Catastrophic and impoverishing impacts of health expenditures: a focus on non-communicable diseases in Pokhara Metropolitan City, Nepal

Abstract

Background

Ensuring equitable access to Universal Health Coverage (UHC) is crucial, particularly in low-resource settings like Nepal, where high out-of-pocket expenditure (OOPE) poses a significant barrier to the utilization of healthcare services. This study examined the catastrophic and impoverishing impact of household-level healthcare expenditures, focusing on whether households with NCDs have a higher likelihood of incurring CHE and experiencing impoverishment.

Methods

We conducted this study in Pokhara Metropolitan City, Nepal, involving 1,276 households. Catastrophic Health Expenditure (CHE) was defined when OOPE was 10% or more of the household’s total expenditure, while impoverishment was measured using the poverty headcount ratio, poverty gap, and squared poverty gap. We used a poverty line of NPR 7,674 (approximately USD 230 in Purchasing Power Parity) per capita per month, as set by the National Statistics Office for the Gandaki urban area in 2024. Total monthly household consumption was the sum of food and non-food expenditures, including healthcare expenditures. Health expenditure was calculated based on self-reported data validated by pertinent documents. Household weight was used in the data analysis.

Results

Out of 1276 households, 853 (66.8%) reported illness in the past month, and 125 households suffered from CHE. This corresponds to 9.8% of all sampled and 14.6% of households that experienced illness. Out of those 125 households, 82 faced CHE due to NCDs, representing 6.4% of all sampled and 9.6% of households experiencing illness. Most health expenditures were primarily due to medication (60%) and curative care (17.3%) in NCD conditions. The poverty rate increased by 1.17%points, from 9.4% to 10.6%, over the past month due to healthcare costs, leading to a 12.3% increase in people living in poverty, with 1.02%points attributed to NCDs. The poverty gap rose from 1.5% to 1.9%, and the squared poverty gap increased from 0.003 to 0.005. Households with more than two members affected by NCDs had 3 times higher odds of experiencing CHE (AOR 3.02, 95% CI 2.59–3.51). Those with a household member/s suffering from heart disease had twice the odds of facing CHE (AOR 2.41, 95% CI 2.22–2.62). Households with diabetic members had 1.13 times higher odds of experiencing CHE (AOR = 1.13, 95% CI: 1.05–1.21). Households in the lowest quintile had twice the odds of incurring CHE than those in the highest quintile (AOR 1.93, 95% CI 1.75–2.15).

Conclusion

NCDs and their associated costs are significant contributors to CHE and impoverishment. As Nepal moves towards UHC, policymakers need to accord the highest priority to enhancing financial protection mechanisms by subsidizing healthcare costs, particularly for medicines and curative care related to NCDs. Furthermore, addressing economic inequalities through targeted support for low-income and marginalized households will mitigate CHE and prevent impoverishment.

Introduction

Universal health coverage (UHC) guarantees access to comprehensive, high-quality health services without causing financial hardship [1]. Despite the commitment to UHC, more than 50% of total health expenditure is still financed by out-of-pocket (OOP) payments in many low- and middle-income countries (LMICs), including Nepal [2]. The 2021 World Bank report indicates that out-of-pocket expenditure (OOPE) in Nepal was 51.26% [3], with the National Health Accounts 2018/19 reporting that more than half of this expenditure was attributed to NCDs [4]. It creates barriers to accessing healthcare and implies that illness can potentially be catastrophic for the household’s economy [5]. This problem could be amplified for people who suffer from NCDs because they need treatment on a recurrent basis, and if they delay care-seeking, the condition may get worse and demand more advanced, expensive treatment [6]. NCDs kill 41 million individuals each year, and 77% are in LMICs, where cardiovascular diseases, cancers, chronic respiratory diseases, and diabetes are responsible for over 80% of premature NCD deaths [6]. In Nepal, NCDs are responsible for 71% of all deaths [7]. Evidence indicates that NCDs’ economic cost tends to impose catastrophic health expenditures (CHE), particularly among low-income households, pushing them deeper into poverty and exacerbating health inequities in LMICs [8]. Furthermore, the lack of adequate healthcare financing and social protection for NCDs disproportionately affects disadvantaged populations, resulting in heightened impoverishment and limited access to necessary care [9].

Nepal has introduced several initiatives to achieve UHC. The Constitution of Nepal (2015), Article 35, mandates that “Every citizen shall have the right to free basic health services from the State, and no one shall be deprived of emergency health services” [10]. In 2010, the program “Medical Treatment of Deprived Citizens” (Bipanna Nagarik Aushadhi Upachar Kosh) was initiated to provide financial assistance for services not included in the basic package to individuals suffering from specific diseases [11]. The National Health Insurance Program (NHIP) is the latest initiative aimed at reducing OOPE and preventing households from falling into poverty [12]. However, low enrollment rates, high morbidity, substantial dependence on OOPE for healthcare, and a notable prevalence of poverty place Nepalese households at a heightened risk of encountering CHE [13–15].

