Does the critical illness insurance reduce patients’ financial burden and benefit the poor more: a comprehensive evaluation in rural area of China

Abstract

Background: Critical illness insurance (CII) is one kind of health insurance that is gradually gaining attention worldwide. China implemented CII in 2012 to decrease patients’ out-of-pocket (OOP) medical payments. The aims of this study were to determine if the project had positive impacts on relieving financial burden and improving health equity.

Methods: A series of questionnaire surveys were undertaken in two counties before and after the intervention in rural China. OOP expenditure, catastrophic Health Expenditure (CHE) incidence, and associated average gap (AG) were assessed across different income groups and project durations, measuring short-term direct medical cost. Medical debt rate and amount were used to measure long-term financial burden; concentration index (CI) was calculated for equity. All data were evaluated by descriptive statistics and multi-variate variance analysis. The linear regression and logit regression with random effect analysis upon area was used to evaluate the effect of CII.

Results: Six hundred and thirteen and 834 patients were surveyed at baseline and final evaluation. After the program, the OOP payments of hospitalizations sharply decreased from RMB 39,363.2 to RMB 28,426.1 (p < 0.001), with the largest decrease for lowest income patients (from RMB 44,507.6 to RMB 29,214.2). With longer duration of CII, more OOP medical payments decreased. The amount of medical debt was decreased by RMB 7,209.4 among all the patients, and the decrease was highest in the highest income group (RMB 8,119.9). The CI of AG changed a lot (from −0.858 to −0.670).

Conclusion: The CII has effectively reduced the financial burden of patients with high medical cost, whether in the short-term or a longer length of time. It also improved health equity in health service utilization and expenditure. However, rich householders still receive more benefits from the policy, government health insurance financing is increased, and the policy needs to further benefit the poor.

Keywords:

• Critical illness insurance
• financial burden
• equity
• health utilization
• poor

JEL CLASSIFICATION CODES:

• I12
• I14

Introduction

Critical illness insurance (CII) is one kind of health insurance, but different from common ones, which specially focus on patients diagnosed with severe and/or critical diseases^1,2. In the early 1980s, CII first came to the scene as “Dread Disease Insurance” in South Africa. It has achieved great success as it filled the gap of no coverage for expensive medical expenses³. CII was also implemented in the UK and began to flourish by combining dread disease cover and income protection in the early 1990s. Since then, more countries have begun to practice CII, including the US, Canada, and Australia. In America and the UK, CII provides a single lump sum payment to the insured, including medical costs and non-medical financial costs, such as mortgages and credits⁴. Patients with critical illness would get $10,000 or $20,000 in Initial Benefits from an insurance company in 2017 in the US⁵. Today, CII cover has a considerable market share in Asian insurance markets⁶.

In rural China, the New Cooperative Medical Scheme (NCMS), a voluntary insurance scheme designed for rural residents, was founded in 2003. The enrollment rate of NCMS gradually increased and had achieved 98.8% in 2015. Although the main objective of NCMS is to prevent rural residents from being impoverished by medical expenses, many problems, reflected on financial burden and equity, still exist. Wagstaff et al.⁷ found that the NCMS took no effects on reducing out-of-pocket (OOP) payments. Cheng et al.⁸ also proved that the financial protection of NCMS against high healthcare expenditures remained very restricted among the poor in rural China. Meanwhile, Ma et al.⁹ found NCMS expanded the gap of income-related inequity in rural China.

To further reduce the financial burden of patients with critical diseases and improve the equity between the rich and the poor, on August 30, 2012, China’s National Development and Reform Commission issued the “Guidance about implementation of residents’ critical illness insurance system”. The guidance points out that critical illness insurance (CII) is an institutional arrangement aimed at providing reimbursement of high medical expenses associated with critical illness. At the beginning, the CII only covered 20 officially critical diseases, such as congenital heart disease, breast cancer, and end-stage renal failure. As of February 2014, 25 provinces had established CII systems¹⁰. China’s government gradually transformed from a disease-based principle to a fee-based principle. In other words, after claiming basis health insurance, all the patients whose OOP still exceeded the deductible, usually the local income per capita, could obtain additional reimbursement, regardless of what the disease is¹¹, the average reimbursement rate was ∼ 50–70%.

