Socio-economic and rural-urban differences in healthcare and catastrophic health expenditures among cancer patients in China: analysis of the China Health and Retirement Longitudinal Study

In China, cancer deaths account for one-fth of all deaths and exert a heavy toll on patients, families, healthcare systems, and society as a whole. This study aims to examine socio-economic and rural-urban differences in treatment, healthcare service utilization and catastrophic health expenditure (CHE) among Chinese cancer patients, and to investigate the relationship between different treatment types and healthcare service use as well as incidence of CHE. We analyzed a nationally representative sample from the China Health and Retirement Longitudinal Study including 17,224 participants in 2011 and 19,569 participants in 2015. Multivariable regression models were performed to investigate the association of cancer treatments with healthcare service utilization and CHE.

To mitigate the healthcare and nancial burden of NCDs including cancers, the central government of China has launched a new round of comprehensive healthcare system reforms since 2009, focusing on healthcare delivery, basic public health service, medical insurance, essential medicine system and public hospital reform. Most of these measures were designed to improve the access to quality and affordable care for patients with NCDs and health outcomes [2,3]. In 2015, URBMI (Urban Residents Basic Medical Insurance) and NCMS (New Rural Cooperative Medical System) were integrated into one urban-rural resident health insurance scheme, which enhance the nancial risk protection for patients with cancers [4,5]. In addition, the Critical Illness Health Insurance and supplementary medical insurance programs were developed to offer extra bene ts packages to citizens with severe chronic diseases. The reform and consolidation of three-level medical institutions (including primary healthcare centers, secondary hospitals and tertiary hospitals) aimed at providing coordinated diagnoses, treatment, and follow-up care to Chinese cancer patients [6]. Despite the launch of China healthcare system reform in 2009, few empirical data have evaluated the impact of the comprehensive health reforms on cancer health expenditure.
We hypothesize that reformed medical insurances and healthcare delivery systems could further reduce catastrophic health expenditures (CHE) among patients with cancer in both rural and urban regions of China. The CHE in this study is de ned as the point at which annual household healthcare expenses exceeded 40% of non-food household expenditures, a measure of degree of nancial risk to patients with cancer. Although the new round of health reforms have been implemented at the national level, it may take time to translate the investments into equitable health service use among people with different socio-economic status (SES) [7].
Previous research had reported the relationship between cancers and socio-economic characteristics, including China, showing that higher socio-economic status is associated with greater likelihood of routine screening, incidence of cancer, cancer treatment utilization and better survival. [8][9][10][11][12]. Other studies in China have examined socio-economic and gender inequalities in the incidence of cancer as well as mortality [17,18], and differences in treatment utilization [19]. Researchers also investigated socio-economic differences among patients with NCDs, such as hypertension, diabetes, stroke, chronic obstructive pulmonary disease as well as multiple chronic diseases [20][21][22][23][24][25]. However, there is limited evidence of the nancial burden of cancer care in China across socio-economic groups. [26] In addition, few studies have evaluated whether CHE varies by treatment types (eg. Chinese traditional or western medicine), or estimated the impact of the 2009 health reforms on health service utilization among cancer patients [13][14][15][16].
Our research is with the aim of: (1) investigating changes of the burden and socio-economic differences in CHE among Chinese adults with cancers; (2) assessing whether CHE varies by treatment type, and (3) examining relationships between different types of cancer therapy and healthcare service use as well as the incidence of CHE.

Data Source
In this study, data was obtained from the China Health and Retirement Longitudinal Study (CHARLS) in 2011 and 2015, and details of the study methodology have been published elsewhere [27]. The CHARLS is a biennial survey, which aimed to be representative of Chinese adults aged 45 years and above. CHARLS study design is similar to the Health and Retirement Study (HRS) and other established related ageing surveys [27]. Study questionnaires covered the domains of demographics, health status and functioning, healthcare and insurance, household income and consumption as well as clinical risk factors (blood pressures) [27].
To ensure ample representativeness at the national level, CHARLS sampled 150 counties and 450 villages/urban communities across 28 provinces, using multi-stage strati ed probability-proportionate-tosize (PPS) sampling. A total of 17,708 and 21,097 individuals were interviewed in 2011 (baseline) and 2015 (3rd wave). Final data were available for 17,224 participants in 2011 and 19,569 participants in 2015, after excluding participants with missing values of socio-demographic and/or cancer treatment information. A total of 234 individuals (in 2011) and 368 participants (in 2015) self-reported clinicallydiagnosed cancers.