CHE is a significant concern globally, particularly in South Asia, where it can account for up to 60%−70% of total expenditure due to inadequate public spending on health [16]. In Nepal, the average per capita out-of-pocket spending on health increased sevenfold between 1995–1996 and 2010–2011, with 13% of households incurring CHE in 2010–2011 [17]. The health-financing system in Nepal has become regressive over the years, disproportionately affecting poorer households and those in the far-western region [18]. While studies conducted in 2010–11 showed Nepal has high levels of CHE [17], a notable gap exists in the updated and in-depth data concerning the incidence of CHE and impoverishment due to healthcare expenditures in Nepal, particularly in the context of NCDs. On the one hand, initiatives to increase financial protection for essential services may have reduced the incidence of CHE and the impoverishing effects of OOPE. On the other hand, the increasing prevalence of NCDs could have had the opposite effect, especially if associated healthcare services are not considered ‘essential.’ This study examines the catastrophic and impoverishing effects of household healthcare expenditures in Nepal, focusing on whether households affected by NCDs face greater odds of CHE and experience increased impoverishment.

Methods

Study design and setting

An analytical cross-sectional design was employed in this study. The study site is Pokhara Metropolitan City, Nepal, with a total population of 599,504 in 120,594 households [19]. The rationale behind choosing this specific location lies in its diverse ward composition, which encompasses both urban and semi-urban areas. Notably, the city is home to a referral-level hospital and several private healthcare facilities, establishing itself as a central hub for medical treatment and expenditures among its residents. Pokhara is the second-largest metropolitan city in Nepal, with growing markets promoting unhealthy lifestyles and diets, making the population prone to NCDs. A 2022 study from this city reported a prevalence of hypertension at 36% [20]. In 2018, the prevalence of type 2 diabetes constituted 11.7% among adults aged 25 years and above [21], while Chronic Obstructive Pulmonary Disease (COPD) prevalence stood at 8.5% in 2020 [22].

Sampling strategy

We applied a two-stage random sampling design in this study. The study aimed to achieve a sample size of 1,276 households, considering an expected 58% OOPE [23], a 95% confidence level, a desired level of precision (e) = 5%, a design effect of 1.5, and a 10% non-response rate. To ensure a diverse and comprehensive sample representation, 11 wards, accounting for 33% of the total 33 wards in Pokhara Metropolitan City (Fig. 1), were selected through Probability Proportional to Size (PPS) sampling. This method was chosen to ensure wards with larger households had a higher chance of being included in the sample. After that, 116 households were selected from each ward’s household list by dividing the overall sample size of 1,276 by the number of chosen wards, ensuring equal representation across all sampled wards and maintaining homogeneity. Household sizes may vary across wards; however, equal allocation sampling provides a methodologically sound approach to standardizing the sample distribution. The sampling frame was constructed based on the 2021 voter and household lists, which were reviewed and finalized in coordination with ward chairpersons and members. Within each selected ward, systematic random sampling was employed to identify specific households for inclusion, ensuring objectivity and reducing selection bias. The primary respondents targeted for this study were household heads, family members with NCDs, or caregivers aged 18 years or older. In cases where these individuals were unavailable after three consecutive visits, the neighboring household was approached for data collection. In this study, wards and households were the sampling units.

Fig. 1.

Pokhara metropolitan city and sampled wards [24]

Data collection

Data was collected from May to October 2023 through face-to-face interviews using a questionnaire. The questionnaire included socio-demographic information, household expenditures on food and non-food items, experiences of illness, and associated expenses. Food consumption encompassed a wide range of items, including cereals, pulses, vegetables, fruits, dairy products, meat, sugar, oil, spices, beverages, water, alcohol, snacks, and miscellaneous food products. Non-food consumption included expenditures on clothing, utilities, education, health services, communication, transportation, entertainment, tobacco, and miscellaneous categories. Moreover, healthcare-related expenses were collected for all household members who were ill, encompassing expenditures associated with outpatient services, emergency care, inpatient treatments, laboratory examinations, diagnostic procedures, medical equipment, pharmaceuticals, transportation, and lodging. For households facing illnesses, expenditures were based on self-reported data validated by relevant documents for tests, treatment, or medications, as available. In contrast, households without any reported illness were assigned zero healthcare expenses. The questionnaire was developed by the researchers’ team through an intensive literature review [25–27] and based on the Nepal Living Standard Survey (NLSS) Questionnaire III (openly accessible) [28], which was modified to align with the study objectives and local context [29].

The fieldwork involved five days of intensive training for research assistants conducted by the researcher and health economics experts. This training included pre-testing the questionnaire through 102 interviews across three rounds in a non-sampled ward of Pokhara Metropolitan City, database preparation, and finalization in the Kobo toolbox. Price lists from three major grocery stores located in three randomly selected sampled wards were obtained to gain an initial understanding of item pricing. This aided in probing and verifying the responses from the interviewees and households. Supervisors maintained scrutiny and supervision through field visits and virtual meetings throughout the data collection process. Ethical approval was obtained from the Ethical Review Board of the Nepal Health Research Council, with reference number 3064, and a support letter was provided by the Pokhara Metropolitan Office, Kaski District (reference number 799).

