DOI: https://doi.org/10.21203/rs.3.rs-1442938/v1
Background: The share of out-of-pocket health spending (OOPS) to household consumption expenditure is very high, around 67 percent, making health care costs catastrophic and leading to impoverishment effects in the medium to longrun. In this study, we have examined the impoverishment effects of OOPS on households with elderly and without elderly.
Methods:Our study extracted households with the hospitalized member in the last 365 days (N=80105) in the first phase and thereafter, it divided households with elderly (N = 23708), and households without elderly (N = 56397). Further, we have categorized a household’s impoverishment as a binary variable (Yes or No) after paying health services. The study has employed bivariate and logistics regression to examine the relationship between outcome variables and selected socio-demographic characteristics of households.
Results. Our findings show that OOPS among elderly households is on the upswing. By controlling for other factors, households with two elderly individuals are 6% more likely to be impoverished than the households without elderly. The prevalence of poverty among households with at least one elderly person is 10.5%, while it is 9.8% among households with elderly. As a result of the greater out-of-pocket expenditure, a higher percentage of elderly households in most Indian states are impoverished than households without elderly.
Conclusion:The findings of the current study call for financial protection and regional level policies for the identified households. There is also a need to expand the insurance coverage, especially the households with elderly.
Ensuring good health to the elderly and providing a healthy environment to live with dignity is become a challenge for many developing economies [1]. Because demand for health care for the elderly has been increasing over the years, it is one of the pertinent determinants of high out-of-pocket spending. Additionally, providing better healthcare without any financial hardship to a household with an elderly population is one of the targets of the United Nation’s Sustainable Development Goals (SDGs) because the elderly without any financial risk protections is treated as vulnerable [2]. The problems related to elderly care is more severe in the country like India, which is a transition towards a become an older country by 2050 [3, 4].
The latest demographic prediction shows India’s elderly population will be around 13% in 2025 and 20% in 2050, respectively [1]. Consider the increment of older adults in any economy that leads to demand more health care because the majority of elderly people suffer co-morbidity and frequent illness in the latest stages of their lives [5]. It has been evident that half of the older adults do not have access to primary health care and who have access to health care hugely suffer from financial stress due to high medical costs [6, 7]. Literature shows that high medical expenditure due to hospitalization cost which is around 25% of total health expenditure and that make them impoverished and force them to below the poverty line [8]. Furthermore, a large portion of the elderly population from developing countries belongs to the poverty-driven society that leads to poor access to health care [9].
At present, one of the pressing challenges for developing countries like India is to provide financial protection to the elderly population, which is one of the goals of universal health coverage [10]. In the International arena, WHO (2018) has also emphasized elderly care to ensure SDGs objective 3.8.2. but many South-East Asia countries have not been able to achieve complete health coverage for those vulnerable people by 2020 [2]. Among the South-East Asia countries, around 95% of the elderly population suffers from one or more ailments, especially in India [11]. As a result, many older people with co-morbidity face substantial healthcare costs, and many families fall into poverty [5, 12].
Few pieces of literature have estimated the financial cost due to out-of-pocket health spending in India shows that there is a direct relationship between higher healthcare costs and increment in poverty [8, 10]. Similarly, few studies have also found a regional variation in India regarding medical cost burden and rate of poverty [13, 14]. Garg and Karan (2009) have found that variation in hospitalization cost in rural and urban that leads to poverty variation across states in India [13]. Similary, Ladusingh & Pandey (2013) have found that the impoverishment effect of out-of-pocket health expenditure is more among rural, poor, and poorest states, namely Odisha, Bihar, Uttar Pradesh, and Madhya Pradesh [14].
Another group of literature argues that government health financing is inadequate to provide health care to all, and especially there are no specific funds for vulnerable people like the elderly, and disabled [1]. As per the latest National Health Account estimates, India spends around 1.4% of GDP on health, and around 67% of people depend on out-of-pocket. Although the spending source from the government is low as compared to other South-East Asia countries, the Indian government has been transforming huge funds since 2005 in terms of the National Health Mission to the regional government [15]. Despite the fund flow to strengthen the health system and improve primary care at the local level, the out-of-spending burden associated with poverty effects remains a challenge for India.
