Introduction
Xuanwei County in Yunnan Province of China was reported to have an abnormally high lung cancer rate by the first mortality survey in China in the 1970s [
1,
2]. Since then, a number of epidemiological studies have been conducted to examine the risk factors of lung cancer in this county [
3−
5]. These studies have suggested that the high lung cancer rate cannot be attributed to smoking because most women are nonsmokers (smoking prevalence lower than 0.21% in women and approximately 42.6% in men [
6]), but they have a similar lung cancer rate to men; industrial pollution is also excluded to be the major reason because no major industry existed in this area before the 1960s [
2]. Indoor air pollution by smoky coal burning for domestic heating and cooking has been suggested as the culprit [
2].
Although numerous studies on lung cancer have been conducted, the epidemiology of other types of cancer in Xuanwei has not yet been documented. Describing the epidemiology of other cancer types, especially their temporal trend, will provide thorough knowledge of the cancer burden among local residents and allow a critical assessment of current strategies for overall cancer control and prevention.
In the United States and several European countries, which have closely monitored the dynamics of cancers, the mortality and distribution pattern of cancers have changed substantially during the past decades, which provided valuable information to guide relevant policy-making and rational health resource allocation about cancer prevention and control [
7,
8]. The overall cancer mortality in China increased rapidly and continuously during the past three decades, from 74.2/100 000 in the 1970s to 108.3/100 000 in the 1990s and to 135.9/100 000 in 2004−2005; such increases were mainly attributable to aging population, tobacco smoking, environmental deterioration, and western lifestyles [
9]. Among various types of cancer, a significant increase was observed in lung cancer, liver cancer, colorectal cancer, and leukemia; meanwhile, mortality of breast cancer among women decreased, but its incidence increased in the past three decades [
9].
However, no such systematical epidemiological study has been conducted in Xuanwei County, although it was included in three mortality surveys in China in the 1970s, 1990s, and 2000s. These surveys provided rich data to have such an analysis. This study aimed to analyze the mortality trends of eight leading cancers in Xuanwei County over the three periods and provide possible explanations.
Materials and methods
Study setting
Xuanwei County is located in the southwestern Chinese province of Yunnan (103°30′−104°40′E, 25°3′−26°41′N). It is 102 km from east to west and 91 km from north to south; its total area is 6257 km
2 [
10]. This county had a population of about 0.8 million people in 1978 (1.2 million in 1995 and 1.3 million in 2003), more than 90% of whom were traditional Chinese farmers. Over 90% of the population belonged to China’s predominant ethnic group, the Han. The population was residentially stable [
3].
Cancer mortality data
The mortality data were obtained from three Chinese national mortality surveys conducted in 1973−1975, 1990−1992, and 2004−2005. The first mortality survey was a nationwide census, which was designed to produce information regarding the mortality profile in the entire China at that time, and it retrospectively collected almost all deaths that occurred during the period of 1973−1975. An atlas was published based on the survey results, which provided a vividly visual geographic distribution profile of different types of cancer in the early 1970s [
11]. An obvious geographical distribution pattern of lung cancer mortality appeared in China, with the highest mortality located in Xuanwei County for both males and females at that time period.
The second survey was a two-stage stratified sampling survey, which retrospectively collected death information during the period of 1990−1992 [
12]. At the first stage, each of the 22 provinces, five autonomous regions, and three municipality directly administered cities represented one of 30 strata. The second stratum was formed at the county or district level according to the first mortality survey. Overall, the survey was designed to cover 10% of all deaths or 242 million baseline person years during the study period. Xuanwei County was selected as one survey site.
The third survey, conducted in 2006, retrospectively collected all deaths that occurred in 2004 and 2005 in 143 million person years, which were equal to about 6% of the entire national population at that time [
13]. The Disease Surveillance Point, which was used by Chinese Center for Disease Control and Prevention, was applied to the survey. A stratified random sample of 160 counties/districts was selected and proven to be representative of the population in the entire country [
13]. Xuanwei County was selected as one survey site.
All the three surveys in Xuanwei covered almost all the local residents who died in corresponding years. The protocol of the surveys has been elucidated elsewhere [
11−
13]. In brief, the surveys were organized by the National Office for Cancer Prevention and Control of China. A standardized death certificate card and relevant instructions were issued. All medical and health facilities in Xuanwei County and nearby cities, including Kunming, the capital city of Yunnan Province, were obliged to notify all deaths from Xuanwei County that occurred in their hospitals during the corresponding period. The death certificate card included information on name, sex, date of birth, residential address, occupation, cause of death, and date and basis of cause of death of the deceased. The certificate cards were sent to the corresponding health departments according to the residential address of the deceased. Based on this information, the health department staff could determine whether a case had already been reported, thereby avoiding duplication and reporting errors. All deaths from outside the territory of Xuanwei were eliminated from the records. Some deceased might not undergo medical treatment or died at their own home, so the hospital reports could not capture these persons; accordingly, a comprehensive survey was conducted in every village to collect the information of these deaths, usually from the village doctors, the spouse, and son or daughter of the deceased [
14].
