Cancer is one of the leading causes of death in the world.1 From 2012 to 2020, the number of new cases of cancer increased by 5.3 million while deaths from cancer increased by 1.8 million.2,3 Low- and middle-income countries are most affected by cancer.2 China has the largest burden of cancer, accounting for more than 23% of new cancer cases and about 30% of cancer deaths worldwide.4,5

Cancer treatment imposes a substantial financial burden on the patients.6 The longer the patients survive, the more likely their households would incur catastrophic health expenditure (CHE).7 In China and Southeast Asia, almost half (48%) of cancer patients experienced CHE events.8,9 Towards the end of life, almost all cancer patients have experienced CHE in China.10

Health insurance has the potential of preventing CHE through subsidizing incurred medical expenditure.11 However, the functions of health insurance programs often vary by cultural contexts, modes of execution, and social institutions in which those programs are embedded.12 If it is not designed and managed properly, consumers can be lured into the trap of CHE if out-of-pocket payments (OOP) are required to contain total health expenditure.13 Existing studies into the role of social health insurance programs in China have generated inconsistent conclusions. Some studies argue that patients with health insurance tend to spend more, increasing the likelihood of CHE; whereas, others argue that patients are likely to receive the same services with lowered OOP expenditure.14–16 Great regional disparities have persisted in China despite the great efforts made by the central government for universal coverage (>95%) of social health insurance.17 The social health insurance system of China is characterized by three major schemes with numerous funds managed by local governments: the basic medical insurance for urban employees (BMIUE) launched in 1998, whose premiums are raised based on a percentage of monthly income and the minimum funding required by the policy is 8%; the rural new cooperative medical scheme (RNCMS) for rural residents launched in 2003; and the basic medical insurance for urban residents (BMIUR) such as students, unemployed, and self-employed launched in 2007, both of which are funded by a annual per capita, different standards could be set in different regions according to different economic levels and can grow year by year.14

Empirical evidence shows that the OOP payments as a proportion of total health expenditure has declined significantly as a result of the expansion of social health insurance in China.18 However, total health expenditure escalated rapidly, increasing from 1.8 trillion in 2009 to 7.2 trillion yuan in 2020,19 which seriously jeopardized the role of social health insurance in alleviating the OOP burden of the patients. Inequalities within and across different insurance schemes and funds are also profound,20 with variations in entitlements depending on the financial capacity of local funds. BMIUE tends to have higher financial capacity than the other two schemes despite local governmental subsidies. Overall, social health insurance slowed down the increase of CHE incidence, but to a less extent in the RNCMS enrollees.21 The Chinese government has introduced multiple strategies to address the inequality concerns. These include supplementary insurance for catastrophic diseases and medical assistance for the households living in poverty. Recently, merge of the RNCMS and the BMIUR programs started.22 The overall effect of these policy initiatives remains unknown.

This study aimed to address the gap in the literature by estimating the effect of social health insurance programs on CHE events experienced by cancer patients. CHE is a highly sensitive indicator for measuring financial burden of consumers, as well as the risk protection effect of health insurance.23

Most of the cancer patients were older than 50 years (73.76%) and did not attend tertiary education (78.97%); 47.75% were BMIUE enrollees. Lung, breast, colorectum, esophageal, liver, and stomach cancers accounted for 70% of the total cancer cases.

About 72% of the households experienced CHE. The cancer patients who were aged between 50 and 69 years (74.7%; p<0.003), received lower levels of education (p <0.001), worked as a peasant (85.5%; p<0.001), and enrolled with the RNCMS (81.9%; p<0.001) were more likely to have CHE. CHE incidence decreased with income: 88.2% in the households with the lowest quintile of income, compared with 54.1% in those with the highest quintile of income (p<0.001). The patients with esophagus cancer had the highest incidence (83.7%, p<0.001) of CHE (Table 1).

