A Study of Catastrophic Health Expenditures in India - Evidence from nationally representative survey data: 2014-2018

Introduction

Goal 3 of the United Nation’s Sustainable Development agenda has the specific goal to provide universal health coverage (UHC) to its population and to improve financial protection. UHC includes securing access to quality healthcare and safe, affordable medicines and vaccines for everyone (Saksena et al., 2014). Resolution 58.33 of the World Health Assembly recommends that all WHO member states should provide UHC to their entire population and protect households from catastrophic health expenditures (CHE) (Obermann et al., 2018). Any household out-of-pocket (OOP) health spending that exceeds a certain proportion of household’s financial ability is defined as CHE (Xu et al., 2003). Globally around 100 countries have achieved UHC or initiated reforms towards UHC (Obama, 2008; Summers, 2015). Even though most countries are striving to enable their citizens to obtain the healthcare they need without financial barriers, 808 million people still experience CHE each year (Wagstaff et al., 2018). More than 90% of the people experiencing CHE live in low-income countries (Xu et al., 2003). Financial protection needed to a population depends on their dependence on OOP health expenditure for paying for healthcare (Xu et al., 2003). Dependence of the households on OOP payments for obtaining healthcare escalates the financial burden of the households (Amaya et al., 2011; Wagstaff et al., 2003; Xu et al., 2003). With OOP health expenditures constituting around 62.6% of total health expenditures, India ranks third in the Southeast Asia region among countries with high OOP health expenditure (Balarajan et al. 2011; Hooda, 2017; Sriram 2018).

Many studies have examined the health expenditures on specific diseases such as diabetes, tuberculosis, cancer, injuries etc., but the problem was that most of these studies were done in small geographical areas of the country and their representativeness for the whole nation was limited (Binnendijk et al., 2012; Yesudian et al., 2014; Rao et al., 2011; Prinja et al., 2015; Muniyandi et al., 2005; Ramachandran et al., 2007). Some studies have examined the determinants of OOP health expenditures for outpatient care in a few districts of India for certain age groups (Brinda et al., 2012; Gupta et al., 2016). Also, other studies have used different National Sample Survey Organization (NSSO) datasets and other nationally available data like National Family Health Survey (NFHS), etc. to study disease-specific OOP health expenditures for hospitalizations (Kastor & Mohant, 2018). Studies have looked at OOP health expenditures due to non-communicable diseases (NCDs) (Tripathy et al., 2016), burden of OOP payments due to medicines (Selvaraj et al., 2018), OOP health expenditure for maternal care (Mohanty & Kastor, 2017), OOP health expenditure for accidental injury (Pradhan et al., 2017), but they did not address the specific research questions related to CHE in general and factors affecting incidence and depth or gap of CHE.

The primary objective of the study is to detect the features of households, health conditions, and health delivery system issues that make people prone to CHE. In particular, the study will examine the association of households’ demographic characteristics, social structure, and healthcare utilization that appear to be associated with relatively high levels of expenditure and also quantify the burden of OOP health expenditures and CHE. In this research, we used the data from two surveys, the 2014 and 2018 NSSO data to assess the level of financial protection in India (NSSO, 2014). To measure the effect of CHE on households, we estimate (i) incidence, intensity, and inequality in CHE in India (ii) incidence and intensity of CHE across different states in India (iii) various aspects influencing the incidence and intensity of CHE in India.

India is taking steps to provide UHC to its citizens. By identifying the incidence, intensity, socioeconomic inequalities in CHE, this study helps the central government provide an appropriate higher budgetary allocation for the groups that have higher OOP health expenditures and helps in designing health insurance programs that benefit the poor. This research will help in reducing catastrophic spending in India. Andersen’s Behavioral Model of Healthcare Utilization as shown in Figure 1 will be used to guide this research (Andersen, 1995). The Andersen model examines the predisposing, enabling, need and healthcare utilization characteristics.

Figure 1. Determinants of household’s OOP health expenditures using Andersen’s behavioral model of healthcare utilization.

Data

Methods

Results

Descriptive statistics presented in Table 1. There were 65,932 households in the sample in the 2014 data and 113,823 households in the 2018 data. 33% of the households (2014) and 25% of households (2018) have at least one child aged 5 years and less; 26.87% households (2014) and 22.92% households (2018) have at least one elderly person. The proportion of households located in rural and urban areas are almost same in 2014 and 2018. About 9.98% of the households (2014) and 15.40% of households (2018) have members who used a private hospital.

