Determinants of catastrophic health expenditure: Evidence from a patient survey at public hospitals in the Northern midlands and mountainous areas,Vietnam

Abstract

Background

Catastrophic health expenditure (CHE) represents a critical barrier to universal health coverage, often pushing households into poverty and deepening financial hardship for vulnerable populations. This study examines determinants of CHE among patients attending public hospitals in the Northern Midlands and Mountainous areas of Vietnam.

Design and methods

A cross-sectional study was conducted from July to December 2021 among 706 patients recruited via multi-stage stratified sampling from 10 public hospitals across five provinces. Data were collected through structured face-to-face interviews. CHE was defined as out-of-pocket health expenditure exceeding 40% of non-food expenditure. Binary logistic regression was employed to identify determinants of CHE.

Findings

Among 706 participants, 147 (20.82%) experienced CHE. CHE cases were concentrated in the poorest income quintile (34.69%) and lowest in the richest (5.44%). Patients experiencing CHE were also disproportionately represented among those diagnosed with cancer (28.57% vs. 5.19% in the non-CHE group), reported higher rates of self-treatment (46.94%), and had more frequent inpatient visits. Key factors positively associated with CHE included cancer (OR = 6.438), self-treatment (OR = 2.798), treatment at provincial hospitals (OR = 2.099), inpatient visit frequency (OR = 1.160), and greater distance to health facilities. Factors negatively associated with CHE included health insurance coverage (OR = 0.579), Kinh ethnicity (OR = 0.319), and higher satisfaction with the availability (OR = 0.594) and quality of medicines (OR = 0.565).

Conclusions

The findings highlight the need to strengthen health insurance coverage for vulnerable households, improve district-level services, and reinforce primary healthcare, with relevance for other resource-constrained settings.

Keywords

catastrophic health expenditure cross-sectional studies hospitals patient satisfaction Vietnam

1. Introduction

The World Health Organization (WHO) promotes Universal Health Coverage (UHC) as a central strategy for ensuring access to essential health services while reducing reliance on out-of-pocket (OOP) payments through prepayment mechanisms such as health insurance.¹ A core objective of UHC is to protect households from financial hardship associated with healthcare utilization. Despite substantial progress, high OOP spending remains a major challenge in many low- and middle-income countries (LMICs), where it continues to undermine financial protection. Globally, an estimated 100 million people are pushed into poverty each year, and approximately 150 million face catastrophic health expenditure (CHE) due to excessive OOP payments.^2–4

OOP health expenditures refer to direct payments made by individuals for healthcare services that are not reimbursed by insurance schemes. Such expenditures are considered catastrophic when they exceed a defined threshold of household resources and threaten a household’s standard of living. When these payments push households below the poverty line or further impoverish those who are already poor, they are classified as impoverishing health expenditures. These indicators are widely used to monitor financial protection and to assess progress toward the health-related Sustainable Development Goals (SDGs).^3,5

Although the WHO recommends limiting OOP payments to 15–20% of total health spending, financial protection remains a global challenge. Wagstaff et al. (2018a) estimated that in 2010, over 800 million people experienced CHE at the 10% consumption threshold, and that the global incidence had increased over the preceding decade.⁶ This burden is disproportionately concentrated in Asia and Africa, where OOP spending in low- and middle-income countries (LMICs) accounts for 20%–60% of total health expenditure, which is substantially higher than in high-income countries.^2,6 As a result, high medical costs continue to push tens of millions of individuals into extreme poverty each year.^3,7

Beyond its financial implications, CHE reflects broader challenges related to equity, access, and health system performance. At the household level, CHE can lead to delayed care, reduced consumption of essential goods, and long-term impoverishment. At the system level, persistent CHE signals gaps in financial protection and inefficiencies in service delivery, particularly in regions facing geographic and socioeconomic barriers. Addressing CHE is therefore central to achieving equitable and sustainable progress toward universal health coverage.

Vietnam has demonstrated a strong commitment to UHC. Social health insurance (SHI) coverage increased from 44% in 2008 to approximately 91% by 2022.⁸ Despite this progress, the healthcare system remains heavily hospital-centered, with hospitals accounting for about 73% of total health expenditure. Patients frequently bypass lower-level facilities in favor of provincial or central hospitals, contributing to overcrowding and service concentration. In addition, the implementation of hospital financial autonomy policies, combined with a rapidly aging population, has increased financial and operational pressures on public hospitals. These dynamics raise concerns about equity and efficiency. They may also inadvertently shift a greater share of healthcare costs onto patients, even under high insurance coverage.^8,9

The Northern Midlands and Mountainous areas of Vietnam comprise 14 provinces and are characterized by persistent socioeconomic disadvantages. Despite relatively strong aggregate economic growth between 2014 and 2021, average per capita income in the region remains below the national level. Poverty incidence also remains high in many localities, with several provinces reporting poverty rates well above the national average.¹⁰ Healthcare utilization is heavily concentrated in public hospitals, particularly for inpatient services, among both urban and rural populations. Although nominal health insurance coverage in the region is high, effective utilization remains limited, especially for outpatient care. Only about 32.5% to 41.1% of insured individuals use their health insurance cards for outpatient services, particularly among poorer households. Consequently, the incidence of CHE in the region remains elevated, reaching 4.6% in 2020 above the national average of 3.3%, and disproportionately affecting low-income households.¹¹

Many studies worldwide and in Vietnam have examined catastrophic medical costs.^11–20 However, most studies in Vietnam rely on secondary data such as the Vietnam Household Living Standards Survey (VHLSS), which lacks detailed clinical information and patient-level perspectives. Beyond financial and demographic determinants, recent international research has begun to incorporate subjective well-being and satisfaction into CHE analyses. For example, Chen et al. (2023) identified life satisfaction as an enabling factor in Andersen’s behavioral model and showed that higher life satisfaction reduced the odds of CHE among older adults in China.²¹ Kim and Hong (2015) demonstrated that Korean households experiencing CHE reported significantly lower levels of life satisfaction, highlighting the link between financial burden and subjective well-being.²² In the context of severe illness, Chino et al. (2014) found that cancer patients reporting heavier financial burdens were significantly less satisfied overall, particularly with technical quality and the financial aspects of care.²³ Similarly, Kang et al. (2018) showed that individuals experiencing CHE reported significantly lower health-related quality of life, with the negative impact being especially pronounced among those with multiple chronic conditions.²⁴ These studies underscore the importance of integrating satisfaction-related variables when analyzing CHE.

Although this emerging literature highlights potential links between subjective experiences and financial hardship, evidence remains limited on how disease type, healthcare utilization patterns, and patient-reported experiences jointly influence CHE risk. This gap is particularly important in socioeconomically disadvantaged and remote regions of Vietnam. Accordingly, this study aims to identify the socioeconomic, clinical, healthcare utilization, and patient-reported factors associated with CHE among patients attending public hospitals in the Northern Midlands and Mountainous areas of Vietnam.

In this context, the present study provides patient-level evidence on the incidence and determinants of CHE in one of Vietnam’s most disadvantaged regions. It focuses on the Northern Midlands and Mountainous areas, where geographic and economic barriers may constrain access to timely and appropriate care. The study makes three contributions. First, it uses primary patient survey data collected across multiple provinces. Second, it accounts for disease type when examining the determinants of CHE. Third, it incorporates patient-reported satisfaction with medicines and services to assess whether perceived service experiences are associated with CHE risk. Taken together, these contributions provide policy-relevant insights for strengthening financial protection and advancing progress toward UHC in Vietnam and comparable LMIC contexts.

