The Impact of Multimorbidities on Catastrophic Health Expenditures among Patients Suffering from Hypertension in China: An Analysis of Nationwide Representative Data

Abstract

Background: Patients with hypertension are sensitive to multimorbidities (i.e., the existence of ≥2 chronic diseases), and the related treatment can create enormous economic burdens. We sought to examine the distribution of multimorbidities, the prevalence and factors of catastrophic health expenditure (CHE), the impact of multimorbidities on CHE, and the variation in this relationship across age groups, work status, and combinations of socioeconomic status and health insurance types. Methods: Socioeconomic-related inequality associated with CHE was estimated by concentration curve and concentration index. We examined the determinants of CHE and the impact of age groups, work status, and combinations of socioeconomic groups and health insurance schemes against the relationship with multimorbidities and CHE using logistic regression. Results: 5693 (83.3%) participants had multimorbidities. In total, 49.8% of families had experienced CHE, and the concentration index was −0.026 (95% confidence interval [CI], −0.032 to −0.020). Multimorbidities were related to the increased odds of CHE (odds ratio [OR], 1.21; 95% CI, 1.18–1.25). The relationship between multimorbidities and CHE persisted across age groups, work status, and combinations of socioeconomic status and health insurance schemes. Conclusions: More than 80% of patients with hypertension had multimorbidities. The protection of health insurance schemes against financial risks is very limited.

1. Introduction

As the most common chronic disease, hypertension is a major contributor to morbidity, early death, and disability, and the number of deaths associated with hypertension totaled 10.4 million worldwide in 2017 [1,2]. According to recent estimates, >1 billion people worldwide suffered from hypertension in 2019, a number which has doubled since 1990 [3], and 2/3 of these patients reside in low- and middle- income countries [4]. Recent evidence demonstrated that up to 27.5% of Chinese adults suffer from hypertension, with prevalence rates of 13.3% among people aged 18–44 years, 37.8% among people aged 45–59 years, and 59.2% among people >60 years, respectively [5].

Patients with hypertension often suffer from multimorbidities [6,7,8], i.e., the existence of ≥2 chronic diseases [9] such as cardiovascular diseases, diabetes, kidney diseases, respiratory diseases, and lung diseases [6,10,11]. A community survey in China found that hypertension complicated by diabetes was the most prevalent chronic disease pairing in middle-aged and elderly individuals, followed by hypertension complicated by bone disease [12]; meanwhile, another study based on the electronic Clinical Management System in Hong Kong found that more than two-fifths of patients with hypertension experienced multimorbidities [6], and a cross-sectional survey of older adults in Bangladesh reported that 95.7% of patients suffering from hypertension had multimorbidities [13]; moreover, a study based on a patient-level primary care database in the UK among patients over 18 years old found that more than half of patients with hypertension had multimorbidities [7]. Multimorbidities among people suffering from hypertension constitutes a huge burden on the healthcare system.

Given that out-of-pocket (OOP) expenditure in low- and middle-income countries are high, patients suffering from multimorbidities may be confronted by economic difficulties [14,15,16]. According to the latest report of the World Health Organization, >35% of China’s health expenditure was OOP expenditure in 2019 [17]. People suffering from hypertension require life-long medical care, causing higher OOP expenditure and an economic burden [18]. In addition, hypertension is becoming the most expensive health condition worldwide, and this burden is only likely to continue to grow further due to an aging society and rising prevalence rates [19].

The health system should allow people’s access to health services without economic sacrifice, which is a goal of universal health coverage [20]. However, more and more families are experiencing catastrophic health expenditures (CHEs) [21]. In 2016, on the basis of 3 major health insurance schemes, the Chinese government announced a decision to integrate health insurance schemes among urban and rural residents that follow a higher reimbursement ratio and wider benefit coverage. However, it was reported that this health insurance integration plan did not decrease the CHE risk [22]. The evidence suggested that the impact of multimorbidities on CHE continued to persist irrespective of health insurance type and socioeconomic group [23]. Additionally, previous research found that people with a higher socioeconomic status have a low risk of experiencing CHEs, which supports the notion that socioeconomic-related inequality exists in CHE [24].

