Gender Inequality in Out-of-pocket Expenditure on Healthcare Among Persons with Disabilities in India

Abstract

Background

People with disabilities (PWDs) often face higher risk of financial burden towards healthcare. Higher out-of-pocket expenditure (OOPE) is a barrier towards accessing medical services, resulting in healthcare inequities between men and women. There is a lack of studies on gender disparities in healthcare burden among PWD in India. The present study examines gender inequality towards healthcare expenditure among PWDs in India.

Methods

Unit level data from the Persons with Disabilities in India Survey under the National Sample Survey, 76th round that covered 118,152 households involving 576,569 individuals of whom 106,894 had at least one disability was used. We performed secondary analysis using descriptive statistics, bivariate estimates, Blinder–Oaxaca decomposition and multilevel log-linear regression to meet the objective of the study.

Results

OOPE as a percent of monthly per capita expenditure was higher among males across all states of India except in Puducherry, Dadra and Nagar Haveli, Lakshadweep and Mizoram. This varied from 0.6% in Jammu and Kashmir to 20% in Nagaland. Multilevel analysis confirms that OOPE was significantly lesser among females (β = –0.070, 95% confidence interval = – 0.129 to 0.01) compared to males after controlling demographic and disability-related predictors. PWDs reported higher healthcare burden as a percent of total household expenditure in India, and relatively less amount was spent on female healthcare compared to male healthcare.

Conclusion

A substantial gender inequality for healthcare expenditure is found among PWDs in India. This makes for an urgent call to health researchers and policymakers towards gender-sensitive research, policy formulation and program implementation for PWDs in India.

Keywords

Disability out-of-pocket expenditure gender decomposition India

Key Messages

Finding clearly indicates the gender gap in both OOPE and share of OOPE on total household consumption expenditure.

Female PWDs have unmet healthcare need due to OOPE and face higher burden in accessing healthcare services in India.

Ensuring healthy life and promoting wellbeing for all at all ages (goal no. 3) and achieving gender equality and empowering all women and girls (goal no. 5) are two of the sustainable development goals (SDGs) in the UN’s agenda. To reach these goals, it is key to recognize the disadvantaged and omit using a gender lens, especially among PWDs.

There is a need for ensuring that women should be benefitted through the various government programs targeted for PWDs in India.

Introduction

Globally, 15–20% population suffers from disability, and there are about 650 million people living with any type of disabilities in the world (World Health Organization, 2011). As populations age, the rate of disability is growing, mainly in the industrialised world (World Health Organization, 2011). As per the 2011 census of India, 2.21% (26.8 million) of the total 1.21 billion population are disabled of which 56% are males (around 15 million) (National Sample Survey Office, 2016). A recent national survey on disability in India found that about 2.2% population have disability in India (NSS 76th, 2018). People with disabilities (PWDs) unduly suffer greater ill health compared to non-disabled populations mainly due to poverty (Yenimahalleli Yasar, 2011).

Review of costs related with disabilities among 10 developed and developing countries shows these are related to expenditure on healthcare including cost for routine activities such as disability-specific aid(s), assistive devices, rehabilitation, personal assistance, etc., and this varies between $1,170 and $6,952 per year (Mitra et al., 2017). Previous studies have shown that households with disabled persons suffer higher percentage of catastrophic expenditure, and proportion of their household income on out-of-pocket expenditure (OOPE) on healthcare compared to households with no disabled persons (Lee et al., 2016). In the USA, healthcare expenditure was observed to be high among PWDs compared to persons without disabilities (Mitra et al., 2009).

Studies in developing countries have shown that disability is often linked with socioeconomic disparities (Mitra & Sambamoorthi, 2006). In the Indian context, only a small segment of the PWDs in India obtain and receive any government support (Mitra & Sambamoorthi, 2006). Socioeconomic and regional inequalities are well documented for prevalence of disability in the Indian context (Bora & Saikia, 2015). Some previous studies have highlighted the role of gender in health and mortality and others have focused on gender inequality in health and disability (Joe & Rajpal, 2018; Kastor & Mohanty, 2018; Sahoo & Madheswaran, 2014; Sangar et al., 2019). However, all these studies have focused on elderly and/or adult populations.

To the best of our knowledge, very few studies in India have so far attempted to understand the gender gap in healthcare expenditure among PWDs in India. High healthcare expenditure on disability may impact overall household consumption expenditure if any member in the household suffers from disability (Lee et al., 2016). Additionally, high OOPE may act as a barrier for the use of healthcare services and further result in inadequate economic safety and healthcare inequities (Kwon, 2009).

Data and Methodology

Data Source

This study used data from the 76th round of the National Sample Survey (NSS) ‘Persons with Disabilities in India’ Survey covering all states and union territories conducted during July–December 2018 (NSS 76th, 2018). The NSS is a nationwide representative population-based survey organization which collects data on several issues like employment status, morbidity, healthcare utilization, household consumer expenditure, housing condition, disabilities conditions and others. NSS 76th round covered 8,992 village/urban blocks (5,378 rural village and 3,614 urban blocks), covering 118,152 households (81,004 rural and 37,148 urban) enumerating 576,569 individuals (402,589 rural and 173,980 urban). Our study sample contains a total of 106,894 persons (74,946 in rural and 31,948 in urban; 61,567 males and 45,305 females) who reported any type of disability (locomotor, visual, hearing, speech and language, mental retardation/intellectual disability, mental illness and other types of disability) at the time of survey (NSS 76th, 2018).

Sampling Design

Census 2011 population of villages was projected by applying suitable growth rates to serve as a sampling frame. A stratified two-stage sampling design has been adopted for the NSS 76th round survey. In the first stage, villages/urban blocks were chosen by probability proportional to size with replacement (PPSWR) while in the second stage, households were selected by simple random sampling without replacement (SRSWOR) scheme in both rural and urban areas. All the households listed in the selected village/Urban Frame Survey (UFS) block/subunit (SU) were stratified into seven second stage strata (SSS). SSS 1 was formed considering the households having person(s) with any of the 11 rare disabilities ((a) acid attack victims, (b) autism spectrum disorder, (c) cerebral palsy, (d) dwarfism, (e) haemophilia, (f) multiple sclerosis, (g) muscular dystrophy, (h) other chronic neurological conditions, (i) Parkinson’s disease, (j) sickle cell disease, (k) thalassemia). SSS 2 was formed from the remaining households (i.e., excluding the households in SSS 1) having at least one person with mental disability. SSS 3 was formed from the remaining households (i.e., excluding the households in SSS 1 and SSS 2) having at least one person with speech disability. SSS 4 was formed from the remaining households (i.e., excluding the households in SSS 1, SSS 2 and SSS 3) having at least one person with visual disability. SSS 5 was formed from the remaining households (i.e., excluding the households in SSS 1, SSS 2, SSS 3 and SSS 4) having at least one person with hearing disability. SSS 6 was formed from the remaining households (i.e., excluding the households in SSS 1, SSS 2, SSS 3, SSS 4 and SSS 5) having at least one person with locomotor disability. SSS 7 was formed considering those households without any disability. A total of 20 households were planned to be selected from each sample village/UFS block/SU for canvassing the detailed schedule. The sample households from each of the SSS were selected by SRSWOR (NSS 76th, 2018).

Outcome Variables

OOPE

The NSS 76th round survey captured both the medical expenditure (surgery, equipment, hospitalization, etc.) and non-medical expenditure (transport, lodging, food, etc.) for infrequent expenditure during the last 365 days from the date of survey, along with usual monthly expenditure (excluding those covered infrequent expenditure during the last 365 days) (NSS 76th, 2018). We calculated OOPE for our study using the following method:

Average out-of-pocket expenditure = \frac{\frac{YME + YNME in last 365 days}{12} + UMME + UMNME}{PWDs reported OOPE on disability}

where, YME = medical expenditure specific to the disability like expenditure for assistive devices, YNME = non-medical expenditure specific to the disability, UMME = usual monthly medical expenditure and UMNME = usual monthly non-medical expenditure

OOPE Burden

We examined the OOPE burden as a percentage of OOPE on the total household consumption expenditure (Mitra et al., 2009; Sahoo & Madheswaran, 2014).