The three mortality surveys witnessed a change in the International Classification of Disease (ICD) catalog, from the 8th to the 10th version. All deaths were re-coded using the three-digit rubrics of the 9th (for the first two surveys) and 10th revisions (for the third one).
The cancers of different sites included in this study were the following eight leading cancers: lung (ICD 9 code: 162; ICD 10 code: C33-C34), stomach (ICD 9 code: 151; ICD 10 code: C16), esophagus (ICD 9 code: 150; ICD 10 code: C15), breast (ICD 9 code: 174-175; ICD 10 code: C50), liver (ICD 9 code: 155; ICD 10 code: C22), rectum (ICD 9 code: 154; ICD 10 code: C19-C20), leukemia (ICD 9 code: 204-208; ICD 10 code: C91-C95), and brain (ICD 9 code: 191; ICD 10 code: C71). Data over the three time periods were extracted from the mortality survey databases for Xuanwei County. Approval to conduct this study was granted by the relevant ethics committees of The Chinese University of Hong Kong (Shatin, New Territories, Hong Kong Special Administrative Region, China). Informed consent was not obtained for each subject because this analysis did not use any individual identifiable information.
Population data
Sex- and age-specific populations for the three time periods were estimated by interpolation based on the third, fourth, and fifth national population censuses in 1982, 1990, and 2000, respectively [
15−
17].
Statistical methods
The sex- and age-specific annualized mortality rates were calculated for the eight major types of cancer over the three time periods. Age-standardized annual mortality rates for the three time periods were also calculated using the Chinese population in 1990 as a standard population. We plotted the mortality rates of the eight common cancers against the three time periods (1973−1975, 1990−1992, and 2004−2005) to visualize the temporal pattern. The mortality rates of lung cancer in Xuanwei were also compared with the national average and that in Yunnan Province at the corresponding periods. The age-specific mortality rates for each cancer over the three time periods were plotted to examine the development trend. The percentage change in mortality of each cancer type was estimated using the mortality at 1973−1975 or 1990−1992 as the reference mortality. Statistical significance was examined using a Poisson model, in which age-specific mortality count was treated as a dependant variable and age and time period were explanatory variables.
All data management and statistical analyses were performed using SAS 9.1 (SAS Institute Inc., Cary, NC, USA) and R software Version 2.14.1 (R Development Core Team, 2012).
Results
Temporal trend of leading cancer mortality
An increasing trend was observed in cancer mortalities for both sexes during the three time periods. Among men, the age-standardized mortality rates for all cancers combined increased from 57.22/105 in 1973−1975 to 65.2/105 in 1990−1992 and to 123.8/105 in 2004−2005, and deaths from cancers accounted for approximately 7.3%, 11.9%, and 24.2% of all causes of death during the three time periods, respectively. However, the evolution trend for specific cancers varied substantially, as shown in Tables 1 and 2.
During 1973−1975, a total of 676 deaths from malignant neoplasm among men and 571 among women were observed. Among men, lung cancer was the leading cause of malignancy deaths (with an annual mortality of 27.9 per 105 population), followed by cancers of liver (7.2 per 105 population), stomach (5.0 per 105 population), and brain (2.5 per 105 population). Among women, lung cancer was also the predominant malignancy (with an annual mortality of 24.5 per 105 population), followed by cancers of liver (4.1 per 105 population), stomach (3.8 per 105 population), breast (2.9 per 105 population), and esophagus (1.1 per 105 population).
During 1990−1992, a total of 1255 deaths from malignant neoplasms among men and 937 among women were reported. Among men, lung cancer was the leading form of malignancy deaths (with an annual mortality of 36.7 per 105 population), followed by cancers of liver (9.6 per 105 population), stomach (3.9 per 105 population), and leukemia (3.1 per 105 population). Among women, the predominant malignancy was also lung cancer (with an annual mortality of 38.8 per 105 population), followed by cancers of liver (5.2 per 105 population), stomach (4.2 per 105 population), leukemia (3.3 per 105 population), and breast (3.1 per 105 population).
During the period of 2004−2005, a total of 2498 deaths from cancers of all sites among men and 1766 among women were recorded in the study area. For both sexes, lung cancer was the leading form of malignancy. For men, the five leading cancers were lung (69.4 per 105 population), liver (18.4 per 105 population), brain (5.9 per 105 population), stomach (5.6 per 105 population), and leukemia (5.3 per 105 population). For women, the five common cancers were lung (74.3 per 105 population), liver (12.6 per 105 population), stomach (5.5 per 105 population), brain (4.3 per 105 population), and breast (3.8 per 105 population).