The research on medical insurance for CII was mostly taken in developing countries, focused on how to improve the health insurance system to alleviate the economic risk of critical disease. Kasemsup¹² showed that the extent of critical illness protection should be adjusted based on factors such as the type of disease, age, treatment effect, emergency, waiting time, and ability to pay in Thailand. Chuma and Maina¹³ took Kenya as an example and recommended that, in addition to inpatient services, outpatient services should be included in the CII service package. In China, due to the short implementation time of CII, the current research was focused on the analysis of the insurance system, from the institutional attributes of the CII to the design of the program and so on¹⁴. However, few studies clearly answer the question: whether the implementation of CII filled the gap caused by basic health insurance on the aspects of financial burden and health equity, especially among rural residents¹⁵.

Financial burden includes two aspects. First, the direct medical cost resulted from the cost of the health services patient utilized including consultations, tests, medicines, and hospitalization, etc.¹⁶ This financial burden is easy to measure. The OOP and catastrophic health expenditure (CHE) are the commonly used indexes. Second, indirect medical cost of income lost and direct non-medical cost of transport and food during healthcare visits¹⁷. This financial burden is hard to measure. In practice, the indirect medical cost and non-medical cost can be bigger than direct medical cost, and may mean a bigger policy gap. Financial burden, in another view, includes short-term burden and long-term burden. The usually used indicators, such as actual health expenditures and amount of expenditures relative to income, respond more to the short-term burden of the patient within 1 year^18–20. Studies indicated that the accumulated long-term burden can be serious and under-researched^21,22. But few studies give a comprehensive evaluation when they focus on the policy’s effect on financial burden.

Equity includes horizontal equity and vertical equity. Kaplow and Musgrave define horizontal equity (HE) as the requirement that equals be treated alike. Both define vertical equity (VE) as requiring an “appropriate” pattern of differentiation among unequals²³. Due to the availability of methods, there is still no mature method to measure vertical equity.

Ensuring everyone can have access to health services without facing financial hardship is the principle of the Universal Health Coverage (UHC). In addition, ensuring equity when people access health services is a major concern in health policy²⁴. In this study, we will give a comprehensive evaluation of CII on financial burden and equity in the rural areas of China. The questions we want to answer are (1) whether the CII decreased the short-term and long-term financial burden of patients; (2) whether the CII decreased the whole burden with patients had less unoccupied health utilization; and (3) if CII do reduce the economic burden, does it have the same effect on the rich and the poor. Our findings quantitatively assessed CII performance relative to the goal of offering financial protection against poverty, and may contribute to improving and adjusting CII policy, thereby further relieving the inequity in healthcare.

Methods

Introduction of sample setting and the policy of CII

Study setting

According to the study by Meng et al.²⁵, households in the central and west region are more likely to suffer from higher rates of catastrophic health spending. The percentage of households experiencing catastrophic health expenses was 13.7% in central China, 13.3% in the west region, compared to 11.9% in the east region. Thus, we selected one county for research in the central and western regions, respectively. Based on the availability of data, we selected Hubei Province in the central part and Guizhou Province in the west as the sampling area. In these two provinces, we randomly selected two sample counties: Xiantao county and Yuqing county.

Hubei province is in central China and the middle reaches of the Yangtze River. It had a gross domestic product (GDP) per capita of Renminbi (RMB)55,665 (8,383 USD) in 2016, which ranks it 11th among the 32 provinces (municipalities and autonomous regions) in China. Xiantao, where this study sample was drawn, is a county in Southwest Hubei. It had a GDP per capita of RMB 56,065 (8,438 USD) in 2016. Guizhou province is located in western China with poor resources. The GDP per capita was RMB 33,242(5,006 USD) in 2016, and ranks it 29th among the 32 provinces. Yuqing is a county with a GDP per capita of RMB 31,503 (4,744 USD).