Moreover, there are hardly any studies that looked at the impoverishment effects of household’s OOPHE due to more health care demand of elderly members that push them into poverty. Therefore, it is necessary to understand the exposure of elderly households in terms of abysmal expenditure and impoverishment effect on household income. Hence, the present study is undertaking to examine the impoverishment effect of out-of-pocket health expenditure in the households with elderly and households without elderly by state, region, and different socio-economic characteristics. The present study also tries to see whether there is a gender differential in terms of health expenditure. This provides an insight into the most vulnerable people and helps to make evidence-based policies and programmes to achieve universal health coverage. The uniqueness of the present study is that it has adjusted the reimbursement effect while calculating poverty and included the amount of medical insurance premium expenditure under the total health expenditure by using National Sample Survey (NSS) data, 2017-18. The study provides a piece of new evidence on how the most vulnerable households and groups have been impoverished due to a lack of access to health care at the state and regional level.
The present study has utilized the unit level, nationally representative large-scale secondary data, National Sample Survey (NSS) 75th round conducted from July 2017 to June 2018. The NSS data is provided rich quantitative information on demographic details, household characteristics, morbidity, nature of the ailment, hospitalization, nature of expenditure incurred on treatment, reimbursement, maternal and older people's health. The NSS data is released by the Ministry of Statistics and Programme Implementation (MOSPI) every five years in India. The latest 75th round survey has covered entire India using multistage stratified sampling technique and collected data from 555114 individuals from 113823 households. It provides information on medical treatment received and expenses incurred during the last 365 days [16].
This study is based on the comparison between two groups of households – households with elderly (coded as HWE) and a household without elderly (coded as HWOE). Further, the study has categorized the Indian states into four groups based on the epidemiological transition level (ETL) as suggested by [18]. Our study has extracted households with a hospitalized member in the last 365 days (N = 80105) in the first phase (Appendix Table A1). Further it divided households with elderly members (N = 23708), and households without elderly (N = 56397)
The present study has used poverty headcount ratio approach to define poverty line based on monthly per capita consumption expenditure. We have adopted the per capita household monthly consumption value of year 2011–2012 as suggested Dr. C. Rangarajan committee to calculate the impoverishment effect of OOPS.
The committee has decided a monthly per capita consumption expenditure for rural areas is INR 972 while in urban areas is INR 1407 for an individual member. These amounts have calculated for a household with five members that includes INR 4860 for rural and INR 7035 for urban respectively. We have followed the following procedure to calculate poverty headcount ratio.
Let xi be the household i's total expenditure. PL is the poverty line. Thereafter, we have defined \({P}_{i}^{gross}\)=1 if xi<\(PL\) and zero otherwise. The gross of health payments poverty headcount ratio is then expressed as
\({H}_{i}^{gross}\) = \(\frac{\sum _{i=1}^{N}{s}_{i}{P}_{i}^{gross}}{\sum _{i=1}^{N}si}\)
Where, \({s}_{i}\) is the size of the household, and N is the number of households in the sample? Then this is compared with poverty criteria to arrive at poverty levels without the impact of health expenditure. An estimate of the net of health payments poverty headcount ratio is, as expressed as
\({H}_{i}^{net}\) = \(\frac{\sum _{i=1}^{N}{s}_{i}{P}_{i}^{net}}{\sum _{i=1}^{N}si}\)
For the computation of net of health payments poverty headcount ratio, the process is the same as gross of health payments poverty headcount ratio, and here we have just replaced \({P}_{i}^{gross}\) by\({P}_{i}^{net}\). And the \({P}_{i}^{net}\)= 1 if (xi –Ti) < PL and zero otherwise. Where, Ti= {(Total Health Expenditure + Insurance Premium)-Total Reimbursement}. Thereafter, these values have been compared with the poverty criteria to arrive at the poverty level due to the impact of expenditure on in-patient care. The difference between \({H}_{i}^{net}\)-\({H}_{i}^{gross}\) gives the percentages or numbers of households that have been affected by OOPS.
The study has adopted one outcome variable - whether household impoverished or not (1 = Yes and 0 otherwise) after paying health services through out-of-pocket health spending (OOPS). We have selected certain predictor variables based on the past literature (Pal, 2012; Lara et al., 2011 and Rubin and Koellin, 1993) [19, 20, 21]. Predictor variables include elderly composition, elderly gender composition, religion, social group, principal activity, monthly per capita expenditure (MPCE), scheme coverage, place of hospitalization, households with at least one non-communicable diseases (NCDs) suffer member and households with at least one under-five children.
