Our research collected Chinese gastric cancer incidence data from 1989 to 2018 via the NCC, which showed declining incidence rates in rural males (AAPC -3.47%), urban males (AAPC -1.51%), rural females (AAPC -3.18%), and urban females (AAPC -1.73%). Gastric cancer incidence rates were higher in rural than urban areas, and higher in males than females, but the urbanrural gap narrowed yearbyyear in observed 1989-2018 data; BAPC-modelprojected estimates suggested nearequalization by 2022.

According to the Sixth National Consensus on the Management of H. pylori Infection in China and the international Maastricht VI Consensus[56,57], noncardia gastric cancer arises from a multifactorial etiology involving H. pylori infection, environmental influences, and genetic predisposition. The population attributable risk percentage of H. pylori infection in noncardia gastric cancer is estimated to be 75%-89%, whereas only 1%-3% of gastric cancer cases demonstrate a genetic predisposition. However, this does not explain the past trends in gastric cancer incidence in China. Our research data indicated a significant positive correlation between the incidence of gastric cancer and the infection rate of H. pylori only in rural areas of China.

According to Shichijo et al[58] and Rugge[59], eradication of H. pylori infection can only reverse some cases of atrophic gastritis, but it is difficult to reverse intestinal metaplasia. H. pylori eradication began only in the late 20^th century, yet before this time, the incidence and mortality rates of gastric cancer were already declining. For example, Japan once had the highest gastric cancer incidence and mortality rate globally[4]. Despite studies in 1992 showing that the H. pylori infection rate among people aged > 40 years exceeded 70%, the gastric cancer mortality rate declined by 60% from 1965 to 1995[5]. In urban areas of China, following the promotion of H. pylori eradication, the infection rate has declined. However, the incidence and mortality rates of gastric cancer have not shown a significant reduction and have even risen since 2009. In contrast, Wen et al[60] reported that, due to improved economic conditions, better living environments, and reduction of unhealthy habits such as drinking untreated water, the H. pylori infection rate and incidence of gastric cancer have significantly decreased in rural areas of China, showing a significant positive correlation between the two. Collectively, our ecological analysis revealed temporal correlation between H. pylori prevalence and gastric-cancer incidence only in rural China. Economic, environmental, and population-migration variables exhibited ecological temporal correlations with incidence changes across urban and rural settings, though causal links cannot be inferred from this study.

Although the incidence and mortality rates of gastric cancer in China have declined, there is a significant urban-rural disparity, with the rate in rural areas being more than double that in urban areas. However, this gap is gradually narrowing and has basically converged since 2010, closely related to economic factors. Our results indicated that the daily per capita consumption of high-salt preserved foods and processed meats was positively correlated with the incidence of gastric cancer in rural areas. In contrast, the annual per capita consumption of fresh fruit, the annual per capita disposable income, and the annual ownership of refrigerators per 100 households were negatively correlated with the incidence of gastric cancer in both urban and rural areas, with stronger correlations observed in the latter. These findings show ecological temporal correlations between economicrelated variables and gastriccancer incidence in both urban and rural China.

In the past, due to poor economic conditions in rural areas, there was a lack of effective refrigeration methods. To ensure that food did not spoil rapidly, preservation often relied on pickling. However, according to the Cancer Report published in the United States in 2018, long-term consumption of high-salt pickled foods and processed meats increases the risk of gastric cancer[40]. Consequently, Ferlay et al[61] proposed that East Asian regions, where traditional diets heavily feature pickled foods, have always had a high incidence of gastric cancer, with a stronger correlation in rural areas. Nevertheless, with economic growth and the promotion of national policies such as the “Home Appliances Going to the Countryside” initiative, refrigerators have rapidly gained popularity in rural China, leading to a gradual decline in the demand for traditional preserved foods. Yu[55] noted that the dietary habits of residents have become more diversified, with a decrease in consumption of low-priced foods such as grains and vegetables, and an increase in consumption of high-priced foods such as meat, dairy products, eggs, seafood, and fruit. Over the past 30 years, rural areas have experienced more significant economic development and changes in dietary structure compared to urban areas, co-occurring with a more pronounced decrease in gastric cancer incidence. These temporal changes have coincided with a narrowing of the urbanrural gap in gastric cancer incidence, and ecological correlations were stronger in rural areas.

Apart from economic factors, environmental pollution, particularly contamination of drinking water, is closely associated with the incidence of gastric cancer. The primary indicators for monitoring water pollution include COD and ammonia nitrogen. COD reflects the level of organic pollution in water bodies. Wilkins et al[62] showed that organic pollutants can accumulate in aquatic organisms, and consumption of these contaminated organisms by humans can lead to carcinogenesis, teratogenesis, and mutagenesis. Ye et al[63] proposed that in the presence of sufficient oxygen, ammonia nitrogen can be oxidized by microorganisms into nitrate nitrogen, which further decomposes into nitrite nitrogen. When nitrite nitrogen combines with proteins, it produces nitrosamines, which are carcinogenic.