CII Policy in sampled cities

Both sampled cities customized their own CII project under the local context. In Xiantao, CII started in 2015, the official premium is 12,000 yuan, and the reimbursement rate is 55% for OOP located for RMB 12,000–30,000, 65% for RMB 30,000–100,000, and 70% for over RMB 100,000. For example, if a patient’s medical expenses were 60,000 in Xiantao, and supposed all medical expenses were included in the benefit package, and NCMS could cover 30,000. In the baseline period, the patient’s out-of-pocket expenses would be 30,000 yuan (60,000 – 30,000), while in the final stage the patient's out-of-pocket expenses would be RMB 20,100 (60,000 – 30,000 – (30,000 – 12, 000) *55%). Yuqing started CII in 2016. Due to the different local economic conditions, the deductible was 8,000 in Yuqing, and the reimbursement rate was 50% in RMB 8,000–RMB 60,000 and 60% when OOP was over 60,000.

Data collection

Data were obtained from cross-sectional surveys at the baseline of the program in 2014 and a final evaluation in 2017. The inclusion criterion for the for the final evaluation that patients benefited from CII in the past year (2016). In order to ensure the comparability of the research objects before and after the implementation of CII, the deductible in final stage was discounted to calculate the hypothetical deductible in baseline, and then select the patients who exceeded the deductible in the current year. In 2016, the number of CII beneficiaries of Xiantao and Yuqing were 4,137 and 935, respectively. According to the economic development of the sample county, all townships are divided into good, middle, and poor categories. Patients were selected from the stratified cluster sampling of each type of township in the extraction of 2–5 townships. Five hundred patients were surveyed in two regions before and after the implementation of CII, for a total of 2,000 patients.

The survey was done with a structured questionnaire, which was constructed with reference to the National Health Services Survey Questionnaire and China Health and Retirement Longitudinal Study (CHALS) Questionnaire. Face-to-face interviews were conducted by trained investigators using a structured survey questionnaire which collected basic demographic information and data on socio-economic status, treatment procedures, and expenditures. All questionnaires were checked on site for completeness and internal logic, and they were captured for analysis using a double-entry procedure in Epi-Data.

Financial burden measurement

Measuring financial burden

• OOP. Comprised the expenses below the deductible, the expenses above the deductible co-paid by patients, and the non-reimbursable amount beyond the NCMS benefit packages.

• Ratio of OOP to income. Household income was the aggregate of household income from production, wage incomes of household members, transfer income (pensions, remittances, welfare), and property income (interest, rent)²⁵. Each family’s income was calculated by summing incomes across individual household members.

• Prevalence of Catastrophic health expenditure (CHE). The criteria to define the CHE varies from 10% of income, 10% of household consumption, to 40% of disposable income^26–28. In this paper we regarded OOP in excess of 10% of annual family income as CHE, because the daily consumption of rural residents, such as food expenditure, cannot be accurately measured. Prevalence of CHE is the ration of the number of families with CHE divided by total amount of sampled families.

• Average gap of CHE. Reflected the extent of the impact of CHE on society as a whole. It describes how much a patient’s medical expenditure (as a percentage of their annual income) is in excess of the catastrophic threshold of 10% of annual income²⁹. $${\mathrm{AG}}_{\mathit{CHE}}=\sum \left(P\right.-\left.10\mathrm{\%}\right)/N$$ where AG refers to the average gap of CHE, p is the proportion of OOP cost as a percentage of the annual income of patients who incur CHE, and N is the total number of patients in the sample.

• Relative gap of CHE. Reflected the extent of the impact of CHE on families who had CHE³⁰. The equation was: $${\mathrm{RG}}_{\mathit{CHE}}=\sum \left(P\right.-\left.10\%\right)/N_r$$

  where RG refers to the relative gap of CHE, p is the proportion of OOP cost as a percentage of the annual income of patients who incur CHE; and $N_r$ is the total number who had CHE.

• Medical debt. The above indicators respond more to the short-term burden of the patient within 1 year. Medical debt rate and the average amount of medical debt were adopted to measure long-term financial burden. Borrowing can be at high rates of interest, so choosing that may sacrifice future income and put the family in persistent poverty³¹.

Measuring patient health utilization

In this paper, we use health service utilization indicators to indirectly reflect the whole financial burden. The principle behind this is that the use of health services and the economic burden are closely related. Patients may have forgone medical care due to high cost³². Therefore, in the case of constant health needs, the increase of health service utilization must result from the reduction of financial burden. The indicators adopted are frequency of hospital admissions per patient per year, average length of stay per patient per year, and the percentage of forgoing treatment or discharge early because of financial burden (underuse of impatient services) in the past year.

