Increasing social inequalities in health in China, coupled with growing inequalities in income and wealth, have focused attention on social class as a key determinant of population health. Our study focused on breast cancer stage at diagnosis as an indicator of health care access and quality, and we found a significant relationship between cancer stage and area-based SES. To our knowledge, ours is the first study to tie SES to health status within a developing country.
In this nationwide multicenter study of breast cancer in China, area-based SES was categorized into three levels: highest, high, and low. The Beijing district in north China was the only highest SES area, and also had the highest GDP per capita, percentage of health-service expenditure in the regional/provincial public affairs general budget, and ratio of urban to rural population, and the lowest percentage of illiteracy in females aged 15 and over. Northeast, south, and east China were high SES areas, while central, northwest, and southwest China were low SES areas. This classification accords with the pattern of economic growth and regional inequality in China during the reform era [25
]. We found that cases living in low SES areas were more likely to be diagnosed at a later breast cancer stage than were those in high SES areas.
Exactly how area-based socioeconomic conditions influence the stage at which an individual is diagnosed with breast cancer is complex. One explanatory factor is that individuals residing in areas with high GDP per capita are likely to have more personal income compared with those in low GDP per capita areas. In the United States, lack of ability to pay for screening services is often implicated as the reason why individuals with low incomes, or those living in the poorest areas, have lower screening rates and are more likely to be diagnosed with cancer at a late stage [34
]. In the present study, we observed that the patients from hospitals located in Beijing and Guangdong were diagnosed at an earlier stage than others. This may partly be explained by the routine breast cancer screening programs that have been implemented in these two cities for several years [32
]. However, the screening programs in a city may not reflect the screening situation of the whole region. To evaluate the effect of routine screening on early diagnosis, a comprehensive prospective study is needed to compare districts with and without screening programs.
Local governments with a high percentage of health-service expenditure in their regional/provincial public affairs general budgets (HSE percentage) may provide better medical services, giving residents more access to health care such as breast cancer screening, diagnosis, and treatment, which in turn results in earlier diagnosis. Barrya and Breen found that residence in economically and socially distressed or medically underserved neighborhoods tends to increase the likelihood of late-stage cancer diagnoses [19
]. However, we observed no clear trend association between HSE percentage and breast cancer stage at diagnosis. HSE percentage may not be a direct indicator of the level of medical service of an area, since HSE percentage may be inconsistent with an area's economic development. For instance, while northeast China had high GDP per capita, it was the lowest HSE percentage of the areas studied.
In contrast, we found an association between earlier breast cancer stage at diagnosis and higher ratio of urban to rural population (PU/PR ratio). In China, there is wide SES gap between cities and rural areas, and so urban/rural status is highly correlated with SES. A higher PU/PR ratio hints that more individuals in the area live in cities, and thus may have more access to medical services and ability to afford medical insurance than do residents of rural areas. This would in turn make them more likely to seek prompt medical care for any health problems. These differences may contribute to disparities in breast cancer stage at diagnosis. Several studies have shown that patients with health insurance are less likely to be diagnosed at a late stage of breast cancer [22
], an effect similar to that found for other cancers, such as colorectal cancer [34
In this study, area-based education was measured as percentage of illiteracy in females aged 15 and over (FPI percentage), emphasizing education of females. We found a trend association between FPI percentage and stage at diagnosis: the higher the FPI percentage in an area, the later the stage of breast cancer at which women in the area were diagnosed. After stratified analysis by area-based SES, individual education level appeared to influence the stage at diagnosis in high SES areas, while in low SES areas, working women were less likely to be diagnosed at late breast cancer stages than were homemakers. These results highlight the importance of education in high SES areas. Education may be a proxy variable representing individual knowledge and behaviors toward prevention and screening of breast cancer. Women with lower education level may also have higher risk of alcohol-related death and diseases [37
]. Low education itself may also create barriers to receiving recommended screening, since low health literacy, low general literacy, and language barriers impact an individual's ability to navigate the medical service system, understand screening options and recommendations, and communicate with healthcare professionals [38
]. Working women in low SES areas were more likely to be diagnosed at early stages, which may result from their high breast cancer screening attendance relative to that of homemakers. The results from our previous study [39
] and a study by Damiani et al. [40
] show that education and occupation were positively associated with breast cancer screening attendance, which in turn may influence breast cancer stage at diagnosis.
However, after adjusting for area-based SES, we found no significant associations between other individual demographic characteristics and breast cancer stage at diagnosis. Thus, area-based SES appeared to be the main, underlying factor influencing breast cancer stage at diagnosis. Baquet et al. similarly found that SES predicted the likelihood of a group's access to education, certain occupations, and health insurance, as well as income level and living conditions, all of which are associated with a person's chances of being diagnosed with late stage disease and of surviving cancer [41
]. Moreover, Singh et al. [42
] and Launay et al. [43
] found that at every stage of diagnosis, breast cancer patients from lower-income areas had lower 5-year relative survival rates did than did those from higher-income areas. The presence of additional illnesses and treatment disparities may contribute to these differences [44
]. Compared with less developed areas, more developed areas possess better detection techniques, and their cases have the ability to pay more for more accurate tests and more effective therapy. For instance, we found that in Beijing, the highest SES area, more than 95% of cases were tested for ER/PR/Her-2. In most of the low SES areas, however, only about half of cases were tested for Her-2. Even so, this pattern was in agreement with the guidelines of the Breast Health Global Initiative (BHGI), which contend that Her-2 measurement is problematic in limited-resource settings due to the high cost of immunohistochemical analysis, fluorescence in situ hybridization, and trastuzumab therapy, and recommend introducing this test only at the maximal-resource level [46
The present study is the first geographically epidemiologic study of breast cancer in China. Although we only included women with breast cancer attended to these 7 regional referral hospitals, we expect the effect of SES and later stage of diagnosis will be much stronger than we found in this study (if we included those women attended at local hospitals, such county hospitals). Our findings thus provide a baseline for understanding the country's patient and tumor characteristics. Our evaluation of associations between breast cancer stage and area-based SES, as well as individual demographic characteristics stratified by area-based SES, may help to identify high risk areas and the most important and controllable regional and individual factors influencing breast cancer stage at diagnosis. However, this was an ecological study, so we must recognize the possibility of ecological fallacy. In addition to the individual-level variables examined in this study, it would be useful to examine the effects of other variables, such as access to breast cancer screening services, alcohol use, family history of breast cancer, and lack of exercise, as they may affect breast cancer stage at diagnosis by influencing women's attitudes and behaviour to breast cancer prevention and screening. However, their effects could not be analyzed in this study due to the high proportion of missing data; we plan to test the effects of these variables on breast cancer stage in a future study. Another limitation is that we lacked data on individual SES, which was likely strongly correlated with both education level and area-based SES. Although our method of calculating area-based SES may not have assessed actual socioeconomic conditions with complete accuracy, our results for area-based SES accord with the pattern of economic growth and regional inequality in China during the reform era [25
]. Thus, we think this indicator has some validity. However, a more comprehensive, validated area-based SES measurement is needed that takes into consideration factors such as concentration of poverty, health insurance coverage, proportion of the population with blue collar jobs, unemployment rate, median household income, and median value of owner-occupied houses [47
]. Finally, area-based SES was measured at a large geographic scale in our study; it would be instructive to study area-based SES differences among smaller geographic areas in a future study.