Provincial variations in catastrophic health expenditure and medical impoverishment in China: A nationwide population-based study
Introduction
Transforming our world: the 2030 agenda for sustainable development [internet]. New York: Seventieth United Nations General Assembly.
Many countries have incorporated UHC objectives into their national health policies and reforms.
Universal health coverage: moving towards better health: action framework for the Western Pacific Region [internet]. Manila: WHO Regional Office for the Western Pacific.
,
Tracking universal health coverage: 2021 global monitoring report [internet]. Washington, DC: World Bank.
The aim of UHC is to ensure that all citizens have access to essential healthcare services without incurring financial hardship. Therefore, alongside improvements in healthcare service coverage, financial protection is a key component of UHC that many countries have made great efforts to strengthen. It refers to how far people are protected from the financial consequences of illness. A target of 100% coverage of financial protection by 2030 was set by the WHO and World Bank.
Tracking universal health coverage: first global monitoring report [internet]. Geneva: World Health Organization.
Catastrophic health expenditure (CHE) and impoverishment due to out-of-pocket health payments are two commonly used indicators to measure financial protection in health.
Tracking universal health coverage: 2021 global monitoring report [internet]. Washington, DC: World Bank.
,
- Wagstaff A.
- van Doorslaer E.
- Yip W.C.M.
- Hsiao W.C.
- Chen W.
- Hu S.
- Ma J.
- Maynard A.
This reform can be broadly classified into two phases: the first phase (2009–2011) emphasized expansion of social health insurance coverage, and the second phase (2012 onwards) prioritized reforming its hospital-centric and fragmented healthcare delivery system.
- Yip W.
- Fu H.
- Chen A.T.
- et al.
To achieve these goals, the Chinese government has made great investments in healthcare, with an increase in government health expenditure (GHE) as a share of total health expenditure (THE) from 24.7% in 2008 to 30.4% in 2020. The share of GHE in total government expenditures also increased from 4.4% to 7.8% during the same period. As a result, China experienced significant reductions in the incidences of CHE and medical impoverishment (MI) between 2010 and 2016,
marking China’s laudable progress towards UHC objectives.
- Fan S.
- Kanbur R.
- Zhang X.
In addition, achieving healthcare equity is an important goal imbedded in the implementation of UHC. Several previous studies have raised concerns with respect to large subnational variations that may be detrimental to progress towards UHC.
- Pulok M.H.
- Uddin J.
- Enemark U.
- Hossin M.Z.
,
- Zhou M.
- Wang H.
- Zhu J.
- et al.
,
- Liang J.
- Li X.
- Kang C.
- et al.
Consequently, some studies have pointed out the importance of regional analysis at the subnational level for framing healthcare policies.
- Han S.M.
- Rahman M.M.
- Rahman M.S.
- et al.
,
- Pandey A.
- Kumar G.A.
- Dandona R.
- Dandona L.
,
- Mohanty S.K.
- Kim R.
- Khan P.K.
- Subramanian S.V.
Although some studies have examined the overall socioeconomic disparities in financial protection in China,
,
- Meng Q.
- Xu L.
- Zhang Y.
- et al.
few have investigated disparities in financial protection at the province level except two previous studies. One study used nationally representative survey data to explore provincial variations in catastrophic health expenditures in 2003 when China’s health system reform was in its early stage,
- Liu Y.
- Rao K.
- Wu J.
- Gakidou E.
and the other one investigated variation across only five provinces in 2014.
An updated investigation into provincial differences in financial protection is of great importance because it will provide more detailed information to researchers and policymakers, and help them to better understand and monitor China’s progress towards UHC. Province-specific results on financial protection will also enable policymakers to form localized healthcare strategies and improve the efficiency of public investments in health.
Using data from the 2017 China Household Finance Survey (CHFS), which is a nationally and provincially representative survey, this study aimed to estimate the provincial variations in the incidence and intensity of CHE and MI. In addition, this study investigated the provincial variations in the urban-rural gap and income-related inequality across provinces. Findings from this study may have implications for setting future healthcare policy priorities in China, and may provide lessons for other countries in the Western-Pacific region.