OOPE burden = \frac{Monthly healthcare expenditure on disability}{Total monthly household consumption expenditure} \times 100

Household Monthly Consumption Expenditure (HMCE)

We used monthly household consumption expenditure as a proxy variable for household income similar to previous studies (Joe & Rajpal, 2018; Kastor & Mohanty 2018; Sangar et al., 2019). This is a preferred measure as it is less prone to disparity and of being underreported or overreported when compared to income (Kastor & Mohanty, 2018). Based on the NSS 76th round, we calculated the monthly consumption expenditure using the following formula (NSS 76th, 2018):

Household monthly consumption expenditure = A + B + C + D / 12

where, A = usual consumer expenditure in a month for household purposes out of purchase, B = imputed value of usual consumption in a month from homegrown stock, C = imputed value of usual consumption in a month from wages in kind, free collection, gifts, etc. and D = expenditure on purchase of household durables during the last 365 days

Defining Predictor Variables

The OOPE varies across socioeconomic and disability-related characteristics of disabled person and household. For the present study, we used appropriate socioeconomic predictors at individual level, namely, age as four broad age groups (0–14, 15–35, 36–59 and 60 years and above), gender as male and female, educational status (illiterate, up to primary, middle and secondary and above), marital status (never married, currently married and others), relation to head of the household (self/spouse of head, married child and spouse of child, unmarried child, grandchild and other family members and relatives) and household level, religion (Hindu, Muslim and others), caste (Scheduled Tribe (ST), Scheduled Caste (SC), Other Backward Class (OBC) and others) and monthly per capita expenditure (MPCE) (poorest, poorer, middle, richer and richest), while at community level, variable includes as place of residence (rural and urban) and geographic region (north, central, east, northeast, west, south and union territories). In addition, we included disability-related variables such as types of disability (locomotor, visual, hearing, speech and language, mental retardation/intellectual disability, mental illness and other types of disability) and certificate of disability (yes/no). These predictor variables are found to be relevant for determining the OOPE and healthcare burden due to disability in several previous studies done in India and other countries (Lee et al., 2016; Moradhvaj & Saikia, 2019; Palmer et al., 2011).

Statistical Analysis

We used descriptive statistics, bivariate estimates, Blinder–Oaxaca decomposition and multilevel log-linear regression analysis for the present study. In the first step of the analysis, the descriptive statistics and bivariate estimate were carried out to compare healthcare burden, and the mean OOPE for males and females separately by socioeconomic and disability-related characteristics. In the second step of the analysis, we used Blinder–Oaxaca decomposition method to examine the gender inequality in OOPE. The Blinder–Oaxaca decomposition method is widely used to explain the gap between two groups like poor and non-poor, male and female, rural and urban, etc. (Blinder, 1973; O’Donnell et al., 2007; Oaxaca, 1973). The Blinder–Oaxaca decomposition model can be defined as

Y^{m a l e} = β^{m a l e} X^{m a l e} + e^{m a l e}

Y^{f e m a l e} = β^{f e m a l e} X^{f e m a l e} + e^{f e m a l e}

where, Y is OOPE, X is the vector of predictor variables, βs are the regression coefficients and e is the error term. In this study, we are using the same vector of predictors for males and females, that is, X^male = X^female. Hence, the gap in out-of-pocket payments among males and females can be calculated by

Y^{m a l e} - Y^{f e m a l e} = Δ x β^{m a l e} + Δ {β x}^{f e m a l e} + Δ x Δ β = C + E + C E

where, $Δ x β^{m a l e}$ and $Δ {β x}^{f e m a l e}$ are the vectors of predictor variables which estimated the means for males and females, respectively. Therefore, the gap in OOPE between male and female is due to C as the coefficient effect, E as the endowment effect and CE as the interaction effect of coefficient effect and endowment effect.

In the third step of the analysis, we carried out the multilevel regression model to examine the role of the gender after controlling socioeconomic and disability-related predictor variables for OOPE. The multilevel model is appropriate for estimating separately the variance and determining the extent of heterogeneity in OOPE on disability at different levels, for example, in the first level (individual), nested within the second level (village), nested within the third level (state), which was used by a previous study (Leone et al., 2013). The NSS 76th round data was collected using a hieratical structure (individual nested within village/urban block, village/urban block within the state). However, the NSS 76th round data do not have information at district level; hence, we have not included the district as a level in the multilevel modelling.

y_{i v s} = β_{0 i v s} + β_{1 i v s}; w h e r e, β_{0 v s} = β_{0} + v_{0 s} + u_{v s} + e_{0 i v s}

Where, i level 1 (individual), v level 2 (village), s level 3 (state)

$v_{0 s}$ is the random effect at the state level and allows to vary departure from the grand mean; $u_{v s}$ is the random effect at the village, a departure from the state effect; is the random effect at the individual level, a departure from the village effect within the state. To observe the different levels of correlation of the outcome variable, variance partition coefficients (VPCs) were calculated (Kiros & White, 2004).

F o r t h r e e l e v e l V P C_{i =} (\frac{σ_{i}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}}), V P C_{v =} (\frac{σ_{v}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}}) a n d V P C_{s =} (\frac{σ_{s}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}})

F o r t h r e e l e v e l V P C_{i =} (\frac{σ_{i}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}}), V P C_{v =} (\frac{σ_{v}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}}) a n d V P C_{s =} (\frac{σ_{s}^{2}}{σ_{i}^{2} + σ_{v}^{2} + σ_{s}^{2}})

where, $σ_{i}^{2} d e n o t e s t h e i n d i v i d u a l l e v e l v a r i a n c e, σ_{v}^{2}$ denotes the village-level variance and $σ_{s}^{2}$ denotes the state-level variance.

This study contains a range of predictors hence screened variables were adjusted for multicollinearity using variance inflation factor (VIF). Wherever VIF was found to be more than 10, that variable data was excluded from the model. Since the OOPE data was found to be skewed, we converted the OOPE into a log scale (Eeckhaut, 2020; Manning, 1998).

In the fourth step of the analysis, we carried out a test of interaction effects between gender (i.e., male vs. female) using selected socioeconomic and disability-related characteristics (such as economic status, education, age, place of residence, geographic region and types of disability) on OOPE. The interaction effects offer us an opportunity to examine the multidimensional nature of inequalities on OOPE of gender and subgroup of disabled person (Eeckhaut, 2020).

Results

Of the total 106,896 persons identified from the data, 57.31% were males and 42.69% were females (Table 1). Around 61% of females were illiterate as compared to 40% of males. 11% of females and 21% of males were educated up to secondary and above, respectively. Majority of disabled (around 60%) were above 35 years and one-fourth was in the 15–35 years age group. One-fourth of disabled belonged to the richest MPCE quintile, and 15% belonged to the poorest MPCE quintile. Majority of disabled persons resided in rural areas. Around one-fourth of disabled resided in the central region, and more than one-fifth in the east region. Higher rates of disability were seen in females across visual, hearing and mental illness disability types. Percentage of disability was found to be more in females as compared to males.

Table 1.

Percentage Distribution of Person Living with Disability by Their Selected Socioeconomic and Disability-related Characteristics: Persons with Disabilities in India Survey, National Sample Survey (NSS), 2018.