Cancer Care Indicators
We identi ed six types of cancer treatments/therapies: Chinese traditional medication (TCM) only, western modern medication only (refers to taking western medication excluding TCM and chemotherapy medications for cancer treatment), TCM & western medication, surgery, chemotherapy and radiation therapy. The overall treatment was de ned as receipt of any TCM or Western medicine treatment (having one or more types of the ve therapies). In terms of health services utilization, this study included 1) outpatient care (participants were asked whether they had any outpatient visit during last month); 2) inpatient care (participants were asked whether they had any hospitalization and how many nights of hospital stay during last year). Medical expenditures were also collected in CHARLS, which included total health expenditure, reimbursed part of expenditure, and out-of-pocket spending for outpatient services in the past month and the inpatient care in the past year.
We apply the indicator of CHE to measure the nancial risk or economic burden for the household with family member suffering cancers. There were two common criteria of CHE de nition: 1) out-of-pocket payments (OOP) over 40% of the household's capacity to pay (using household's expenditure on non-food consumption as proxy), and 2) over 10% of total household income/expenditure [28][29][30]. Following previous studies, we de ned CHE as OOP for healthcare equaling or exceeding 40% of the household's capacity to pay [29]. The OOP health expenditure (household's capacity to pay) was considered as the numerator (denominator). CHE was coded as "yes" if the proportion over 40% and "no" if not.
The SES was assessed using the Socio-economic Index score, and calculated based on educational attainment, occupation and household consumption expenditure. This research used Li's scale for Chinese residents (version 2010) [31] as a standard scale. This scale, commonly used in social science research in China [32][33][34][35], was modi ed based on the scale rst proposed by Duncan [36]. Educational level, occupation and household consumption expenditure were classi ed to assign scores and summarized as a comprehensive Socio-economic Index score. Based on the Socio-economic Index scores, all subjects were classi ed as low (< 10 scores) and high SES class (≥ 10 scores).

Statistical Analysis
This study applied Chi-square tests to examine the SES differences in types of cancer treatment, outpatient and inpatient service use as well as incidence of catastrophic medical expenditure. For continuous variables including nights of hospitalization and OOPE of health care, we used t-tests to analyze SES group differences. Based on the pooled two-wave data of cancer patients, we performed multivariable logistic regressions to investigate the association of cancer treatment with outpatient visit, hospitalization as well as incidence of CHE, adjusting for socio-demographic factors Covariates in the regression analyses included gender, age, marital condition, education status, location residence, region in China and social health insurance enrolled.
To explore differential relationships across the wealth groups, we also conducted subgroup analyses with logistic regression models strati ed by the SES Index. The adjusted odds ratio (OR) and 95% con dence intervals (CI) were reported for logistic regression analyses in this study. The weighted prevalence of cancers were also reported considering nonresponse data and the complex, multistage design of CHARLS study. P values less than 0.05 were considered as statistically signi cant. Statistical analyses were conducted using STATA software (version 15.0; StataCorp LLC College Station, Texas, United States). Table 1 presents the socio-demographic characteristics of participants and cancer patients among Chinese people aged 45 years and above in 2011 and 2015. The prevalence of cancer increased from 1.36% (234 of 17,224) in 2011 to 1.88% (368 of 19,569) in 2015. The age-adjusted prevalence of cancer is 1.37% for 2011 and 1.84% for 2015. The prevalence of cancer was higher in people who: were female; had social health insurance; were located in the eastern region and unemployed individuals than those who: were male; without health insurance, lived in the western region and being employed Note: % (1), the unweighted prevalence of cancers; % (2), the weighted prevalence of cancers. The age-adjusted prevalence of cancer is 1.37% for 2011 and 1.84% for 2015. Table 2 shows the treatment\therapy condition among Chinese adults with cancers. Overall, approximately half of the cancer patients utilized treatment for their disease, with a higher proportion of urban residents (54%) than rural residents (46%) receiving cancer treatment in 2015. In addition, a higher proportion of high versus low SES patients utilized cancer treatment. Taking western medication and surgery were the two main type of treatment for cancers. People with a high SES level received more medication treatment, surgery and chemotherapy than low SES patients in China.    There are likely several factors contributing to the observed urban-rural differences in cancer care that needs to be addressed to increase the equitable access to healthcare in future. First, cancer care is provided as specialized services, and generally secondary or higher levels hospitals have the capacity to provide such care. Rural residents have less geographic access to cancer care comparing with urban areas, and some patients need extended travel and additional expenses to urban cities to get specialized cancer services [37]. During the period of the Coronavirus Disease 2019 (COVID-19) pandemic, patients may face more challenges in accessing cancer care. COVID-19 pandemic could disrupt the access to essential healthcare and likely exacerbate existing disparities since rural residents are more likely to be negatively by travel restrictions for access to specialized health services [38-40]. Public health efforts will be needed to address the impact of delayed diagnosis and suboptimal care in these rural populations to minimize adverse health outcomes among patients with cancer.