2. Methods

2.1. Data collection and sampling

A cross-sectional study was conducted using a multi-stage sampling strategy to select hospitals and patients. In the first stage, five provinces in the Northern Midlands and Mountainous areas of Vietnam (Thai Nguyen, Bac Giang, Bac Kan, Yen Bai, and Phu Tho) were purposively selected to reflect the socioeconomic and ethnic diversity of the region. Within each province, one provincial hospital and one district hospital were randomly selected from the list of public hospitals provided by the provincial health authorities, resulting in a total of ten hospitals included in the study. In the second stage, patients were recruited using stratified sampling to ensure representation across hospital levels (provincial versus district), patient types (inpatient versus outpatient), and clinical departments (such as internal medicine, surgery, pediatrics, and obstetrics and gynecology). Within these strata, eligible patients were identified through hospital records during the survey period, and investigators approached them directly to invite them to participate.

The sample size was estimated using Cochran’s formula for determining a representative sample for proportions²⁵:

n = z^{2} * \frac{p (1 - p)}{ε^{2}}

where n is the required sample size, p is the expected prevalence of CHE, Z is the standard normal deviate corresponding to the desired confidence level (1.96 for 95%), and ε is the desired margin of error. The prevalence rate (p) was based on the findings of Wagstaff et al. (2018a),⁶ which reported the CHE rate in Vietnam as 10% ± 2.22%. Substituting these values into the formula gave:

n = {1.96}^{2} * \frac{0, 1 (1 - 0, 1)}{{0, 0222}^{2}} = 702

The Cochran-based estimate was interpreted as the required number of completed and valid interviews. Given the on-site recruitment process, refusals or ineligible cases were replaced by the next eligible patient within the same sampling stratum, defined by hospital level, patient type, and clinical department, until the target number of completed questionnaires was obtained in each province. This approach was used to preserve the intended sample structure across the selected hospitals.

To ensure balanced geographic coverage, we surveyed 141 patients per province, resulting in a total of 706 respondents. For the sample size calculation, we used a conservative prevalence estimate of 10% based on Wagstaff et al. (2018a).⁶ In the absence of specific regional data for the 40% capacity-to-pay threshold, this approach was deemed appropriate, as using a higher prevalence estimate yields a larger sample size than strictly required, thereby strengthening the validity of the analysis.

Inclusion criteria required participants to be at least 18 years of age, to have utilized healthcare services at the study hospitals during the survey period, and to provide informed consent. For inpatients, inclusion was restricted to those with a hospital stay of at least three days. This criterion was established to ensure sufficient exposure to inpatient care processes (e.g., diagnostics, medication regimens, and facility use) to allow for reliable reporting of costs and satisfaction; and a minimum level of clinical severity and resource utilization, distinguishing substantial inpatient episodes from short-term observation stays. In contrast, outpatients were interviewed immediately following their consultation to minimize recall bias, given that outpatient encounters are typically concluded within a single visit. Participants were excluded if they were under 18 years of age, declined or were unable to provide informed consent, or were too clinically unstable to participate in the interview. The survey was conducted between July and December 2021.

Data were collected exclusively through structured face-to-face interviews using a predesigned questionnaire. The structured questionnaire was developed based on a review of relevant literature^11–24,26 and adapted from the conceptual framework of Kumara and Samaratunge (2016) to reflect the specific socioeconomic context of the region.²⁷ To ensure standardization and consistency in data collection, all interviewers received comprehensive training on the survey protocol and interview techniques prior to fieldwork. The instrument covered four main components: socio-demographic profiles (age, gender, ethnicity, marital status, household size, household composition, residential area, and distance to the nearest health facility), household socioeconomic conditions (income, household expenditure patterns, and OOP expenditures for outpatient and inpatient services), health-related information (self-reported illness, presence of chronic diseases, healthcare utilization patterns, and health insurance coverage), and patient satisfaction with the availability and quality of medicines and healthcare services. Before the main survey, the questionnaire was pilot-tested with a small group of patients to assess the clarity, feasibility, wording, sequencing, and appropriateness of the questions in the local context. Based on participants’ feedback, minor revisions were made to improve the clarity, comprehensibility, and contextual relevance of the questionnaire before formal data collection. For the patient satisfaction items measured on a 5-point Likert scale, internal consistency was also assessed using Cronbach’s alpha, which yielded a value of 0.75, indicating acceptable reliability.

Expenditure data were obtained entirely from self-reported information, as hospital billing records were not consistently available across facilities. To minimize recall bias, respondents were encouraged to cross-check their responses with any receipts or payment records they had retained. Only direct medical costs were considered, defined as all OOP payments made by patients at the hospital, including consultation fees, diagnostic tests, medical procedures, surgical fees, and pharmaceuticals prescribed by doctors during the previous 12 months. Indirect costs such as transportation, accommodation, and income lost were excluded.

Inpatient and outpatient costs were aggregated to estimate the total annual household health burden. Total direct medical costs were computed as the sum of OOP payments over the last 12 months. Although shorter recall periods are sometimes recommended to minimize bias for outpatient care, a 12-month period was employed to align with standard CHE measurement methodologies and to ensure capture of infrequent but high-cost episodes.^6,28 Following the actual food spending method developed by the World Bank, as reviewed by Cylus et al. (2018), a household was classified as experiencing CHE when OOP health expenditures exceeded 40% of total household expenditure net of actual food spending.²⁹ This threshold is widely used in the financial protection literature and allows comparison with previous studies of CHE in Vietnam.^11,13,14 It was considered appropriate for the present study because it reflects household capacity to pay after basic subsistence needs have been met, which is particularly relevant in a socioeconomically disadvantaged region.^10,11 Although alternative thresholds have also been used in the literature,⁶ the 40% criterion was adopted here to ensure comparability with commonly used approaches to CHE measurement.⁴

2.2. Ethical approval

The research protocol was reviewed and approved for scientific and ethical compliance by Thai Nguyen University of Economics and Business Administration, Vietnam, as documented in the university’s official approval letter (Approval No. 68/ĐHKT&QTKD-KHCN, dated 21 May 2021). Participation was entirely voluntary, and written informed consent was obtained from all participants prior to data collection.

2.3. Measurements and variables

The study employed a Binary Logistic Regression model to identify the factors influencing CHE in the Northern Midlands and Mountainous areas. The following equation represents the model:

L o g i t Y = \log (\frac{Y}{1 - Y}) = β_{0} + β X_{i} + e

where Y represents the probability of a household experiencing CHE. Specifically, Y=1 if the household experiences CHE, and Y=0 otherwise. CHE for patients in hospitals in the Northern Midlands and Mountainous areas is defined as medical examination and treatment costs exceeding 40% of household expenditure net of food spending. Total direct medical costs were calculated as the sum of inpatient and outpatient OOP expenditures over the 12 months preceding the interview, consistent with the standard construction of CHE, which assesses total annual OOP health spending relative to household resources.

X_{i}

denotes the independent variables. The independent variables were chosen based on a thorough review of previous studies.^{2,4,11–24,26,27,30–34}

To guide the selection and classification of explanatory variables, a conceptual framework was developed (Figure 1). This framework was adapted from the conceptual framework of Kumara and Samaratunge (2016), which distinguishes between demand-side and supply-side determinants of financial burden.²⁷

Figure 1.