Methods
Data sources
,
,
For example, the 2017 CHFS wave covered 356 counties (including districts and county-level cities), 1428 urban and rural communities, and 40,011 households, and was much larger than the 2011 wave in sample size. Specifically, a province’s counties, ranked by per capita GDP, have been dealt with systematic sampling and weighted by population. This sampling design has guaranteed its representativeness at provincial level. The representativeness at the provincial level, which was not common in other Chinese household surveys, allowed us to investigate province-level differences in financial risk protection across China.
Indicators
- Wagstaff A.
- van Doorslaer E.
,
- Saksena P.
- Hsu J.
- Evans D.B.
We calculated these indicators using annual expenditure data from the CHFS. Details of the calculation of each indicator are shown in supplementary notes in Appendix A.
,
- Meng Q.
- Xu L.
- Zhang Y.
- et al.
,
- van Doorslaer E.
- O’Donnell O.
- Rannan-Eliya R.P.
- et al.
,
- Xu K.
- Evans D.B.
- Kawabata K.
- Zeramdini R.
- Klavus J.
- Murray C.J.
We used “health expenditure was at least 40% of the total household expenditure net of food consumption in a year” as our main indicator so that we could produce comparable results. In addition, as earlier studies have found some variability across different calculation methods,
- Cylus J.
- Thomson S.
- Evetovits T.
we used the two thresholds to explore the robustness of our results: 10% and 25% of total household consumption. These methods and corresponding thresholds have been widely used in the UHC monitoring report
Global monitoring report on financial protection in health 2021 [Internet]. Washington, DC: World Bank.
and related literature.
,
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
The intensity of CHE is also called the catastrophic payment overshoot. It was measured as the average gap of CHE, which is the proportion of healthcare expenditures in total expenditures exceeding the CHE threshold.
- O’Donnell O.
- van Doorslaer E.
- Wagstaff A.
- Lindelow M.
We also used the thresholds of 40%, 25%, and 10% when calculating the intensity of CHE.
- Wagstaff A.
- Flores G.
- Smitz M.F.
- Hsu J.
- Chepynoga K.
- Eozenou P.
Using the poverty line set by the World Bank (1.9 USD per person-day in 2010) and the purchasing power parity between China and the United States (3.308 RMB/USD in 2010), we defined the poverty line in 2010 as 2300 RMB per person-year. This poverty line is consistent with the national poverty line set by the Chinese government in its anti-poverty campaign. We adjusted this poverty line using the Consumer Price Index (CPI) and obtained a threshold for the survey year 2017, equaling 2735 RMB per person-year. The intensity of MI was measured as the poverty gap, referring to the change under the poverty line due to out-of-pocket health expenditures. When a household was impoverished by out-of-pocket healthcare expenditures, the change in the poverty gap was the amount by which out-of-pocket healthcare expenditures pushed the household below the poverty line. When a household was below the poverty line, the change in the poverty gap equaled the household’s out-of-pocket health expenditures. We used the poverty gap divided by the poverty line to obtain a normalized poverty gap.
Statistical analysis
- Gan L.
- Yin Z.
- Jia N.
- Xu S.
- Ma S.
- Zheng L.
we performed statistical analyses using the sampling weights of the households to ensure that our results were nationally and provincially representative. We presented variable means with 95% confidence intervals (CIs). Given the large differences in social and economic development between urban and rural areas in China, we also stratified data by urban/rural residence status to examine whether variations in financial protection across provinces differ by urban and rural areas.
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
health care expenditure as a share of GDP,
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
,
- Xu K.
- Evans D.B.
- Carrin G.
- Aguilar-Rivera A.M.
- Musgrove P.
- Evans T.
health financing structure such as the share of government health expenditure in total health expenditure and prepayment mechanism,
- Xu K.
- Evans D.B.
- Kawabata K.
- Zeramdini R.
- Klavus J.
- Murray C.J.
,
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
,
- Xu K.
- Evans D.B.
- Carrin G.
- Aguilar-Rivera A.M.
- Musgrove P.
- Evans T.
health system capability,
- Xu K.
- Evans D.B.
- Kawabata K.
- Zeramdini R.
- Klavus J.
- Murray C.J.
and population characteristics such as age structure and population health.
- Xu K.
- Evans D.B.
- Carrin G.
- Aguilar-Rivera A.M.