Background Characteristics	Male		Female		Total
Background Characteristics	n	%	n	%	n	%
Age (in years)
0–14	8,274	12.5	5,387	10.8	13,664	11.8
15–35	16,360	25.9	10,073	21.2	26,440	23.9
36–59	18,764	30.7	13,026	28.7	31,796	29.8
60 and above	18,169	30.9	16,819	39.3	34,994	34.5
Education
Illiterate	23,995	40.0	26,736	61.0	50,745	49.0
Up to primary	16,341	25.7	9,746	20.2	26,092	23.3
Middle	8,095	13.0	3,694	7.5	11,790	10.7
Secondary and above	13,136	21.3	5,129	11.3	18,267	17.0
Marital status
Never married	23,050	34.9	13,636	26.7	36,697	31.4
Currently married	33,083	55.9	16,938	39.3	50,025	48.8
Others	5,434	9.2	14,731	34.0	20,172	19.8
Relation to head of the household
Self/spouse of head	32,005	54.2	20,670	47.6	52,683	51.4
Married child and spouse of child	3,611	6.0	1,910	4.4	5,522	5.3
Unmarried child	17,516	26.4	10,355	20.4	27,878	23.9
Grandchild	2,171	3.3	1,402	2.8	3,574	3.1
Other family members and relatives	6,263	10.0	10,968	24.8	17,236	16.3
Religion
Hindu	48,869	81.7	35,672	81.4	84,561	81.5
Muslim	8,488	13.1	6,142	12.8	14,632	12.9
Others	4,210	5.3	3,491	5.9	7,701	5.6
Caste
ST	6,524	8.6	5,173	8.8	11,698	8.7
SC	12,108	20.6	8,768	20.2	20,881	20.5
OBC	26,783	44.9	19,330	44.8	46,121	44.9
Others	16,152	25.9	12,034	26.1	28,194	26.0
MPCE quintile
Poorest	8,627	15.6	6,066	15.1	14,695	15.4
Poorer	11,671	19.8	8,286	19.0	19,961	19.4
Middle	12,095	19.9	8,627	18.8	20,727	19.4
Richer	14,072	22.0	10,324	21.9	24,400	22.0
Richest	15,102	22.8	12,002	25.2	27,111	23.8
Place of residence
Rural	43,610	73.5	31,323	71.9	74,946	72.8
Urban	17,957	26.5	13,982	28.2	31,948	27.2
Geographic region
North	7,628	13.0	5,681	13.1	13,309	13.0
Central	14,068	26.9	9,406	24.9	23,479	26.0
East	13,729	22.4	9,290	21.1	23,025	21.8
Northeast	4,908	2.7	3,914	2.9	8,824	2.8
West	7,721	12.4	5,675	12.1	13,398	12.3
South	12,392	21.7	10,554	25.0	22,953	23.1
Union territories	1,121	1.0	785	1.0	1,906	1.0
Type of disability
Locomotor	33,374	57.6	22,596	54.3	55,977	56.2
Visual	5,567	8.6	5,459	11.3	11,031	9.8
Hearing	5,856	9.0	5,500	11.6	11,360	10.1
Speech and language	6,086	8.8	4,218	8.0	10,305	8.4
Mental retardation/intellectual disability	5,110	7.6	3,313	6.4	8,424	7.1
Mental illness	3,813	5.9	2,864	6.0	6,679	5.9
Others	1,761	2.6	1,355	2.4	3,118	2.5
Certificate of disability
No	40,705	67.6	33,549	75.9	74,271	71.2
Yes	20,862	32.4	11,756	24.1	32,623	28.9
Percentage of disability
None	329	1.5	182	2.0	511	1.7
40% or more but less than 60%	8,477	42.4	4,413	39.4	12,890	41.3
60% or more but less than 80%	7,605	36.0	4,364	36.6	11,973	36.2
80% or more	4,451	20.2	2,797	22.1	7,249	20.8
Total	61,567	100	45,306	100	106,896	100

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Note: All ‘n’ are unweighted. Total may not be equal due to some missing cases.

Abbreviations: MPCE, monthly per capita expenditure; OBC, Other Backward Class; SC, Scheduled Caste; ST, Scheduled Tribe.

Gender Difference in OOPE and OOPE Burden

Table 2 presents the share of OOPE on total household consumption expenditure by selected socioeconomic and types of disability between males and females. Overall, 20% of the household’s consumption expenditure was spent on healthcare expenditure on disability, with higher proportion of OOPE burden among males (23%) compared to females (17.2%). OOPE burden across various MPCE quintiles showed a similar trend with a higher proportion of male contributing to this burden. The gender gap for OOPE was highest among 15–25 years age groups. This gap was narrower among SC/ST and OBC as compared to other castes and more pronounced in rural areas (7.7%) than urban areas (3.8%).

Table 2.

Gender Differences in Average Monthly Out-of-pocket Expenditure (OOPE) and Healthcare Burden (HCB) of Person Living with Disability by Their Selected Socioeconomic and Demographic Characteristics: Persons with Disabilities in India Survey, National Sample Survey (NSS), 2018.