Based on the past literature and the characteristics of available data, the present study has employed bivariate analyses and logistics regression. Bivariate analyses have been utilized to examine the bi-variate relationships between outcome variables and selected socio-demographic characteristics of households. Logistics regression has been applied to assess the determinants of impoverishment. P-value < 0.05 has been considered statistically significant. All the analyses have been done based on the analytical weight, and sometimes, in sporadic cases, individual weight has been used. The present study has also utilized ArcGIS to prepare maps to show the regional variations among the 86 regions covered by the NSS.
Our results section is divided into four sub-sections. First, estimate households OOPS in both households with and without elderly. Second, impoverishment effects of OOPS in the Indian sample. Third, estimate the state-wise prevalence of impoverishment effects. Fourth, estimate the region-wise prevalence of impoverishment. Finally, examine the determinants of impoverishment effects of OOPS using socio-demographic characteristics of sample households.
Appendix Table A2 presents cross-tabulation results between monthly per capita consumption expenditure (MPCE) patterns with patterns of households that include households with elderly (HWE) and households without elderly (HWOE). Further, MPCE patterns include household’s total consumption expenditure (HCE), household’s health spending (HHS), household’s out-of-pocket health spending (HOOPS). Concerning household usual expenditure, it shows that average MPCE on HE is INR 2467, (INR) has been spent from a HE, while MPCE on HWE is INR 2237. MPCE (INR) from an HWE. The second row shows that the average MPCE on HHE health is INR 578 (INR) spent from a HE, while the average MPCE on HWE is INR 334 (INR) from an HWE. Similarly, in the last row, Table A2 shows that the average MPCE on HOOPS spent as out-of-pocket spending on health is INR 537 (INR) from a HE, while the average MPCE on HWE is INR 315 (INR) from an HWE. Significantly, in all three cases, the expenditure patterns have been observed higher in the households with elderly than households without elderly; even it is higher than the national average.
Figure 1 presents the prevalence of impoverishment due to out-of-pocket health spending among the households with elderly and households without elderly and the National level. Our results show that the prevalence of impoverishment among HWE is 10.5 percent, while it is 9.8 percent among HWOE and 10 percent at the National level. It is seen that the maximum percentage of households that fall into the Below Poverty Line (BPL) has been found higher among the HWE compared to HWOE and the National level. It is also observed that the percentage of households affected by impoverishment due to OOPHE has been increasing with the number of elderly persons increasing.
[Please Insert Fig. 1 here]
Table 1 presents the prevalence of households affected by impoverishment at the state level. The result shows that a higher percentage of households with elderly in most Indian states are impoverished due to higher OOPE than the households without elderly. There is a significant gap observed among the states regarding the pattern of impoverishment. For instance, states namely Chhattisgarh, Madhya Pradesh, Nagaland, Uttarakhand, Gujarat, Andhra Pradesh, Andaman & Nicobar Island, Dadra & Nagar Haveli, Daman & Diu, Punjab, and Kerala are found that households with elderly were having more impoverished. On the contrary, other states are showing the opposite result that the impoverishment effect is higher among the HWE than the HWOE. But at the national level, the impoverishment effects of OOPS is higher among household with the elderly than a household without the elderly.