Table 1 shows the temporal trend of the mortality of leading cancers during the three time periods in Xuanwei County. Table 2 presents the percentage change and statistical significance of the temporal trend. From 1973−1975 to 1990−1992, a slight increase was found in the mortalities from cancers of brain, liver, lung, rectum, and leukemia in men and cancers of lung, brain, liver, breast, and stomach in women, whereas a decrease was found in cancers of esophagus and stomach in men and cancers of esophagus in women. By contrast, most cancers had an upward trend from 1990−1992 to 2004−2005 for males and females. During the three time periods, mortalities from most cancers showed a relatively lower level than the Chinese national average (Supplementary Table S1) [
3,
6,
13,
18], with lung cancer as an exception for both sexes, which was the leading malignancy death in Xuanwei and among the highest in China [
3,
18]. An obvious increase and a high rate were observed in mortality from liver cancer for men (92.4% increase) and women (143.7% increase) from 1990−1992 to 2004−2005, although the rates were lower than the national average (for men: 19.4 vs. 37.6; for women: 13.2 vs. 14.5).
Age-specific cancer mortality
Fig. 1 presents the age-specific mortality rates of the leading cancers among men and women in Xuanwei County over the three time periods. For most cancers, the mortality was higher in older age groups than that in younger age groups, with leukemia as an exception. A slight increase was observed for most malignancies from 1973−1975 to 1990−1992, and a more pronounced increase existed from 1990−1992 to 2004−2005. An increase in age groups of 25 years and above was generally found in 2004−2005, but the increase was not consistent across the age groups. In particular, the increase was more obvious in the age group of 60 years and above.
Comparison with Yunnan Province and the national levels
We compared the lung cancer mortality in Xuanwei County with that in Yunnan Province and the national average during the three time periods. As shown in Fig. 2, an increasing trend occurred during the past three survey periods in Xuanwei, Yunnan, and the entire China; lung cancer mortality was higher in Xuanwei than that in Yunnan Province and China. The pattern was similar for males and females (Supplementary Fig. S1).
Discussion
Xuanwei County has been a focus of research on lung cancer during the past decades [
18−
20]. This study might be the first to examine the mortality pattern and temporal trend for other types of cancer in this area. Results revealed that the spectrum of malignancies gradually changed in the past decades in Xuanwei. Increasing trends were observed for most cancers, but they were still at relatively lower level than the national average [
12,
13]. For all malignancies, the annual mortality increased from 52 per 10
5 in 1973−1975 to 61 per 10
5 in 1990−1992 and to 164 per 10
5 in 2004−2005.
Although several increases in the mortality might be partly due to diagnosis improvement, as in the early 1970s and 1990s when the socio-economic status was relatively lower, the residents might not have consulted a doctor when they were sick, so they might not be correctly diagnosed. The local policy on financial subsidy also might have been partly responsible for the high cancer mortality in 2004−2005; since 2003, local residents who went to consult a doctor in the hospital can acquire 40%−50% compensation of hospitalization expenditures, which caused soaring hospitalization [
21]. However, the temporal trends were unlikely to be fully explained by improvements in cancer detection or completeness of mortality surveys. The observed trends in cancer mortalities may reflect, to some extent, the changes in the patterns of relevant risk factors in this area, such as cigarette smoking, air pollution, and dietary factors [
20,
22].
Over the past decades, lung cancer has remained the leading killer among all cancer types; its mortality has shown an obvious increasing trend over the past decades, with annual mortality rates of 69.4 and 74.3 per 10
5 in 2004−2005 for males and females, respectively, which were the highest in China [
13].
Lung cancer is one of the preventable cancers because more than 80% of cases can be attributed to tobacco smoking worldwide [
23]. However, Xuanwei is an exception, where lung cancer is mainly caused by indoor air pollution from smoky coal combustion [
5]. The observed temporal trend may be partly due to improvements in disease diagnosis and access to health care and probably to smoking and air pollution caused by indoor smoky coal burning for heating and cooking, which has been linked to the excess lung cancer risk in this area in previous studies [
2,
24,
25]. Different lung cancer risks have been reported for different coal types in Xuanwei in previous studies, and smoky coals have been found to have more carcinogens than smokeless coals and woods [
4]; an intervention, namely, stove improvement, was implemented in the late 1970s, and a cohort study reported a significant reduction in lung cancer risks after about 10 years of the intervention [
26]. Nevertheless, our study found an increasing trend of lung cancer mortality. This trend likely occurred after the installation of chimney. The vented emissions were released outdoors and may contribute to high neighborhood air pollution even when all houses used vented stoves. The neighborhood pollution created by local household emissions led to a significant re-entry of pollution into households [
27]; it was also possible that smoky coals were used more often and widely in the study area. Other possible explanations pointed to industrial pollution, which was excluded as the reason for the high lung cancer risk in the 1970s. However, several industries have been located in this area since then.