Conceptual framework.

The measurement of independent variables is summarized as follows. Patient age was measured in years, while household size was recorded as the total number of household members and distance to the nearest health facility was measured in kilometres. Binary variables included gender, ethnicity, marital status, health insurance coverage, visits to private facilities, area of residence, and the presence of children under six years old or elderly household members (aged 60+). Self-treatment was defined as a binary variable, coded as 1 if the patient reported purchasing medicines or medical equipment on their own, without a doctor’s prescription or professional medical consultation, prior to the first hospital visit for the current illness, and 0 otherwise. Household economic status was classified into five income quintiles, ranging from the poorest (Q1) to the richest (Q5). Health-related variables included disease categories, healthcare utilization intensity (number of visits in the past 12 months), and level of health facility (district versus provincial). Finally, patient satisfaction was measured using 10 item-level indicators based on the conceptual dimensions of Kumara and Samaratunge (2016) and the empirical determinants identified by Akthar et al. (2023).^26,27 Each item was assessed on a 5-point Likert scale ranging from 1 (“very dissatisfied”) to 5 (“very satisfied”). These items were included in the regression analysis as separate indicators. Detailed descriptions of the dimensions and corresponding survey items are provided in Appendix A.

The reporting of this study conforms to the STROBES statement guidelines for cross-sectional studies.³⁵

3. Results

3.1. Descriptive results

Figure 2 illustrates the distribution of OOP health expenditure relative to household capacity to pay, measured as non-food household expenditure, among patients and their households in the study population. The largest proportion of patients (30.30%) allocated between 5% and 10% of their household non-food budget to healthcare, while smaller shares spent between 10-15% or 15-25%. Notably, more than one-fifth of patients (20.82%) incurred OOP health expenditures exceeding 40% of non-food spending during the 12 months preceding the interview, indicating a substantial burden of CHE at the household level in this patient-based sample.

Figure 2.

Distribution of out-of-pocket health expenditure as a share of non-food household expenditure Source: Authors’ calculation based on survey data.

The social-demographic profile of patients surveyed at public hospitals in the Northern Midlands and Mountainous areas is detailed in Table 1. Of the patients surveyed, 147 experienced CHE, representing 20.82%, while 559 patients did not experience CHE, accounting for 79.18%.

Table 1.

Social-demographic profile of patients.

	Total (n = 706)	CHE = 1 (n = 147)	CHE = 0 (n = 559)	p-value
Age (years)	53.95	54.57	52.79	0.082
Gender (male)	331 (46.88%)	81 (55.10%)	250 (44.72%)	0.025
Ethnicity (Kinh)	479 (67.85%)	92 (62.59%)	387 (69.23%)	0.125
Marital status (Married)	563 (79.75%)	110 (74.83%)	453 (81.04%)	0.094
Household size (Number of household members)	4.53	4.38	4.57	0.285
Household with children (under 6 years)	505 (71.53%)	113 (76.87%)	392 (70.13%)	0.085
Household with elderly (age 60+)	574 (81.30%)	115 (78.23%)	459 (82.11%)	0.283
Distance from home to the nearest health facility, km	24.54	37.71	21.08	0.000
Settlement (Urban)	194 (27.48%)	25 (17.01%)	169 (30.23%)	0.001
Household income (million VND/year)	181.27	122.31	196.25	0.000
Minimum–Maximum (million VND/year)	(42.54 – 715.21)	(56.97 – 597.56)	(42.54 – 715.21)
Economic status (Income Quintile)
Q1 (poorest)	166 (23.51%)	51 (34.69%)	115 (20.57%)	0.000
Q2	145 (20.54%)	39 (26.53%)	106 (18.96%)	0.000
Q3	197 (27.90%)	37 (25.17%)	160 (28.62%)	0.000
Q4	98 (13.88%)	12 (8.16%)	86 (15.38%)	0.000
Q5 (richest)	100 (14.16%)	8 (5.44%)	92 (16.46%)	0.000
Health Insurance status	599 (84.84%)	124 (84.35%)	475 (84.97%)	0.900
Type of disease
Acute diseases	236 (33.43%)	41 (27.89%)	195 (34.88%)	0.000
High Blood Pressure	79 (11.19%)	8 (5.44%)	71 (12.70%)	0.000
Cardiovascular	108 (15.30%)	14 (9.52%)	94 (16.82%)	0.000
Diabetes	55 (7.79%)	11 (7.48%)	44 (7.87%)	0.000
Chronic Pulmonary/Asthma	34 (4.82%)	13 (8.84%)	21 (3.76%)	0.000
Cancer	71 (10.06%)	42 (28.57%)	29 (5.19%)	0.000
Hemodialysis	28 (3.97%)	14 (9.52%)	14 (2.50%)	0.000
Diseases related to the nervous system	95 (13.46%)	4 (2.72%)	91 (16.28%)	0.000
Number of outpatient visits in the last 12 months	5.40	7.70	4.79	0.026
Number of inpatient visits in the last 12 months	2.49	4.73	1.90	0.000
Self-treatment	259 (36.69%)	69 (46.94%)	190 (33.99%)	0.004
Visiting private health facility	237 (33.57%)	51 (34.69%)	186 (33.27%)	0.746
Type of hospital (district hospital)	451 (63.88%)	110 (74.83%)	341 (61.00%)	0.003

Source: Authors’ calculation based on survey data. Values are presented as means for continuous variables and n (%) for categorical variables.

Table 1 summarizes the socio-demographic and contextual characteristics of the 706 participants. The mean age was 53.95 years, 46.88% of participants were male, and 67.85% belonged to the Kinh ethnic group. Most patients were recruited from district hospitals (63.88%) and lived in rural areas (72.52%). The average annual household income was 181.3 million VND, with the largest proportion of households in the third income quintile (27.90%). Participants were predominantly married (79.75%), with an average household size of 4.53 members; many households included elderly members (81.30%) and children under six years of age (71.53%). Health insurance coverage was high (84.84%). Acute diseases were the most common diagnoses (33.43%), followed by cardiovascular conditions (15.30%). On average, participants lived 24.54 km from the nearest health facility and reported 5.40 outpatient visits and 2.49 inpatient admissions in the previous year; 36.69% also reported self-treatment.

The results in Table 1 show that patients who experienced CHE in the 12 months prior to the survey were, on average, older than those who did not (54.57 vs. 52.79 years). The CHE group also had a higher proportion of male patients (55.10% vs. 44.72%) and households with children under six years of age (76.87% vs. 70.13%). In contrast, a smaller proportion of patients in the CHE group lived in urban areas compared with the non-CHE group (17.01% vs. 30.23%). Geographic accessibility and hospital level also differed between the two groups. Patients with CHE lived substantially farther from the nearest health facility than those without CHE (37.71 km vs. 21.08 km), and the type of hospital attended was also associated with CHE status. Marked socioeconomic differences were also observed. CHE was more concentrated among poorer households: 34.69% of patients in the CHE group belonged to the poorest quintile, compared with only 5.44% in the richest quintile. In the non-CHE group, the corresponding shares were 20.57% and 16.46%, respectively. Figure 3 indicates that the income disparity between the richest and poorest groups was substantially larger in the CHE group than in the non-CHE group, at 29.25 and 4.11 percentage points, respectively. By contrast, health insurance coverage was similarly high in both groups, with only a marginal difference between patients with CHE and those without CHE (84.35% vs. 84.97%).