Equity measurement

Concentration index (CI) is a widely used index to measure equality, and need standardized CI is commonly used to measure horizontal equity. High income people usually have low needs, while low income people usually have high needs. Due to the unavailability of need data, in this paper, we just calculated CI. The effect of CII on the horizontal equity was speculated by the CI and income. CI was calculated using the following equation: $$\mathrm{CI}=\frac{2}{N\mu }{\sum }_{i=1}^N{h}_i{r}_i-1-\frac{1}{N}$$ where $h_i$ represents the variable we cared, $\mu$ is its mean, and $r_i=i/N$ is the fractional rank of income, ranging from 0–1. For one individual ranked$i,$ $r_i$= (i − 0.5)/N, in which N is the total number of individuals.

The CI is bounded between −1 and 1. It takes positive (negative) values when the concentration curve lies below (above) the diagonal. This means that a positive (negative) value corresponds to inequalities favoring the rich (poor)³³.

Statistical analysis

General characteristics, including gender, marriage, education, jobs, and length of stay (LOS), impatient frequency, etc. were examined at baseline and at final evaluation using descriptive statistics. Considering the susceptibility of disease, illness category was determined by the patient’s primary diagnostic ICD10 code in the visit record. In order to explore the CII’s potential differential impact on patients from different income groups, especially the effect on the poorest patients that are most susceptible to financial difficulty, patients were categorized based on quartiles of their household income as low income group, middle income group, and high income group. We performed a normality test on all variables and selected appropriate statistical method based on the data distribution. Chi-square tests, T-tests, rank sum test, and Fisher’s exact probability were used to assess the differences in variables upon sub-groups. p < 0.05 was considered to be statistically significant. All currency-related metrics have been discounted

Linear regression and logit regression were used to evaluate the impact of CII for OOP with control of other independent variables. Since OOP is a right skewed distribution, we take the logarithm of OOP in linear regression. Random effect analysis upon counties was included in regression that would have taken into consideration the correlated nature of samples measured from the same county. Considering that two chosen counties have a different duration of experiencing CII, OOP changes upon the duration of CII implementation was also assessed. We used Stata 12.0 to conduct the quantitative data analyses.

Results

Six hundred and thirteen and 834 patients were surveyed at baseline and final evaluation. The socio-economic characteristics of patients were similar in both samples, except for their employment, income, hospital level, and area (Table 1). At baseline, 46% patients had jobs including farming, but only 27.9% of patients worked in the final stage. Meanwhile, the average household income of patients was RMB 53,219.5 (RMB 650 = US$100) in baseline compared to RMB 31,986.4 at final evaluation (p < 0.001). After program implementation, 74.3% of patients went to the higher level medical institutions outside the county for treatment. Chronic diseases accounted for the most part of all patients.

Table 1. Patient characteristics before and after intervention (%).

	Baseline (n = 613)	Final (n = 834)
Variables	n (%)	n (%)	P value
Gender
Male	320 (52.2)	439 (52.6)	0.870^a
Female	293 (47.8)	395 (47.4)
Age
≤17	63 (10.3)	63 (7.6)	0.189^a
18–64	427 (69.7)	595 (71.3)
≥65	123 (20.0)	176 (21.1)
Marriage
Married	480 (78.3)	657 (78.8)	0.828^a
Others^d	133 (21.7)	177 (21.2)
Education level
None	130 (21.2)	192 (23.0)	0.687^a
Primary	244 (39.8)	309 (37.1)
Secondary	188 (30.7)	257 (30.8)
High school and above	51 (8.3)	76 (9.1)
Family member
≤2	163 (26.6)	242 (29.0)	0.537^a
3–5	355 (57.9)	460 (55.2)
≥6	95 (15.5)	132 (15.8)
Employment
Currently working	282 (46.0)	233 (27.9)	<0.001^a
Others^e	331 (54.0)	601 (72.1)
Income
Income	53,219.5 (53,120.5)^c	31,986.4 (39,363.8)	<0.001^b
Hospital level
In county	258 (42.1)	214 (25.7)	<0.001^a
Out county	355 (57.9)	620 (74.3)
Illness category
Acute illness	133 (21.7)	188 (22.5)	0.702^a
Chronic illness	480 (78.3)	646 (77.5)
Area
Xiantao	270 (44.0)	434 (52.0)	0.003^a
Yuqing	343 (56.0)	400 (48.0)

^a Chi-square tests.