Incidence and intensity of catastrophic health expenditures by state

Figure 2 and Figure 3a show the incidence of CHE for the states in India for the years 2014 and 2018. Figure 2 and Figure 3b show the intensity of CHE for the states in India for the years 2014 and 2018. Figure 4 shows the change in CHE between 2014 to 2018 across the different states. In 2018, Kerala (33.77%) has the highest incidence of CHE, while Meghalaya (1.83%) has the lowest incidence of CHE. In 2014, Kerala (15.71%) has the highest incidence of CHE, while Chandigarh (2.80%) has the lowest incidence of CHE. In 2018, Arunachal Pradesh (53.86%) has the highest intensity of CHE, while Daman and Diu (6.45%) has the lowest intensity of CHE. In 2014, Chhattisgarh (47.38%) has the highest intensity of CHE, while Daman and Diu (10.55%) has the lowest intensity of CHE. Kerala (18.06%) experienced the highest increase in incidence of CHE from 2014 to 2018, while Goa (5.85%) experienced a decline in incidence of CHE from 2014 to 2018. Manipur (24.75%) experienced the highest increase in intensity of CHE from 2014 to 2018, while Uttaranchal (24.67%) experienced a decline in intensity of CHE from 2014 to 2018. In 2014, households in the richest expenditure quintile have the highest incidence of CHE, while in 2018, households all the income expenditure quintiles incur almost more or less same incidence of CHE as shown in Table 4.

Figure 2. Incidence and intensity of CHE by state.

Figure 3. a: Map showing incidence of CHE by states, 2014 and 2018. b: Map showing intensity of CHE by states, 2014 and 2018.

Figure 4. Change in incidence and intensity of CHE by states: 2014-2018.

Table 4. Headcount and overshoot of CHE by socioeconomic status and location.

Threshold	2014			2018
	Rural	Urban	Total	Rural	Urban	Total
Headcount
Poorest	9.61%	7.33%	9.22%	15.98%	13.86%	15.64%
Second	10.51%	8.68%	10.05%	17.28%	16.52%	17.12%
Third	10.93%	10.24%	10.71%	16.54%	17.75%	16.94%
Fourth	14.24%	12.66%	13.50%	18.09%	16.24%	17.18%
Richest	18.61%	11.70%	13.82%	20.70%	14.57%	16.00%
CI	0.0910	0.0904	0.0848	0.0268	-0.0078	0.0113
SE (CI)	0.0081	0.0096	0.0062	0.0048	0.0058	0.0037
Overshoot
Poorest	60.20%	43.92%	58.03%	47.68%	54.70%	48.67%
Second	31.12%	43.78%	33.84%	31.20%	39.04%	32.80%
Third	23.75%	28.21%	25.10%	30.46%	26.30%	29.04%
Fourth	21.76%	28.81%	24.86%	25.09%	24.24%	24.70%
Richest	28.37%	32.92%	31.04%	23.36%	21.27%	21.90%
MPO	36.91%	33.78%	35.94%	35.82%	30.25%	34.08%
CI	-0.1328	0.0277	-0.0882	-0.1089	-0.1836	-0.1440
SE (CI)	0.0242	0.0239	0.0177	0.0128	0.0220	0.0111

Discussion

In 2014, among the households which experienced CHE, the mean positive overshoot indicates that on average, the OOP health expenditures was 35.94% higher, but the overshoot decreased to 34.08% in 2018. This shows that although the intensity is very high among the households experiencing CHE, but it is decreasing. Our study showed that both in 2014 and 2018, a higher odds of incidence of CHE among the households with children, elderly people and those who used a private hospital for treatment. This was consistent with literature which showed that households which consisted of members at extremes of age (Mohanty et al., 2014; da Silva et al., 2015), members utilization of private health facility (Alam & Mahal 2014; Saksena et al., 2012; Somkotra & Lagrada, 2009) had higher OOP and CHE. The likelihood of incidence and intensity of CHE in our study increased progressively with the increase in duration of hospital stay. However, the effect of duration of hospitalization on incidence of CHE was much lower in 2018 compared to 2014, but the intensity of CHE was much higher in 2018 compared to 2014. This shows that health insurance and other health financing protection mechanisms may have been successful in reducing the number of households being pushed to experience CHE, but among the households that experienced CHE, the burden (overshoot) has increased greatly over this time-period. A World Bank study showed that hospitalizations are the major drivers of OOP health expenditures (McIntyre et al., 2006). Also, the presence of chronic illness among members in the household increased odds of CHE incidence and also increased the intensity among households experiencing CHE. Similar results were found in India (Mondal et al., 2014), Bangladesh (Molla et al., 2017), and China (You & Kobayashi, 2011) that showed that chronic illness is an important determinant for experiencing CHE. Our study showed that the presence of health insurance coverage among members in the household reduced the likelihood of CHE incidence and even among the households that experienced CHE, the intensity was lesser for households that had health insurance coverage. Other studies from India (Fan et al., 2012), Indonesia (Aji et al., 2013), Laos (Alkenbrack & Lindelow; 2015), and Vietnam (Sepehri 2013) supported this finding of the protective effect of health insurance from CHE.

