Medical Debt and Catastrophic Health Expenditures among Middle- and Low-Income Families in China

Background Medical debt is a persistent global issue and a crucial and effective indicator of long-term family medical financial burden. This paper fills a research gap on the incidence and causes of medical debt in Chinese low and middle-income households. Method Data was obtained from the 2015 China Household Finance Survey and medical debt measured as borrowings from families, friends and third parties. Tobit regression models were used to analyze the data. The concentration index was employed to measure the extent of socioeconomic inequality in medical debt incidence. Results We found that 2.42% of middle-income families had a medical debt, averaging US$6278.25, or 0.56 times household yearly income; and 3.92% of low-income families had medical debts averaging US$5419.88, which was equivalent to 2.49 times household yearly income. The concentration index for low and middle-income families’ medical debt was significantly pro-poor. Medical debt impoverished about 10% of all non-poverty households and pushed poor households deeper into poverty. While catastrophic health expenditure was the single most important factor in medial debt, age, education, and health status of householder, hospitalization and types of medical insurance were also significant factors determining medical debt. Conclusions Using a narrow definition of medical debt, the incidence of medical debt Chinese and middle-income households was relatively low. But, once medical debt happened, it imposed a long-term financial burden on medical indebted families, tipping many low and middle-income households into poverty and imposing