^b Rank sum test.

^c Exchange rate = RMB 650 = US$100 at the end of 2017.

^d Others includes unmarried, divorced, and widowed.

^e Others includes unemployed, school students, and retired.

After program implementation, the out-of-pocket payments of all patients in hospitalizations sharply decreased from RMB 39,363.2 to RMB 28,426.1, with lowest income patients having the largest decrease (from RMB 44,507.6 to RMB 29,214.2). All changes were statistically significant (p < 0.001). High income patients had a greater decrease in CHE incidence, with rates decreases of 2.4 and 0.4 percentage points, respectively. The average gap of CHE decreased in the most poverty group (−11.8), but increased in the middle-income group (0.7). The result of the relative gap of CHE was similar to the average gap (Table 2).

Table 2. Patients financial burden at baseline and final evaluation (by income group).

	Baseline	Final
Income level	Mean (SD)	Mean (SD)	Change	p-value
OOP(RMB)
Low	44,507.6 (44,507.5)	29,214.2 (22,345.4)	−15,293.4	<0.001^a
Middle	38,594.0 (27,424.9)	29,385.0 (27,783.9)	−9,209.0	<0.001^a
High	35,780.6 (29,182.0)	25,784.6 (16,324.5)	−9,996.0	<0.001^a
Total	39,363.2 (33,013.1)	28,426.1 (24,021.5)	−10,937.1	<0.001^a
OOP/Income (%)
Low	2,292.6 (13,811.1)	1,106.5 (1,350.5)	−1,186.1	0.275^a
Middle	103.3 (90.6)	164.6 (201.9)	61.3	<0.001^a
High	36.4 (38.1)	40.8 (32.0)	4.4	0.052^a
Total	582.7 (6625.4)	322.5 (741.1)	−260.2	0.019^a
CHE incidence (%)
Low	1 (0)	1 (0)	0	–
Middle	1 (0)	97.6 (15.4)	−2.4	0.006^b
High	90.9 (28.8)	90.5 (29.4)	−0.4	0.900^c
Total	97.7 (15.0)	96.4 (18.6)	−1.3	0.020^b
Average gap of CHE
Low	22.8 (138.1)	11.0 (13.5)	−11.8	0.275^a
Middle	0.9 (0.9)	1.6 (2.0)	0.7	<0.001^a
High	0.3 (0.4)	0.3 (0.3)	0	0.050^a
Total	5.7 (66.3)	3.1 (7.4)	−2.6	0.018^a
Relative gap of CHE
Low	22.8 (138.1)	11.0 (13.5)	−11.8	0.275^a
Middle	0.9 (0.9)	1.6 (2.0)	0.7	<0.001^a
High	0.3 (0.4)	0.3 (0.3)	0	0.019^a
Total	5.9 (67.0)	3.3 (7.5)	−2.6	0.003^a

Income was divided in three levels: min–Q25 was low income level, Q25–Q75 means middle income level; and Q75–max was high income level. Q means quantile.

^a Rank sum test.

^b Fisher’s exact probability.

^c Chi-square tests.

Table 3 shows the proportion of having medical debt increased from 62.8% to 69.1% (p < 0.001), where the lowest income group had the least change (from 64.7% to 64.9%, p = 0.969). But the amount of medical debt was decreased by RMB 7,209.4 among all patients, and the highest income group had the largest change (decreased by RMB 8,119.9, p = 0.756).

Table 3. Medical debt rate and amount at baseline and final evaluation (by income group).