|
States |
Percentage of households |
||
|---|---|---|---|
|
With elderly |
Without elderly |
Total |
|
|
Low ETL group |
10.8 |
10.6 |
10.7 |
|
Bihar |
10.5 |
12.6 |
12.3 |
|
Jharkhand |
9.4 |
9.7 |
9.6 |
|
Uttar Pradesh |
11.6 |
11.7 |
11.7 |
|
Rajasthan |
6.5 |
6.5 |
6.5 |
|
Meghalaya |
1.7 |
0.6 |
0.7 |
|
Assam |
9.4 |
7.5 |
7.9 |
|
Chhattisgarh |
16.3 |
11.6 |
12.5 |
|
Madhya Pradesh |
11.0 |
8.2 |
8.9 |
|
Orissa |
13.2 |
15.1 |
14.5 |
|
Lower-middle ETL group |
6.9 |
4.3 |
5.0 |
|
Arunachal Pradesh |
4.4 |
4.2 |
4.2 |
|
Mizoram |
2.4 |
1.5 |
1.7 |
|
Nagaland |
5.8 |
2.8 |
3.2 |
|
Uttarakhand |
9.3 |
7.4 |
8.0 |
|
Gujarat |
6.9 |
3.4 |
4.3 |
|
Tripura |
4.6 |
6.9 |
6.3 |
|
Sikkim |
4.0 |
5.3 |
5.0 |
|
Manipur |
8.7 |
8.8 |
8.7 |
|
Higher-middle ETL group |
10.7 |
10.0 |
10.2 |
|
Haryana |
4.7 |
4.2 |
4.3 |
|
Delhi |
3.4 |
4.2 |
3.9 |
|
Telangana |
8.1 |
12.1 |
11.4 |
|
Andhra Pradesh |
15.3 |
12.6 |
13.4 |
|
Jammu & Kashmir |
4.2 |
5.5 |
5.1 |
|
Karnataka |
9.9 |
8.7 |
9.1 |
|
West Bengal |
10.1 |
10.0 |
10.0 |
|
Maharashtra |
12.9 |
11.6 |
12.1 |
|
Lakshadweep |
5.8 |
5.1 |
5.4 |
|
Pondicherry |
7.1 |
9.5 |
8.7 |
|
Andaman & Nicobar Isl |
9.0 |
6.0 |
6.8 |
|
Dadra & Nagar Haveli |
6.6 |
2.6 |
3.1 |
|
Daman & Diu |
9.0 |
4.3 |
4.9 |
|
Chandigarh |
5.2 |
6.4 |
6.0 |
|
High ETL group |
10.7 |
9.2 |
9.8 |
|
Himachal Pradesh |
9.5 |
9.9 |
9.7 |
|
Punjab |
11.5 |
7.2 |
8.7 |
|
Tamil Nadu |
9.5 |
9.5 |
9.5 |
|
Goa |
4.0 |
13.4 |
8.6 |
|
Kerala |
11.5 |
9.5 |
10.5 |
|
India |
10.5 |
9.8 |
10.0 |
| Source: Computed from NSS 75th round data file | |||
Our results have also estimated the impoverishment effects of OOPS of Individual states and India. States that belong to a higher percentage of impoverished households with elderly than the national figure (10.5%) are Andhra Pradesh (15.3%), Orissa (13.2%), Maharashtra (12.9%), Uttar Pradesh (11.6%), Kerala (11.5%), Punjab (11.5%), Madhya Pradesh (11.0%) and Bihar (10.5%). Similarly, in the household without elderly, the impoverishment effect is higher of those states than the national level (9.8%) is Goa (13.4%), Bihar and Andhra Pradesh (12.6%), Telangana (12.1%), Rajasthan, Maharashtra, and Chhattisgarh (11.6%), and West Bengal (10.0%). But our results have found fewer variations in the rate of impoverishment among the ETL groups in India.
Regional variation has been observed between the mean out-of-pocket health expenditure and the impoverishment effect in the geographical regions of India (See Fig. 2–4). Figure 2–4 shows a significant variation in expenditure on health as a ratio of OOPS across regions in India. We have found a gender gap in health spending among households with the elderly. For instance, any regions spent regions namely Coastal Northern Tamil Nadu, Northern Bihar, Mahanadi Chhattisgarh, Central Himachal Pradesh, Coastal north Andhra Pradesh, Northern Madhya Pradesh, Lakshadweep, Coastal Tamil Nadu, Coastal Tamil Nadu, and Ranchi Plateau are shown a higher OOPE on male elderly than the female elderly.
Finally, Fig. 2–4 shows the regional variation of impoverishment effect due to OOPE. In the previous analysis, we have figured out that the impoverishment effect is more among the households with elderly than households without elderly (See Table 2). But here, we have found significant regional variations in impoverishment rate due to OOPS (See Figure A3a)., For instance, regions namely Ladakh of J&K (34.8%), South-eastern Rajasthan (21.8%), Northern Chhattisgarh (21.0%), Inland Eastern of Maharashtra (19.5%), Inland Northern of Maharashtra (18.8), Malwa of Madhya Pradesh (18.6%), Coastal Odisha (17.7%), south and inland south of Andhra Pradesh (16.7%), Eastern Maharashtra (16.3%), Mahanadi of Chhattisgarh (15.9%) are showing a significant percentage of households impoverished due to OOPHE. On the contrary, regional variations are also observed in the households without the elderly (See Table 2 and Fig. 3. b). The list of regions without elderly households which shows a higher rate of impoverishment due to OOPHE includes Inland North-Eastern Telangana (19.8%), Inland Eastern of Maharashtra (19.4%), Coastal Odisha (18.5), Himalayan West Bengal (18.0%), Malwa of Madhya Pradesh (16.4%), Northern Chhattisgarh (15.8%), Northern and southern Bihar (14.6%), Coastal Tamil Nadu (14.1%), Northern upper Ganga Plains of UP (13.6%).