Results
Second, previous studies have documented the urban-rural disparity in availability of medical resources in China [41,42]. Research has generally found that residents living in rural areas would be less likely to use healthcare services with same quality due to the unbalanced economic development [43,44]. Consistent with other studies, people in rural areas are more likely to have a shortage of healthcare service providers and lack social supports [45][46][47]. Previous research suggested that the number of licensed doctors and nurses, medical-technical personnel as well as beds per 1000 population in urban communities increased more than the number in rural communities in China from 2005 to 2017. [48] Registered doctors per thousand populations in urban areas were 2.57 times more than in rural areas in China in 2015 [49]. For this reason, cancer patients in rural areas may prefer using the services from a nearby urban area, even at the cost of higher nancial burden.
Third, patients in the rural areas face potentially more nancial barrier to cancer care compared with patients in urban areas, likely due to more barriers in physical access to healthcare services, and high percapita payment for cancer care. The social medical insurances likely contributes to urban-rural disparity in CHE through gaps in coverage and bene t packages [50]. While over 95% Chinese population are covered by basic medical insurances, the bene ts packages can vary signi cantly across main social health insurances [5,6,23]. The per-capita fund of UEBMI is US$424.7, whereas only $61.2 for NCMS in 2015 [25]. The co-payment rates of NCMS (73.4%) was higher than UEBMI (36.8%) and URBMI (50.7%) in 2008 [51]. Herein, cancer patients in rural regions have lower reimbursement rates and lower annual maximum payment from insurance. Additionally, patients seek care from other county/prefectures usually have reduced reimbursement rates. In this case, if cancer patients from rural areas have to utilize cancer care from another city, they have to pay more proportion of their bill out-of-pocket. Such a heavy nancial burden likely compels low SES and rural patients away from cancer treatment utilization, possibly explaining the large urban-rural disparity in inpatient care use than outpatient visits. Paying more attention to out of pocket expenses and nancial implications of cancer treatment across health insurance schemes is need in future.
Finally, while the expenditures for cancer care are increasing in rural areas as well as urban areas, the uneven nature of economic development and low-level disposable income of rural residents may further contribute the urban-rural disparities in nancial risk and treatment burden. The disposable income of urban citizens was 31,195 RMB per capita, which was three times higher than that of residents living in rural areas (11,422 RMB) in China in 2015 [52]. Medical expenditures due to cancer treatment, speci cally surgery and chemotherapy, is likely to have signi cantly greater impact on rural patients, leading to increased risk of CHE and impoverishment [52]. In addition, since NCMS and URBMI were mainly nanced by local county-level governments at the early stage of China new health system reform, the quality of bene t packages likely depends on local economic development [49].
We observed that those cancer patients with high socio-economic level were less likely to experience CH than low SES individuals, and by 2015, rural cancer patients had almost double the prevalence of CHE compared with urban cancer patients, suggesting that a potential unintended consequence of the health reform is widening of rural-urban disparities in CHE. The ndings are consistent with previous research on trends in nancial burden among residents with NCDs in China [25,53]. Recent studies have documented a rapid rise in healthcare cost for cancer patients in China, however data on the population-level economic burden of cancers is limited and the reported expenditure per patient may be underestimated [54][55][56][57][58][59][60]. For example, a systematic review of the economic burden of liver cancer shows an increase in expenditure indicators (direct medical expenditure, annual expenditure per visit and annual expenditure per diem) from 1996 to 2015, with medication costs accounting for more than half of the overall expenditure (56. 6%) [57]. For colorectal cancer, the annual growth rate for medical expenditure per patient, per visit and per day increased from 6.9-7.8% from 1996 to 2015, respectively [58].