Background Characteristics	Male			Female			Total			Gap in HCB		Gap in OOPE
	OOPE(in ₹)	MCE(in ₹)	Healthcare Burden	OOPE(in ₹)	MCE(in ₹)	Healthcare Burden	OOPE(in ₹)	MCE(in ₹)	Healthcare Burden	Absolute Gap	Ratio	Absolute Gap	Ratio
	Mean(in ₹)	Mean(in ₹)	%	Mean(in ₹)	Mean(in ₹)	%	Mean(in ₹)	Mean(in ₹)	%	(M –F)	M/F	(M –F)	M/F
Age (in years)
0–14	2,648	12,054	22.0	2,312	11,777	19.6	2,522	11,950	21.1	2.4	1.1	336	1.2
15–35	3,233	11,321	28.6	2,035	11,978	17.0	2,809	11,553	24.3	11.6	1.7	1,198	1.6
36–59	2,791	11,273	24.8	2,214	12,308	18.0	2,524	11,754	21.5	6.8	1.4	577	1.3
60 and above	2,502	12,609	19.8	2,115	12,989	16.3	2,307	12,800	18.0	3.5	1.2	387	1.2
Education
Illiterate	2,174	10,521	20.7	1,810	11,007	16.4	1,965	10,800	18.2	4.3	1.3	364	1.2
Up to primary	2,340	10,963	21.3	2,066	12,789	16.2	2,231	11,684	19.1	5.1	1.3	274	1.1
Middle	2,788	11,604	24.0	2,497	13,598	18.4	2,691	12,267	21.9	5.6	1.3	291	1.1
Secondary and above	3,961	15,037	26.3	3,625	18,303	19.8	3,860	16,018	24.1	6.5	1.3	335	1.1
Marital status
Never married	2,553	12,027	21.2	2,171	12,420	17.5	2,418	12,166	19.9	3.7	1.2	383	1.2
Currently married	2,903	11,855	24.5	2,351	12,970	18.1	2,699	12,267	22.0	6.4	1.4	551	1.2
Others	2,321	11,859	19.6	1,915	12,097	15.8	2,011	12,040	16.7	3.8	1.2	407	1.2
Relation to head of the household
Self/spouse of head	2,800	11,302	24.8	2,147	12,009	17.9	2,524	11,600	21.8	6.9	1.4	653	1.3
Married child and spouse of child	3,391	14,493	23.4	2,580	13,275	19.4	3,088	14,038	22.0	4.0	1.2	811	1.3
Unmarried child	2,435	11,289	21.6	2,090	11,306	18.5	2,315	11,294	20.5	3.1	1.2	346	1.2
Grandchild	3,146	14,937	21.1	2,717	16,353	16.6	2,984	15,473	19.3	4.5	1.3	429	1.2
Other family members and relatives	2,639	14,661	18.0	2,083	13,747	15.2	2,249	14,022	16.0	2.8	1.2	556	1.3
Religion
Hindu	2,786	11,570	24.1	2,152	12,078	17.8	2,501	11,798	21.2	6.3	1.4	634	1.3
Muslim	2,539	12,388	20.5	1,992	12,608	15.8	2,298	12,486	18.4	4.7	1.3	548	1.3
Others	2,682	14,391	18.6	2,402	16,302	14.7	2,544	15,328	16.6	3.9	1.3	280	1.1
Caste
ST	1,950	9,250	21.1	1,504	8,622	17.4	1,749	8,967	19.5	3.7	1.2	445	1.3
SC	2,239	9,834	22.8	1,727	9,955	17.3	2,012	9,888	20.3	5.5	1.3	513	1.3
OBC	2,559	11,188	22.9	1,941	11,278	17.2	2,281	11,228	20.3	5.7	1.3	618	1.3
Others	3,427	14,551	23.5	2,770	16,225	17.1	3,124	15,321	20.4	6.4	1.4	656	1.2
MPCE quintile
Poorest	1,397	6,000	23.3	1,149	6,093	18.9	1,288	6,041	21.3	4.4	1.2	248	1.2
Poorer	1,941	7,715	25.2	1,439	7,484	19.2	1,726	7,616	22.7	6.0	1.3	502	1.4
Middle	2,154	9,289	23.2	1,610	8,774	18.3	1,918	9,066	21.2	4.9	1.3	544	1.3
Richer	2,913	10,945	26.6	1,907	11,014	17.3	2,480	10,974	22.6	9.3	1.5	1,006	1.5
Richest	3,742	17,834	21.0	3,059	18,594	16.5	3,409	18,203	18.7	4.5	1.3	683	1.2
Place of residence
Rural	2,418	9,449	25.6	1,735	9,709	17.9	2,117	9,563	22.1	7.7	1.4	682	1.4
Urban	3,351	16,496	20.3	2,843	17,198	16.5	3,112	16,827	18.5	3.8	1.2	508	1.2
Geographic region
North	3,059	15,462	19.8	2,434	18,047	13.5	2,760	16,700	16.5	6.3	1.5	625	1.3
Central	2,833	10,522	26.9	2,135	10,434	20.5	2,529	10,484	24.1	6.4	1.3	698	1.3
East	1,954	9,234	21.2	1,678	9,326	18.0	1,832	9,275	19.8	3.2	1.2	277	1.2
Northeast	2,103	10,900	19.3	1,640	10,907	15.0	1,887	10,904	17.3	4.3	1.3	462	1.3
West	3,511	13,277	26.4	2,549	14,303	17.8	3,097	13,718	22.6	8.6	1.5	962	1.4
South	2,723	12,098	22.5	2,165	12,068	17.9	2,461	12,084	20.4	4.6	1.3	558	1.3
Union territories	3,876	23,092	16.8	3,298	24,956	13.2	3,621	23,915	15.1	3.6	1.3	577	1.2
Type of disability
Locomotor	2,986	11,696	25.5	2,322	12,791	18.2	2,675	12,208	21.9	7.3	1.4	664	1.3
Visual	1,491	10,582	14.1	1,243	11,032	11.3	1,367	10,810	12.6	2.8	1.2	249	1.2
Hearing	1,594	12,630	12.6	1,189	12,194	9.8	1,389	12,410	11.2	2.8	1.3	405	1.3
Speech and language	3,098	12,605	24.6	2,612	12,433	21.0	2,942	12,549	23.4	3.6	1.2	486	1.2
Mental retardation/intellectual disability	2,525	12,991	19.4	2,225	13,294	16.7	2,418	13,099	18.5	2.7	1.2	301	1.1
Mental illness	2,258	11,456	19.7	1,926	12,024	16.0	2,115	11,699	18.1	3.7	1.2	332	1.2
Others	3,718	13,402	27.7	2,963	12,640	23.4	3,401	13,083	26.0	4.3	1.2	755	1.3
Certificate of disability
No	2,843	11,917	23.9	2,175	12,649	17.2	2,525	12,264	20.6	6.7	1.4	668	1.3
Yes	2,399	11,872	20.2	2,032	11,896	17.1	2,271	11,881	19.1	3.1	1.2	367	1.2
Percentage of disability
None	3,164	12,198	25.9	870	8,741	10.0	1,870	10,248	18.2	15.9	2.6	2,294	3.6
40% or more but less than 60%	2,100	11,737	17.9	1,781	11,612	15.3	1,991	11,695	17.0	2.6	1.2	319	1.2
60% or more but less than 80%	2,436	11,631	20.9	2,254	12,180	18.5	2,371	11,825	20.1	2.4	1.1	182	1.1
80% or more	2,782	12,477	22.3	2,417	12,791	18.9	2,664	12,578	21.2	3.4	1.2	365	1.2
Total	2,743	11,907	23.0	2,154	12,538	17.2	2,477	12,192	20.3	5.8	1.3	589	1.3

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Note: Healthcare burden = [Average monthly out-of-pocket expenditure (in ₹)/Monthly consumption expenditure (MCE) (in ₹)] × 10.

Abbreviations: MPCE, monthly per capita expenditure; OBC, Other Backward Class; SC, Scheduled Caste; ST, Scheduled Tribe.

The absolute and relative gaps in OOPE were higher in currently married, in Hindus, in richer MPCE quintile and in disabled persons residing in rural areas. Those with a certificate of disability had lower OOPE. The OOPE burden was highest in other disabilities (male 27.7% and female 23.4%) and locomotor disabilities (male 25.5% and female 18.2%), while it was lower in hearing disability (male 12.6% and female 9.8%) and visual disability (male 14.1% and female 11.3%). Gender gaps in healthcare burden were higher in locomotor disability (7.3%), while it was lower in visual disability (2.8%) and hearing disability (2.8%). OOPE and healthcare expenditure were higher (21.2%) if disability percentage was more than 80%, while it was lower (17.0%) among disabled persons who reported disability percentage between 40% and 60%. Further gender gap also shows among severity of disability. Here, it is noted that the percentage of disability ‘none’ means they do not have the certificate of disability. Region-wise OOPE was found to be less in the east, northeast, west, south and union territories regions as compared to north, but was higher in central region when compared to north (Table 3).

Table 3.

Gender Differences in Average Households Expenditure and Healthcare Burden (HCB) of Person Living with Disability Across States: Persons with Disabilities in India Survey, National Sample Survey (NSS), 2018.