Some studies have examined the economic effect of multimorbidities; however, most were conducted among non-representative populations [25], and the findings have not been systematically validated in low- and middle-income countries [23]. Moreover, while the prevalence of multimorbidities and the incidence of CHE is high among people with hypertension [26], little evidence supports the association between multimorbidities and CHE among patients with hypertension, and the influence of different combinations of socioeconomic groups and health insurance types on said association remains unclear.

The purpose of this study was to address the evidence gap using a national population-based data sample to assess the distribution of multimorbidities, the prevalence rate and factors of CHE, and the impact of multimorbidities on CHE and whether this relationship varies under different combinations of socioeconomic groups and health insurance schemes among middle-aged and elderly people suffering from hypertension in China. In addition, since this study is a part of a retirement study, we also performed stratified regression models to examine the effect of multimorbidities on CHE by age and work status.

The remaining part of the paper proceeds as follows: Section 2 describes the data sources and study methodology. The Section 3 presents the findings of the study, focusing on the relationship between multimorbidities and CHE and whether this relationship varied under different subgroups. The Section 4 provides a discussion of the findings and attempts to give the implication for further research in this area. Finally, the conclusion gives a brief summary and critique of the findings.

2. Materials and Methods

2.1. Data Sources

We used a nationwide population-based dataset from the China Health and Retirement Longitudinal Study (CHARLS) in 2018, which gathered basic individual and household information from 28 provinces, 150 countries or districts, and 450 villages or communities in China using a stratified multistage probability proportional to population size random sampling method. CHARLS collected data on middle-aged and elderly people aged 45 years and above in China. In 2018, 19,507 respondents from 11,595 households were recruited. We included 7114 patients diagnosed with hypertension and then excluded 278 patients who had insufficient information on health or household expenditure to calculate CHE. Thus, the final study sample was 6836 participants (Figure 1). In addition, we compared the discrepancy between the data before deleting the missing value and after deleting the missing value using the chi-squared test or t-test and found no significant difference between the two groups (Supplemental Table S1).

For this study, information concerning main respondents and spouses’ demo-graphic, socioeconomic, and health-related behavioral characteristics, including sex, birth date and place, family structure, education, work status, marital status, income and expenditure, health, health insurance, and health care utilization, was gathered.

2.2. Multimorbidities

The survey measured chronic diseases by diagnosis. We used 14 chronic diseases to define multimorbidities, including hypertension, dyslipidemia, diabetes, malignant neoplasm, chronic lung disease, liver disease, heart disease, stroke, chronic kidney disease, digestive disease, mental problems, memory-related disease, arthritis or rheumatism, and asthma. We counted the number of chronic diseases in each patient with hypertension to determine who had multimorbidities.

2.3. Catastrophic Health Expenditure

In a base case analysis, a definition from the World Bank was used to calculate CHEs, i.e., CHE was defined when OOP spending accounts for ≥25% of the house-hold’s capacity to pay [27]. We calculated the household’s capacity to pay as the total amount of non-food household expenditure. We defined CHE as a binary variable and the equation is as follows [28]:

$$E=\{\begin{array}{c}1, if\frac{T}{x-f\left(x\right)}>Z\\ 0, otherwise\end{array}$$

where E indicates whether the patient’s household has experienced CHE or not, T represents OOP spending, x represents the total consumption expenditure of a household, f(x) represents food-based household expenditure, and Z represents the threshold of 25%.

2.4. Socioeconomic Status

We adopted the annual per-capita household consumption expenditure amount as a measure of socioeconomic status [28]. Taking into account the varying levels of economic development in different areas, we defined four socioeconomic groups in each city according to per-capita household consumption expenditure quartiles, then combined all 4 groups.

2.5. Health Insurance

The study participants were covered by 4 major public health insurance schemes in 2018. The Urban Employee Basic Medical Insurance (UEBMI) scheme, which aims to insure urban workers and retirees, was implemented by the Chinese government in 1998 and is funded by the payroll income of employees. The target population for the Urban Resident Basic Medical Insurance (URBMI) scheme is other urban residents who are not enrolled in UEBMI, and this scheme is funded by government subsidies and individual contributions. The New Rural Cooperative Medical Scheme (NRCMS) scheme was implemented in 2003 to meet the needs of rural residents, and, similar to URBMI, the NRCMS is financed by subsides from governments and premiums from individuals. Since 2016, the Chinese government has integrated URBMI and NRCMS as the Urban and Rural Resident Basic Medical Insurance (URRBMI) scheme, which covers residents from these two main public health insurance schemes. This integration has further expanded the pool of funding and narrowed the gap in access to medical services across health insurance types.