	Baseline	Final
Income level	Mean (SD)	Mean (SD)	Change	p-value
Debt rate
Low	64.7 (47.9)	64.9 (47.8)	0.2	0.969^a
Middle	69.6 (46.1)	72.3 (44.8)	2.7	0.432^a
High	47.4 (50.1)	66.7 (47.3)	19.3	<0.001^a
Total	62.8 (48.4)	69.1 (46.3)	6.3	<0.001^a
Debt value
Low	48,883.6 (58,477.2)	41,146.0 (60,625.5)	−7,737.6	0.194^b
Middle	43,532.7 (53,443.7)	37,083.9 (46,985.0)	−6,448.8	0.504^b
High	38,583.2 (59,410.1)	30,463.3 (43,333.4)	−8,119.9	0.756^b
Total	43,624.8 (56,289.1)	36,415.4 (49,965.5)	−7,209.4	0.192^b

^a Chi-square test.

^b Rank sum test.

After program implementation, utilization of inpatient services increased, but to different extents in different income groups. The frequency of hospital admissions significantly increased from 2.6 to 4.6 among all patients (p < 0.001), and the increase was greatest for patients with low income (p = 0.001). Length of stay showed the same change. The underuse of impatient services decreased from 16.2% to 14.9%, with the least change in low income group (from 20.9% to 19.9%, p = 0.778) (Table 4).

Table 4. Inpatient service utilization at baseline and final evaluation (by income group).

	Health service utilization, mean (SD)
Income level	Baseline	Second-line	Change	p-value
Frequency of hospital admissions
Low	2.8 (3.4)	4.6 (5.6)	1.8	0.001^a
Middle	2.5 (1.8)	5.1 (7.9)	2.6	<0.001^a
High	2.5 (2.1)	3.7 (5.3)	1.2	0.012^a
Total	2.6 (2.4)	4.6 (6.8)	2	<0.001^a
Length of stay (days)
Low	37.9 (48.8)	52.5 (70.6)	14.6	0.026^a
Middle	32.5 (35.9)	51.9 (68.4)	19.4	<0.001^a
High	29.2 (29.3)	41.0 (57.1)	11.8	0.020^a
Total	33.0 (38.2)	49.3 (66.5)	16.3	<0.001^a
Underuse of inpatient services (%)
Low	20.9 (40.8)	19.9 (40.0)	−1	0.778^b
Middle	16.7 (37.3)	15.6 (36.3)	−1.1	0.717^b
High	10.4 (30.6)	8.5 (28.1)	−1.9	0.546^b
Total	16.2 (36.8)	14.9 (35.6)	−1.3	0.445^b

^a t-test.

^b Chi-square tests.

The CI of OOP payments and AG were all negative in baseline and final stage, indicating that all cases OOP payments and average gap were pro-poor. After program implementation, the CI of OOP payment changed from −0.045 to −0.035, the CI of average gap increased by 0.188 (Table 5). This change clearly shows that harsh economic effects of critical illness on poorer populations were better than before with a lower absolute value after the intervention. But the CI of AG was −0.670 in final stage, indicating a strong, pro-poor inequity.

Table 5. Concentration Index of indicators before and after the intervention.

	Baseline	Final	Change
OOP (RMB)	−0.045	−0.035	0.010
OOP/Income (%)	−0.843	−0.650	0.193
CHE incidence (%)	−0.022	−0.024	−0.002
Average gap	−0.858	−0.670	0.188
Relative gap	−0.854	−0.659	0.195
Debt rate	−0.053	0.006	0.059
Debt value	−0.038	−0.064	−0.026
Frequency of hospital admissions	−0.009	−0.031	−0.022
Length of stay (days)	−0.041	−0.035	0.006
Underuse of inpatient services (%)	−0.128	−0.138	−0.01

Table 6. The linear regression of log out-of-pocket and logit regression of CHE and debt incidence with random effect analysis upon area.

	OOP	CHE	Debt rate
	Coefficient	Coefficient	Coefficient
CII	−0.558***	−1.609***	−0.088
Gender
Female	0.034	0.365	0.073
Age
18–64	0.001	−1.548	0.670*
≥65	0.101	−1.371	−0.390
Marriage
Married	−0.056	−0.731	−0.076
Education level
Primary	0.137**	0.019	0.187
Secondary	0.136*	0.270	0.297
High school and above	0.056	−0.029	0.067
Family member
3–5	−0.004	−0.632	1.137***
≥6	−0.087	−0.386	1.335***
Employment
Currently working	−0.008	0.097	−0.404
Income
Middle	−0.049	−14.836	−0.081
High	−0.056	−16.919	−0.897***
Length of stay (LOS)	0.001**	0.001	0.001
Impatient frequency
2–4	0.152***	0.282	0.270
≥5	0.237***	1.547*	0.622**
Hospital level
Out of county	0.510***	1.210**	0.307*
Illness category
Chronic illness	−0.185***	−1.173*	−0.072
Cons	10.281	23.726	−0.773

*p < 0.05, **p < 0.01, ***p < 0.001.