Data analysis

First, total monthly household consumption was calculated as the sum of food and non-food consumption. We did not use income data directly; instead, we approximated income (or available resources) by analyzing household spending [30]. The reported out-of-pocket expenses (OOPE) associated with acute illnesses or injuries (within the past 30 days) or non-communicable diseases (NCDs) (within the past 12 months) were converted into monthly healthcare expenditures and aggregated. Healthcare expenses were categorized under non-food consumption, encompassing acute and NCD expenditures. We calculated the expenditures in Nepali rupees (NPR) and also converted them to U.S. Dollars (USD) for 2023 using the World Bank’s Purchasing Power Parity (PPP) conversion factor.

Outcome measures (dependent variables)

Financial hardship was measured through two outcome measures [31]: the presence of CHE and impoverishment. This study used the well-known methodology developed by Wagstaff and Doorslaer to estimate the incidence of CHE and its impoverishment effect [32].

CHE

CHE was calculated using the threshold of whether the OOPE equaled or exceeded 10% of the total household expenditure [30]. Out-of-pocket (OOP) expenses on healthcare are defined as payments made at the point of service after deducting any reimbursement.

Impoverishment

Impoverishment was defined as when a household’s expenditure level was above the poverty line before health payments (pre-payment) but fell below the poverty line after healthcare expenditures (post-payment) [33]. OOP payments can also deepen poverty for those who are already below the poverty line. We used the poverty line of the NPR. 7674 per month per capita (equivalent to US$57.48) for the Gandaki urban area, as published by Nepal’s National Statistics Office in 2024. This figure is based on a study conducted in Pokhara Metropolitan City (within Gandaki urban) in 2023, which considered the localized poverty lines that account for cost differentials between regions [34].

We evaluated whether OOPE led to poverty using the poverty headcount ratio [35] to measure the extent of impoverishment. The incidence of impoverishment was estimated by summing the number of impoverished households among all sampled households (n = 1276) for the month preceding the survey. Additionally, NCD-related expenditures were analyzed to evaluate their impoverishing impact.

To assess the depth of poverty, we measured the poverty gap and the squared poverty gap. Foster, Greer, and Thorbecke’s (FGT) poverty estimation was used to estimate income poverty, where PL represents the poverty line income, and YPi represents the total monthly income. The values of λ = 0, 1, and 2 correspond to the headcount index, the normalized deficit (poverty gap) ratio, and the severity of poverty, respectively [36].

$$\mathrm{Poverty}\mathrm{index}=\frac{1}{N}\sum _{i=1}^n{\left[\frac{P_L-Y_{P_i}}{P_L}\right]}^{\lambda }$$

The index is sensitive to changes in income when λ > 0 and to the transfer to income when λ > 1. The poverty gap expresses the gap between the poverty-line income and the income of below-poverty-line people, which was measured by the ratio of the difference between poverty-line income and below-poverty-line income and poverty-line income. The square of the poverty gap measures poverty severity by emphasizing the disparities among the poor through a weighted sum of poverty gaps, giving more weight to observations significantly below the poverty line.

Key independent variable of interest—presence of NCD

The key variable of interest in this study was the presence of NCDs. If any household member has been suffering from a chronic condition or NCD for the last three months or more, it is considered a household affected by NCD (yes = 1, otherwise 0). The number of household members suffering from NCD was also obtained from interviews. The types of NCD used as independent variables were the presence (yes = 1; no = 0) of cardiovascular or heart disease, hypertension, chronic respiratory disease, cancer, and diabetes among any household member. The association with cancer could not be calculated because it was observed in a few households.

Covariates

Independent variables were derived from previous studies [37–41], which found various factors to be associated with CHE and NCDs: households with acute illness, households with elderly individuals aged 60 years and above and children under five, enrollment in national health insurance programs, expenditure quintiles, caste/ethnicities, and family/household size. These variables were included as covariates to adjust for potential confounding. Caste and ethnicities have been categorized into Brahmin/Chettri, Janajati, and Dalit groups according to the 2021 Nepal Census [42]. Janajati included Gurung, Newar, Magar, Rai, and Thakali.

Statistical analyses

The data were analyzed using descriptive statistics (frequency, percentage, mean, standard deviations, and percentiles) and inferential analysis (logistic regression) via SPSS version 29. Households served as the study units, which were subsequently used as the units of analysis. However, for the analysis of coping strategies, the unit of analysis shifted to individual household members. This is because coping strategies vary across different household members, with responses recorded as multiple-choice (Table 8). Descriptive analysis was performed for all sampled households and those experiencing illness. Inferential analysis focused on households with illness, as they are more likely to incur healthcare expenditures. A multivariate binary logistic regression was conducted to examine the factors determining catastrophic payments, with a focus on the impact of NCDs. All variables identified in the univariate logistic regression were included as covariates in the multivariate logistic regression analysis to ensure controls for confounding, identify independent effects, and enhance model accuracy. Sampling weights were applied in the data analysis to ensure that the findings represent the overall households of Pokhara Metropolitan City, accounting for variations in household distribution across wards. We report unweighted counts (n) to indicate actual sample sizes and weighted percentages to provide representative estimates for 1276 all-sampled and 853 households with illness.