2.6. Variables

The following variables were included in the study as covariates: age, sex, education (no education and primary school, secondary school, and college and above), work status (employed, jobless, unemployed, and retired), smoking status (non-smoker and smoker), frequency of drinking (>1 time/month, <1 time/month, and never), physical examination (including physical examination, routine blood test, routine urine test, liver function test, kidney function test, lipids profile test, blood glucose test, and so on), health insurance (UEBMI, URRBMI, URBMI, NRCMS, other insurance, and without insurance), socioeconomic quartiles, area (east, central, west, and northeast), number of family members, and impoverishment status (i.e., registered poverty-stricken households, a poverty alleviation method in China’s poverty alleviation policy that aims to accurately identify poor households and poor villages).

2.7. Statistical Analysis

The distribution of multimorbidities among patients suffering from hypertension was displayed by descriptive statistics using mean, standard deviation, and percentage values. The chi-squared test or analysis of variance was adopted to explore the discrepancy between groups. To estimate socioeconomic-related inequalities associated with CHEs, a concentration curve and concentration index were adopted [28]. The concentration curve showed the cumulative percentage of health outcome variable (i.e., CHE) against the cumulative percentage of households ranked based on living standard (i.e., per-capita household consumption expenditure). The concentration index, which represents the extent of inequality, is defined as twice the area between the concentration curve and the line of equality and was produced by the CONCINDC [29], which is the most standard method to generate the concentration index for individual or grouped data [30]. If people with a low socioeconomic status are more likely to experience CHE, the curve will lie above the equality line and the index will be a negative value [28].

The concentration index was calculated as follows [28]:

$$C=\frac{2}{\mu }cov\left(h,r\right)$$

where C represents the concentration index, h represents the health outcome variable, µ represents its mean, and r represents the fractional rank according to the living standard.

We used binary logit regression to examine the correlation between each variable and CHE and then used logistic regression to investigate the determinants of CHE and the relationship between multimorbidities and CHE. In addition, the impact of age groups (60+ or not), work status (working or not), and different combinations of socioeconomic groups and health insurance schemes on the relationship between multimorbidities and CHE was also examined via logistic regression. We reported odds ratios (ORs) adjusted for age, sex, education, work status, smoking status, frequency of drinking, physical examination, area, number of family members, and impoverishment status, with 95% confidence intervals (CIs).

Data analyses were conducted by SPSS version 25.0 (IBM Corp., Armonk, NY, USA) and Stata version 13.0 (Stata Corp., College Station, TX, USA). p values were regarded as statistically significant with the threshold of 0.05.

2.8. Sensitivity Analyses

We used the other threshold of CHE by the definition of the World Bank to investigate the effect of multimorbidities on CHE in the sensitivity analyses—that is, 40% of non-food household expenditure [27].

3. Results

3.1. Basic Characteristics of the Sample Population

In total, 6836 participants suffering from hypertension were selected for this analysis.

Table 1. Relationship of multimorbidities with demographic, socioeconomic, and health-related behavioral characteristics.