Table 6 clearly shows that implementation of the CII was associated with a significant decrease in OOP medical expenditure (p < 0.001). It is also notable that having chronic illness and seeking healthcare out of the county was associated with high OOP medical payments. The CII also has intensified the occurrence of CHE (p < 0.05). Seeking healthcare out of the county and having acute disease would attribute to high CHE incidence, more family members, and hospitalization times leading to higher medical debt incidence. For areas CII implemented for 2 years (Xiantao), the OOP payments declined by RMB 18,016.2, accounting for 35.69% of baseline OOP, which is significantly more than that in Yuqing (Figure 1).

Figure 1. Out-of-pocket payments at baseline and final evaluation (by CII durations).

Discussion

In this study, we have evaluated the effect of the CII systems of two counties in China based on the financial burden and health utilization before and after the implementation of the CII system. The results suggest that CII effectively reduced the OOP payments and medical debt amount, and improved inpatients health utilization with higher frequency of hospital admissions and length of stay. However, the equity improvements provided by CII is somewhat limited.

After the CII, the OOP significantly decreased with adjusting for the covariates. Nearly 10% of total medical expenditure were additionally payed by CII³⁴, which resulted in lower OOP. A systematic review on interventions to reduce illness and injury related financial burden also proved that eliminating or largely reducing copayments in current insurance schemes is effective in reducing OOP medical payments³⁵. In all groups, low-income patients had the most OOP reduction after CII. The reason is people with hard economic status usually had higher OOP payments because of limited private health coverage and existing co-morbidities¹, and CII implemented segmentary reimbursement. The higher the OOP payment after NCMS, the higher the reimbursement rate. Thus, poor patients would get more subsidies in CII.

With a longer duration of CII, more OOP medical payments decreased. This may be because many patients did not understand the policy content when was CII implemented in the first year. With the promotion of CII, the demand for medical services is further released, and more patients benefited from it. The duration of intervention was also an important moderating variable in other studies^36,37, with durations of more than 2 weeks or 12 months being significantly more effective than interventions of shorter duration. But the effect on CHE incidence was limited, only 1.3% patients changed after the CII, and over 95% patients still had catastrophic medical payments. One of the main reasons was lower household income in the final stage. Although CII effectively reduced OOP payments, the household income of patients after the project was less than before (decreased by RMB 20,000). So the CHE incidence did not change a lot. Why had the income decreased after CII? We used the same sampling method for the same sampling area before and after the project. Under the condition of ensuring the scientific study design, the possible reason is that poorer patients would seek health utilization if they had less health expenditure. Many poor people actually give up medical services because of economic shortages^38,39. With the additional compensation of CII, their OOP has dropped significantly, and more poor people will choose treatment, so the average income of the late stage of the project is lower. Our results about the health utilization also support this discussion.

There was an interesting finding in long-term financial burden in this study: after the program, debt incidence increased, but debt value decreased. More patients had medical debt after CII. This may be caused by less OOP payments. With a catastrophic medical cost, households would have to rely on informal healthcare payment arrangements, such as using savings, borrowing money, or selling assets⁴⁰. With less medical payments, more patients may try to borrow money for healthcare because they think repaying money is not very difficult. Furthermore, lower OOP payments may also attribute to a lower borrowing amount. Once they had less medical expenditure, they would reduce the debt amount to avoid further negative effects. The CII effectively increases the utilization of health services and reduces the abandonment of treatment due to economic reasons. This is because CII can further reduce financial barriers. Health insurance increases the need for healthcare among patients through the reduction of effective prices⁴¹.