Background Characteristics	Male			Female			Total			Gap in HCB		Gap in OOPE
	OOPE	MCE	Healthcare Burden	OOPE	MCE	Healthcare Burden	OOPE	MCE	Healthcare Burden	Absolute Gap	Ratio	Absolute Gap	Ratio
	Mean (in ₹)	Mean (in ₹)	%	Mean (in ₹)	Mean (in ₹)	%	Mean (in ₹)	Mean (in ₹)	%	(M–F)	M/F	(M–F)	M/F
Andaman and Nicobar Islands	5,142	21,827	23.6	1,680	16,902	9.9	3,929	20,102	19.5	13.6	2.4	3,463	3.1
Andhra Pradesh	2,618	10,419	25.1	2,171	10,595	20.5	2,410	10,496	23.0	4.6	1.2	447	1.2
Arunachal Pradesh	1,860	9,953	18.7	1,168	10,378	11.3	1,612	10,106	15.9	7.4	1.7	693	1.6
Assam	2,081	10,765	19.3	1,724	10,751	16.0	1,911	10,758	17.8	3.3	1.2	357	1.2
Bihar	1,273	8,374	15.2	942	8,179	11.5	1,147	8,303	13.8	3.7	1.3	331	1.4
Chandigarh	3,585	27,072	13.2	2,778	40,464	6.9	3,074	35,542	8.6	6.4	1.9	807	1.3
Chhattisgarh	2,702	9,462	28.6	2,076	9,324	22.3	2,451	9,407	26.1	6.3	1.3	627	1.3
Dadra and Nagar Haveli	1,509	10,305	14.6	2,045	10,036	20.4	1,703	10,208	16.7	–5.7	0.7	–536	0.7
Daman and Diu	2,634	12,784	20.6	1,485	12,202	12.2	2,344	12,637	18.6	8.4	1.7	1,149	1.8
Goa	2,440	15,840	15.4	2,564	17,817	14.4	2,499	16,782	14.9	1.0	1.1	–124	1.0
Gujarat	2,771	14,501	19.1	2,144	15,350	14.0	2,494	14,876	16.8	5.1	1.4	628	1.3
Haryana	3,524	17,296	20.4	2,369	22,255	10.6	2,956	19,734	15.0	9.7	1.9	1,156	1.5
Himachal Pradesh	3,236	12,267	26.4	3,011	12,619	23.9	3,149	12,403	25.4	2.5	1.1	226	1.1
Jammu and Kashmir	2,377	13,227	18.0	2,237	12,842	17.4	2,318	13,065	17.7	0.6	1.0	140	1.1
Jharkhand	2,620	10,323	25.4	1,771	10,108	17.5	2,242	10,227	21.9	7.9	1.4	849	1.5
Karnataka	2,778	12,681	21.9	1,728	12,281	14.1	2,324	12,511	18.6	7.8	1.6	1,050	1.6
Kerala	2,819	13,675	20.6	2,465	13,628	18.1	2,635	13,650	19.3	2.5	1.1	355	1.1
Lakshadweep	4,530	15,719	28.8	4,778	15,748	30.3	4,648	15,732	29.5	–1.5	0.9	–248	0.9
Madhya Pradesh	2,786	10,082	27.6	2,354	11,834	19.9	2,609	10,798	24.2	7.7	1.4	432	1.2
Maharashtra	3,748	12,869	29.1	2,677	13,899	19.3	3,292	13,308	24.7	9.9	1.5	1,071	1.4
Manipur	3,296	10,210	32.3	1,964	10,575	18.6	2,735	10,364	26.4	13.7	1.7	1,332	1.7
Meghalaya	954	11,275	8.5	784	11,877	6.6	866	11,586	7.5	1.9	1.3	171	1.2
Mizoram	1,243	24,522	5.1	1,202	22,781	5.3	1,226	23,827	5.1	–0.2	1.0	41	1.0
Nagaland	4,477	12,716	35.2	1,934	12,793	15.1	3,424	12,748	26.9	20.1	2.3	2,543	2.3
Delhi	4,060	24,010	16.9	3,223	24,690	13.1	3,698	24,304	15.2	3.9	1.3	837	1.3
Puducherry	1,746	12,758	13.7	5,442	13,534	40.2	3,438	13,113	26.2	–26.5	0.3	–3,696	0.3
Punjab	2,388	16,759	14.3	1,895	18,960	10.0	2,124	17,941	11.8	4.3	1.4	493	1.3
Rajasthan	3,454	14,295	24.2	3,340	15,851	21.1	3,406	14,949	22.8	3.1	1.1	114	1.0
Sikkim	1,649	10,286	16.0	1,015	10,386	9.8	1,350	10,333	13.1	6.3	1.6	634	1.6
Tamil Nadu	2,573	11,218	22.9	2,071	10,971	18.9	2,332	11,100	21.0	4.1	1.2	502	1.2
Telangana	2,824	12,049	23.4	2,084	11,694	17.8	2,536	11,911	21.3	5.6	1.3	740	1.4
Tripura	2,182	10,246	21.3	1,592	10,354	15.4	1,933	10,301	18.8	5.9	1.4	590	1.4
Uttar Pradesh	2,858	10,736	26.6	2,090	10,199	20.5	2,515	10,497	24.0	6.1	1.3	768	1.4
Uttarakhand	4,884	13,751	35.5	3,038	13,385	22.7	3,984	13,573	29.4	12.8	1.6	1,846	1.6
West Bengal	2,264	10,827	20.9	2,073	11,113	18.7	2,179	10,953	19.9	2.3	1.1	191	1.1
Odisha	1,796	7,020	25.6	1,552	7,338	21.1	1,677	7,175	23.4	4.4	1.2	244	1.2
Total	2,743	11,907	23.0	2,154	12,538	17.2	2,477	12,192	20.3	5.9	1.3	589	1.3

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Note: Healthcare burden = [Average monthly out-of-pocket expenditure (OOPE) (in ₹)/Monthly consumption expenditure (MCE) (in ₹)] × 100.

Gender Inequality in OOPE

To explain the effect of variables at different levels on the gender gap, we further explored the contribution of selected socioeconomic and disability-related predictors in explaining the gender gaps in OOPE using Blinder–Oaxaca decomposition method (Table 4). The gender gap in OOPE was decomposed into three parts. The first part is endowment effect, which shows the gap due to variances in the distribution of factors between male and female; the second part is the coefficient effect, which demonstrates the gap due to the variances in the effect of factors between male and female; the third is the interaction effect, which shows the interaction between both endowment effect and coefficient effect. The average of log OOPE was 6.98 for male and 6.84 for female. The gap between the male and female was 0.14 of which 0.04 was due to endowment, 0.11 was due to coefficient and –0.01 was due to interaction. The results show that about 28.80% of the male–female gap in OOPE was explained by the differential distribution of socioeconomic and disability-related predictors.

Table 4.

Result of the Blinder–Oaxaca Decomposition Analysis Showing the Male–Female Differences in Out-of-pocket Expenditure (OOPE) for People with Disabilities (PWDs) by Their Selected Socioeconomic and Demographic Characteristics: Persons with Disabilities in India Survey, National Sample Survey (NSS), 2018.

Background Characteristics	Mean Health Expenditure Gap	% Contributed	95% CI
Background Characteristics	Mean Health Expenditure Gap	% Contributed	Lower	Upper
Overall
Mean prediction high (male)	6.980		6.922	7.038
Mean prediction low (female)	6.836		6.743	6.928
Raw differential (R) (male–female)	0.144		0.036	0.253
Endowments	0.042	28.803	–0.033	0.116
Coefficients	0.109	75.267	0.002	0.215
Interaction	–0.006	–4.07	–0.075	0.063
Total explained	0.144	100
Endowments
Older age	0.001	1.223	–0.008	0.009
Literate	0.016	38.771	–0.003	0.036
Currently married	–0.011	–25.545	–0.035	0.013
Non-Hindu	0	–0.17	–0.002	0.002
SC	–0.001	–3.436	–0.012	0.009
Rich	0	–0.483	–0.021	0.02
Urban	0.009	21.892	–0.004	0.023
Locomotor	0.021	49.772	–0.005	0.046
Visual	0	0.777	–0.002	0.003
Hearing	0	0
Speech and language	0.002	4.542	–0.01	0.013
Mental retardation/intellectual disability	–0.01	–24.599	–0.033	0.012
Mental illness	–0.006	–13.952	–0.02	0.008
Others	–0.008	–20.094	–0.022	0.005
Percentage of disability (none)	0.03	71.302	–0.02	0.08
Total	0.042	100
Coefficients
Older age	0.036	–9.47	–0.021	0.093
Literate	–0.002	0.53	–0.12	0.116
Currently married	–0.087	22.781	–0.197	0.152
Non-Hindu	–0.079	20.645	–0.31	0.037
SC	–0.026	6.928	–0.09	0.251
Rich	0.124	–32.536	–0.002	–0.06
Urban	–0.324	84.893	–0.589	0.12
Locomotor	–0.045	11.893	–0.211	0.043
Visual	0.006	–1.616	–0.03
Hearing	0	0		0.03
Speech and language	–0.004	1.104	–0.038	0.072
Mental retardation/intellectual disability	–0.015	3.964	–0.102	0.022
Mental illness	–0.018	4.594	–0.057	0.025
Others	–0.005	1.297	–0.034	0.121
Percentage of disability (none)	0.057	–15.007	–0.007
Total	–0.382	100
Interaction
Older age	0.007	–113.962	–0.005	0.018
Literate	0	3.192	–0.011	0.011
Currently married	0.029	–496.362	–0.008	0.067
Non-Hindu	0.001	–25.174	–0.004	0.006
SC/ST	0	–5.683	–0.002	0.003
Rich	0	1.693	–0.01	0.01
Urban	–0.007	121.707	–0.019	0.004
Locomotor	–0.005	81.492	–0.023	0.013
Visual	0	7.707	–0.003	0.003
Hearing	0	0
Speech and language	0	6.023	–0.004	0.003
Mental retardation/intellectual disability	0.001	–16.623	–0.005	0.007
Mental illness	0.002	–27.002	–0.004	0.007
Others	0.001	–13.217	–0.004	0.006
Percentage of disability (none)	–0.034	576.209	–0.092	0.025
Total	–0.006	100

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Abbreviations: CI, confidence interval; SC, Scheduled Caste; ST, Scheduled Tribe.