Figure 3.

Difference between the poorest and richest income quintiles by CHE status, %. Source: Authors’ calculation based on survey data.

The prevalence of acute illness was lower among patients experiencing CHE than among those without CHE (27.89% vs. 34.88%). Similarly, conditions such as cardiovascular disease, hypertension, diabetes, and neurological disorders were less common in the CHE group. In contrast, cancer and hemodialysis were more prevalent among CHE patients (28.57% vs. 5.19% and 9.52% vs. 2.50%, respectively). Patients with CHE also reported more frequent outpatient visits and inpatient admissions during the past 12 months (7.70 vs. 4.79 and 4.73 vs. 1.90 times per year, respectively). Additionally, the rate of purchasing medicines or medical equipment for self-treatment was higher among CHE patients, at 46.94% compared with 33.99% among non-CHE patients (Table 1).

Detailed descriptive statistics for patient satisfaction indicators are presented in Appendix B. Overall, satisfaction levels were relatively similar between the CHE and non-CHE groups, although some differences were observed for diagnosis and treatment results and administrative procedures.

To provide further insight into the financial burden across different socio-economic profiles, Table 2 presents the distribution of OOP spending, the OOP share in total household expenditure net of actual food spending, and the incidence of catastrophic expenditure stratified by residence and income quintiles.

Table 2.

Urban–rural distribution of out-of-pocket spending and catastrophic health expenditure by income quintile.

Variable	Rural (n =512)						Urban (n =194)
Variable	Q1	Q2	Q3	Q4	Q5	Total	Q1	Q2	Q3	Q4	Q5	Total
Means of OOP (000′ VND)	5640.56	6930.51	6211.93	8196.60	7443.09	6601.38	3449.44	8042.50	6648.26	5590.50	8951.96	6807.67
OOP payments as a share of non-food household expenditure (%)	77.49	58.37	40.19	23.43	13.54	50.43	72.80	28.18	19.69	20.19	12.13	30.95
Households with catastrophic health expenditure (%)	33.57	27.64	20.00	14.29	8.33	23.83	22.73	15.38	15.38	9.52	7.69	12.89

The data reveal a clear contrast between absolute spending levels and financial burden. While higher-income households (Q5) in both rural and urban areas tended to incur absolute OOP expenditures that were higher than or comparable to those of poorer households (Q1), the relative financial burden was disproportionately concentrated among the poor. In rural areas, households in the lowest income quintile spent, on average, 77.49% of their non-food expenditure on healthcare, compared with only 13.54% among the richest quintile. A similar pattern of vertical inequity was observed in urban settings, where the OOP share among the poorest households (72.80%) was substantially higher than that of the richest group (12.13%). Correspondingly, the incidence of CHE was highest among the poorest households and declined progressively with rising income, highlighting pronounced socioeconomic disparities in financial protection.

Taken together, the descriptive findings indicate that the burden of CHE in the Northern Midlands and Mountainous areas is unequally distributed. It is disproportionately concentrated among socioeconomically vulnerable populations, particularly those facing low income, severe chronic conditions, geographical barriers, and intensive healthcare needs. Notably, the observation that CHE persists despite high health insurance coverage suggests that while enrollment is widespread, the depth of financial protection could be further strengthened. This is particularly relevant for vulnerable groups requiring repeated inpatient care or specialized treatment at higher-level facilities. These descriptive patterns provide an important empirical foundation for the multivariate analysis that follows, highlighting the interaction between socioeconomic disadvantage, disease severity, and health system characteristics in shaping financial risk.

3.2. Determinants of catastrophic health expenditure

Table 3 presents the results of the binary logistic regression model on factors influencing CHE among patients visiting health facilities in the Northern Midlands and Mountainous areas. Because patients were nested within hospitals, the model was estimated using cluster-robust standard errors at the hospital level to account for potential intra-hospital correlation. The Hosmer–Lemeshow goodness-of-fit test yielded a p-value of 0.470, indicating that the model fits the data adequately. In addition, the pseudo R² was 0.2677, providing further information on the overall performance of the estimated model. The multicollinearity test results showed that most of the independent variables had a variance inflation factor (VIF) of less than 2, while the satisfaction-related variables ranged from 2.45 to 3.58. The average VIF was 1.95, suggesting that multicollinearity was not a serious concern in the estimated model (Appendix C).

Table 3.

Determinants of catastrophic health expenditure for patients at health facilities in Northern Midlands and Mountainous areas of Vietnam.

Variable	OR	95% CI
Variable	OR	LL	UL
Age	1.014*	0.999	1.029
Gender (Reference: Female)	1.435*	0.943	2.185
Ethnicity (Reference: Minority)	0.319***	0.195	0.523
Marital status (Reference: Unmarried)	1.022	0.591	1.766
Household size	1.001	0.882	1.135
Household with children (under 6 years) (Reference: No)	0.765	0.439	1.332
Household with elderly (age 60+) (Reference: No)	1.043	0.540	2.018
Distance from home to the nearest health facility, km	1.006**	1.000	1.012
Settlement (Reference: Rural)	0.753	0.417	1.360
Economic status (Income Quintile) (Reference: Q1 poorest)
Q2	0.483**	0.272	0.858
Q3	0.506**	0.290	0.882
Q4	0.484*	0.219	1.068
Q5 (richest)	0.348**	0.145	0.838
Health Insurance status (Reference: No)	0.579*	0.304	1.013
Type of disease (Reference: Acute diseases)
High Blood Pressure	0.526	0.210	1.319
Cardiovascular	0.851	0.388	1.866
Diabetes	0.990	0.415	2.365
Chronic Pulmonary/Asthma	1.423	0.543	3.731
Cancer	6.438***	2.956	14.022
Hemodialysis	4.260***	1.518	11.956
Diseases related to the nervous system	0.179***	0.058	0.551
Number of outpatient visits in the last 12 months	1.006	0.992	1.020
Number of inpatient visits in the last 12 months	1.160***	1.079	1.248
Self-treatment (Reference: No)	2.798***	1.698	4.612
Visiting private health facility (Reference: No)	0.905	0.545	1.503
Type of hospital (Reference: district hospital)	2.099***	1.274	3.456
Satisfaction level with medical supplies and equipment	1.284	0.660	2.496
Satisfaction level with the availability of medicines	0.594*	0.346	1.019
Satisfaction level with the availability of skilled professionals	1.145	0.613	2.139
Satisfaction level with diagnosis and treatment results of doctors	1.597	0.622	2.765
Satisfaction level with doctors prescribing appropriate medications	0.890	0.463	1.708
Satisfaction level with the quality of prescription drugs	0.565*	0.292	1.093
Satisfaction level with explanation, advice and health care guidance from doctors	0.877	0.476	1.613
Satisfaction level with the attitude in communication, gestures and interactions of doctors and nurses	0.831	0.445	1.550
Satisfaction level with waiting time	1.420	0.746	2.704
Satisfaction level with administrative procedures	1.566	0.782	3.137
Prob > chi2	0.000
Pseudo-R2	0.2677
Hosmer-Lemeshow (Prob) Test	0.470
Number of observations	706

Source: Authors’ calculation based on survey data. *, ** and *** show significance levels at 10%, 5% and 1%, respectively.