on households several years of debt repayments. Further studies need to use broader definitions of medical debt to better assess the long-term financial impact of medical debt on Chinese families. Policy makers need to modify basic medical insurance schemes to manage out-of-pocket medical expenses better and take account of pre-existing medical debt when addressing this problem. background It is no secret that bad health and bad debts often coincide. A large and growing worldwide problem, medical debt imposes financial hardship on household budgets.
In the United States, the Kaiser Family Foundation reported that roughly a quarter of non-elderly American adults had difficulties paying their medical bills. [1] In many cases, health care costs substantially exceed patients' ability to pay for that care, even among families with adequate health insurance coverage. [2] A U.S. national survey found that about 29 million people had a recent or accrued medical debt [3], and such debts were a leading contributor to personal bankruptcy. Large medical bills impose an additional sizable burden on those who are already economically fragile, where financially stretched households, and households that are uninsured or underinsured, can be pushed into long-term personal debt, bankruptcy and poverty. [4] One survey found that 44% of American families with medical debt exhausted their savings to service outstanding medical bills, and many of them traded medical debt for other types of debt. [3] Medical debt can be more damaging than other types of consumer debt because medical bills are frequently incurred through illness or injury that limits one's ability to work. The work loss-medical debt problem confronts millions of middle-and low-income American families, where family savings are depleted paying medical debts, leaving families vulnerable to unanticipated economic shocks. [4] In Canada, the financial strain of medical costs arose because of medical expenditures and the loss of labor-related income due to illness. [5] The 2003 UK Commonwealth Fund Biennial Health Insurance Survey revealed that 37% of British adults had difficulty paying their medical bills, or had accrued medical debt, or both. [2] Among developing economies, medical expenses impose the greatest financial pressures on low-income families and people suffering prolonged poor health, especially where families have accumulated unpaid medical bills. [6] Medical debt hardship in Romania sprang from the social health burden that grew from 64.6% of the total health services costs in 1998 to 82.7% of costs in 2004. [7] Due to funding inadequacies, India's public health infrastructure suffers perennial shortages, impacting the rural and poorer population segments the most. As a consequence of these health infrastructure shortages, nearly 70% of Indian health spending was borne as household medical debt, frequently to private health care facilities, leaving many families impoverished [8]. A Vietnamese survey revealed that the poor generally delay treatment and paid more for each process of illness than the rich.
Further, poor Vietnamese families were compelled to cut consumption of essential goods or to borrow to meet their medical costs. [9] As in other developing countries, Cambodian households used a combination of savings, disposing of consumables and assets, and borrowed money to manage medical debt. [10] The vicious circle linking health and poverty and poverty and health remains a challenge in China. More than one in three Chinese households experienced difficulties in affording health care, or go without health care, or are impoverished by the health care costs. [11,12] Avoiding health care because of medical debts contribute to poor individual and national health outcomes. [10] A recent study found that medical debts of end-of-life Chinese cancer patients forced 83.92% of urban and 91.1% rural households below the poverty line. [13] After housing, business, education and cars, the 2012 China Household Finance Survey (CHFS) revealed that medical expenses were the next most important factor causing household debt. [14] Of course, the relationship between health debts and poverty is a global, not just a Chinese, phenomenon, [15] where out-of-pocket (OOP) expenses for public and private health care services drive many families into poverty and aggravate the poverty level of those who are already poor. China launched an ambitious national health reform plan in 2009 reaffirming the government's central role in financing essential health care services, which focused on expanding health insurance coverage, strengthening the primary care system, providing public health services, and conducting public hospital reform. [16,17] As part of the reform process, the Chinese government established basic medical insurance schemes, which aimed to protect all families and individuals, particularly poor households, from financially catastrophic unaffordable medical care and medical debt induced poverty. [ [20]. To protect families from heavy medical financial burden and impoverishment from medical expenses, health reform in China broadened the coverage of the medical insurance schemes and increased reimbursements, reducing OOP expenses.
While the total OOP medical expenses increased rapidly from RMB657.11 (US$95.2) billion in 2009 to RMB1513.36 (US$213.5) billion in 2017, the share of the OOP medical expenditure in total health expenditure decreased from 37.46% in 2009 to less than 30% in 2015, and remained stable to 2017 [21]. But, OOP medical expenses as a proportion of medical expenditure varied greatly across provinces, ranging from 16.29% in Beijing to 36.46% in Liaoning province, and the average cost of a single hospital admission could be equivalent to average person's annual income, and more than twice the average annual income of those in the lowest 20% of the income distribution. As a result, many Chinese families with basic medical insurance face health cost financial burdens. Did China's medical insurance schemes protect families from medical debt and poverty?
Surprisingly, few studies have explicitly examined medical service costs and medical insurance in relation to medical debt and poverty. Various Chinese studies on family finance classify medical debt as part of household consumption-fueled borrowing [22] or household debt [15,23] ignoring the particularities of medical Scheme was limited in reducing medical debt. [15] In a study from two poor Chinese counties, Jiang et al. (2019) found that the medical debt incidence was 60%-70% among low-and middle-income households and the debt was not decreased by the CII between 2013 and 2016. [24] Using data on the poorest families in 29 provinces in China, Jiang (2018) found more than 60% of families in poverty suffered from medical debt, no matter whether from urban or rural areas. [23] There is, however, a paucity of nationwide studies on medical debt of middle-and low-income Chinese families and medical debt induced poverty. To address these gaps in the Chinese

Dependent variable
There is no agreed definition of medical debt. Some American research, such as Batty et al (2018), referred to cases where a patient did not pay a bill for an unspecified period, but one long enough that the provider sends the medical debt to a collection agency. [25] Some research defined medical debt as unpaid medical bills. [26][27][28] Several studies measured medical debt by the response to the following type of survey question: "If you added up all medical bills, about how much would they amount to right now? Including unpaid balance(s), or medical bills that are outstanding." [29] Several studies have found these measures of outstanding medical bills to be highly reliable medical debt proxies. [4,30] Carlson et al (2006) defined medical debt as 'how much money they currently owed health care providers, credit cards, or other loan companies for medical expenses' [31] while Altice et al. (2017) used the depletion of assets to measure medical debt. [32] Some studies give medical debt a narrow definition, such as 'borrow money or going into debt because of disease, its treatment, or the lasting effects of that treatment. We take a narrow definition of medical debt, based on the response to the 2015 CHFS question: 'How much in total did your household borrow to pay medical debts?' Similar to other studies, the answer included bank and credit union loans and private loans from friends and family [15,33]. Since our definition of medical debt does not take into consideration the fall in family savings or selling assets to cover medical expenditures, it provides a lower bound estimate of the level and amount of medical debt and medical debt induced poverty.