Table 5 presents the result of the Blinder–Oaxaca decomposition analysis of OOPE among males and females by PWD groups (individual, household, community and disability).

Table 5.

Result of the Blinder–Oaxaca Decomposition Analysis Showing the Male–Female Difference in Out-of-pocket Expenditure (OOPE) for People with Disability (PWD) Groups.

Group-wise Decomposition	OOPE on Disability
	Mean Health Expenditure Gap	% Contributed	95% CI
	Mean Health Expenditure Gap	% Contributed	Lower	Upper
Endowments
Individual	0.006	14.449	–0.028	0.040
Household	–0.002	–4.090	–0.026	0.023
Community	0.009	21.892	–0.004	0.023
Disability status	0.028	67.748	–0.027	0.083
Total	0.042	100.000
Coefficients
Individual	–0.053	13.841	–0.231	0.125
Household	0.019	–4.962	–0.217	0.255
Community	–0.324	84.893	–0.589	–0.060
Disability status	–0.024	6.228	–0.374	0.326
Total	–0.382	100.000
Interaction
Individual	0.036	–607.122	–0.005	0.077
Household	0.002	–29.162	–0.009	0.013
Community	–0.007	121.705	–0.019	0.004
Disability status	–0.036	614.579	–0.094	0.022
Total	–0.006	100.000

Source: Based on the author computation from Persons with Disabilities in India Survey, National Sample Survey Office (NSSO), 2018.

Note: Individual: Age, education and married; Household: Religion, caste and wealth quintile; Community: Place of residence; Disability status: Type of disability and percentage of disability.

Abbreviation: CI, confidence interval.

Factors Associated with OOPE

Table 6 presents the multilevel model to know the variation in average OOPE between place of residence, between households and between individual of the household in the place of residence. The variation in OOPE is higher at individual level (Ωe2 = 1.170) than that of household level (Ωu2 = 0.618) at the two level model (level 1—individual, level 2—first sampling unit (FSU)/village) than three level model (level 1—individual, level 2—FSU/village, level 3—state). Variance partition coefficient (VPC) shows 35% of the variation in disability cost is due to the household level clustering of the individual without controlling the socioeconomic and disability-related predictors (model 1 two level empty model). In the full model controlled with socioeconomic and disability-related characteristics, the variances endorsed to the differences in the village 34.56–42.97 and state increased by 4.41–10.25. The variation in OOPE is higher at individual level (Ωe2 = 1.170) than that of household level (Ωu2 = 0.584) and place of residence level (Ωu2 = 0.035) at three level analysis. The variation in the OOPE is higher at individual level (Ωe2 = 1.172) than that of household level (Ωu2 = 0.564) and place of residence level (Ωu2 = 0.080) at three level analysis.

Table 6.

Random-effects Parameters for the Multilevel Model for Out-of-pocket Expenditure (OOPE) for People with Disability (PWD) Groups as per Residence with Disability by Their Selected Socioeconomic and Demographic Characteristics: Persons with Disabilities in India Survey, National Sample Survey (NSS), 2018.

Random Effects	Without Explanatory Variables (Null Model)
	Model 1				Model 2
	Random Variance	Standard Error	95% CI		Random Variance	Standard Error	95% CI
	Random Variance	Standard Error	Lower	Upper	Random Variance	Standard Error	Lower	Upper
Individual	1.170	0.011	1.150	1.192	1.172	0.011	1.151	1.193
Community (FSU)	0.618	0.017	0.585	0.652	0.564	0.016	0.533	0.596
Individual VPC (%)	65.436				64.540
Community (FSU) VPC (%)	34.564				31.047
State					0.080	0.023	0.045	0.141
State VPC (%)					4.413
Random Effects	With Explanatory Variables (Full Model)
	Random Variance	Standard Error	95% CI		Random Variance	Standard Error	95% CI
	Random Variance	Standard Error	Lower	Upper	Random Variance	Standard Error	Lower	Upper
Individual	0.905	0.022	0.862	0.949	0.950	0.012	0.928	0.973
Community (FSU)	0.682	0.033	0.620	0.750	0.799	0.020	0.760	0.839
Individual VPC (%)	57.026				48.765
Community (FSU) VPC (%)	42.974				40.985
State					0.200	0.039	0.136	0.293
State VPC (%)					10.250

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Abbreviations: CI, confidence interval; FSU, first sampling unit; VPC, variance partition coefficient.

The results of multilevel analysis showing the factors associated with OOPE are provided in Table 7. After adjusting the socioeconomic and disability-related predictors, we found that the average OOPE is significantly lesser among females (β = –0.070, 95% confidence interval (CI)= –0.129 to 0.01) compared to males at two levels (individual level and household level), and similar pattern shows at three levels also (individual level, household level and state level). These clearly indicate the discrimination in the genders for OOPE due to disability-related healthcare expenditure. The associations between OOPE and selected socioeconomic and disability-related predictors were also in quite an expected direction. The probability of OOPE was less with increasing age. Economic status of household had a significantly positive association with OOPE. PWDs from richer households had higher OOPE for disability care.

Table 7.

Results of the Multilevel Analysis Showing Factors Associated with Out-of-pocket Expenditure (OOPE) for People with Disabilities (PWDs) by Their Selected Socioeconomic and Demographic Characteristics.