Policy Implications
The health insurance programs in China has had some positive impacts on healthcare utilization. For instance, we observed that outpatient visits and admissions in the past months increased between 2011 and 2015 in all socio-demographic groups examined. However, there are still challenges especially related to costs. Overall, the burden of cancer among Chinese adults is increasing and about one-fourth cancer patients experienced CHE. Yet disparities among urban-rural, and different SES still exist, even after the implementation of the national health insurance scheme.
To reduce nancial burden of cancer and bridge the SES gaps, comprehensive changes to health insurance bene t packages and healthcare resource allocation are needed to ensure universal, affordable and patient-centered health coverage. First, the URBMI and the NCMS need to be further integrated in contribution, bene t package, as well as nancial risk protection to accelerate the equitable access to health service in both urban and rural areas. Secondly, social health insurance bene t packages need to be further expanded. Health services, including medicines, with proved cost-effectiveness evidence can be added to the National Insurance Reimbursable List (NIRL). For instance, 17  should play an increasing role in providing nancial support on catastrophic expenses, including cancer patients. In particular, the current CIMI in most regions followed the NIRL which prioritize the essential care [63]. To provide better protection on catastrophic expenses, CIMI should explore additional coverage on other therapies with proven health bene ts. Furthermore, enhancing the capacity of the National Public Health Initiative can enhance cancer prevention strategies such as routine screening and case management. This might lead to early detection of cancer, reduced nancial burden and improved cancer outcomes if diagnosed at early stages when cancer treatments are most effective. This approach might be especially critical given healthcare disruptions due to the COVID-19 pandemic.
Given unprecedented effects of the COVID-19 on global healthcare systems, it would be inevitable that the pandemic would substantially negatively in uence the cancer patients, health practitioners, and healthcare systems [64,65]. Due to the COVID-19 outbreak, several healthcare services have to be postponed and even cancelled in many countries, leaving patients without access to essential health services [66]. Globally, 42% countries have partially or completely disrupted services for cancer treatment/therapy [67]. During the COVID-19 period, the waiting time for surgery and days in hospital was longer, and total expenditures for inpatient care increased in China [68]. To cancer patients, this means delayed diagnosis or treatment and further lead to disease progression. In many countries, teleconsultations became an emerging form to provide essential healthcare while limit the spread of COVID- 19 [69]. However, the capacity for providing telemedicine varied greatly, and there is limited coverage on telemedicine by China's social health insurances, which may potentially make further disparities in health service use. In the post COVID-19 era, how to balance the health system emergency reaction with essential healthcare, how to best utilize new technology to improve access and equity will be emerging health policy issues.

Strengths and Limitations
This research utilized data from a China nationally representative study to investigate the trends and disparity in cancer treatments, healthcare service utilization and CHE from 2011 to 2015. Our study could contribute to a deeper understanding of socio-economic and rural-urban disparities in cancer treatment, health service utilization and expenditures. There are several signi cant limitations. This study used the measurements based on self-reported information of diagnosed cancer, treatment type and healthcare service use, which may lead to underestimate the prevalence of cancer and an under-reported utilization due to the recall bias [57]. Medical information regarding the severity of cancers had not been collected. The CHARLS survey only considered those middle-aged and elderly Chinese residents. Future research should also focus on the younger adults to explore the trends and disparities in cancer care and nancial burden. Moreover, about 20% of total participants in CHARLS are with missing values of key dependent and independent variables during the survey conducted in 2011 and 2015.

Conclusions
The burden of cancer among Chinese adults is increasing. Socio-economic and urban-rural disparities in cancer treatment and health service use were largely determined by patient nancial capability in China.
The current social health insurance schemes are insu cient to address these disparities. A comprehensive health insurance policy of further expanding bene t packages and strengthening the Public Medical Assistance System, is essential for Chinese people with cancers.

Declarations
Ethics approval and consent to participate The Biomedical Ethics Review Committee of Peking University approved the CHARLS study, and all interviewees were required to provide informed consent. The ethical approval number was IRB00001052-11015.