- Musgrove P.
- Evans T.
,
,
- Zhao Y.
- Atun R.
- Oldenburg B.
- et al.
Based on these findings, we developed a theoretical framework to guide our selection of influencing factors and summarized variables into dimensions of population characteristics, economic development, and health system characteristics. Appendix Fig. 2 shows the detailed theoretical framework.
(1)
where the subscript
indicates a province.
represents the dependent variables (i.e., incidence of CHE, incidence of MI).
indicates disposable income per capita, and
indicates two variables of population characteristics, including Disability-Adjusted Life Year (DALY) rates and the proportion of the population aged 65 years and above.
is a vector of variables measuring health system characteristics at the province level, including the share of total healthcare expenditures in GDP, the share of public financing in total healthcare expenditure, the basic health insurance coverage, and the density of physicians and hospital beds. All these independent variables were taken from the China statistics yearbook 2018, except for the proportion of the aging population, which was retrieved from the Seventh National Population Census of China, and the DALY rates which were collected from a Global Burden of Diseases study in China.
- Zhou M.
- Wang H.
- Zeng X.
- et al.
refers to the robust error term.
- O’Donnell O.
- van Doorslaer E.
- Wagstaff A.
- Lindelow M.
Following previous literature,
- Kjellsson G.
- Gerdtham U.G.
,
- Erreygers G.
- Van Ourti T.
the concentration index was defined as follows:
where
is the dichotomous indicator (i.e., CHE and MI),
is its mean, and
is the fractional rank in the living standards distribution, which is estimated using household annual income per capita in this study. Because the concentration index is derived from the Gini coefficient of income inequalities, it requires the variables of interest to be on a ratio-scaled measure without an upper bound. As the incidence indicators were bounded, ranging from 0 to 1, Erreygers’s weighting method was applied to deal with the concentration index of bounded variables.
- Erreygers G.
- Van Ourti T.
We used the same method to calculate concentration indexes for CHE and MI intensities.
- O’Donnell O.
- van Doorslaer E.
- Wagstaff A.
- Lindelow M.
,
- Koolman X.
- van Doorslaer E.
:
which holds approximately for large samples. The value of
suggests the percentage of the CHE and MI indicators that would need to be linearly redistributed from the richer half to the poorer half of the population, in which case that health inequality favors the rich, to arrive at a distribution with an index value of zero. Sampling weights were applied throughout the analysis.
Role of the funding source
This study is financially supported by National Natural Science Foundation of China (Grant Number: 72074049) and the Shanghai Pujiang Program (2020PJC013). The funding sources did not play any role in the study design, data analysis, data interpretation, or in the writing of the paper, or the decision to publish.
Discussion
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
,
- Wagstaff A.
- Flores G.
- Smitz M.F.
- Hsu J.
- Chepynoga K.
- Eozenou P.
The estimated incidences of CHE and MI in this study are also consistent with findings from previous studies in China. A previous study using data from the China Family Panel Study applied the same calculation methods as in our study and reported that the incidences of CHE under the 40% threshold and MI were 10.73% and 3.19%, respectively, in 2016.
Another study using data from the China Health and Nutrition Survey applied total net household income as the denominator, and showed that the incidence of CHE using the 40% threshold was around 9% in 2015.
- Xu Y.
- Zhou Y.
- Pramono A.
- Liu Y.
- Jia C.
All these estimates suggest large margins for improvements in financial protection in China, although China has achieved laudable progress during the past 20 years. More importantly, we found large variations in financial risk protection within the nation. Eastern provinces, such as Jiangsu and Zhejiang, had much lower incidences and intensities of CHE and MI, while the populations in central and western provinces were more likely to suffer financial hardship due to out-of-pocket health expenditures. For example, the incidences of CHE using the 40% threshold in Qinghai and Heilongjiang provinces were more than twice as large as that in Beijing. These findings imply that the provincial variation in financial protection is significant, and an overall figure for CHE or MI may not be sufficient to frame specific interventions.
- Wagstaff A.
- Flores G.
- Hsu J.
- et al.
,
- Wagstaff A.
- Flores G.
- Smitz M.F.
- Hsu J.
- Chepynoga K.
- Eozenou P.
China’s health system has been long criticized for its fragmented financing system because of decentralization in the governance structure.