	Hypertension Only N = 1143 (16.7%)	Hypertension and 1 Other Chronic Disease N = 1582 (23.1%)	Hypertension and 2 Other Chronic Diseases N = 1493 (21.8%)	Hypertension and 3 Other Chronic Diseases N = 1067 (15.6%)	Hypertension and ≥4 Chronic Diseases N = 1551 (22.7%)	Overall N = 6836	p Value
Age							<0.001
Age, years	63.1 (10.5)	64.4 (10.2)	65.2 (10.0)	65.3 (9.5)	66.2 (9.3)	64.9 (9.9)
Sex							<0.001
Male	612 (53.5%)	800 (50.6%)	692 (46.3%)	488 (45.7%)	647 (41.7%)	3239 (47.4%)
Female	531 (46.5%)	782 (49.4%)	801 (53.7%)	579 (54.3%)	904 (58.3%)	3597 (52.6%)
Education							0.049
No education and primary school	746 (65.3%)	1063 (67.2%)	1012 (67.8%)	743 (69.6%)	1107 (71.4%)	4671 (68.3%)
Secondary school	377 (33.0%)	485 (30.7%)	447 (29.9%)	301 (28.2%)	415 (26.8%)	2025 (29.6%)
College and above	20 (1.7%)	34 (2.1%)	34 (2.3%)	23 (2.2%)	29 (1.9%)	140 (2.0%)
Work status							<0.001
Employed	757 (66.3%)	950 (60.1%)	788 (52.8%)	519 (48.7%)	648 (41.9%)	3662 (53.6%)
Jobless †	5 (0.4%)	1 (0.1%)	3 (0.2%)	2 (0.2%)	6 (0.4%)	17 (0.2%)
Unemployed	356 (31.2%)	580 (36.7%)	649 (43.5%)	496 (46.5%)	800 (51.7%)	2881 (42.2%)
Retired	24 (2.1%)	50 (3.2%)	52 (3.5%)	49 (4.6%)	94 (6.1%)	269 (3.9%)
Smoking status							<0.001
Non-smoker	806 (70.6%)	1160 (73.3%)	1143 (76.6%)	853 (79.9%)	1252 (80.8%)	5214 (76.3%)
Smoker	336 (29.4%)	422 (26.7%)	350 (23.4%)	214 (20.1%)	297 (19.2%)	1619 (23.7%)
Frequency of drinking							<0.001
>1 time/month	368 (32.2%)	421 (26.6%)	359 (24.0%)	251 (23.5%)	253 (16.3%)	1652 (24.2%)
<1 time/month	80 (7.0%)	96 (6.1%)	96 (6.4%)	86 (8.1%)	97 (6.3%)	455 (6.7%)
Never	694 (60.8%)	1065 (67.3%)	1038 (69.5%)	730 (68.4%)	1199 (77.4%)	4726 (69.2%)
Physical examination							<0.001
No	745 (65.2%)	956 (60.5%)	842 (56.5%)	554 (52.0%)	781 (50.4%)	3878 (56.8%)
Yes	397 (34.8%)	625 (39.5%)	649 (43.5%)	512 (48.0%)	768 (49.6%)	2951 (43.2%)
Health insurance							0.020
No public health insurance	42 (3.7%)	65 (4.1%)	49 (3.3%)	34 (3.2%)	55 (3.6%)	245 (3.6%)
UEBMI	140 (12.3%)	220 (13.9%)	243 (16.3%)	171 (16.0%)	256 (16.5%)	1030 (15.1%)
URRBMI	169 (14.8%)	203 (12.8%)	200 (13.4%)	124 (11.6%)	184 (11.9%)	880 (12.9%)
URBMI	51 (4.5%)	65 (4.1%)	59 (4.0%)	45 (4.2%)	87 (5.6%)	307 (4.5%)
NRCMS	728 (63.7%)	1006 (63.6%)	926 (62.1%)	673 (63.1%)	936 (60.4%)	4269 (62.5%)
Other *	12 (1.1%)	22 (1.4%)	15 (1.0%)	19 (1.8%)	31 (2.0%)	99 (1.4%)
Socioeconomic status							<0.001
Quartile 1 (lowest)	320 (28.0%)	425 (26.9%)	327 (21.9%)	247 (23.2%)	319 (20.6%)	1638 (24.0%)
Quartile 2	332 (29.1%)	407 (25.7%)	398 (26.7%)	252 (23.6%)	396 (25.5%)	1785 (26.1%)
Quartile 3	257 (22.5%)	398 (25.2%)	410 (27.5%)	295 (27.7%)	399 (25.7%)	1759 (25.7%)
Quartile 4 (highest)	232 (20.3%)	351 (22.2%)	358 (24.0%)	272 (25.5%)	437 (28.2%)	1650 (24.2%)
Area							<0.001
East	450 (39.4%)	539 (34.1%)	494 (33.1%)	307 (28.8%)	365 (23.5%)	2155 (31.5%)
Central	307 (26.9%)	437 (27.6%)	423 (28.3%)	316 (29.6%)	445 (28.7%)	1928 (28.2%)
West	316 (27.6%)	507 (32.0%)	469 (31.4%)	367 (34.4%)	591 (38.1%)	2250 (32.9%)
Northeast	70 (6.1%)	99 (6.3%)	107 (7.2%)	77 (7.2%)	150 (9.7%)	503 (7.4%)
Number of family members							0.002
Population	2.8 (1.5)	2.7 (1.4)	2.7 (1.4)	2.6 (1.4)	2.6 (1.4)	2.7 (1.4)
Impoverished							0.251
No	1037 (90.7%)	1435 (92.6%)	1345 (91.6%)	960 (91.5%)	1375 (90.5%)	6152 (91.7%)
Yes	87 (7.6%)	114 (7.4%)	123 (8.4%)	89 (8.5%)	145 (9.5%)	558 (8.3%)