Table 8.

Coping strategies of household members to manage NCD health care expenses^a

Coping strategies	Household members with NCDs in households with CHE (n = 363)	Household members with NCDs in households without CHE (n = 1437)
	Percent of cases
Saving/Income	86.9	88.6
Loan & Borrowing	18.2	8.0
National Health Insurance Program	12.7	13.9
Relatives, pensioners fund	9.8	10.8
Government Social Security Program	2.9	3.2
Selling of Assets	1.1	0.4
Private Hospital Insurance	0.4	0.3

^aMultiple responses

Results

In total, 1276 households participated in the study. The non-response rate was 2%, but we replaced the sampled non-response households with adjacent households. Among these 1276 households, 853 (67%) reported illness in the past month.

Socio-demographic characteristics of households

Table 1 presents the socio-demographic characteristics of all sampled households (n = 1276), households reporting illness (n = 853), and households with NCDs only (n = 745). Caste/ethnicity distribution revealed that Brahmin/Chettri households represented the majority across all three groups, comprising 54.8% of all households, 55.5% of households with illness, and 58.3% of those with NCDs. Family size data showed that 57.6% of all households had four or fewer members, with slightly lower proportions in households with illness (53.9%) and NCDs (51.9%). The presence of elderly household members (≥ 60 years) was more common in households with NCDs (61.3%) compared to all households (47.5%) and households with illness (55.8%). Enrollment in the NHIP was higher in households with illness (44.4%) and NCDs (48.7%) compared to the total sample (39.2%). Households with acute illnesses comprised 28.2% of those with illness, while 19.1% of households with NCDs reported such conditions. For households with NCDs, 61.5% had hypertension, 30.7% had diabetes, 11.9% had heart disease, and 8.2% had chronic respiratory diseases. Lastly, most households with NCDs were in the highest expenditure quintile (22.4%).

Table 1.

Socio-demographic characteristics of households, and households with illness among total households, and households with NCDs (N = 1276)

Variables	All households (n = 1276)		Of 1276 households, households with illness (n = 853)		Households with NCDs only (n = 745) among all households
	Unweighted frequency (yes)	Weighted (column %)	Unweighted frequency (yes)	Weighted (column %)	Unweighted Frequency (yes)	Unweighted (column %)
Caste/ethnicities
Brahmin/Chettri	735	54.8	489	55.5	434	58.3
Janajatis	313	28.0	218	28.5	189	25.4
Dalit and others	228	17.2	146	15.9	122	16.4
Family size
< = 4 (median)	727	57.6	453	53.9	387	51.9
> 4	549	42.4	400	46.1	358	48.1
Households with elderly ≥ 60 years	617	47.5	485	55.8	457	61.3
Households with < 5 children	212	16.2	136	14.5	114	15.3
Households enrolled in NHIP	534	39.2	405	44.4	363	48.7
Households with acute illnesses/injuries	250	19.2	250	28.2	142	19.1
Number of NCD members in the household
None	521	40.8	108	12.7	0	0.0
One	355	29.4	353	43.0	355	47.7
Two	219	16.3	218	23.8	219	29.4
More than two	171	13.5	174	20.4	171	22.9
Households with members having
Hypertension	458	36.4	458	53.6	458	61.5
Diabetes	229	18.9	229	27.9	229	30.7
Heart disease	89	7.5	89	11.1	89	11.9
Chronic respiratory problems	61	4.4	61	6.5	61	8.2
Expenditure quintiles (after HE)
Q1 (Lowest)	255	20.6	170	20.8	138	18.5
Q2 (Second)	255	19.6	171	20.5	135	18.1
Q3 (Third)	256	20.6	171	20.6	159	21.3
Q4 (Fourth)	255	20.0	171	19.7	146	19.6
Q5 (Highest)	255	19.0	170	18.4	167	22.4

Economic characteristics of households

The findings indicate that households experiencing illness have higher total and non-food expenditures than the overall sample, likely due to increased healthcare-related costs (Table 2). While food consumption expenditure remains relatively stable across both groups, with a weighted mean of NPR 30,234 (approximately USD 906) for households affected by illness and NPR 29,054 (approximately USD 870) for all households, non-food expenditures show substantial differences. Households experiencing illnesses spent NPR 31,665 (approximately USD 949) on non-food items, compared to NPR 26,671 (approximately USD 799) for all households, indicating increased spending on healthcare among households with illnesses. The higher standard deviations in non-food and total expenditures indicate variability in financial burden across households, with some incurring significantly higher costs.

Table 2.