Compared with poor counterparts, patients with middle-income benefited more from health insurance in terms of using inpatient services. This is because the better-off are better equipped to respond to increased medical needs by diverting resources from expendable consumption, whereas the poor are limited in their ability to divert resources from basic and subsistence needs⁴². Why isn’t it the highest income group or the lowest that have the most growth? Because, for the wealthiest people, there is less potential health services demand. The price elasticity of health service demand is small, and CII will not cause huge changes in service utilization. For low-income patients, although CII may release demand, the constraints, such as income, service quality, access and direct user charge, would prevent the poor from taking advantage of this services⁴³. Other studies have similar findings. Hedda Flatø and Huafeng Zhang⁴⁴ found that richer households enjoyed more increase in healthcare utilization before and after universal health coverage reforms in China, and the CI was increased from 0.065 to 0.069.

With positive change of most indicators, CII improved economic fairness. High OOP and high reimbursement ratio lead to more insurance subsidies for low-income patients, and equity in financing improved. But this ability is limited, and several factors are attributed to this. First, the deductible of CII was RMB 12,000 in Xiantao and RMB 8,000 in Yuqing. This standard was based on local per capita income, but many poor families can’t even pay for the threshold. It was too high for patients to enjoy the subsidies. And this is particularly true when CII only cover the part above the threshold, households must still meet the other part of the costs of care themselves⁴⁵. On the other hand, the NCMS benefit package excluded some new and expensive drugs and examinations. But in the fee-for-service payment system, doctors still have strong incentives to prescribe these expensive drugs and examinations. Thus, the expenditure within the benefit package amounted to much less than actual expenditure⁴⁶. Moreover, policies mandating deductibles, co-payments and ceilings further increased the CHE incidence. There is rich international literature suggesting that public subsidies for health programs frequently benefit richer more than poorer people^47–49.

Taking into account the economic burden of diseases in low-income people, the government began to enact some pro-poor CII policies. Some regions have tried to lower the deductible line for low-income patients and increase the proportion of reimbursement⁵⁰. These measures can further improve economic fairness, and Medical Financial Assistance (MFA) specifically for low-income families can alleviate the economic difficulties of these poor people. But fairness is still a problem that cannot be ignored in China and other low- and middle-income countries^24,51, and more measures need to be taken.

This study has several limitations. First, the study was not a randomized controlled study. We evaluated the impact of CII with pooled cross-section data before and after intervention. However, various biases may occur in implementation. National statistics show that the average inpatient cost per admission increased 15.6% over the intervention period⁵². The effect of CII on lowering OOP may be under-estimated. Furthermore, China is implementing health system reform, there were multiple concurrent policy interventions that may be synergistic or antagonistic, these confounding factors may make us estimate the role of CII with bias. Second, we only analyzed the patients in the NCMS system who enjoyed the CII in this study, but many poor patients may give up treatment because of poverty. They simply do not reach the deductible line and are not included in the sample. Therefore, we did not get the data of these patients. Third, we assessed the OOP change between two stages upon CII durations, but counties and CII durations are highly collinear, and we cannot judge the determinants of differences in OOP changes over the two periods.

Conclusion

Through the comprehensive measurement of medical burden, this study found that the implementation of CII effectively reduced the burden of patients, including the short-term actual burden of OOP, CHE incidence, the long-term actual burden of medical loans, and the underuse of hospitalization services. It has further improved economic equity. However, the effect of CII on the improvement of fairness is relatively limited. This study suggests that the implementation of CII is necessary, and the Chinese government urgently needs to increase CII funds and extend the healthcare insurance fund investment channel. In the follow-up work, the government should consider more the release of health needs and how to lean toward the poor.

Transparency

Declaration of funding

This study is supported by the National Natural Science Foundation of China: “Study on the Dynamic Optimization of Catastrophic Health Insurance Reimbursement Modes and the Scale of Fund Expenditure in the Perspective of UHC” (Grant No. 71573095).

Declaration of financial/other interests

JJ, YX, LZ, ZZ, XW, and XL are employed by Huazhong University of Science and Technology, Wuhan, China. SC is employed by the Chinese University of Hong Kong, Hong Kong, China. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgment

We would like to thank the staff of the Health and Planning Commission of Xiantao county and Yuqing county for their support on the research. We also want to thank the associate editor and reviewers for their careful review and insightful comments, which have led to significant improvement of the manuscript.