- Meng Q.
- Mills A.
- Wang L.
- Han Q.
Increased fiscal investments from the central government may contribute to narrowing the disparities in financial protection.
,
- Meng Q.
- Xu L.
- Zhang Y.
- et al.
We also found that the urban-rural gap in financial protection within a province was associated with economic development. Compared with provinces in the central and western regions, provinces in Eastern China had much a smaller urban-rural gap in financial protection within them. Moreover, we found large variations in incidence and intensity of CHE and MI among rural households across provinces, while the provincial disparity in these indicators among urban households was much smaller. This finding suggests that the larger urban-rural gap in central and western provinces was largely attributable to their much higher incidence and intensity of CHE and MI among rural residents.
This study also investigated the provincial variation of income-related inequality in financial protection. We found that there was a substantial pro-rich inequality in all CHE and MI indicators. More importantly, we found that pro-rich inequality in all CHE and MI indicators was more pronounced in central and western provinces such as Qinghai, Anhui, and Henan. This finding was consistent with a much larger urban-rural gap in central and western provinces. Given that central and western provinces had much higher incidence and intensity of CHE and MI, the higher inequality and urban-rural gap in these provinces suggest that policymakers should pay special attention to poor households in central and western provinces, and that provision of better financial protection for these vulnerable groups is key to achieving UHC in China.
- Meng Q.
- Xu L.
- Zhang Y.
- et al.
,
Given the contextual complexities within China, the investigation of financial protection at the province level would provide more detailed information to policymakers and could contribute to developing province-specific strategies. The large variations in the extent of financial protection and its inequality in China suggest that a one-size-fits-all approach would not be sufficient to provide the needed insight into addressing challenges in achieving UHC. A subnational analysis may be crucial to address the diverse challenges faced by provincial governments, and province-specific healthcare policies will be essential for an overall success at the national level. We recommend that China, as well as other Western-Pacific countries, should monitor subnational trends in financial protection in addition to the nationwide data. Thus, we call for greater investment in data collection, and more detailed studies on factors that explain the subnational variations within a country. We believe that results of these studies would better guide health policy making in Western-Pacific countries, especially in countries with large variations in social and economic developments such as Indonesia and Vietnam.
- Zang J.
- Luo B.
- Wang Y.
- et al.
,
- Adams J.
- Goffe L.
- Brown T.
- et al.
,
- Lund T.B.
- Kjærnes U.
- Holm L.
the incidence of CHE in developed provinces may be overestimated more compared to underdeveloped provinces. Therefore, the provincial variation in CHE incidence may be underestimated. Even with this potential downward bias, we still derive a substantial provincial variation. In addition, we find similar results in provincial variations in CHE incidence when using different calculation methods. All these results suggest that inclusion of eating out consumption may not be a major threat for our conclusion. Fourth, when estimating financial hardship, we only considered households incurring healthcare expenditure, and this study fails to include the situation when poor patients do not have a chance to utilize healthcare services. In addition, due to data limitations, we did not include other non-medical costs related to healthcare utilization, such as transportation and accommodation expenses. A previous study showed that these non-medical costs accounted for approximately 18% of the total inpatient costs, and that the proportion was highest for those in the lowest wealth groups.
These are important issues to be addressed in future research.
Conclusion
Our study indicates considerable room for improvements in financial protection in China, despite the laudable achievements in the past 20 years. More importantly, we found the existence of substantial variations in the incidence and intensity of CHE and MI across provinces. Developed provinces in Eastern China have much lower incidence and intensity of CHE and MI compared with central and western provinces. Also, these provinces in general have smaller income-related inequality and urban-rural gaps in financial protection within them. These findings imply that poor households in central and western provinces should be the key targeted population, and that provision of better financial protection for these vulnerable groups is needed. Province-specific healthcare policies and increased public health spending in these provinces should be put in place to reduce between-province and within-province inequalities and they are crucial for the nationwide success in achieving UHC in China.
Ethics committee approval
The study (IRB approval number: IRB#2022-04-0962) was ethically approved by Institutional Review Board of Fudan University School of Public Health, which was registered with the Office for Human Research Protections, IRB00002408, and has a Federal wide Assurance, FWA00002399.