The results indicate that several factors were associated with CHE. These included age, gender, ethnicity, distance from home to the nearest health facility, household economic status, health insurance status, disease category, number of inpatient visits, self-treatment, type of hospital, and satisfaction with the availability and quality of prescribed drugs (Table 3).

For every one-year increase in age, the odds of experiencing CHE increased by 1.014 times (OR = 1.014; 95% CI: 0.999–1.029). Male patients were more likely to experience CHE than female patients (OR = 1.435; 95% CI: 0.943–2.185). Patients of Kinh ethnicity were less likely to experience CHE than ethnic minority patients (OR = 0.319; 95% CI: 0.195–0.523). Distance to the nearest health facility was positively associated with CHE, with each additional kilometre slightly increasing the odds (OR = 1.006; 95% CI: 1.000–1.012).

Household economic status showed a protective factor. Compared with patients from the poorest households, those in the second, third, fourth, and richest quintiles had lower odds of CHE, with ORs of 0.483 (95% CI: 0.272–0.858), 0.506 (95% CI: 0.290–0.882), 0.484 (95% CI: 0.219–1.068), and 0.348 (95% CI: 0.145–0.838), respectively. Similarly, health insurance played a protective role in reducing the risk of CHE associated with medical examination and treatment. Compared with patients without health insurance, those with health insurance were less likely to experience CHE, with OR = 0.579; 95% CI: 0.304-1.013.

In the regression model, disease status was represented by specific disease categories, with acute diseases as the reference group. The results indicated that patients with diseases related to the nervous system were significantly less likely to experience CHE than those with acute diseases, with OR = 0.179; 95% CI: 0.058-0.551. In contrast, patients with cancer and those undergoing hemodialysis were found to be at a significantly higher risk of experiencing CHE compared with patients with acute diseases. The odds ratios for these groups were OR = 6.438 (95% CI: 2.956-14.022) for cancer patients and OR = 4.260 (95% CI: 1.518-11.956) for those receiving hemodialysis.

Healthcare utilization was also important. The frequency of inpatient treatments significantly increased the likelihood of patients experiencing CHE. Specifically, for each additional treatment received in the 12 months prior to the survey, the odds of experiencing CHE increased by 1.16 (OR = 1.160; 95% CI: 1.079–1.248). Moreover, patients who reported self-medicating or purchasing medical equipment for their illness were at a much higher risk of experiencing CHE than those who did not (OR = 2.798; 95% CI: 1.698–4.612). Additionally, our findings indicate that patients who sought treatment at provincial hospitals had a higher risk of experiencing CHE compared with those who received care at the district level, with OR = 2.099; 95% CI: 1.274-3.456.

Among the patient satisfaction variables, two variables were significantly associated with CHE. The estimation results showed that higher satisfaction with the availability of medicines was associated with lower odds of CHE (OR = 0.594; 95% CI: 0.346–1.019), and higher satisfaction with the quality of prescribed drugs was also associated with lower odds of CHE (OR = 0.565; 95% CI: 0.292–1.093).

4. Discussion

This study provides patient-level evidence on the determinants of CHE among patients attending public hospitals in the Northern Midlands and Mountainous areas of Vietnam. The findings show that CHE was more likely among older patients, men, ethnic minorities, those living farther from health facilities, poorer households, patients without health insurance, those with cancer or those undergoing hemodialysis, those with more inpatient admissions, those reporting self-treatment, and those treated at provincial hospitals. By contrast, Kinh ethnicity, higher household income, health insurance coverage, and higher satisfaction with the availability and quality of medicines were associated with lower odds of CHE.

The positive association between age and CHE suggests that older patients may face greater financial vulnerability because they often require more frequent healthcare and more intensive treatment. This result is consistent with the findings of Bedado et al. (2022).² Male patients also showed a higher risk of CHE than female patients. This can be explained by the fact that men may be more exposed to risk factors such as alcohol and tobacco use, which increase the likelihood of developing severe chronic diseases. It is important to note that this finding differs from the findings of Thuong et al. (2022), who found that female-headed households were more likely to experience CHE.¹¹ This discrepancy likely reflects the differences in the unit of analysis. The earlier study focused on household heads, whereas the present study focuses on individual patients. At the patient level, our finding aligns with the results of Fu and Chen (2022),³⁶ who also reported a higher probability of CHE among male patients.

Kinh ethnicity appeared to be protective relative to ethnic minority status. This may be explained by several contextual factors. Ethnic minorities often live in remote areas with difficult socio-economic conditions and may need to travel long distances to access medical care. These areas may also lack adequate medical infrastructure, including facilities, equipment, medical personnel, and essential medicines. Such limitations can affect both the quality and effectiveness of care. As a result, when patients from these communities seek treatment at higher-level facilities, such as district or provincial hospitals, their conditions may already have worsened, resulting in higher treatment costs and a greater likelihood of CHE.

Distance to the nearest health facility was positively associated with CHE. This finding is consistent with Masiye et al. (2016) who reported that greater distance to health facilities increased the odds of CHE for households in Zambia by 1.015.³⁷ Distance is an important indicator of healthcare accessibility and may shape treatment-seeking behavior. When facilities are far away, patients may delay seeking care and present with more severe conditions. Consequently, they may require more intensive treatment, longer hospital stays, and more expensive medicines. This can substantially increase direct medical expenditures and the likelihood of CHE. In mountainous and geographically isolated settings, this finding underscores that improving access to care is not only a matter of service availability but also of financial protection.

Household economic status was an important determinant of CHE, with wealth acting as a protective factor for patients and their households. These findings align with previous studies, which also demonstrated that wealthier households are less vulnerable to CHE.^31,32,38 The sharp contrast between the poorest and richest quintiles underscores the continuing importance of income-related inequality in financial protection. Health insurance was negatively associated with CHE, suggesting that insurance coverage plays an important protective role against financial hardship. This finding is consistent with previous research.³⁹ In line with this, Gheinali et al. (2023) reported a comparable protective effect among hospitalized patients, showing that those with adequate insurance coverage were significantly less likely to incur catastrophic expenditures than those without such financial protection.⁴⁰

Disease category was strongly associated with CHE. Compared with acute diseases, cancer and hemodialysis showed particularly large positive associations with catastrophic spending, whereas diseases related to the nervous system were associated with lower odds. These findings are consistent with those of Minh and Bach (2012), who reported that households with members suffering from non-communicable diseases in Vo Nhai district, Thai Nguyen province, had 3.2 times higher odds of experiencing CHE than other households.⁴¹ Similarly, Lee et al. (2019) found that in Quoc Oai district, Ha Tay province, the risk of CHE increased with the number of household members suffering from chronic diseases, with an OR of 1.16.⁴² Our results are also in line with those of Giang et al. (2020) in Hanoi, Dien Bien, and Dong Thap provinces.⁴³ In addition, Yadav et al. (2021) showed that the likelihood of impoverishment due to health expenditures was significantly higher among hospitalized cancer patients in India, with an OR of 7.670.⁴⁴

Recent international findings further reinforce the link between severe health conditions and financial hardship. Meshkat et al. (2024) highlighted that non-communicable diseases significantly drive catastrophic expenditures, especially among older populations.⁴⁵ Likewise, Dehaghi et al. (2025) found that severe conditions such as ischemic heart disease place a heavy financial burden on households.⁴⁶ These studies collectively highlight the substantial financial burden associated with chronic diseases, particularly when treatment is long term and intensive. The elevated odds ratios observed for cancer and hemodialysis underscore the pronounced financial vulnerability associated with these conditions in the study region. For these patients, long-term management, frequent travel to provincial hospitals, and the costs of specialized care can place substantial strain on household resources. These findings suggest that while health insurance provides a baseline of support, gaps may remain in the level of financial protection available for households facing high-cost chronic illnesses.