Data are displayed as mean (standard deviation) or N (%). † Jobless refers to participants who have no ability to work, and unemployed means that participants are able to work but without work opportunities. * Other refers to Government Employee Health Insurance. Abbreviations: SD, standard deviation; NRCMS, New Rural Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical insurance; URBMI, Urban Resident Basic Medical insurance; URRBMI, Urban and Rural Resident Basic Medical Insurance.

shows the baseline characteristics of multimorbidities groups and overall patients with hypertension. Most patients (5693, 83.3%) suffered from multimorbidities. A total of 1143 participants (16.7%) reported having only hypertension, whereas 1582 participants (23.1%) suffered from two chronic diseases, 1493 participants (21.8%) suffered from three chronic diseases, 1067 participants (15.6%) suffered from four chronic diseases, and 1551 participants (22.7%) suffered from ≥5 chronic diseases. There were significant differences (p < 0.05) in age, sex, education, work status, smoking status, frequency of drinking, physical examination, health insurance, socioeconomic group, area, and number of family members between the multimorbidities groups.

3.2. Prevalence and Factors Associated with CHE

We found that 49.8% of households experienced CHE. Figure 2 shows the concentration curve of CHE incidence, which lies above the equality line, with a concentration index of −0.026 (p < 0.05; 95% CI, −0.032 to −0.020), showing that CHE were mainly concentrated among the poor. Moreover, the incidence of CHE increased with the number of chronic diseases. Among participants suffering from hypertension only, the incidence rate of CHE was 35.1%, whereas those for patients with two, three, four, and ≥5 chronic diseases were 45.5%, 50.3%, 57.1%, and 64.7%, respectively.

According to the result of binary logit regression (

Table 2. The correlation between each variable and CHE.

	Coef. (95% CI)	p Value
Number of chronic diseases	0.22 (0.19, 0.24)	<0.001
Age, per 5 years	0.16 (0.13, 0.18)	<0.001
Sex
Male	1 (ref)
Female	0.19 (0.10, 0.29)	0.015
Education
No education and primary school	1 (ref)
Secondary school	−0.43 (−0.53, −0.32)	0.048
College and above	−0.71 (−1.05, −0.36)	0.143
Work status
Employed	1 (ref)
Jobless †	0.05 (−0.91, 1.00)	0.839
Unemployed	0.46 (0.36, 0.56)	<0.001
Retired	0.25 (0.00, 0.50)	0.151
Smoking status
Non-smoker	1 (ref)
Smoker	−0.31 (−0.42, −0.20)	0.014
Frequency of drinking
>1 time/month	1 (ref)
<1 time/month	0.18 (−0.03, 0.39)	0.296
Never	0.50 (0.38, 0.61)	<0.001
Physical examination
No	1 (ref)
Yes	0.19 (0.09, 0.28)	0.074
Health insurance
No basic medical insurance	1 (ref)
UEBMI	−0.23 (−0.50, 0.05)	0.782
URRBMI	0.18 (−0.10, 0.47)	0.005
URBMI	−0.02 (−0.36, 0.31)	0.747
NRCMS	0.12 (−0.14, 0.38)	0.012
Other *	−0.24 (−0.71, 0.23)	0.568
Socioeconomic status
Quartile 1 (lowest)	1 (ref)
Quartile 2	−0.15 (−0.28, −0.02)	0.115
Quartile 3	−0.16 (−0.29, −0.02)	0.039
Quartile 4 (highest)	−0.13 (−0.27, 0.00)	0.161
Area
East	1 (ref)
Central	0.13 (0.01, 0.26)	0.206
West	0.25 (0.13, 0.36)	0.023
Northeast	0.14 (−0.05, 0.34)	0.307
Number of family members	−0.21 (−0.24, −0.17)	<0.001
Impoverished
No	1 (ref)
Yes	0.61 (0.43, 0.79)	<0.001

† Jobless refers to participants who have no ability to work, and unemployed means that participants are able to work but without work opportunities. * Other refers to Government Employee Health Insurance. Abbreviations: CI, confidence interval; NRCMS, New Rural Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical insurance; URBMI, Urban Resident Basic Medical insurance; URRBMI, Urban and Rural Resident Basic Medical Insurance.

