Access to Social Protection Among People with Disabilities: Mixed Methods Research from Tanahun, Nepal

While people with disabilities are often targeted as key beneficiaries of social protection, there is little evidence available on their participation in existing programmes. This study uses mixed methods to explore access to disability-targeted and non-targeted social protection programmes in Nepal, particularly the District of Tanahun. In total, 31% of people with disabilities had Disability Identification Cards, which entitles them to a range of different social protection benefits depending on the card level, including disability-targeted social assistance (received by 13% of people with disabilities). Overall, 37% of people with disabilities received social assistance, which was higher than for people without disabilities (21%). The most commonly accessed form of social assistance was the Old Age Allowance, which had universally high coverage amongst both people with and without disabilities. Uptake of disability-targeted social protection entitlements other than social assistance (e.g. scholarships, discounted transportation and health services) was generally low. Factors impacting upon access included the geographic and financial accessibility of the application process, procedures for determining eligibility and compliance of service providers.


Introduction
Social protection systems are an integral component to most governments' strategies for poverty reduction, as well as broader social and economic development (World Bank 2012). Although conceptualisations vary, social protection may be defined as "public actions taken in response to levels of vulnerability, risk, and deprivation which are deemed socially unacceptable within a given polity or society" (Conway et al. 2000). Aims of social protection similarly differ, with some focussing narrowly on protecting minimum living standards, while more "transformational" approaches view social protection as a tool to develop stronger livelihoods, tackle chronic poverty and address social inequalities (Merrien 2013).
Almost all countries, including low-and middle-income countries (LMICs), have put in place some social protection initiatives, with many continuing to focus on expanding their content and coverage, as well as improving their impact (International Labour Organization 2017). There are variety of instruments that may fall under the banner of social protection (Gentilini and Omamo 2011). Core components of social protection are social assistance and social insurance (Barrientos 2018). Social assistance refers to non-contributory transfers of cash or kind to groups deemed vulnerable to or currently in poverty (e.g. conditional/unconditional cash transfers, food assistance). Social insurance programmes in turn are typically contributory, and are designed to mitigate risks that may be encountered throughout the life-course (e.g. illness and injury, unemployment, loss of income in old age, natural disasters). Under broader definitions of social protection, labour market regulations (e.g. minimum wage, non-discrimination legislation) or interventions to promote equitable access to services and enjoyment of basic rights may be considered as social protection (Gentilini and Omamo 2011;Devereux and Sabates-Wheeler 2004). In LMICs, however, social assistance has been the main pillar of social protection (Barrientos 2011).
Developing and strengthening social protection systems is a core aim of the International Labour Organization's Social Protection Floors Recommendation (2012) and the 2015-2030 Sustainable Development Goals (SDGs) (United Nations 2017); For example, SDG Target 1.2 highlights social protection as a key input to "end poverty in all its forms", while other SDGs and their targets specify social protection as central for achieving a range of goals, such as universal health coverage, gender equality, reducing inequality and decent work for all (International Labour Organization 2017). While acknowledging that all citizens should be provided with certain nationally appropriate guarantees-such as income security and access to essential services-it is acknowledged that social protection is particularly important for groups and individuals at a higher risk of poverty and facing other forms of marginalisation (Gentilini and Omamo 2011;Devereux and Sabates-Wheeler 2004).
People with disabilities-who account for upwards of 15% of the global population-are often considered as key beneficiaries in national and international social protection strategies (Yeo 2001;Elwan 1999;World Health Organization & World Bank 2011). Disability is defined in the United Nations Convention on the Rights of Persons with Disabilities (UNCRPD) as including people "who have long-term physical, mental, intellectual or sensory impairments which in interaction with various barriers may hinder their full and effective participation in society on an equal basis with others" (United Nations Office of the High Commissioner on Human Rights 2006). People with disabilities may be targeted for inclusion in social protection due to high levels of economic and multidimensional poverty, as well as frequent social exclusion and marginalisation (United Nations 2015;Gentilini and Omamo 2011;Holzmann and Jørgensen 2001). For example, in a recent systematic review of 150 studies in LMICs, 81% found that people with disabilities were economically poorer compared with people without disabilities (Banks et al. 2017). Similarly, other studies have found that people with