Inpatient utilization was another strong predictor of CHE. This vulnerability among hospitalized patients is consistent with recent observations by Gheinali et al. (2023) and Jafaei et al. (2025), who identified substantial rates of CHE in hospitalized cohorts for COVID-19 and maternal care, respectively.^40,47 Hospital admission generally involves higher direct costs and may also indicate greater disease severity. Both factors can increase the likelihood of CHE. Self-treatment was also associated with markedly higher odds of CHE. One possible explanation is that patients may initially try to manage illness on their own in order to avoid immediate care costs, especially in resource-constrained and remote settings. However, delayed formal treatment may allow conditions to worsen and eventually require more intensive and costly care. This pattern is consistent with health-seeking behaviors observed in remote Vietnamese settings.⁴⁸ The finding suggests that early diagnosis, timely care-seeking, and stronger primary healthcare outreach may play an important role in reducing financial risk.

Patients treated at provincial hospitals were more likely to experience CHE than those treated at district hospitals. This result is consistent with Lee et al. (2019), who showed that greater use of higher-level health facilities was associated with a higher likelihood of CHE, with an OR of 1.22.⁴² One explanation may relate to differences in insurance coverage and treatment-related costs across hospital levels. At the district level, patients with health insurance typically have a co-payment rate of 0%, and shorter travel distances further reduce additional expenses. However, at the provincial level, patients may encounter higher co-payments for treatments not fully covered by health insurance, along with increased indirect costs such as travel, food, and accommodation for accompanying family members. Together, these costs can increase the overall financial burden and heighten the risk of CHE. This finding underscores the financial vulnerability associated with accessing higher-level healthcare services, particularly in contexts where insurance coverage and healthcare costs vary substantially across levels of care.

Among the satisfaction variables, higher satisfaction with the availability and quality of medicines was associated with lower odds of CHE. This finding is consistent with Haakenstad et al. (2022), who suggested that reducing expenditures on medicines purchased from private pharmacies could significantly lower the incidence of CHE in India.⁴⁹ These results highlight the importance of medicine availability and perceived quality at healthcare facilities. When medicines are unavailable or perceived as low quality, patients may purchase them out of pocket from external sources or outside the scope of insurance coverage. In Vietnam, where medicine prices can exceed international averages and self-medication accounts for a notable share of OOP health expenditures (up to 35%),⁵⁰ these dynamics may further increase the financial burden on patients.

This study has several limitations that should be considered when interpreting the findings. First, the cross-sectional design does not permit causal interpretation, and the hospital-based sample from five provinces may limit generalizability, particularly to patients using private facilities or those not seeking care. In addition, reverse causality cannot be ruled out, especially for patient satisfaction and self-treatment variables, which may both influence and be influenced by financial burden. Second, the measurement of OOP spending relied on patient self-reports over the past 12 months, which may be subject to recall bias, particularly for outpatient visits and medication costs. This may have resulted in under- or over-reporting of healthcare expenditures and, consequently, may have affected the precision of the CHE estimates. Only direct medical costs were included, while indirect costs such as transportation, accommodation, caregiver expenses, and lost income were excluded. As a result, the incidence of CHE may have been underestimated, particularly in geographically isolated and mountainous areas. In addition, inpatient and outpatient expenditures were aggregated into a single annual measure; while consistent with standard CHE measurement, this aggregation may obscure differences in their respective contributions to financial risk. Third, because refusals or ineligible cases were replaced by the next eligible patient within the same sampling stratum, some potential for selection bias may remain. Fourth, medicine-related variables were captured through patient satisfaction measures rather than objective assessments of drug prices or stock availability, and face-to-face interviews may have introduced response bias. Despite these limitations, this study provides valuable patient-level evidence on the determinants of CHE in a disadvantaged region of Vietnam and offers insights that may be relevant for other low- and middle-income countries or settings with similar health system characteristics. Future studies using longitudinal or panel data would strengthen causal inference by clarifying the temporal ordering between predictors and CHE and by tracking changes in healthcare use and financial burden over time. Additionally, incorporating qualitative methods (such as in-depth interviews) could provide deeper insights into patient coping strategies and the specific reasons behind self-medication behaviors.

5. Conclusion

The study identified the factors influencing CHE among patients visiting medical facilities in the Northern Midlands and Mountainous areas. The results showed that older age, male gender, greater distance to health facilities, cancer, hemodialysis, more inpatient visits, self-treatment, and treatment at provincial hospitals were associated with higher odds of CHE. By contrast, the study found that certain variables were associated with lower odds of CHE, including being of Kinh ethnicity, higher household income, having health insurance, and greater satisfaction with the availability of drugs, and satisfaction with the quality of prescribed drugs.

Based on the findings of the study, several recommendations can be made to mitigate the risk of CHE for patients in the Northern Midlands and Mountainous areas. First, strengthening financial protection remains essential, particularly through expanding effective health insurance coverage for vulnerable households. Second, because self-treatment and intensive healthcare use were associated with greater financial risk, strengthening primary healthcare, preventive care, and early management of chronic conditions is important to encourage timely use of formal services. Third, improving both the availability of essential medicines and the quality of prescribed drugs at public hospitals may help reduce additional OOP spending. Finally, the effect of distance to health facilities highlights the importance of geographic access in shaping CHE risk. Improving the functional capacity and service quality of district hospitals may reduce referrals to higher-level facilities and help lower the risk of CHE, especially in remote and mountainous areas.

Footnotes

Acknowledgement

We gratefully thank the Ministry of Education and Training of Vietnam, Thai Nguyen University (TNU), and Thai Nguyen University of Economics and Business Administration for their support in facilitating this research.

ORCID iD

Nguyen Thi Thu Thuong

Ethical considerations

Author contributions

Thuong Nguyen Thi Thu was responsible for conceptualization, formal analysis, data curation, writing – original draft, and writing – review & editing. Linh Dinh Hong contributed to data curation, supervision, and validation. Nhu Dinh Vien was in charge of methodology, data curation, and visualization.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data are available upon request from the corresponding author.*

Appendix

Appendix A.

Measurement scale for patient satisfaction.

Dimensions (adapted from kumara & samaratunge, 2016)	Items (survey variables) (adapted from akthar et al., 2023)
1. Availability	1. Satisfaction with medical supplies
	2. Satisfaction with medicine availability
	3. Satisfaction with skilled professionals
2. Accessibility	4. Satisfaction with waiting time
2. Accessibility	5. Satisfaction with administrative procedures
3. Interaction Quality	6. Satisfaction with attitude and interactions
3. Interaction Quality	7. Satisfaction with explanation and advice
4. Technical Quality	8. Satisfaction with diagnosis results
	9. Satisfaction with appropriate medication
	10. Satisfaction with prescription drug quality

The satisfaction scale comprises 10 items adapted from the conceptual dimensions of Kumara and Samaratunge (2016) and the empirical determinants identified by Akthar et al. (2023). Items were measured on a 5-point Likert scale (1 = Very Dissatisfied to 5 = Very Satisfied).