Independent variables From the CHFS, our independent variables on each
household comprised all types of medical insurance, whether suffering noncommunicated diseases (NCD), hospitalization rates, per capita household income, food and non-food household expenditures, CHE, household assets, OOP medical expenditure, and the marriage, age, education, working and health status of the householder. Since children and students were only covered by URBMI, and families bringing up children are more likely to face CHE, this study analyzed the ability of family members of different age groups insured by basic medical insurance to protect against a long-term medical burden. Based on the 2015 CFHS question "How much was paid or reimbursed by your family's medical insurances?", OOP medical payments were calculated, which included all types of health-related medical expenses that were not reimbursed through any health insurance. [29] The 2015 CFHS also included a question on medical expenditure before any health insurance reimbursements: "What was the amount your family spent on healthcare last year, including medical treatment expenses?", which allows us to calculate the copayments and deductibles.
We define CHE as OOP healthcare payments comprising more than 40% of non-food household consumption, which reduces non-health household expenditure below the level required for necessities [34,35]. Per capita household income was calculated as the total family income divided by their household size. [34] Households in poverty were defined as when total household expenditure was smaller than household subsistence spending, where subsistence spending was defined as when the share of food expenditure in total household expenditure was at the 50th percentile [36]. Our household specific poverty measure varies from poverty measures based on a fixed poverty line for all households. Impoverishment households were defined as when household expenditure was higher than subsistence spending, but was lower than subsistence spending net of out-of-pocket health payments [36]. Impoverished households were households tipped into poverty by medical expenses. The concentration index (CI) was employed to measure the extent of socioeconomic inequality (per capita income) in medical debt incidence. [37,38] Per capita income, household expenditure and assets, OOP medical expenses and medical debt were converted to US dollars at the 2014 average exchange rate.
Household medical debt modeling Given the censored medical debt data, and that the influencing factors of medical debt of middle-income and low-income households may differ, a hierarchical Tobit model was selected. [39] We investigated the error distributions of the models to determine the level of heteroscedasticity, which was not a problem. The analysis was weighted using appropriate sampling weights to account for the complex sample design and selection. The level of significance was set at α = 0.05, with an alpha level of P <0 .05 considered statistically significant. Stata SE 15.1 was used to analyze the data. results Table 1 displays the indicators of medical financial burden and medical debt of middle-and low-income families. The medical debt incidence of middle-income families was 2.42% of all middle-income families and 3.92% of low-income families, but only 0.93% for high-income families. The average medical debt of middleincome families was US$6278.25 and US$5419.88 among low-income families, with the average medical debt of middle-income households 15.8% higher than that of low-income households. The median medical debt of low-income households was roughly 2.5 years their net household income and about 6 months' median yearly income for medium-income households. Moreover, the CI in Table 1 for low and middle-income families' medical debt was significantly pro-poor. The geographical distribution of the incidence of medical debt in Figure 1 shows that the incidence of medical debt was higher in northwest provinces (Heilongjiang, Inner Mongolia, Gansu, Qinghai, Yunnan) and Henan Province, and lower in eastern coastal provinces (Beijing, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian). The higher medical debt incidence in the poorer more economically backward provinces, was consistent with the CI results in Table 1.