The regression results show that the households from all other expenditure quintiles had lesser odds of incurring CHE compared to poorest households. Among the households that experienced CHE, the intensity was also highest among the poorest households. For the poorer households, high level of intensity or overshoot may be due to low level of absolute income. It is expected that the poor people are more prone to experience CHE, since they have lower level of income and any expenditure that incur for healthcare will easily make it “catastrophic” since the proportion of the health expenditure will become relatively high for them because of low total consumption expenditure (low value of denominator). Thus, poorest people have higher risk of facing CHE even with a relatively small adverse health event. Globally there is mixed evidence on the relationship between socio-economic status and CHE. These results are similar with the findings from research done in India (Bhojani et al., 2012), Thailand, Paraguay, and Burkina Faso (Makinen et al., 2000) which showed that low-income households were associated with a higher likelihood of CHE. Other studies in Nigeria, Namibia, Albania, Kenya, Bangladesh, and India show that poorer households have lower absolute OOP health expenditures compared to richer individuals and households, but the relative proportion of OOP health expenditures to non-food household expenditures was higher in poor households (Chuma & Maina 2012; Gustafsson-Wright et al., 2011; Hotchkiss et al., 2005; Karan et al., 2014; Onwujekwe et al., 2014; Rahman et al., 2013). However, studies from 13 low-income Asian countries (O'Donnell O et al. 2008), Sri Lanka, South Africa and Guatemala (Makinen et al., 2000) showed that richer households spent more on OOP health expenditures and also enjoyed a wide range of services. There is an increase in both the incidence and intensity of CHE with higher duration of hospital stay. Higher duration of hospital stay increases the chance of experiencing CHE. When the higher health expenditures are inadequately covered by health insurance, OOP health expenditures may become catastrophic for many households.

There are wide differences among states of India in CHE experienced by them. Epidemiological transition and economic growth also lead to an increase in CHE. Higher economic growth leads to an increase in demand for healthcare services and increase in technology used in healthcare delivery. For example, Kerala being an economically advanced state experienced higher incidence of CHE. States like Meghalaya and Nagaland which are economically poor experienced lower incidence of CHE. In the developed states of India, the higher CHE may be due to increased awareness of health benefits and increased utilization of private healthcare facilities over the free public health facilities. The epidemiological and disease burden in a state also determines the healthcare utilization, type of healthcare used, and the volume of healthcare services used in the state. For example, some economically better states may suffer from higher burden of non-communicable diseases, while in contrary some of the poorer states may suffer from maternal and child health problems and infectious diseases. The type of disease experienced by the states directly determines the health expenditures experienced by them (Wagstaff et al., 2018; Kea et al., 2011; Kien et al., 2016).

Also, in India there is a very poor non-communicable disease service provision in the public sector, and this makes more people choose the private sector hospitals for non-communicable disease treatment which increases their OOP costs when not adequately covered by any health insurance. Different state health insurance programs increased the usage of more private healthcare facilities by increasing the access to them, they have not actually protected the households from the financial burden that arises due to the usage of private health facilities (Karan et al., 2017; Selvaraj & Karan, 2012; Ranjan et al., 2018).

In conclusion, this research helps policymakers to design health insurance programs to better serve the population. Health insurance benefit packages and coverage limits may be adjusted based on the income levels of poor households with the poorest group receiving the highest level of protection. This type of targeting is also difficult to implement in practice, but it is not impossible with help from community organizations representing the poor and extremely poor households. Households with children less than 5 years and elderly more than 60 years have higher CHE incidence. Children and elderly are the vulnerable age groups who are prone to higher level of health risks. They have higher healthcare utilizations and thus experience higher healthcare expenditures which make the expenditure levels catastrophic in many cases. This implies that policymakers should also consider age as an important factor for health insurance coverage. Health insurance coverage limits in India are restricted and are not adequate especially when the patients stay for longer duration in the hospitals. Thus, the coverage limits for hospital insurance needs to be increased to protect households from CHE. Also, better cost regulation of the private sector hospitals must be done. Chronic illness increases both CHE incidence and intensity. Steps should be taken for early diagnosis and treatment, to reduce the severity of illness, reduce the cost of services, and implementation of better approaches to treat them in the ambulatory settings. This research will help in developing policies to reduce OOP health expenditures in India.

Author contributions

SS was involved in Conceptualization, Data Curation, Formal Analysis, Investigation, Methodology, Project Administration, Resources, Software, Supervision, Validation, Visualization, Writing – Original Draft Preparation, Writing – Review & Editing. MA was involved in Writing – Original Draft Preparation, Writing – Review & Editing.

Data availability

The data from the National Sample Survey Organization (NSSO) of the Government of India were used for the study. Social Consumption (Health), NSS 71^st Round for 2014 and NSS 75^th Round for 2018 were used for this analysis. Both the surveys covered whole of the Indian Union. Data can be obtained from the Government of India or from the corresponding author by reasonable request.