On average, each household with CHE (n=1824) accrued a mean positive overshoot equivalent to 28.3% daily household consumption. The same risk factors were identified for catastrophic mean positive overshoot as those for CHE incidence. Higher mean positive overshoot was also found in the central developing region compared to others (Table 1).

The BMIUE enrollees were more likely to have the highest quintile of income. By contrast, the RNCMS enrollees were more likely to have the lowest quintile of income (p<0.001) (Table 2).

The marginal contribution of health insurance in alleviating CHE varied by household income and across insurance schemes. Those with higher income benefited more from insurance, with the top income group experiencing a 18.75-29.01% drop in CHE incidence, more than doubled that (8.33-11.36%) of the bottom income group, also the same general trend in CHE intensity. Higher pre-post compensation differences in CHE incidence and intensity were found in the BMIUE enrollees, respectively 24.13% and 16.36%, compared with their RNCMS counterparts with 10.64% and 9.56% (Table 3).

The logistic regression model confirmed that patient age, occupation, household income and site of cancer were significant predictors of CHE incidence. Those aged between 50 and 69 years had higher odds (AOR=1.340; p=0.011) of CHE than their younger counterparts. Peasants had higher odds (AOR=1.944; p=0.002) of CHE than those working in the public sector. Higher household income was associated with lower odds of CHE (p<0.001). Patients with breast cancer had lower odds (AOR=0.631; p=0.002) of CHE than those with lung cancer. The differences in CHE incidence across different insurance programs became statistically insignificant after adjustment for variations in other variables (Table 4). No significant multicollinearity was identified in the multivariate modelling (see Supplementary Table S1 in online Appendix).

About 12.7% (n=322) respondents reported incomplete cancer treatment. Although the RNCMS enrollees were less likely to experience incomplete treatment than their BMIUE counterparts, 42.7% RNCMS enrollees identified financial hardship as a major reason compared with 13.2% in the BMIUE enrollees (p<0.001). Higher levels of financial hardship in the RNCMS enrollees was further confirmed by self-rating: 61.98% RNCMS enrollees felt financial burden of cancer treatment “stressful” and 14.2% “unaffordable”, compared with 49.55% and 12.84%, respectively, in the BMIUR enrollees, and 42.48% and 4.88%, respectively, in the BMIUE enrollees (p<0.001) (Table 5).

This study revealed high levels of CHE incidence in the households with cancer patients and a limited role of social health insurance in alleviating the CHE events in China. Over 71% cancer patients experienced CHE events, with 28.3% catastrophic mean positive overshoot. About 60% reported financial burden of cancer treatment as “stressful” or “unaffordable”. More than 12.7% did not complete their cancer treatments, with 21.1% reporting financial hardship as the major reason.

The CHE incidence revealed in this study is higher compared with the findings of other studies using the same threshold. In Vietnam, cancer treatment resulted in 64.7% of CHE between 2012 and 2014.26 The average CHE incidence of cancer treatment in the Association of Southeast Asian Nations countries was around 48% in 2012.8 A 2011 study in China's Liaoning province found 42.8% CHE incidence in cancer patients.9 The sample of this study were all patients diagnosed with cancer and paid for cancer treatment in tertiary hospitals with higher technology and higher cost, not general disease population or general population, which mainly explain the higher level of CHE incidence. In addition to expensive technologies, those hospitals can also charge more for the same interventions.

The CHE intensity associated with cancer treatment was found to be 28.3% as measured by mean positive overshoot, which also appears to be higher in comparison with other studies using the same threshold. The 2011 Liaoning study in China reported 24.3% catastrophic mean positive overshoot for cancer treatment in the households with CHE.9 In Bangladesh, the average catastrophic overshoot was reported to be as low as 1.9%.23