Monthly household expenditures (in NPR and USD)

Economic characteristics	n = 1276 (all sampled households)			n = 853 (households reporting illnesses)
	Weighted			Weighted
	Mean	Median	SD	Mean	Median	SD
Food consumption	29054 (870)	29888 (895)	11455 (343)	30234 (906)	31041 (930)	11739 (352)
Non-food consumption	26671 (799)	23400 (701)	16327 (489)	31665 (949)	27038 (810)	21879 (656)
Total consumption	55725 (1669)	54403 (1630)	24614 (737)	61899 (1854)	58869 (1764)	29457 (882)

PPP conversion factor = 33.39 [43]

Healthcare expenditures among households who paid for health services

Table 3 shows a significant proportion of healthcare expenditures was on medicine (approximately 60%), and transportation costs consumed 20% of total health expenditures. Curative care-including all health facility expenses except medicine, transportation, and accommodation, accounted for 17.3% and 18.8% of total health expenditures for households that paid for NCD and acute illnesses, respectively.

Table 3.

Healthcare expenditures among households who paid for health services (%)

Expense category	NCDs (n = 732)	Acute illness/injuries (n = 238)
Consultation fee	3.5	5.2
Emergency	0.1	0.1
Inpatient	1.2	2.5
Laboratory	8.3	6.1
X-ray/USG	3.9	3.8
Medicines	63.2	59.7
Medical equipment	0.3	1.1
Transportation	18.9	21.5
Accommodation	0.6	0.1

Catastrophic health expenditures

One hundred twenty-five households faced CHE in the past month (Table 4). This corresponds to 9.8% of all sampled households and 14.6% of those with an illness. Among them, 82 households incurred CHE due to NCD-related expenses, representing 6.4% of the total sample and 9.6% of households that reported an illness.

Table 4.

CHE faced by households in the past month

Categories	n	Frequency	Proportion
All sampled households	1276	125	9.8
Households experiencing illnesses (acute or NCDs or both)	853	125	14.6
Households with NCD expenditures only	745	82	11.0

Descriptive analysis of CHE among households with illness (n = 853)

Table 5 presents a descriptive analysis of CHE among households with illness (n = 853). Among caste/ethnic groups, the proportion of households facing CHE was highest among Brahmin/Chhetri households (16.0%), but no statistically significant association was observed (p = 1.0). Regarding family size, households with fewer than or equal to four members (15.7%) exhibited a slightly higher proportion of CHE compared to larger households (more than four members; 13.5%), although the association was marginally non-significant (p = 0.052). The presence of elderly members (≥ 60 years) was significantly associated with CHE, with 16.9% of households with elderly members experiencing CHE, compared to 13.5% in those without elderly members (p = 0.008). Households enrolled in the NHIP were significantly associated with CHE, with 17.0% of insured households reporting such expenditures, compared to 13.5% among non-insured households (p = 0.026). The association between CHE and the presence of acute illnesses or injuries was not statistically significant (p = 0.084). Households with more than two members affected by NCDs showed a substantial association with CHE (p = 0.005). Specific NCD conditions showed varying associations: diabetes (p = 0.016) and heart disease (p < 0.001) were significantly linked to CHE, whereas hypertension did not show a statistically significant association (p = 0.075). Households in the lowest quintile (Q1) have the highest proportion of CHE (20.6%, p = 0.016).

Table 5.

Descriptive analysis of CHE among households with illness (n = 853)

Variables	Total HHs with illness (N = 853)	Households with CHE (n = 125)	Yes (row%)	p values
Caste/ethnicities
Brahmin/Chettri	489	78	16.0	1
Janajatis	218	29	13.3	0.365
Dalit and others	146	18	12.3	0.285
Family size
< = 4 (median)	453	71	15.7	0.052
> 4	400	54	13.5	1
Households with elderly ≥ 60 years	485	82	16.9	0.008*
Households with < 5 children	136	15	11.0	0.104
Households enrolled in NHIP	405	68	17.0	0.026*
Households with acute illnesses/injuries	250	53	21.2	0.084
Number of NCD members in the household
None	108	15	13.9	1
One	353	31	8.8	0.124
Two	218	30	13.8	0.884
More than two	174	49	28.1	0.005*
Households with members having
Hypertension	458	66	14.4	0.075
Diabetes	229	42	18.3	0.016*
Heart disease	89	30	33.7	0.000*
Chronic respiratory problems	61	9	14.8	-
Expenditure quintiles (after HE)
Q1 (Lowest)	170	35	20.6	0.016*
Q2 (Second)	171	27	15.8	0.446
Q3 (Third)	171	16	9.4	0.375
Q4 (Fourth)	171	22	12.9	0.647
Q5 (Highest)	170	25	14.7	1

An asterisk (*) denotes statistical significance at the 95% confidence level (p < 0.05) *. The p-value for chronic respiratory problems was not calculated because the cell frequency was less than 10 and was also excluded from the logistic regression analysis