Our data indicates annual industrial-wastewater COD emissions were positively correlated with gastric cancer incidence in both urban and rural areas, while annual ammonia nitrogen emissions were positively correlated with gastric cancer incidence only in rural areas. In contrast, there was no correlation between the annual emissions of COD and ammonia nitrogen from municipal domestic sewage and the changes in gastric cancer incidence in urban and rural areas. These findings indicate an ecological temporal association between industrial-wastewaterrelated water pollution and gastric-cancer incidence; however, causal relationships cannot be established based solely on these ecological correlations.

During the initial stages of China’s reform and opening up, economic development was primarily extensive, leading to the entry of many highly polluting industries into the country. A significant amount of industrial wastewater was discharged into rivers, lakes, and seas, resulting in pollution. In the past, due to their backward economic conditions, rural areas lacked wastewater treatment plants and centralized drinking water supplies, making them the primary regions affected by water pollution. Conversely, urban areas are less affected due to the presence of wastewater treatment plants and centralized drinking water supplies. Subsequently, as the country prioritized environmental protection, the economic development model shifted from extensive to intensive. Industrial wastewater discharge gradually decreased annually, and enterprises increased their investments in wastewater treatment. Meanwhile, rural areas gained access to centralized tap water supplies. These factors contributed to the decline in the incidence of gastric cancer in rural areas.

With the continuous development of China’s economy and the expansion of cities, China has rapidly transformed from a traditional rural to a modern urban society. According to data from the National Bureau of Statistics, China’s urbanization rate of permanent residents reached 66.16% at the end of 2023, an increase of 55.52% compared with the end of 1949, with an average annual increase of 0.75%[27]. This marks one of the largest and fastest urbanization processes in history.

Based on the urban and rural population proportion data extracted from the China Statistical Yearbook, our analysis revealed that the incidence rate of gastric cancer in urban areas was negatively correlated with the proportion of the urban population, while the incidence rate of gastric cancer in rural areas was positively correlated with the proportion of the rural population. These results demonstrate ecological temporal correlations between urbanization-related indicators and gastric-cancer incidence in urban and rural populations. Urban improvements in economy, environment and medical conditions co-occurred with declining gastric-cancer incidence. Meanwhile, according to economic research data published by Cao et al[64], the growing urban population mainly comprises young and middle-aged rural laborers and highly educated rural adolescents. This results in a negative correlation between the incidence of gastric cancer in urban areas and the proportion of the urban population. In contrast, rural areas lag behind in terms of economy, environment, and medical conditions. Rural-to-urban population migration coincided with falling gastric-cancer incidence in rural areas. Therefore, the incidence of gastric cancer in rural areas is positively correlated with the proportion of the rural population.

Our study had several limitations. First, the updated gastric cancer screening program in China, particularly the 2024 national guidance emphasizing risk-based endoscopic screening, may have affected recent incidence rates[65]. However, its nationwide coverage remains limited and it was primarily implemented in high-risk regions, making it insufficient to explain the three-decade long-term declining trend. Second, advances in endoscopic technology, pathological diagnostic criteria, and disease classification systems may have altered case detection and registration practices, rendering it difficult for incidence fluctuations to truly reflect epidemiological changes. Third, data on H. pylori infection rates were derived from segmented literature reviews[41-48], lacking continuous surveillance. Fourth, the effects of influencing factors on gastric cancer incidence may have been delayed. Fifth, data on other potential factors, such as EBV infection, the proportion of workers exposed to high-temperature and dusty environments, and hereditary gastric cancer, were unavailable. Sixth, bivariate statistical analysis can identify associations but not causal relationships; such correlations among strongly co-trending annual variables are susceptible to spurious associations and do not control for time, autocorrelation, latency, confounding, or multiplicity. Finally, as an ecological study, we cannot rule out residual confounding from screening intensity, diagnostic practices, and other unmeasured factors; therefore, the results should be regarded as exploratory findings rather than definitive causal evidence.

CONCLUSION

Gastric cancer incidence rates have been declining in China for > 30 years. Although H. pylori infection is an important factor in gastric cancer development, our ecological analysis revealed temporal correlation between H. pylori prevalence and incidence only in rural settings. In contrast, economic growth, improved environmental conditions, and population migration patterns showed ecological temporal correlations with the declining incidence in both urban and rural areas. Notably, 2020-2022 data are modelderived estimates rather than realworld observations, and ecological correlation analysis cannot establish causality. The persistently high incidence in rural areas indicates the need for targeted prevention interventions.

ACKNOWLEDGMENTS

We thank Dr Yong-Quan Shi from the Xijing Hospital of Digestive Diseases for his contributions to the study.