disabilities experience high levels of multidimensional poverty, such as malnutrition and lower levels of health, barriers to accessing education and healthcare, and exclusion from decent work and social participation Hume-Nixon and Kuper 2018;Mizunoya and Mitra 2013;Mizunoya et al. 2018). Poverty may even be underestimated amongst people with disabilities, due for example to high out-of-pocket spending on disability-related items (e.g. personal assistance, rehabilitation and assistive devices), intra-household discrimination in the allocation of resources and opportunities, as well as structural inequalities that prevent equal participation in society World Health Organization & World Bank 2011;United Nations 2015). The combination of spending on extra disability-related costs, social exclusion and opportunity costs from caregiving provided by other household members can lead to increased vulnerability to poverty amongst people with disabilities and their households (Schneider et al. 2011b;Kidd 2017).
In addition to this needs-based argument, the right of people with disabilities to inclusion in social protection is established in international treaties such as Article 28 of the UNCRPD and Articles 22 and 25 of the Universal Declaration of Human Rights (United Nations 2016;United Nations General Assembly 1948). As mandated in the UNCRPD's Article 28, people with disabilities have the right to equitable access to mainstream social protection programmes-such as health insurance, pensions and other benefits where eligibility is not dependent on disability status (Devandas Aguilar 2017;United Nations 2015). Further, disability-targeted programmes may be required to address disability-specific concerns such as the need for rehabilitation and assistive devices, workplace support or specialist education. People with disabilities also frequently face additional costs, such as for extra transportation or medical expenses, which can deepen inequalities in standards of living compared with people without disabilities (Devandas Aguilar 2017; International Labour Organization 2017; Mitra et al. 2017). Subsidising these additional costs and promoting access to services and support required for full and equal participation are typically the main objectives of disability-targeted programmes. Overall, a central aim of disability-inclusive social protection across both mainstream and targeted schemes is addressing social exclusion (United Nations 2015).
For people with disabilities to partake in any of the intended benefits of social protection, they must first be able to access programmes. A recent systematic review found little evidence across LMICs on whether people with disabilities were actually participating in social protection programmes, despite the emphasis placed on targeting people with disabilities as key recipient groups (Banks et al. 2016). Available evidence, however, suggests that coverage is well below need and many eligible individuals are not accessing programmes (Palmer and Nguyen 2012;Mitra 2005;Saloojee et al. 2007). Low coverage may be linked to broader social exclusion of people with disabilities; For example, policymakers may not adequately take into account the needs of people with disabilities when designing and delivering social protection-such as in aligning programmes to address disability-specific vulnerabilities to poverty or ensuring application procedures are accessible-due to discrimination, lack of understanding or insufficient political will (Kidd 2017). For disability-targeted programmes, disability assessment procedures are a frequent cause of exclusion if eligibility criteria are poorly defined or administrative capacity is insufficient to properly implement them (Kidd 2017;Schneider et al. 2011b).
High levels of exclusion are reflected in recent efforts by the International Labour Organization to estimate coverage of people with disabilities in disability-targeted programmes, using regional prevalence estimates of disability from the 2002-2004 World Health Surveys combined with country-level programme enrolment figures. Under this approach, it was estimated that 27.8% of people with "severe" disabilities are recipients of a disability-targeted cash benefit, with access differing substantially across regions, with coverage lowest in Asia and the Pacific at 9.4% and highest in Eastern Europe (97%) (International Labour Organization 2017). However, these figures are based on modelled estimates, rather than direct survey, and the definition of "severe" disability is unclear. Further, little evidence is available on how enrolment differs amongst people with disabilities, enrolment in non-targeted schemes, or factors affecting access to both mainstream and targeted programmes (Banks et al. 2016).
Consequently, this study explores participation of people with disabilities in social protection programmes, with Tanahun District of Nepal as the study setting. This research uses mixed methods to assess coverage (through direct survey), how coverage varies amongst people with disabilities (e.g. by gender, impairment type), as well as challenges and facilitators to enrolling in or using relevant social protection programmes. This research benefits from a population-based study design and from the use of the Washington Group question sets for measurement of disability, which are internationally validated and recommended for robust and comparable disability statistics, including in the tracking of progress towards the SDGs (Washington Group on Disability Statistics 2009).