Background Characteristics	Model 3				Model 4
	Coefficient	p Value	95% CI		Coefficient	p Value	95% CI
	Coefficient	p Value	Lower	Upper	Coefficient	p Value	Lower	Upper
Age (in years)
0–14^®
15–35	–0.329	0.000	–0.418	–0.240	–0.352	0.000	–0.441	–0.263
36–59	–0.259	0.000	–0.380	–0.138	–0.284	0.000	–0.405	–0.163
60 and above	–0.209	0.005	–0.354	–0.064	–0.242	0.001	–0.387	–0.097
Education
Illiterate^®
Up to primary	0.017	0.659	–0.057	0.090	–0.001	0.976	–0.075	0.072
Middle	0.051	0.299	–0.045	0.146	0.046	0.348	–0.050	0.141
Secondary and above	0.248	0.000	0.161	0.334	0.221	0.000	0.134	0.307
Gender
Male^®
Female	–0.070	0.021	–0.129	–0.011	–0.066	0.027	–0.125	–0.008
Marital status
Never married^®
Currently married	0.302	0.000	0.134	0.470	0.327	0.000	0.160	0.495
Others	0.067	0.460	–0.110	0.243	0.085	0.341	–0.090	0.261
Relation to head of the household
Self/spouse of head^®
Married child and spouse of child	–0.063	0.349	–0.195	0.069	–0.043	0.520	–0.175	0.088
Unmarried child	0.235	0.008	0.062	0.408	0.259	0.003	0.088	0.431
Grandchild	0.315	0.002	0.111	0.519	0.352	0.001	0.149	0.554
Other family members and relatives	0.042	0.549	–0.096	0.181	0.066	0.346	–0.072	0.205
Religion
Hindu^®
Muslim	–0.010	0.834	–0.104	0.084	–0.001	0.981	–0.097	0.095
Others	0.039	0.516	–0.080	0.158	0.010	0.880	–0.114	0.133
Caste
ST^®
SC	0.124	0.079	–0.015	0.262	0.169	0.018	0.028	0.309
OBC	0.274	0.000	0.145	0.403	0.313	0.000	0.183	0.444
Others	0.365	0.000	0.233	0.496	0.370	0.000	0.237	0.504
MPCE quintile
Poorest^®
Poorer	0.154	0.014	0.031	0.277	0.180	0.001	0.073	0.287
Middle	0.322	0.000	0.197	0.447	0.299	0.000	0.191	0.408
Richer	0.364	0.000	0.243	0.485	0.462	0.000	0.351	0.573
Richest	0.739	0.000	0.616	0.862	0.827	0.000	0.711	0.943
Geographic region
North^®
Central	0.086	0.304	–0.078	0.249	0.019	0.910	–0.316	0.354
East	–0.406	0.000	–0.552	–0.260	–0.449	0.003	–0.747	–0.151
Northeast	–0.405	0.000	–0.592	–0.217	–0.377	0.013	–0.675	–0.079
West	–0.128	0.099	–0.279	0.024	–0.291	0.092	–0.628	0.047
South	–0.215	0.002	–0.351	–0.079	–0.297	0.036	–0.575	–0.020
Union territories	–0.162	0.217	–0.421	0.096	–0.269	0.138	–0.625	0.087
Type of disability
Locomotor^®
Visual	–0.405	0.000	–0.511	–0.299	–0.411	0.000	–0.516	–0.305
Hearing	–0.483	0.000	–0.620	–0.345	–0.487	0.000	–0.624	–0.351
Speech and language	–0.208	0.000	–0.324	–0.092	–0.205	0.001	–0.321	–0.090
Mental retardation/intellectual disability	0.102	0.026	0.012	0.191	0.090	0.047	0.001	0.180
Mental illness	0.101	0.063	–0.005	0.208	0.101	0.062	–0.005	0.207
Others	0.327	0.000	0.194	0.460	0.316	0.000	0.184	0.448
Percentage of disability
None^®
40% or more but less than 60%	0.088	0.382	–0.109	0.284	0.067	0.500	–0.128	0.262
60% or more but less than 80%	0.235	0.019	0.038	0.432	0.206	0.039	0.011	0.402
80% or more	0.396	0.000	0.194	0.598	0.350	0.001	0.149	0.551
Constant	6.197	0.000	5.870	6.523	6.297	0.000	5.935	6.658

Source: Based on the author computation from Persons with Disabilities in India Survey, NSS, 2018.

Note: Likelihood ratio (LR) tests for multilevel linear regression model versus ordinary linear regression model: For model 1 ( $x^{2} =$ 5,164.49, p value = 0.00); for model 2 ( $x^{2} =$ 5,396.78, p < 0.000); for model 3 ( $x^{2} =$ 894.02, p < 0.000); for model 4 ( $x^{2} =$ 894.02, p < 0.000); ® = Reference

Abbreviations: CI, confidence interval; MPCE, monthly per capita expenditure; NSS, National Sample Survey; OBC, Other Backward Class; SC, Scheduled Caste; ST, Scheduled Tribe.

Finally, we present the analysis of the adjusted multivariable models with interaction by OOPE according to gender and socioeconomic and disability-related predictor (Figure 1). Figure 1 shows results for age, education, place of residence, region of residence, MPCE quintile and types of disability (models were adjusted with marital status, relation to head of the household, religion, caste, having certificate of disability and percentage of disability). Moreover, after using the interaction, the gender gap remains constant across the socioeconomic and disability status, and shows that OOPE among males was higher than female irrespective of predictor variables.

Figure 1.

Source: Author’s calculations based on NSS, 2018

Discussion

This study examines the gender gaps in OOPE among PWDs in India. This study describes the gender inequality in OOPE and OOPE burden among PWDs. For our analysis, we have used the recent national representative cross-sectional National Sample Survey Office (NSSO) survey data on the theme ‘Persons with Disabilities in India Survey’. We defined monthly OOPE due to disability as the total medical expenditure and non-medical expenditure monthly on disability healthcare. And also examined the OOPE burden as a percentage of total household consumption expenditure similar to previous studies (Mitra et al., 2009; Sahoo & Madheswaran, 2014). Our study focused on the gender gap in OOPE through the use of various methods such as relative and absolute gap of OOPE between male and female populations, Oaxaca decomposition method and multilevel models.

There are several notable findings. First, almost 20% of the total household income was spent on healthcare expenditure for PWD treatment care with a higher share of OOPE among males (23.0%) as compared to females (17.2%). The male–female differences in mean OOPE also vary largely across states (Table 7). The highest gaps in male–female differences in mean OOPE were observed in Andaman and Nicobar Islands (₹3,463), followed by Nagaland (₹2,543), Uttarakhand (₹1,846), Manipur (₹1,332), Haryana (₹1,156), Daman and Diu (₹1,149) and Maharashtra (₹1,071) while some of the states like Puducherry (₹–3,696), followed by Dadra and Nagar Haveli (₹–536), Lakshadweep (₹–248) and Goa (₹–124). Younger adults, the poorest quintile household and rural areas showed a wider gender gap with 23–25% of males spending more on healthcare as compared to 17–19% of females.

Second, the mean OOPE was found to be higher in males compared to females. Male–female differentials among younger age group (15–35 years) were higher compared to other age groups. The gender gap was higher among rural residents as compared to urban residents. Rural households had spending of 25% of total household expenditure on healthcare expenditure if the disabled person was male while only 17% of total household expenditure was spent on healthcare if disabled person was female. These findings are consistent with the wisdom that the reporting of poor health by women has been found to be significantly higher than men, hence the lower mean OOPE reported in our study (Bora & Saikia, 2015).

The actual cost on disability was higher in the richest households but the OOPE burden was the highest in poor households. The lower expenditure on disability-related healthcare by the poor households could be linked to the households’ ability to pay due to low income (Seth & Mohanty, 2017). Further, among the poorest, the share of household consumption on disability-related OOPE on health was higher in males (23.3%) than females (18.9%). This could be due to the fact that in poorer households, women have lesser power than men due to absence of property ownership, lack of income continuing social norms and feelings (Ajala, 2017; Saikia et al., 2016).

Thirdly, the gender gap in OOPE using decomposed analysis clearly indicates that after adjusting the socioeconomic and disability-related predictors, the average OOPE is significantly lesser among females (β = –0.070, 95% CI= –0.129 to 0.01) compared to males at two levels (individual level and household level) and at three levels also (individual level, household level and state level). Findings clearly indicate the gender gap in both OOPE and OOPE burden.

Our findings were consistent with the findings of previous studies (Palmer et al., 2011) and suggest that OOPE and OOPE burden on healthcare in household have gender inequalities among PWDs in India, and will require long-term social, cultural and policy changes. The findings of this study suggest the need for a coherent and gender-sensitive health agenda for disabled populations in India. Focusing on the prevention and early treatment, efforts to mitigate disability conditions are essential steps along with increased access to health insurance and social security for promoting gender-sensitive health and well-being of PWDs in India. We also suggest future in-depth studies focused on complex gender dynamics in health and disability at states and district levels in India. Since the NSS does not report expenditure of non-PWD, future rounds of NSS could focus on this aspect.

Limitations of the Study

The present study has a few limitations. First, we could not provide insights on gender differentials on disability-related healthcare OOPE and OOPE burden at the district level. Though the data is from a nationally representative survey, recall bias along with self-reporting of healthcare expenditure for detailing healthcare expenditure data by the respondents could be a major limitation. The cross-sectional nature of the survey limits its use for causal inference and long-term healthcare burden. Lastly, the NSS did not report health facility utilization such as private and public among sampled populations, so we were unable to present this analysis. This is an important limitation in our study as it is known that populations incur higher expenditure in private health facilities compared to public (Kumar et al., 2015; Sahoo & Madheswaran, 2014; Yadav et al., 2020, 2021).