Determinants of CHE among households experiencing illness

Table 6 presents univariate and multivariate logistic regression of independent variables against the dependent variable, CHE. After adjusting for confounding variables, households having more than two NCD members faced substantially higher odds of experiencing CHE (AOR = 3.02, 95% CI: 2.59–3.51). Households with heart-related diseases had significantly higher odds of experiencing CHE (AOR = 2.41, 95% CI: 2.22–2.62). Households with members who have diabetes showed a significant association with CHE (AOR = 1.13, 95% CI: 1.05–1.21), while hypertension was linked to lower odds of facing CHE (AOR = 0.51, 95% CI: 0.47–0.56). Households with acute illness had 1.17 times higher odds of experiencing CHE (95% CI: 1.09–1.25). Larger family sizes (more than four members) were associated with reduced odds of experiencing CHE (AOR = 0.81, 95% CI: 0.75–0.87), as were households with children under five years of age (AOR = 0.80, 95% CI: 0.75–0.87). Janajati and Dalits/others exhibited lower odds of facing CHE compared to Brahmin/Chettri (AOR = 0.78, 95% CI: 0.72–0.84 and AOR = 0.60, 95% CI: 0.54–0.66, respectively). Moreover, households with elderly members (aged 60 years or older) did not show an association with CHE (AOR = 1.05, 95% CI: 0.98–1.13). Households enrolled in the NHIP had 1.18 odds of incurring CHE (AOR = 1.18, 95% CI: 1.11–1.26). Households in the lowest expenditure quintile had nearly double the odds of experiencing CHE (AOR = 1.93, 95% CI: 1.75–2.15) compared to those in the highest expenditure quintile.

Table 6.

Determinants of catastrophic health expenditures using logistic regression (n = 853)

Independent variables	Crude OR (95% CI)	AOR (95% CI)
Number of NCD members in the household
None	1	1
One	0.52 (0.47–0.57)a	0.52 (0.47–0.58)a
Two	0.90 (0.81–1.00)	1.10 (0.96–1.25)
More than two	2.54 (2.31–2.80)a	3.02 (2.59–3.51)a
Households with acute illnesses/ijuries	0.99 (0.93–1.06)	1.17 (1.09–1.25)a
(Yes = 1, No = 0)	1	1
Households with NCDs type
(Yes = 1, No = 0)
Heart-related	3.76 (3.49–4.05)a	2.41 (2.22–2.62)a
Diabetes	1.68 (1.58–1.78)a	1.13 (1.05–1.21)a
Hypertension	1.08 (1.02–1.14)a	0.51 (0.47–0.56)a
Family size	0.80 (0.75–0.85)a	0.80 (0.75–0.87)a
(More than 4 = 1, 4 or less = 0)	1	11
Households’ caste/ethnicity
Brahmin/Chettri	1	1
Janajatis	1.61 (1.47–1.77)a	0.78 (0.72–0.84)a
Dalits and others	1.25 (1.13–1.39)a	0.60 (0.54–0.66)a
Households with elderly > = 60 years	1.46 (1.37–1.55)a	1.05 (0.98–1.13)
(Yes = 1, No = 0)	1	1
Households with under five children	0.85 (0.77–0.92)a	0.79 (0.72–0.86)a
(Yes = 1, No = 0)	1	1
Households’ enrollment in the National Health Insurance Program	1.44 (1.35–1.52)a	1.18 (1.11–1.26)a
(Yes = 1, No = 0)	1	1
Households’ expenditure quintiles (after health expenses)
Q1 (Lowest)	1.68 (1.54–1.84)a	1.93 (1.74–2.15)a
Q2 (Second)	0.83 (0.75–0.92)a	0.95 (0.86–1.06)
Q3 (Third)	0.52 (0.46–0.58)a	0.57 (0.51–0.64)a
Q4 (Fourth)	1.12 (1.02–1.23)a	1.21 (1.10–1.34)a
Q5 (Highest)	1	1

^aIndicates statistical significance at the 5% level. The Variation Inflation Factor (VIF) is less than 5 for all independent variables. 1 denotes reference

Impoverishment impact

The poverty headcount before deducting total healthcare costs was 9.48%, rising to 10.65% after deducting total healthcare costs (Table 7). The study identified a 1.17%-point increase in the number of households impoverished over the past month directly attributable to total healthcare expenditures, with a 1.02%-point increase due to NCD-related expenses. Thus, the proportion of people living in poverty increased by 12.3% due to health care expenditures.

Table 7.

Impoverishment impact on households (n = 1276)

Poverty index	Impoverishment impact of total health expenditures		Impoverishment impact of NCD-associated health expenditures
	Frequency	Proportion	Frequency	Proportion
Pre-payment	121	9.48	121	9.48
Post-payment	136	10.65	134	10.50
Impoverishment impact	1.17		1.02

Before deducting total healthcare costs, the poverty gap was 0.015, indicating an average income shortfall of 1.5% relative to the poverty line among those below it. After accounting for healthcare expenditures, the poverty gap increased to 0.019, reflecting a 1.9% income shortfall. Similarly, the squared poverty gap rose from 0.003 to 0.005 after deducting healthcare costs, indicating a heightened intensity of income shortfall among impoverished households.