), the number of chronic diseases, age, sex, employment status, frequency of alcohol consumption, physical examination, region, and the impoverishment status all had a significant positive impact on the likelihood of CHE, whereas the level of education, smoking status, socioeconomic status, and the number of people in the household all had a significant negative impact on the likelihood of CHE.

Additional chronic diseases correlated with an increased risk of CHE (OR, 1.21; 95% CI, 1.18–1.25;

Table 3. Determinants of catastrophic health expenditure among patients with hypertension.

	Odds Ratio (95% CI)	Robust Standard Error	p Value
Number of chronic diseases	1.21 (1.18, 1.25)	0.019	<0.001
Age, per 5 years	1.09 (1.06, 1.13)	0.017	<0.001
Sex
Male	1 (ref)
Female	0.85 (0.75, 0.97)	0.056	0.015
Education
No education and primary school	1 (ref)
Secondary school	0.88 (0.78, 1.00)	0.056	0.048
College and above	0.75 (0.51, 1.10)	0.143	0.143
Work status
Employed	1 (ref)
Jobless †	0.90 (0.33, 2.48)	0.496	0.839
Unemployed	1.23 (1.09, 1.39)	0.074	<0.001
Retired	1.23 (0.93, 1.64)	0.180	0.151
Smoking status
Non-smoker	1 (ref)
Smoker	0.84 (0.73, 0.97)	0.059	0.014
Frequency of drinking
>1 time/month	1 (ref)
<1 time/month	1.13 (0.90, 1.41)	0.131	0.296
Never	1.33 (1.16, 1.53)	0.093	<0.001
Physical examination
No	1 (ref)
Yes	1.10 (0.99, 1.22)	0.059	0.074
Health insurance
No basic medical insurance	1 (ref)
UEBMI	0.96 (0.70, 1.31)	0.154	0.782
URRBMI	1.55 (1.14, 2.10)	0.245	0.005
URBMI	1.06 (0.74, 1.52)	0.199	0.747
NRCMS	1.42 (1.08, 1.88)	0.205	0.012
Other *	0.86 (0.52, 1.44)	0.232	0.568
Socioeconomic status
Quartile 1 (lowest)	1 (ref)
Quartile 2	0.89 (0.77, 1.03)	0.066	0.115
Quartile 3	0.86 (0.74, 0.99)	0.064	0.039
Quartile 4 (highest)	0.90 (0.77, 1.04)	0.070	0.161
Area
East	1 (ref)
Central	1.09 (0.95, 1.24)	0.073	0.206
West	1.16 (1.02, 1.32)	0.076	0.023
Northeast	1.12 (0.90, 1.38)	0.118	0.307
Number of family members	0.83 (0.80, 0.86)	0.017	<0.001
Impoverished
No	1 (ref)
Yes	1.54 (1.28, 1.87)	0.154	<0.001

† Jobless refers to participants who have no ability to work, and unemployed means that participants are able to work but without work opportunities. * Other refers to Government Employee Health Insurance. Abbreviations: CI, confidence interval; NRCMS, New Rural Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical insurance; URBMI, Urban Resident Basic Medical insurance; URRBMI, Urban and Rural Resident Basic Medical Insurance.

). In addition, the determinants of CHE (p < 0.05) included age, sex, level of education, work status, smoking status, frequency of drinking, health insurance, socioeconomic group, area, number of family members, and the impoverishment status.

The likelihood of CHE increased with age (OR, 1.09; 95% CI, 1.06–1.13) and was lower among women than among men (OR, 0.85; 95% CI, 0.75–0.97). Compared with that among patients with no education and primary school education, the likelihood of CHE was lower among those with a higher level of education (OR, 0.88; 95% CI, 0.78–1.00). The likelihood of CHE was higher among patients who were unemployed than among those who were employed (OR, 1.23; 95% CI, 1.09–1.39). Smokers were associated with a lower probability of experiencing CHE than non-smokers (OR, 0.84; 95% CI, 0.73–0.97). Non-drinkers were associated with a higher probability of experiencing CHE than drinkers (OR, 1.33; 95% CI, 1.16–1.53). URRBMI (OR, 1.55; 95% CI, 1.14–2.10) and NRCMS (OR, 1.42; 95% CI, 1.08–1.88) enrollments were associated with a higher probability of experiencing CHE. The likelihood of CHE was lower among those from higher socioeconomic status than among those with the lowest socioeconomic status (OR, 0.86; 95% CI, 0.74–0.99).