|
Rank |
Regions |
Impoverishment effect |
Rank |
Regions |
Impoverishment effect |
|
|---|---|---|---|---|---|---|
|
Households with elderly |
Households without elderly |
|||||
|
1 |
Ladakh of J&K |
34.8 |
1 |
Inland North Eastern Telangana |
19.8 |
|
|
2 |
South-eastern Rajasthan |
21.8 |
2 |
Inland Eastern of Maharashtra |
19.4 |
|
|
3 |
Northern Chhattisgarh |
21.0 |
3 |
Coastal Odisha |
18.5 |
|
|
4 |
Inland Eastern of Maharashtra |
19.5 |
4 |
Himalayan West Bengal |
18.0 |
|
|
5 |
Inland Northern of Maharashtra |
18.8 |
5 |
Malwa of Madhya Pradesh |
16.4 |
|
|
6 |
Malwa of Madhya Pradesh |
18.6 |
6 |
Northern Chhattisgarh |
15.8 |
|
|
7 |
Coastal Odisha |
17.7 |
7 |
Bihar |
14.6 |
|
|
8 |
south and inland south of Andhra Pradesh |
16.7 |
8 |
Coastal Tamil Nadu |
14.2 |
|
|
9 |
Eastern Maharashtra |
16.3 |
9 |
Coastal north Andhra Pradesh |
14.1 |
|
|
10 |
Mahanadi of Chhattisgarh |
15.9 |
10 |
Northern upper Ganga Plains of UP |
13.6 |
|
The prevalence of impoverishment effect at the socio-economic and demographic characteristics has been presented in Table 3. The impoverishment effect is found to be higher among HWE as compared to HWOE across all demographic and socio-economic characteristics except the households with regular wages, from Schedule Caste (SC) category, sale physical assets to meet health care, use public hospital and Charitable/Trust/NGO. With the consideration of HE, impoverishment effect has been found higher among the households belonging to a small size (16.6%), another type (14.0%), rural area (11.0%), Christian religion (11.8%), other social categories (11.0%), lowest wealth quintile (14.6%), practising selling physical assets (43.3%), preferring private hospital (16.1%), no insurance coverage (10.7%), not having at any under-five children (13.2%) and having at least one non-communicable disease suffered member (16.6%) and Central geographic region (12.2%). Additionally, we have found a similar pattern of impoverishment in the context of HWOE, except households belonging to the Hindu religion (10.5%), lower wealth quintile (12.5%), and east geographic region (11.7%).
|
Household Characteristics |
Description |
Percentage of Households |
||
|---|---|---|---|---|
|
With Elderly |
Without Elderly |
Total |
||
|
Household Size |
1–3 |
16.6 |
13.8 |
14.5 |
|
4–6 |
11.6 |
9.8 |
10.3 |
|
|
7–9 |
6.4 |
5.8 |
6 |
|
|
10 and above |
3.8 |
2.4 |
3.2 |
|
|
Household Type |
Self-Employed |
10.4 |
9.8 |
10 |
|
Regular Wage/Salary |
7.9 |
8.8 |
8.5 |
|
|
Labour |
11.3 |
10.5 |
10.6 |
|
|
Other |
14 |
11 |
12.9 |
|
|
Household Head |
Male |
10.8 |
9.9 |
10.1 |
|
Female |
8.4 |
9.3 |
9.1 |
|
|
Place of Residence |
Rural |
11 |
10.9 |
10.9 |
|
Urban |
9.6 |
7.4 |
8.1 |
|
|
Religion |
Hinduism |
10.9 |
10.5 |
10.6 |
|
Islam |
7.7 |
7.3 |
7.4 |
|
|
Christianity |
11.8 |
7.6 |
9.3 |
|
|
Other |
9.9 |
7.1 |
8.1 |
|
|
Social Category |
ST |
8.8 |
8 |
8.1 |
|
SC |
9.8 |
10 |
9.9 |
|
|
OBC |
10.6 |
10 |
10.2 |
|
|
Others |
11 |
10.1 |
10.4 |
|
|
Wealth Quintile |
Lowest |
14.6 |
11.3 |
12.2 |
|
Lower |
12.2 |
12.5 |
12.4 |
|
|
Middle |
8.4 |
9.6 |
9.3 |
|
|
Higher |
8.7 |
10 |
9.6 |
|
|
Highest |
9.6 |
6.4 |
7.4 |
|
|
Major Source of Finance |
Savings |
8.9 |
8.5 |
8.7 |
|
Borrowing |
23.7 |
21.2 |
22 |
|
|
Sale Physical Assets |
43.3 |
48.1 |
46.2 |
|
|
Contribution from Other |