Conclusion

Ensuring healthy life and promoting wellbeing for all at all ages (goal 3) and achieving gender equality and empowering all women and girls (goal 5) are two of the sustainable development goals (SDGs) in the UN’s agenda. To reach these goals, it is key to recognize the disadvantaged and omitted using a gender lens, especially among PWDs. This study identifies the general differentials in OOPE and OOPE burden on healthcare among PWDs in India and highlights the need for gender-sensitive programs and policies targeted at PWDs by the Government of India.

Author Contributions

JY, GRM and DJ conceived the idea for the study and developed the analysis plan. JY contributed to the data analysis. GRM, JY, RS and DJ wrote the initial draft and all authors were involved in commenting on subsequent revisions.

Footnotes

Declaration of Conflicting Interests

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

Ethical Approval

Not required.

Ethics Statement

The present study used the data from secondary sources which is freely available to individuals. Therefore, no ethical approval is required separately for the present study.

Funding

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

References

Ajala

(2017). Gender discrimination in land ownership and the alleviation of women’s poverty in Nigeria: A call for new equities. International Journal of Discrimination and the Law , 17(1), 51–66. https://doi.org/10.1177/1358229117700028

Blinder

A. S.

(1973). Wage discrimination: Reduced form and structural estimates. The Journal of Human Resources , 8(4), 436. https://doi.org/10.2307/144855

Bora

J. K.

, & Saikia

(2015). Gender differentials in self-rated health and self-reported disability among adults in India. PLoS One , 10(11), 1–14. https://doi.org/10.1371/journal.pone.0141953

Eeckhaut

M. C. W.

(2020). Intersecting inequalities: Education, race/ethnicity, and sterilization. Journal of Family Issues , 41(10), 1905–1929. https://doi.org/10.1177/0192513X19900529

Joe

, & Rajpal

(2018). Unravelling the socioeconomic gradient in the incidence of catastrophic health care expenditure: A comment. Health Policy and Planning , 33(5), 699–701. https://doi.org/10.1093/heapol/czy026

Kastor

, & Mohanty

S. K.

(2018). Disease-specific out-of-pocket and catastrophic health expenditure on hospitalization in India: Do Indian households face distress health financing? PLoS One , 13(5), e0196106. https://doi.org/10.1371/journal.pone.0196106

Kiros

G.-E.

, & White

M. J.

(2004). Migration, community context, and child immunization in Ethiopia. Social Science & Medicine , 59(12), 2603–2616. https://doi.org/10.1016/j.socscimed.2004.04.009

Kumar

, Singh

, Kumar

, Ram

, Singh

, Ram

, Negin

, & Kowal

P. R.

(2015). Socio-economic differentials in impoverishment effects of out-of-pocket health expenditure in China and India: Evidence from WHO SAGE. PLoS One , 10(8), e0135051. https://doi.org/10.1371/JOURNAL.PONE.0135051

Kwon

(2009). Thirty years of national health insurance in South Korea: Lessons for achieving universal health care coverage. Health Policy and Planning , 24(1), 63–71. https://doi.org/10.1093/heapol/czn037

10.

Lee

J. E.

, Shin

H. I.

, Do

Y. K.

, & Yang

E. J.

(2016). Catastrophic health expenditures for households with disabled members: Evidence from the Korean health panel. Journal of Korean Medical Science , 31(3), 336–344. https://doi.org/10.3346/jkms.2016.31.3.336

11.

Leone

, James

K. S.

, & Padmadas

S. S.

(2013). The burden of maternal health care expenditure in India: Multilevel analysis of national data. Maternal and Child Health Journal , 17(9), 1622–1630. https://doi.org/10.1007/s10995-012-1174-9

12.

Manning

W. G.

(1998). The logged dependent variable, heteroscedasticity, and the retransformation problem. Journal of Health Economics , 17(3), 283–295. https://doi.org/10.1016/S0167-6296(98)00025-3

13.

Mitra

, Findley

P. A.

, Rd

E. F.

, & Hall

(2009). Fordham University, Department of Economics, Discussion Paper Series. Healthcare expenditures of living with a disability: Total expenditures, out of pocket expenses and burden, 1996-2004 . Discussion Paper No: 2008-18, September 2008 (Updated: February 2009). Fordham University.

14.

Mitra

, Palmer

, Kim

, Mont

, & Groce

(2017). Extra costs of living with a disability: A review and agenda for research. Disability and Health Journal , 10(4), 475–484. https://doi.org/10.1016/j.dhjo.2017.04.007

15.

Mitra

, & Sambamoorthi

(2006). Measurement of disability disability estimates in India. Economic & Political Weekly , 41(38), 4022–4024.

16.

Moradhvaj , & Saikia

(2019). Gender disparities in health care expenditures and financing strategies (HCFS) for inpatient care in India. SSM - Population Health , 9, 100372. https://doi.org/10.1016/j.ssmph.2019.100372

17.

National Sample Survey Office. (2016). Disabled persons in India: A statistical profile 2016 . https://mospi.gov.in/sites/default/files/publication_reports/Disabled_persons_in_India_2016.pdf

18.

NSS 76th. (2018). Persons with disabilities in India , May, 1–22.

19.

O’Donnell

, van Doorslaer

, Wagstaff

, & Lindelow

(2007). Analyzing health equity using household survey data . The World Bank. https://doi.org/10.1596/978-0-8213-6933-3

20.

Oaxaca

(1973). Male–female wage differentials in urban labor markets. International Economic Review , 14(3), 693–709. https://doi.org/10.2307/2525981

21.

Palmer

, Nguyen

, Neeman

, Berry

, Hull

, & Harley

(2011). Health care utilization, cost burden and coping strategies by disability status: An analysis of the Viet Nam National Health Survey. International Journal of Health Planning and Management , 26(3), e151–e168. https://doi.org/10.1002/hpm.1052

22.

Sahoo

A. K.

, & Madheswaran

(2014). Socio-economic disparities in health care seeking behaviour, health expenditure and its source of financing in Orissa: Evidence from NSSO 2004–05. Journal of Health Management , 16(3), 397–414. https://doi.org/10.1177/0972063414539614

23.

Saikia

, Moradhvaj , & Bora

J. K.

(2016). Gender difference in health-care expenditure: Evidence from India human development survey. PLoS One , 11(7), 1–15. https://doi.org/10.1371/journal.pone.0158332

24.

Sangar

, Dutt

, & Thakur

(2019). Comparative assessment of economic burden of disease in relation to out of pocket expenditure. Frontiers in Public Health , 7, 9. https://doi.org/10.3389/fpubh.2019.00009

25.

Seth

, & Mohanty

S. K.

(2017). Poor and non-poor differentials in household health spending in India. Journal of Public Health (Germany) , 25(1), 75–86. https://doi.org/10.1007/s10389-016-0765-0

26.

World Health Organization. (2011). Summary world report on disability. World Health , 1–24. https://doi.org/10.1136/ip.2007.018143

27.

Yadav

, Allarakha

, Menon

G. R.

, John

, & Nair

(2021). Socioeconomic impact of hospitalization expenditure for treatment of noncommunicable diseases in India: A repeated cross-sectional analysis of National Sample Survey Data, 2004 to 2018. Value in Health Regional Issues , 24, 199–213. https://doi.org/10.1016/j.vhri.2020.12.010

28.

Yadav

, Menon

G. R.

, Mitra

, Allarakha

, & John

(2020). Burden and cost of communicable, maternal, perinatal and nutrition deficiency diseases in India. Journal of Public Health , 44(2), 217–227. https://doi.org/10.1093/pubmed/fdaa173

29.

Yenimahalleli Yasar

(2011). Health transformation programme in Turkey: An assessment. International Journal of Health Planning and Management , 26(2), 110–133. https://doi.org/10.1002/hpm.1065