Appendix B.

Descriptive statistics for patient satisfaction indicators by catastrophic health expenditure status.

	Total (n =706)	CHE=1 (n=147)	CHE=0 (n=559)	p-Value
Satisfaction level with medical supplies and equipment	4.18	4.25	4.17	0.192
Satisfaction level with the availability of medicines	4.10	4.10	4.10	0.972
Satisfaction level with the availability of skilled professionals	4.21	4.28	4.19	0.164
Satisfaction level with diagnosis and treatment results of doctors	4.16	4.25	4.13	0.055
Satisfaction level with doctors prescribing appropriate medications	4.15	4.21	4.13	0.188
Satisfaction level with the quality of prescription drugs	4.19	4.22	4.19	0.615
Satisfaction level with explanation, advice and health care guidance from doctors	4.28	4.30	4.27	0.629
Satisfaction level with the attitude in communication, gestures and interactions of doctors and nurses	4.29	4.32	4.28	0.517
Satisfaction level with waiting time	4.15	4.21	4.13	0.240
Satisfaction level with administrative procedures	4.15	4.26	4.12	0.029

Appendix C.

Variance inflation factor.

Variable	Variance inflation factor (vif)	Tolerance
Age	1.13	0.7516
Gender	1.04	0.9627
Ethnicity	1.36	0.7333
Marital status	1.25	0.8115
Household size	1.22	0.9413
Household with children (under 6 years)	1.26	0.8226
Household with elderly (age 60+)	1.29	0.7944
Distance from home to the nearest health facility, km	1.23	0.7731
Settlement	1.20	0.8144
Economic status (Income Quintile)	1.33	0.8311
Health Insurance status	1.15	0.7529
Type of disease	1.16	0.8721
Number of outpatient visits in the last 12 months	1.07	0.8611
Number of inpatient visits in the last 12 months	1.07	0.9362
Self-treatment	1.18	0.9341
Visiting private health facility	1.16	0.8478
Type of hospital	1.15	0.8325
Satisfaction level with medical supplies and equipment	3.49	0.8605
Satisfaction level with the availability of medicines	3.07	0.2863
Satisfaction level with the availability of skilled professionals	3.14	0.3257
Satisfaction level with diagnosis and treatment results of doctors	2.71	0.3181
Satisfaction level with doctors prescribing appropriate medications	3.58	0.3683
Satisfaction level with the quality of prescription drugs	3.18	0.2797
Satisfaction level with explanation, advice and health care guidance from doctors	3.30	0.3142
Satisfaction level with the attitude in communication, gestures and interactions of doctors and nurses	3.25	0.3029
Satisfaction level with waiting time	2.70	0.3076
Satisfaction level with administrative procedures	2.45	0.4082
Average	1.95

References

Jowett

Brunal

Flores

, et al. Spending Targets for Health: No Magic Number. Health Systems Governance & Financing. World Health Organization, 2016.

Bedado

Kaso

Hailu

. Magnitude and determinants of out of pocket health expenditure among patients visiting outpatients in public hospitals in East Shoa Zone, Ethiopia. Clinical Epidemiology and Global Health 2022; 15: 101066. https://doi.org/10.1016/j.cegh.2022.101066

Sharma

Pavlova

Groot

. Catastrophic health care expenditure and impoverishment in Bhutan. Health Policy and Planning 2023; 38: 228–238. https://doi.org/10.1093/heapol/czac107

Vahedi

Rezapour

Khiavi

, et al. Decomposition of socioeconomic inequality in catastrophic health expenditure: an evidence from Iran. Clinical Epidemiology and Global Health 2020; 8: 437–441. https://doi.org/10.1016/j.cegh.2019.10.004

WHO. Global Monitoring Report on Financial Protection . Health 2021. World Health Organization, 2021.

Wagstaff

Flores

Hsu

, et al. Progress on catastrophic health spending in 133 countries: a retrospective observational study. The Lancet Global Health 2018a; 6: e169–e179. https://doi.org/10.1016/S2214-109X(17)30429-1

Wagstaff

Flores

Smitz

, et al. Progress on impoverishing health spending in 122 countries: a retrospective observational study. The Lancet Global Health 2018b; 6: e180–e192. https://doi.org/10.1016/S2214-109X(17)30486-2

Cai

. A robust health system to achieve universal health coverage in Vietnam. The Lancet Regional Health - Western Pacific 2023; 37: 100854. https://doi.org/10.1016/j.lanwpc.2023.100854

Nguyen

Tariq

Hinchcliff

, et al. Contribution of private health services to universal health coverage in low and middle-income countries: factors affecting the use of private over public health services in Vietnam. The International Journal of Health Planning and Management 2023; 38: 1613–1628. https://doi.org/10.1002/hpm.3689

10.

Nguyen

TTT

. Research on catastrophic health expenditure, influencing factors and the role of health insurance in reducing catastrophic medical costs in the Northern Midlands and Mountainous Areas. Thai Nguyen University Publishing House, 2023.

11.

Thuong

NTT

Huy

. Catastrophic health expenditure in the Northern Midlands and Mountainous Areas and its determinants, Vietnam from 2014 to 2020: a cross-sectional study. BMJ Open 2022; 12: e058849. https://doi.org/10.1136/bmjopen-2021-058849

12.

Kavosi

Rashidian

Pourreza

, et al. Inequality in household catastrophic health care expenditure in a low-income society of Iran. Health Policy and Planning 2012; 27: 613–623. https://doi.org/10.1093/heapol/czs001

13.

Löfgren

. Catastrophic health expenditure in Vietnam: studies of problems and solutions. Umeå University, 2014.

14.

Falconi

Bernabé

. Determinants of catastrophic healthcare expenditure in Peru. International Journal of Health Economics and Management 2018; 18: 425–436. https://doi.org/10.1007/s10754-018-9245-0

15.

Ghimire

Ayer

Kondo

. Cumulative incidence, distribution, and determinants of catastrophic health expenditure in Nepal: results from the living standards survey. International Journal for Equity in Health 2018; 17: 23. https://doi.org/10.1186/s12939-018-0736-x

16.

Aregbeshola

Khan

. Determinants of catastrophic health expenditure in Nigeria. The European Journal of Health Economics 2018; 19: 521–532. https://doi.org/10.1007/s10198-017-0899-1

17.

Yazdi-Feyzabadi

Mehrolhassani

Darvishi

. Measuring catastrophic health expenditures and its inequality: evidence from Iran’s Health Transformation Program. Health Policy and Planning 2019; 34: 316–325. https://doi.org/10.1093/heapol/czz031

18.

Sepehri

. Severe injuries and household catastrophic health expenditure in Vietnam: findings from the Household Living Standard Survey 2014. Public Health 2019; 174: 145–153. https://doi.org/10.1016/j.puhe.2019.06.006

19.

Liu

Nicholas

, et al. Trends and determinants of catastrophic health expenditure in China 2010–2018: a national panel data analysis. BMC Health Services Research 2021; 21: 526. https://doi.org/10.1186/s12913-021-06533-x

20.

Cui

Liu

, et al. Catastrophic health expenditure and health-related quality of life among patients with chronic diseases: the moderation effect of family doctor contract services. Pharmacoeconomics and Policy 2025; 1: 84–91. https://doi.org/10.1016/j.pharp.2025.05.001

21.