We alarmingly found that the households with higher income benefited more from insurance. The households in the top income group experienced the highest level (18.8-29.0%) of drop in CHE incidence, more than doubled that (8.3-11.4%) of those in the bottom income group, which is consistent with the results of other studies.23 This may be due to the fact that richer people could afford medical services and were willing to pay large medical bills compared to poorer people. Cost containment is critical for the sustainable development of health insurance. OOP requirements are often considered as one of the most effective measures to deter over-use of healthcare services and contain the rapid growth of health expenditure. Unfortunately, this can lead to serious financial pressure on the patients. In China, OOP as a proportion of total health expenditure dropped from 59.97% in 2001 to 28.36% in 2019.19,27 However, such a level of OOP is still deemed high using the benchmark of some low- and middle-income countries such as Cuba, Romania and Thailand, let alone in comparison with the high income countries.19

According to the World Health Organization, health insurance should promote equal access to health care and provide financial protection for the most vulnerable.28–30 However, differences of impact on CHE still existed among different types of health insurance, although it was found to be statistically significant. The RNCMS is the insurance category with the lowest compensation level for both richer and poorer individuals. In 2014, insurance reimbursement for hospital inpatient services only reached 56% for RNCMS enrollees, compared with 73% for BMIUE enrollees.31 This may be due to the relatively low level of funding. The available RNCMS funds equal less than 5% of the net income of rural residents. By contrast, the available BMIUE funds have reached at least 8% of the salary of urban employees.32,33 Since the RNCMS enrollees are more likely to live with lower income compared with their BMIUE counterparts, more favorable policies are needed to enhance the ability of the RNCMS in providing financial protections. With the integration of BMIUR and RNCMS as well as the continuously increasing of the financing and compensation ratio of medical insurance in recent years, the medical insurance will be more and more effective.17,34

There are several limitations in this study. Self-report data were used in this study, which may contain recall bias. The data collection strategy was designed for ensuring data consistency, but it may lead to challenges for the earlier diagnosed patients to recall data accurately (>2 years). However, patients usually have an accurate record about the cancer treatment expenses as they are a large expense (covering multiple providers in and outside of hospitals) that would usually need support/subsidize from others such as insurance funds and family members. The CHE events might be underestimated as we did not consider the source of income for household consumption. Some households experiencing financial difficulties might borrow money to pay for daily consumption. The nature of cross-sectional design prevented us from making causal conclusions. This study did not intend to draw a nationally representative sample. Instead, study participants were evenly distributed despite variations in population size across provinces for the purpose of estimating CHE incidence and intensity.

Cancer treatment is associated with high incidence of CHE events in China. The households with lower income are more likely to suffer from CHE. The alleviating effect of social health insurance on CHE events is limited. The health insurance programs have failed to address income-related inequality problems in CHE resulting from cancer treatment. Given that supplementary insurance for catastrophic diseases and medical assistance programs have already been tried, China may need to consider some pro-poor insurance compensation policies, such as higher reimbursement rate for patients with lower income. Under a system that has yet to achieve universal health coverage, targeted interventions (such as those on the older people and rural residents) may be more effective and efficient for improving health equity. Further studies are also needed to better understand the underlying reasons of CHE inequality, including the role of the site of cancer.

David Cantarero.

The corresponding author on behalf of the other authors guarantee the accuracy, transparency and honesty of the data and information contained in the study, that no relevant information has been omitted and that all discrepancies between authors have been adequately resolved and described.

Conceptualization and design: G. Liu, J. Shi, W. Fu, W. Chen and J. He. Data acquisition and analysis: W. Fu and C. Liu. Drafting of the manuscript: W. Fu and C. Liu. Overall supervision: G. Liu. All authors read and approved the final manuscript.

The authors appreciate support from the National Cancer Center of China, the seven participating provinces, the Health Economic Evaluation Working Group, and the Cancer Screening Program in Urban China. We thank all of the study participants.

This work was supported by National Key R&D Program of China grant number (2017YFC1308700, 2017YFC1308705), National Natural Science Foundation of China grant number (71603065,71673071), and the National Key Public Health Program of China (Cancer Screening Program in Urban China) (Grant number: N/A).

None.