Coping strategies of household members to manage NCD healthcare expenses

Table 8 compares the coping strategies of household members with NCDs in households that experienced CHE versus those that did not. Both groups primarily rely on savings or income, with 86.9% of CHE households and 88.6% of non-CHE households using this strategy. About 18% of CHE households and 8% of non-CHE households relied on loans and borrowing. The proportion of government health insurance is similar in both groups. The reliance on government social security programs is low for both (2.9% for CHE vs. 3.2% for non-CHE), and few households sell assets, with slightly higher asset sales among CHE households (1.1% vs. 0.4%).

Discussion

The primary aim of this study was to investigate the financial hardship, in terms of CHE and impoverishment, experienced by households due to healthcare expenditures, with a focus on NCDs. We will discuss the extent of financial hardship, the determinants contributing to CHE, the implications of our findings for Nepal’s decentralized health system, and the strengths and limitations of our study.

Catastrophic impact of health expenditure and NCDs

The findings indicate that, out of total healthcare expenditures for NCDs, over 60% were allocated to medication costs and 17.3% to curative care. This aligns with national data from the Nepal National Health Accounts (2019/20), which reported that pharmaceuticals and medical supplies account for 65.6% of out-of-pocket payments, followed by curative care at 26.3% [44]. Our study revealed that among all sampled households and those experiencing illnesses, the proportions experiencing CHE were 9.8% and 14.6%, respectively. Among 125 households with CHE, 82 incurred it due to NCD expenditures, representing 6.4% of the total sample and 9.6% of households with reported illnesses. As NCD-related costs are recurrent, they impose a higher long-term economic burden compared to acute illnesses. Cross-country comparative studies also highlight that NCDs account for over half of CHE cases [45]. The choice of CHE thresholds matters in estimating financial hardship. This study uses a 10% total household expenditure threshold, which is more sensitive to economic vulnerability. However, a higher cut-off, such as 40% of non-food expenditure, yields a more conservative estimate of catastrophic financial burden [46]. While a 10% cut-off will tend to overestimate CHE by including situations where households rely on coping mechanisms such as savings or borrowing, the 40% threshold could underestimate financial strain by excluding essential food expenditures [47].

Impoverishment impact of health care expenditures and NCDs

Healthcare expenditures were found to have a significant impact on poverty levels. The poverty headcount increased from 9.48% to 10.65% after accounting for healthcare costs, with a 1.02%-point attributable to NCD-related spending. Overall, these expenditures resulted in a 12.3% increase in the proportion of people living in poverty. Earlier research conducted in Nepal revealed a downward trend in the incidence of impoverishment due to health expenditures, with a reported rate of 1.6% in 2010 [48]. Yet, current regional estimates indicate that 10.7% of Nepalese households allocate over 10% of their total expenditure to healthcare, with an associated impoverishment rate of 1.7% [49]. In Nepal, 87% of households affected by CHE relied on savings, and 18% borrowed money to cover healthcare costs, pointing to the unsustainable nature of these coping strategies. Studies from Nepal [50] and India [51] confirm these coping strategies. While occasional borrowing may suffice for acute illnesses, recurrent NCD-related expenses can lead to financial depletion. Evidence of the impact of OOPE on poverty is also available from global and regional studies. A survey based on macroeconomic data from 145 countries (2000–2017) found that direct healthcare costs contribute to increased poverty rates [52]. Furthermore, the study reveals an increase in both the poverty gap (from 1.5% to 1.9%) and the squared poverty gap (from 0.003 to 0.005) after accounting for healthcare expenditures. This highlights the role of healthcare costs in perpetuating poverty, a trend consistent with the findings in India [53].

CHE and NCDs as the key independent variable

Households with more than two members suffering from NCDs were found to have 3 times higher odds of experiencing CHE. This finding is consistent with previous studies [45, 54–56] and emphasizes the cumulative financial burden of managing multiple NCDs within a household. Specifically, households with members diagnosed with heart disease had double the odds of incurring CHE, corroborating findings from other studies conducted in similar settings [57, 58]. After adjusting for other variables, households with members suffering from hypertension exhibited an inverse relationship with CHE, contrasting with findings from other studies [59, 60]. This discrepancy may be attributed to the low cost of diagnosing, medicating, and treating hypertension, which are provided as free services by the Government of Nepal [61]. Moreover, studies have shown a higher prevalence of hypertension among the wealthiest quintiles in Nepal [62]. Similar trends have been observed in neighboring India [63]. Households with members who have diabetes showed a significant association with CHE (AOR 1.13, 95% CI: 1.05–1.21), as consistent with findings from other studies [64, 65].