Liver cancer, the second leading malignancy in Xuanwei for both sexes over the three time periods, has been closely associated with hepatitis B infection and aflatoxin intake [
28,
29]. The improved coverage of hepatitis B vaccine and the improvement in living standards have decreased liver cancer in urban Shanghai, China [
30,
31]; however, in Xuanwei, a stable increase trend has been observed. A large-scale hepatitis B vaccination survey in China has revealed that the vaccination coverage is above 95% among neonates in urban area, but it is only about 50% in rural areas [
32]. Xuanwei is a typical Chinese rural county, where 90% of the residents are farmers [
2]; although no data are available on the hepatitis B vaccination coverage in this area, the low hepatitis B vaccination coverage may be responsible for the high mortality from liver cancer in Xuanwei County.
Liver cancer often occurs after liver cirrhosis. In the study area, the mortality rates of cirrhosis during the first two periods (11 and 9 per 105) were approximately two times of that in 2004−2005 (5 per 105); the average death ages of cirrhosis were 54, 56, and 62 years during the three time periods, respectively. Therefore, the large decrease in cirrhosis mortality may be mainly due to the improvement in the disease treatment and partly due to the transfer of deaths from cirrhosis to deaths from liver cancer; the improvement in liver cirrhosis survival provides sufficient time for liver cancer development. Changes in cirrhosis management would then be a possible reason for the increase in liver cancer mortality.
Stomach cancer, the most common malignancy in China [
33], has presented a decreasing trend in both rural and urban areas in China [
30,
31]. By contrast, a continued slight increase has been noted in Xuanwei, although the rates are relatively lower than the national average. Though the increase is partly due to the completeness of data record, improvements in people’s access to health care, and clinical diagnosis; this increasing trend may be a reflection of the real situation. Further studies are needed to explore this research question in the future. Studies have suggested the important roles of diet and infection with
Helicobacter pylori in the etiology of stomach cancer [
34,
35]. Several local lifestyles, such as consumption of the locally famous tofu and vegetables, have been suggested to have a possible protective effect on stomach cancer risk [
36], which may cause the low rate in this county.
Esophageal cancer showed a decline in mortality from the 1970s to the 1990s, with rates in 1990−1992 about half of those in the 1970s. This trend was consistent with the findings in urban Shanghai [
30], which might be related to improvements in nutritional status. Cardiac cancer was classified into stomach cancer in the 1990s, but it was classified into esophageal cancer in the 1970s [
33]; the decline in mortality rate for esophageal cancer might be partly explained by this classification change. Smoking and alcohol consumption are considered risk factors of esophageal cancer [
37,
38]. An increasing trend in esophageal cancer was found for both sexes in 2004−2005, which might be attributed to the prevalence of smoking and alcohol consumption, although no data on the smoking and alcohol consumption are available.
Although breast cancer is the most frequent cancer in women worldwide [
39], it has relatively low mortality rates in Xuanwei, ranking fifth in the female malignancy deaths. The evolutionary trend of female breast cancer mortality in Xuanwei was similar to that in overall China, with a slight increase over the three time periods; the trend remained at a low level, but female breast cancer will probably experience an increase trend in future [
31,
39,
40]. The slight increase in mortality might indicate a large increase in incidence because of the improvement in cancer treatment over the past decades [
23]. Risk of breast cancer is strongly associated with reproductive life and ovarian function [
41]. Menstrual and reproductive patterns among Xuanwei women may have changed, as reported in other areas in China; women born more recently have an earlier age at menarche and later age at first pregnancy than earlier birth cohorts [
31], which may be responsible for the change trends in mortality from female specific cancers. This pattern may also be related to trends in dietary fat intake and obesity, as suggested by a study in Shanghai [
42].
Cancers of brain, rectum, and leukemia appeared to be rare in this county and remained rather stable. A slight increase was observed, which was believed to be mainly due to the continuous improvements in diagnosis and access to health services, although some underlying risk factors should not be neglected [
31].
Cause-specific mortality rates derived from vital statistics are highly valuable for the evaluation of health status; a critical evaluation of data quality is necessary because several sources of errors can affect mortality statistics [
43]. Our study results should be treated with caution because of the uncertainties associated with the temporal changes in medical diagnosis, health care access, and survey methods.
In summary, our analyses of temporal trend of mortality from the eight leading cancers in Xuanwei County revealed for the first time that most cancers demonstrated an increase trend during the past three decades, with lung cancer as the predominant malignancy. Further epidemiologic studies in this area are warranted to systematically examine the underlying reasons for the temporal trend of the major cancers in Xuanwei and advise strategies for overall cancer control and prevention.
Higher Education Press and Springer-Verlag Berlin Heidelberg