Compared with patients from Eastern China, the CHE risk was higher among those from Western China (OR, 1.16; 95% CI, 1.02–1.32). The probability of experiencing CHE decreased with an increasing number of family members (OR, 0.83; 95% CI, 0.80–0.86). Finally, the likelihood of CHE was higher in impoverished families (OR, 1.54; 95% CI, 1.28–1.87).

3.3. Impact of Age, Work Status and Combinations of Socioeconomic Groups and Health Insurance Schemes on the Relevance of Multimorbidities and CHE

In regression models stratified by age and work status (Figure 3), we found that the probability of experiencing CHE increased with an increase in the number of chronic diseases for the older age group, younger age group, working group, and no-work group.

An increasing number of chronic diseases correlated with an increased risk of CHE among all patients suffering from hypertension who were covered by UEBMI (OR, 1.16; 95% CI, 1.08–1.26; Figure 4), URRBMI (OR, 1.26; 95% CI, 1.16–1.38), URBMI (OR, 1.14; 95% CI, 1.01–1.30), and NRCMS (OR, 1.22; 95% CI, 1.18–1.27). The relationship between multimorbidities and CHE persisted among patients with different insurance types across every socioeconomic status group.

We found a similar relationship of multimorbidities and CHE at the other CHE thresholds using the definition of the World Bank in the sensitivity analyses (Supplemental Table S2). The CHE risk increased by 17% (OR, 1.17; 95% CI, 1.14–1.21) with each additional chronic disease at the threshold of 40%.

4. Discussion

To our knowledge, the study is the first analysis of a population-based survey to assess the distribution of multimorbidities, the prevalence and factors associated with CHE, and the relationship of multimorbidities and CHE among middle- to older-aged adults with hypertension in China. We also examined the impact of combinations of socioeconomic groups and insurance schemes on the relationship of multimorbidities and CHE. We found that patients who suffered from hypertension were sensitive to multimorbidities, and almost half of families experienced CHE. The factors associated with an increase in the odds of CHE were increasing number of chronic diseases, older age, male sex, less education, unemployment, non-smoker, non-drinker, covered by URRBMI or NRCMS, lower socioeconomic status (although sensitivity analysis did not support this finding), residing in western China (the sensitivity analysis also did not support this finding), decreasing number of family members, and living in an impoverished family.

In our study, 83.3% of patients suffering from hypertension had multimorbidities. There are few reports on multimorbidities among patients with hypertension, and some previous studies that also examined patients suffering from hypertension reported a lower prevalence of multimorbidities among them [6,7]. This may be because these studies included adults of all ages; in contrast, we enrolled only middle-aged and elderly patients, and the prevalence of multimorbidities is known to increase with age [31]. In addition, the prevalence of multimorbidities varied between different countries and areas, which depends on diagnoses and the selection of chronic diseases, measurements, and the sampling frame [6,32].

In the base case analysis, 49.8% of families experienced CHE. There is no consensus on the threshold and very few reports on CHE and its inequality among patients with hypertension. According to two reports of adult patients with hypertension, in which a threshold of 10% was used, the incidence rates of CHE were 43.3% and 42%, respectively [33,34]. Five studies using a threshold of 40% estimated that the incidence of CHE ranged from 11.8–46.9% [26,35,36,37,38]. The possible reasons for this discrepancy include variations in sample population recruitment, settings and period of estimation, and definition and calculation of CHE. Additionally, consistent with previous studies [26,36], the concentration index in our study was −0.026, indicating that socioeconomic-related inequalities existed in the distribution of CHE—that is, the poor are more likely to experience CHE.