CHE and covariates

Households affected by acute illness had 1.17 higher odds of experiencing CHE. Studies have revealed that the prevalence of CHE and impoverishment was higher with tertiary care hospitals and increased duration of hospitalization for acute injuries [66, 67]. Economic disparities also played a crucial role, with households in the lowest expenditure quintile facing twice the odds of experiencing CHE compared to those in the highest quintile, as also depicted in other studies [68, 69]. This disparity highlights the inequitable distribution of healthcare costs and underscores the vulnerability of low-income households. Additionally, age-specific analyses revealed a significant association between CHE and households with members aged 60 years and older in the crude analysis, but this association weakened upon adjustment. This finding is consistent with previous studies [70, 71]. Insignificant associations may be due to the presence of pensions and social security allowances provided to the elderly population in Nepal, which could have mitigated the financial impact of health expenditures for this age group. Moreover, considering the correlation between age and NCDs [72], this factor may already be accounted for through the inclusion of other variables. Similarly, an inverse association was observed for households with children under 5 years of age, unlike other studies [73, 74]. This could be attributed to the Government of Nepal’s provision of free healthcare services to this age group through the Maternal and Child Health Program [75, 76]. Lastly, our study found a protective association between household size and CHE, inconsistent with another study [77]. This could be because household members, regardless of their number, contribute financially to the family, thereby spreading the financial burden across multiple earners within the household. Janajati and Dalit/other households faced lower odds of experiencing CHE, respectively, compared to Brahmin and Chettri households. This disparity can be attributed to the financial vulnerability faced by marginalized communities, which often results in limited access to healthcare services [78, 79]. Households enrolled in the NHIP had 1.18 higher odds of experiencing CHE. This could be because households with greater healthcare needs are more likely to enroll in NHIP, as shown by other studies [80, 81].

Strengthening financial protection in Nepal’s decentralized health system

Nepal’s decentralized health system presents a strategic opportunity to alleviate the financial burden of healthcare, particularly for NCDs, through targeted policies and localized interventions. This study identifies households with multiple NCD cases, heart disease, diabetes, acute illness/injuries, and within the lowest socioeconomic quintiles as being at greater risk of CHE. Out-of-pocket expenses, especially for medication and curative care, impose substantial financial hardship. Expanding subsidies for NCD medications and treatments across all healthcare levels could ease this burden. The findings also highlight socioeconomic disparities, emphasizing the need for targeted financial support to prevent further impoverishment. Incorporating these insights into local health planning and resource allocation can strengthen decentralized governance, promoting equitable and sustainable financial risk protection.

Strengths and limitations

A key strength of this research is its focus on NCD, which has substantial policy implications. The study employed a standardized tool based on the validated third NLSS questionnaire, which was adapted to align with the project’s objectives and the local context of Nepal, ensuring cultural sensitivity and relevance to Nepalese households. This adaptation enhanced the quality and applicability of the findings. To maintain data quality, the fieldwork included comprehensive training for research assistants, numerous pilot interviews, and daily oversight of data verification. With a robust sample of 1276 households and a rigorous sampling strategy, the findings are highly generalizable in Pokhara Metropolitan City. By employing a context-specific poverty threshold, the study ensures a nuanced and regionally relevant assessment of healthcare expenditure-induced financial hardship.

While potential recall bias may affect the accuracy of reported healthcare expenses over the past year, efforts to mitigate this included using probing techniques and referencing specific timeframes, examples, and case scenarios. Missing data on healthcare expenses, primarily due to costs covered by family members or relatives outside the household, was minimal and addressed during data validation. Analysis proceeded by excluding these missing expenses. Additionally, healthcare expenditures may differ significantly between households seeking care at private versus public health facilities, as private healthcare tends to be more expensive. The lack of distinction between public and private care expenditures is a limitation and warrants further research. Lastly, the generalizability of the findings to diverse socio-cultural and economic contexts within the country may be limited. Considering that the nationally established poverty threshold for Pokhara metropolitan city and urban municipalities in Gandaki Province, referred to as ‘Gandaki Urban,’ is the second highest in Nepal after Kathmandu, it is presumed that the incidence of CHE in other regions would be higher than in the study area.

Conclusion

In conclusion, NCDs and their associated costs are major contributors to CHE and impoverishment. Given Nepal’s commitment to UHC, policymakers should consider enhancing financial protection mechanisms, such as subsidizing healthcare costs, particularly for medications and curative care related to NCDs. Additionally, addressing economic disparities by implementing targeted support for low-income and marginalized households could help alleviate CHE and prevent households from falling into poverty.

Authors’ contributions

SK and UE conceptualized the study while SK, UE, and SRA designed the methods and data collection tools. SK conducted statistical analysis and wrote the first draft of the paper. UE, SRA, DN, and PK reviewed the drafts and provided critical inputs on the paper. All authors read and approved the final manuscript.

Funding

This paper is a part of the corresponding author’s PhD study, supported by the Department of Public Health at Aarhus University, Denmark.

Data availability

Data is provided within the manuscript.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethical Review Board of Nepal Health Research Council (number 3064, 10 May 2023). Written informed consent was obtained from all participants. The study was conducted in accordance with the principles outlined in the Declaration of Helsinki [82].

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.