Figure1 Geographical Distribution of Medical Debt Incidence in 2015
The mean OOP medical payments of low-income households were US$886.78 and US$925.71 for middle-income households. We found that 51.27% of low-income families and 33.90% of middle-income families with OOP payment faced CHE, with the share of OOP medical expenses in non-food expenditures of low-income families higher than that of middle-income families. Therefore, it was not surprising that We found that 9470, or 38.08% of middle-and low-income households fell below the poverty line because their total household expenditure was smaller than their food subsistence spending. As shown in  We calculated that 1534 (or 9.96%) non-poverty household were impoverished by CHE, when these households were tipped into poverty after incurring OOP expenditures for health services. As Table 3 shows, among these impoverished families, the incidence of CHE reached 79.01%, and the incidence of medical debt was 11.08%, which was significantly higher than non-poverty households (P < 0.001), where CHE incidence was only 16.76% and medical debt was 2.05%.
Impoverished households and non-poverty households had similar amounts of medical debt, but the former had much higher OOP medical expenditure.   Our analysis found that the medical debt incidence of middle-income families was only 2.42% of all middle-income families and 3.92% for low-income families.
Second, once medical debt occurs, medical debt brings serious financial and longterm negative impacts on families with medical debt representing 6-month household income for middle-income families and 2.5 years' household income for low-income families.
In comparison with medical debt studies in America [1], Britain [2], India [8] and Cambodia [10], medical debt incidence of Chinese households was low. We took a narrow measure of medical debt, defined as debt incurred by outside borrowing for medical expenses. [15,33] Using household savings and disposal of assets to pay for medical expenses were not included, resulting in a lower bound estimate of the incidence of medical debt. With the largest per capita savings rate in the world, Chinese families' frugal habits would have seen them dispose of their assets or use their accumulated savings to pay medical expenses before seeking external borrowed funds. We can get some measure of households selling assets to meet medical debt by comparing the household assets of debt and non-debt households.
The average household assets of medical debt households was US$34157.60 compared to US$86998.18 for non-medical debt households, which suggests assets were sold before borrowing to pay off medical debt. These non-borrowing debt payment strategies should be categorized into a broader definition of medical debt.
Since data on non-borrowing for medical debt strategies were not covered in existing CFHS surveys, new surveys need to be undertaken to measure these alternative strategies, especially using accumulated saving and assets dispersal for covering medical debt. Also, the Chinese government has a scheme of Medical Financial Assistance and CII, providing part medical financial support for impoverished households due to illness, which would reduce the medical debt of low-income families to a certain extent. These schemes should be promoted, expanded and made accessible to a large share of Chinese households.
Not surprisingly, medical debt was one of the most common debts for middle-and low-income families, which is similar to finding from other Chinese and international studies. Although the overall incidence of medical debt in China is low, once medical debt happens, it had a long-term negative impact on the financial sustainable development of middle-and low-income families. This contrasts with an American study that found most medical debts were relatively modest in size [25] with more than 50% of them less than $600 annually. [40] According to other US surveys, the average debt load per person was about half of their annual reported income [41,42]. However, low-income families in China required several years' annual income to repay their medical debts, and even medium-income households medical debt accounted for roughly half their yearly income. Given the need to use part of annual household income for day-to-day living expenses, medical debts took many years to be repaid, indebting households for years as medical cost repayments were made.
Therefore, further studies need to investigate medical debt as a powerful long-term financial burden on families and to expand the definition of medical debt from borrowings to broader definitions of medical debt. We suggest that medical insurance and other income relief funds should identify existing medical debts when compensating patients for new medical expenses. By providing appropriate financial support to families with medical debts, access and equity in health will be improved and families further relieved from the economic burden of disease caused debt. Awareness of existing medical debt and household health-related impoverishment will also help to minimize the likelihood that policy reforms create unintended consequences on these vulnerable groups. Policymakers should define low-income households as a sub-population "at risk" of financial catastrophe from CHE and medical debt and implement insurance scheme support.
Further, "at risk" families require safety nets to increase access to healthcare without incurring additional medical debt. [44] This study has a number of limitations. We used a narrow definition of medical debt, which did not include using savings and disposal of assets to pay-off medical expenditures, leading to a lower bound estimate of medical debt. A study of households in western China found households in these provinces were likely to incur medical debt, but the CHFS did not cover two western provinces, Tibet and

Consent for publication
All the participants gave their consent to publications (signed form).

Availability of data and materials
The data that support the findings of this study are available from the Survey and Research Center for China Household Finance, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the Survey and Research Center for China Household Finance on the website http://chfs.swufe.edu.cn/datas/.

Conflict of interest
The authors declare that they have no competing interests.