12.5 |
11 |
11.5 |
|
|
Place of Hospitalized |
Public |
4 |
4.7 |
4.5 |
|
Charitable/Trust/NGO |
8.8 |
12.5 |
11 |
|
|
Private |
16.1 |
16.4 |
16.3 |
|
|
Health Insurance Coverage |
Household with Elderly Coverage |
9.9 |
- |
9.9 |
|
Household without Elderly Coverage |
- |
9.3 |
9.3 |
|
|
Not Coverage |
10.7 |
10 |
10.2 |
|
|
Household with at least 1 Under 5 Children |
No |
13.2 |
13.4 |
13.3 |
|
Yes |
7.2 |
7.2 |
7.2 |
|
|
Household with at least 1 NCD member |
No |
7.1 |
7.6 |
7.5 |
|
Yes |
16.6 |
15.7 |
16.1 |
|
|
Region of India |
North |
9 |
9.1 |
9.1 |
|
East |
10.8 |
11.7 |
11.5 |
|
|
North-East |
7.5 |
6.4 |
6.6 |
|
|
Central |
12.2 |
9.1 |
9.8 |
|
|
West |
11.3 |
8.8 |
9.7 |
|
|
South |
11.2 |
10.3 |
10.6 |
|
|
India |
10.5 |
9.8 |
10 |
|
| Source: Computed from NSS 75th round data file | ||||
Table 4 presents the possible factors for the impoverishment of households due to high OOPS using Indian samples. We have estimated two models. First, estimation done with unadjusted effect, while in the second, we have used with the adjusted impact of impoverishment on elderly composition in households. The first model shows that the probability of the impoverishment effect has increased with the increasing number of older people. For instance, households with two elderly people are 11% more likely to get affected by impoverishment than the HWOE. Our result is statistically significant, but in this model, we have not controlled other socio-economic and demographic parameters.
|
Variables |
AOR |
95% CI |
|
|---|---|---|---|
|
Lower Bound |
Upper Bound |
||
|
Number of Elderly within Households (0) |
|||
|
1 |
0.99*** |
0.99 |
0.99 |
|
2 |
1.06*** |
1.06 |
1.07 |
|
3 and above |
2.05*** |
2.02 |
2.08 |
|
Head of the Household (female) |
|||
|
Male |
1.27*** |
1.27 |
1.28 |
|
Household Size (10 and above) ® |
|||
|
1–3 |
8.39*** |
8.32 |
8.46 |
|
4–6 |
4.68*** |
4.64 |
4.72 |
|
7–9 |
2.21*** |
2.19 |
2.23 |
|
Household Type (Self-Employed) ® |
|||
|
Regular Wage/Salary |
1.00 |
1.00 |
1.00 |
|
Labour |
1.05* |
1.04 |
1.05 |
|
Other |
1.02* |
1.01 |
1.02 |
|
Place of Residence (Urban) ® |
|||
|
Rural |
2.13*** |
2.12 |
2.13 |
|
Wealth Quintile (Highest) ® |
|||
|
Lowest |
4.37*** |
4.36 |
4.39 |
|
Lower |
3.84*** |
3.82 |
3.85 |
|
Middle |
2.17*** |
2.16 |
2.18 |
|
Higher |
1.82*** |
1.82 |
1.83 |
|
Religion (Hindu) ® |
|||
|
Islam |
0.72*** |
0.72 |
0.72 |
|
Christianity |
0.84* |
0.84 |
0.85 |
|
Other |
0.77* |
0.76 |
0.77 |
|
Social Category (ST) ® |
|||
|
SC |
0.95*** |
0.94 |
0.95 |
|
OBC |
0.99*** |
0.98 |
0.99 |
|
others |
1.15*** |
1.15 |
1.16 |
|
Major Source of Finance (Savings) ® |
|||
|
Borrowing |
2.05*** |
2.05 |
2.06 |
|
Sale Physical Assets |
5.65* |
5.55 |
5.74 |
|
Contribution from other |
1.36 |
1.35 |
1.36 |
|
Place of Hospitalization (Public) ® |
|||
|
Charitable/Trust/NGO |
3.15** |
3.13 |
3.17 |
|
Private |
5.19** |
5.17 |
5.2 |
|
Health Insurance Coverage (Govt Sponsored) |
|||
|
Other Than Govt |
0.89*** |
0.88 |
0.89 |
|
Not Covered |
1.11** |
1.1 |
1.11 |
|
Household with at least 1 U5Children (No)® |
|||
|
Yes |
0.85* |
0.84 |
0.86 |
|
Household with at least 1 NCD member (No) |
|||
|
Yes |
1.94*** |
1.93 |
1.94 |
|
Sate based on ETL (Lower ETL) ® |
|||
|
LMETL |
0.63*** |
0.62 |
0.63 |
|
HMETL |
0.99*** |
0.98 |
0.99 |
|
HETL |
1.16*** |
1.16 |
1.17 |