Chen

Zhu

You

, et al. Evaluation of status quo and determinants of catastrophic health expenditure among empty-nest elderly in China: evidence from the China Health and Retirement Longitudinal Survey (CHARLS). European Review for Medical and Pharmacological Sciences 2023; 27: 1398–1412.

22.

Kim

Hong

. Catastrophic health expenditures and life satisfaction: a case in South Korea. Journal of Family and Economic Issues 2015; 36: 369–382. https://doi.org/10.1007/s10834-015-9459-y

23.

Chino

Peppercorn

Taylor

Jr , et al. Self-reported financial burden and satisfaction with care among patients with cancer. The Oncologist 2014; 19: 414–420. https://doi.org/10.1634/theoncologist.2013-0374

24.

Kang

Yoon

, et al. The relationship between catastrophic health expenditure and health-related quality of life. International Journal for Equity in Health 2018; 17: 166. https://doi.org/10.1186/s12939-018-0883-0

25.

Cochran

. Sampling techniques. 3rd ed. : John Wiley & Sons, 1977.

26.

Akthar

Nayak

Pai

. Determinants of patient satisfaction in Asia: evidence from systematic review of literature. Clinical Epidemiology and Global Health 2023; 23: 101393. https://doi.org/10.1016/j.cegh.2023.101393

27.

Kumara

Samaratunge

. Patterns and determinants of out-of-pocket health care expenditure in Sri Lanka: evidence from household surveys. Health Policy and Planning 2016; 31: 970–983. https://doi.org/10.1093/heapol/czw021

28.

Wagstaff

van Doorslaer

. Catastrophe and impoverishment in paying for health care: with applications to Vietnam 1993-1998. Health Economics 2003; 12: 921–933. https://doi.org/10.1002/hec.776

29.

Cylus

Thomson

Evetovits

. Catastrophic health spending in Europe: equity and policy implications of different calculation methods. Bulletin of the World Health Organization 2018; 96: 599–609. https://doi.org/10.2471/BLT.18.209031

30.

Azzani

Yahya

Roslani

, et al. Catastrophic health expenditure among colorectal cancer patients and families: a case of Malaysia. Asia-Pacific Journal of Public Health 2017; 29: 485–494. https://doi.org/10.1177/1010539517732224

31.

Kien

Minh

Ngoc

, et al. Inequalities in household catastrophic health expenditure and impoverishment associated with noncommunicable diseases in Chi Linh, Hai Duong, Vietnam. Asia-Pacific Journal of Public Health 2017; 29: 35S–44S. https://doi.org/10.1177/1010539517712919

32.

Prinja

Jagnoor

Sharma

, et al. Out-of-pocket expenditure and catastrophic health expenditure for hospitalization due to injuries in public sector hospitals in North India. PLoS One 2019; 14: e0224721. https://doi.org/10.1371/journal.pone.0224721

33.

Hailemichael

Hanlon

Tirfessa

, et al. Catastrophic health expenditure and impoverishment in households of persons with depression: a cross-sectional, comparative study in rural Ethiopia. BMC Public Health 2019; 19: 930. https://doi.org/10.1186/s12889-019-7239-6

34.

Shumet

Mohammed

Kahissay

, et al. Catastrophic health expenditure among chronic patients attending Dessie Referral Hospital, Northeast Ethiopia. ClinicoEconomics and Outcomes Research 2021; 13: 99–107. https://doi.org/10.2147/CEOR.S291463

35.

von

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. The Lancet 2007; 370: 1453–1457. https://doi.org/10.1016/s0140-6736(07)61602-x

36.

Chen

. The impact of multimorbidities on catastrophic health expenditures among patients suffering from hypertension in China: an analysis of nationwide representative data. Sustainability 2022; 14: 7555. https://doi.org/10.3390/su14137555

37.

Masiye

Kaonga

Kirigia

. Does user fee removal policy provide financial protection from catastrophic health care payments? Evidence from Zambia. PLoS One 2016; 11: e0146508. https://doi.org/10.1371/journal.pone.0146508

38.

Krishnamoorthy

Rajaa

Sinha

, et al. Equity and extent of financial risk protection indicators during COVID-19 pandemic in rural part of Tamil Nadu, India. Heliyon 2023; 9: e18902. https://doi.org/10.1016/j.heliyon.2023.e18902

39.

Thanh

Anh

BTM

Xiem

, et al. Out-of-pocket health expenditures among insured and uninsured patients in Vietnam. Asia-Pacific Journal of Public Health 2019; 31: 210–218. https://doi.org/10.1177/1010539519833549

40.

Gheinali

Moshiri

Tavani

, et al. Catastrophic health expenditures in hospitalized patients with delta variant of COVID-19: a cross-sectional study. Health Promotion Perspectives 2023; 13: 68–76. https://doi.org/10.34172/hpp.2023.09

41.

Minh

Bach

. Assessing the household financial burden associated with the chronic non-communicable diseases in a rural district of Vietnam. Global Health Action 2012; 5: 18892. https://doi.org/10.3402/gha.v5i0.18892

42.

Lee

Minh

, et al. Use of high-level health facilities and catastrophic expenditure in Vietnam: can health insurance moderate this relationship? BMC Health Services Research. 2019; 19: 318.

43.

Giang

Oanh

TTM

Tuan

, et al.

Is health insurance associated with health service utilization and economic burden of non-communicable diseases on households in Vietnam?

Health Systems & Reform 2020; 6: e1619065. https://doi.org/10.1080/23288604.2019.1619065

44.

Yadav

Menon

John

. Disease-specific out-of-pocket payments, catastrophic health expenditure and impoverishment effects in India: an analysis of National Health Survey data. Applied Health Economics and Health Policy 2021; 19: 769–782. https://doi.org/10.1007/s40258-021-00641-9

45.

Meshkat

SAM

Kassaeian

Danaei

, et al. Households catastrophic health expenditure among older people with noncommunicable diseases: a cross-sectional study from Iran. Health Science Reports 2024; 7: e1992. https://doi.org/10.1002/hsr2.1992

46.

Dehaghi

MAL

Kassaeian

Jandaghi

, et al. Catastrophic health expenditure and its determinants in patients with ischemic heart disease and their households: a cross-sectional study from Iran in 2024. Health Science Reports 2025; 8: e71082. https://doi.org/10.1002/hsr2.71082

47.

Jafaei

Kassaeian

Danaei

, et al. Household catastrophic health expenditures in maternal care: a cross-sectional study from Semnan, Iran. Health Science Reports 2025; 8: e70426. https://doi.org/10.1002/hsr2.70426

48.

Tran

Nguyen

Nong

, et al. Health status and health service utilization in remote and mountainous areas in Vietnam. Health and Quality of Life Outcomes 2016; 14: 85. https://doi.org/10.1186/s12955-016-0485-8

49.

Haakenstad

Kalita

Bose

, et al. Catastrophic health expenditure on private sector pharmaceuticals: a cross-sectional analysis from the state of Odisha, India. Health Policy and Planning 2022; 37: 872–884. https://doi.org/10.1093/heapol/czac035

50.

Japan International Cooperation Agency (JICA)International KRI Corp . Basic Information Survey for Basic Health Service Package and Provider Payment Mechanism in Viet Nam. Japan International Cooperation Agency, 2017.