Overview of Social Protection Entitlements in Nepal
Social protection entitlements in Nepal are gradually expanding, particularly social assistance (World Bank 2017). Most social protection entitlements are targeted to various groups deemed to be at greatest risk of poverty and marginalisation, including people with disabilities (Government of Nepal 2015b). There are also a few schemes open to the broader population-namely for social insurance and healthcare provisions-although these are more limited in scope and coverage.
While specific aims of social protection in Nepal have not been formally outlined, they are meant to "maintain a degree of equity among citizens" and ensure all citizens are able to "live a decent life" (Khadka 2017). Social protection entitlements focus both on alleviating deprivation in income (i.e. social assistance), as well as fostering broader social inclusion (e.g. through educational scholarships, healthcare discounts, employment support) (Khanal 2013). The range of entitlements offered in Nepal recognise both economic and social challenges facing people with disabilities and other groups experiencing poverty or other forms of deprivation (Barrientos et al. 2016).

Disability-Targeted Social Protection Entitlements
Key disability-targeted entitlements are listed in Table 1 (Government of Nepal 2015a, 1982). In order to be eligible for these entitlements, people with disabilities must first undergo an assessment of disability and receive a disability card. To apply for a disability card, an individual must first submit an application to the Women and Children's District Office (WCDO). The application consists of the following: a letter from the applicant's Village Development Committee (one of the lowest administrative units in Nepal) verifying their identity and place of residence, birth certificate or citizenship card, photographs, and completed application form. The application form includes questions on self-reported type and severity of the disability, any difficulties the person faces because of their disability (e.g. in doing daily chores, working), and the need for assistive devices. Medical documentation and references from Disabled Peoples' Organizations (DPOs) can also be included to support the application, although they are not an explicit requirement.
Once the application is processed by the WCDO, the applicant is called for an in-person evaluation, which is typically conducted by the Disability Identification Committee. The Disability Identification Committee comprises representatives from local government offices (e.g. District Health Office, WCDO, District Development Committee) and, if available, from a registered DPO operating in the district. Assessments of disability are informed by an in-person evaluation on the applicant's level of difficulty in performing daily activities and participating in social life, gauged through observation and responses in the application form.
Based on the results of the assessment, an individual is classified into one of four disability card categories (from most to least severe: red, blue, yellow, white). The disability card category is based on an assessment of the level of support needed: • Red-"complete disability"-difficulty in performing daily activities, even with the help of others • Blue-"severe disability"-difficulty in performing daily activities without the help of others • Yellow-"moderate disability"-able to perform daily activities and participate in social life if environment is barrier free, and appropriate training and education are provided • White-"mild disability"-ability to perform daily activities and participate in social life if environment is barrier free The disability card classification determines which social protection benefits a person is eligible to receive. While the Disability Allowance is reserved for people with the two most severe card categories (red, blue), the remainder of the benefits are open to disability cardholders of any colour classification. Nationally, 198,788 people were registered as receiving disability cards in the fiscal year 2014/2015 (Government of Nepal 2016a). Slightly under half (93,858) have yellow or white cards, which means they are not eligible for the Disability Allowance.
For red and blue cardholders to receive the Disability Allowance, they must submit an additional application to their local Village Development Committee (VDC). Separate applications are also needed for educational scholarships, while other benefits (e.g. health and transport discounts) can be accessed through presentation of the disability card at the point of use.