Our study found that patients suffering from hypertension enrolled in URRBMI and NRCMS had a higher CHE risk, but other insurance schemes also did not protect patients from CHE. Due to the defects of a low financing level, low compensatory benefits level, and poor ability of cost-sharing [39], many studies have also found that people covered by NRCMS have a greater CHE risk [40,41,42]. Additionally, there is evidence that many NRCMS funds do not include outpatient fees, rehabilitation, or long-term care for the elderly [39], which can further explain why middle-aged and elderly patients suffering from hypertension covered by NRCMS had a high CHE risk. Meanwhile, for URRBMI, following the integration of URBMI and NRCMS, compared to inpatient services, outpatient insurance policies still covered very limited diseases and had lower reimbursement ratios in most areas [22]. Therefore, medical expenses related to hypertension and most chronic diseases suitable for outpatient treatment can only be reimbursed after the use of inpatient services; thus, these patients either paid high fees or sought hospitalization [22,41].

Consistent with a previous study [43], we demonstrated that an increasing number of chronic diseases correlates with a higher CHE risk among patients suffering from hypertension. This may be because more health care services are needed by patients suffering from multimorbidities [20] and middle-aged and elderly patients are more sensitive to CHE [24]. In addition, OOP spending among patients suffering from a great number and specific combinations of chronic diseases was very high [44].

A statistically significant interaction between multimorbidities and CHE was observed regardless of the age groups and work status. In addition, an increasing number of chronic diseases had a high probability of experiencing CHE, especially among the younger age group and the working group. As far as we know, this is the first study to examine the effect of multimorbidities on CHE across age groups and work statuses. The functions of the health system are to facilitate access to health care services and prevent illness-associated economic difficulties among residents [45]. Additionally, the important goal of insurance is also to protect the insured from financial difficulties. However, we found that health insurance schemes provide very limited protection for patients suffering from hypertension and multimorbidities. Even among patients covered by UEBMI, which has a generous and comprehensive benefits package, the relationship of multimorbidities and CHE persisted. This suggests that existing insurance schemes have largely failed to achieve their goal and financial hard-ship is still experienced by patients. There is evidence that benefits packages have not improved significantly over the last decade [46]. Additionally, to narrow the gap between basic medical insurance schemes, URBMI and NRCMS were integrated into URRBMI [47]; however, this health insurance integration did not seem to achieve the expected effect, with an increasing number of chronic diseases seen among patients suffering from hypertension.

For the same insurance type, we did not see an obvious reduction in the impact of multimorbidities on the CHE risk with an increase in socioeconomic status, indicating that wealthier families have a similar probability of experiencing CHE as poor families with each additional chronic disease. Although there is a common consensus that a family’s strong financial capacity can reduce its probability of experiencing CHE [48], affluent people are more likely to utilize private, high-level health services, making them more vulnerable to CHE [49,50]. In addition, many previous studies also found that demographic variables, such as the number of family members and the socioeconomic group, could alleviate the CHE risk, whereas the type of health insurance did not play a protective role [22,23,46], which suggests that the CHE risk might not be related to the social system but instead to the nature of individuals or families. The social system plays an important role in making up for deficits in families; unfortunately, our study suggested that the protective effect of current health insurance schemes is very limited.

This population-based study is the first analysis of a nationally representative survey to examine the impact of multimorbidities on CHE among middle-aged and older Chinese people suffering from hypertension, and our study provides new evidence for the formulation of targeted policies and interventions to cope with the growing burden of the multimorbidities of hypertension. However, there are some limitations to our study. First, the data on chronic diseases included in our study and the household expenditure were based on self-reported measures, which may be affected by recall bias. Second, this study had a cross-sectional design, which could not establish causal relationships. Third, we used a method of simply counting the number of chronic diseases to estimate multimorbidities without accounting for the correlation between diseases and their severity. Finally, the types of chronic diseases included in the CHARLS questionnaire were limited; thus, further studies examining the prevalence of multimorbidities should expand the types of diseases.

5. Conclusions

Multimorbidities is common in this study of the middle-to-older-aged hypertensive population, and the increasing number of chronic diseases increases CHE risk. The number of chronic diseases, age, unemployment, not drinking, enrollment in URRBMI and NRCMS, and living in an impoverished family are risk factors for experiencing CHE, whereas female sex, higher educational levels, smoking, and the number of family members are protective factors against CHE. Participants aged <60 years who work have a relatively higher probability of experiencing CHE with the increasing number of chronic diseases. In addition, the impact of multimorbidities on CHE persists regardless of health insurance schemes and socioeconomic status. Current health insurance schemes in China do not provide adequate protection for the insured; thus, we look forward to further expansion of benefit packages to guard against financial risks in the future.