|
Region (North) ® |
|||
|
East |
1.26*** |
1.25 |
1.26 |
|
North-East |
1.39*** |
1.38 |
1.4 |
|
Central |
1.02*** |
1.01 |
1.02 |
|
West |
0.98*** |
0.98 |
0.99 |
|
South |
0.92*** |
0.92 |
0.93 |
| Note: ®: reference category; ***p < 0.001, **p < 0.01, *p < 0.05; Sample Size (N) = 80105; CI: Confidence Interval, AOR: Adjusted odds ratio. | |||
However, in the second model, we have found that the households with two elderly people are 6% more likely to get affected by impoverishment than the HWOE by controlling other factors. Table 4 also shows another significant result that the households which have male elderly members are more (OR: 1.17; P > z = 0.00) likely to be affected by impoverishment than the female elderly households. Additionally, we have found a better result in the case of place of resident, wealth quintile, social groups, assets, and geographic regions. It shows that households from rural (OR: 2.13; P > z = 0.00), lowest wealth quintile households (OR: 4.37; P > z = 0.00), other social categories (OR: 1.15; P > z = 0.00), selling physical assets (OR: 5.65; P > z = 0.00), preferring private hospital (OR: 5.19; P > z = 0.00), not insurance coverage (OR: 1.11; P > z = 0.00), having NCD suffered member (OR: 1.94; P > z = 0.00), high ETL group (OR: 1.16; P > z = 0.00), and North-East (OR: 1.39; P > z = 0.00), East (OR: 1.26; P > z = 0.00) and Central regions (OR: 1.02; P > z = 0.00) are more likely to be affected by impoverishment than the reference categories.
The improvement of life expectancy and increase of disease prevalence among older people has forced policymakers to revisit the importance of health care financing in the economy. Therefore, it is necessary to the determinants of OOPS and evaluates its impoverishment effects on households across regions of India. The current study has examined the impoverishment effects of OOPS on the elderly population in India using the latest social consumption survey, 2017-18. Our findings show that OOPS and its impoverishment effect is higher among the households having elderly than the households without any elderly member. So, it implies that the probability of getting impoverished is more among households with the elderly than the households without elderly members. We can infer that households are becoming impoverished due to higher health care spending on older people.
However, states such as Andhra Pradesh, Odisha, Maharashtra, Uttar Pradesh, Kerala, Punjab, Madhya Pradesh, Bihar, etc. have the highest percentages of households affected by impoverishment which is even higher than the national level (10.5%). These results are also consistent with the previous studies [13, 17, 2]. Similar results were also found at the state level that some states are more vulnerable to poverty due to higher OOPHE [14, 2]. Our study has also found regional variations in impoverishment effects of OOPHE. It implies that regions namely Ladakh of J&K, South-eastern Rajasthan, Northern Chhattisgarh, Inland Eastern of Maharashtra, Inland Northern of Maharashtra, Malwa of Madhya Pradesh, Coastal Odisha, south and inland south of Andhra Pradesh, Eastern Maharashtra, Mahanadi of Chhattisgarh are contributed a significant percentage of impoverished households. Further, the impoverishment effect of OOPHE is found to be higher among these regions.
Another interesting finding is the gender difference. Results suggest that the OOPS and the impoverishment effect are found to be higher among those households having male rather than a female older person. Our logistic regression has also found that households having male elderly members are more likely to get impoverished due to OOPS. Regional variation has also been observed on the line of gender differences. Most regions have spent more on male elderly than females. Our results are like past studies in SEAR regions that ageing is a pertinent indicator of a household’s OOPS [2]. Further few literatures are linked the gender difference in impoverishment to cultural barriers and male dominance in Indian society [18, 19, 20]. Past studies also found that household’s OOPS and impoverishment effect will be higher for male elderly than female elderly due to variation in the health-seeking behaviour [18, 19, 20]. So, there is a need to focus on female elderly persons in the households and provide a basic health care service.
Rural households are found to be more suffered from impoverishment than households from urban areas. Our findings corroborate with the findings from other studies [14]. It may be because of unequal spending patterns, increasing medical and hospital care costs, less investment in health insurance coverage, etc. Arora and Gumber, (2005) have mentioned that the Government of India has been spending more (nearly 75% out of total amount) on curative care in secondary and tertiary hospitals which are in urban areas, although most of the people (almost 70%) are residing in the rural area [21]. Kastor and Mohanty, (2018) have reported that reimbursement accounts very small portion of household health spending in India [20]. A study by Jayakrishnan et al., (2016) based on the NSSO survey conducted from January to June 2014, stated that only 12 percent of urban and 13 percent of the rural population received any protection coverage through any policy like "Rashtriya Swasthya Bhima yojana" (RSBY) [22]. Hence, the current study has demanded the attention of such a group of households living in rural areas.
Similarly, a significant gap has been observed between rich and poor from the above findings. Households from the poor wealth quintile are more likely to suffer from the impoverishment effect as compared to households from the rich wealth quintile. As a significant portion of the poor group belongs to rural areas, similar reasons could also be here, like unequal spending patterns, increasing medical and hospital care, and less investment in health insurance coverage. Our finding is like past studies that rural poor are more vulnerable to the impoverishment effect of OOPHE [13, 1]. Hence, much attention is needed on the poor rural household.
Another important finding is the utilization of health care institutes and the impoverishment effect. A higher impoverishment effect has been found among the household that utilized the private hospital rather than public hospitals. It is obvious that urban have a good number of private hospitals; along with, the health facilities are also good in the private hospitals and people are preferred to get good health facilities. Therefore, they have to go to urban or private hospitals and face the enormous health expenditure in private hospitals compared to public hospitals and resultant impoverishment. Furthermore, households with small size, other or labour workers, other castes, Christian religion, selling physical assets or borrowing and have not insured member, etc., are found to be more suffered from impoverishment. Results from logistic regression are also consistent with the bivariate analysis. These results are also consistent with the previous studies [13, 23].
Overall, the study has found that the households with elderly members are more suffered from the impoverishment effect. There are gender differences; particularly, male elderly households are suffered more in terms of impoverishment effect than the households with female elderly. The findings of the current study call for financial protection and regional level policies for the identified households. Along with, improving the health care services and protecting households from catastrophic health care payments and impoverishment effects. It is also essential to improve public hospital facilities on the line of the private institute so that people can get good health facilities at a subsidized rate.
There is also a need to cover most of the population by giving insurance and reimbursement. The present study has demanded significant attention on these grounds and groups to halt households to face the abysmal health expenditure and hidden impoverishment effect. Like other studies, our study is not free from limitations that could be analysed in the future. The present study has utilized only inpatient care which is influenced by recall bias. Hence, one can also use outpatient care. Furthermore, due to lack of data, the present study could not incorporate the non-food expenditure; hence one may also use the non-food expenditure for better examination of the effect of health expenditure. The present study did not adjust the poverty line by state and regional level to estimate of poverty line and impoverishment effect. It has used national criteria to define the poverty line at the state and regional levels.
Acknowledgements:
Not applicable
Authors’ contributions
P.B., H.S. and M.D. conceptualised the study, and P.B. conducted all the data analyses, D.K.B., P.B., and M.D. analysed the results and drafted the manuscript, H.S. guided the literature review, helped draft the manuscript. All authors read and approved the manuscript.
Availability of data and materials
It is open-source government data, easily available on online under the Ministry of Statistics and Programme Implementation (MOSPI). https://www.mospi.gov.in/web/mospi/download-tables-data
Funding
None
Ethics approval and consent to participate
The source of this secondary data analysis was a national survey conducted by the Government of India, and the anonymized dataset is available in the public domain. Therefore, an ethical review was not deemed necessary.
Consent for publication
Not applicable
Competing interests
None of the authors has conflicting interests.