Non-Disability-Targeted Social Protection Entitlements
People with disabilities may also be eligible for programmes aimed at other targeted groups, if they meet their eligibility criteria; For example, children may receive educational scholarships for reasons other than disability, such as if they are a girl living in poverty, are from the Karnali Zone or belong to the Dalit caste. The amounts provided through these scholarships vary, but are typically less than is provided for children with disabilities. Similarly, unconditional social assistance is available to older adults (aged 70+ or 60+ for the Karnali Zone/members of Dalit caste), single women over 60, widows, indigenous groups considered to be endangered and for children under 5 from the Dalit caste or who live in the Karnali Zone. Amounts range from NR 200-1000 per month (US $2-9). It is important to note that an individual cannot receive more than one type of social assistance at any one time. People receiving a pension, or on a government salary, are ineligible to receive social assistance (Johnson and Subedi 2017).
Regarding pension schemes, Nepal has been expanding available schemes, although presently all are restricted to formal-sector employees. Pensions are mandatory for public-sector employees, and can be accessed after 20 years of services (16 for military, 13 for people with disabilities) (Khadka 2017). Voluntary contributory pension schemes are available for formal-sector employees, in which contributions are exempt from income tax. In addition to pensions, public-sector employees and employees of formal-sector businesses with at least ten employees have legal protections covering maternity and sickness leave, as well as provisions for injury, disablement and death due to work-related accidents.
Lastly, all Nepali citizens are entitled to some basic healthcare provisions. This includes coverage of some services at lower-level health posts and 60 prescription drugs free of charge. Older adults aged 70 and above can access additional health services without charge, up to an annual limit of NR 4000 (US $38).

Methods
A mixed methods approach was used to assess access to existing social protection programmes. In-depth qualitative and quantitative research was undertaken in one district in Nepal to measure coverage and uptake of social protection benefits, as well as barriers and enablers to enrolment among people with disabilities. Qualitative interviews were also conducted with national-level stakeholders to contextualise findings within national policies and programmes. For all components, disabilitytargeted entitlements were the main focus, although non-disability-targeted programmes were also explored where possible.
Ethical approval for this research was granted by the Ethics Committee at the London School of Hygiene & Tropical Medicine and the Nepal Health Research Council. Informed written consent was obtained from all study participants prior to beginning any interviews. For children below 16 years (age of consent) and people with impairments that severely limited their ability to understand/communicate, a carer answered on their behalf as a proxy.
Data collection was undertaken between August and September 2016.

Setting
A rapid policy analysis was conducted in 2015 of social protection systems in Asia and the Pacific and the extent to which they were inclusive of people with disabilities (International Centre for Evidence in Disability 2015). From this analysis, Nepal was selected as a study site as it was identified as having a relatively strong social protection system, which had made concerted efforts to be inclusive of people with disabilities, particularly through multiple disability-targeted programmes. Consequently, studying Nepal's system offered an opportunity to identify and describe examples of good practice in disability-inclusive programme design and implementation. Within Nepal, data collection was undertaken in Tanahun. Tanahun is a predominantly rural district in Province No. 4, which is part of the Hills Region. Tanahun was selected as the study setting after consultation with in-country stakeholders who recommended this district due to its strong network of DPOs, adequate availability of disability support services and relatively well-functioning social protection administration. To complement district-level research, qualitative interviews with national-level stakeholders were used to provide an overview of the national context.

Quantitative Research in Tanahun
Quantitative data collection included a population-based survey of disability across Tanahun, with a nested case-control to compare awareness of and enrolment in social protection programmes between people with and without disabilities.
The 2011 national census was used as the sampling frame for the populationbased survey. A two-stage sampling strategy was employed based on methodology used in other surveys (Kuper et al. 2006). In the first stage, probability-proportionate-to-size sampling was undertaken to select 30 clusters (wards of VDCs) in Tanahun. In the second stage, 200 people aged 5 years and older were enumerated and recruited in each cluster through compact segment sampling. The sample size of the population-based survey was set at 6000 to ensure the identification of adequate numbers of people with disabilities for the case-control study. For the case-control, a sample of 240 cases with disabilities and 240 controls without disabilities was powered to detect an odds ratio of 1.9, assuming 80% power and a prevalence of exposure (e.g. poverty, the main measure of the broader study) of 25% among controls.
Disability was measured using two age-specific question sets created by the Washington  Both the adult and child question sets focus on similar activities. For most questions, participants can select one of four response options describing their level of difficulty in performing each activity: none, some, a lot or cannot do at all. Anxiety and depression in the adult set were assessed through a two-part question on intensity and frequency of symptoms. For the purpose of this study, disability was defined as: • Adults (16+): • Reported "a lot of difficulty" or "cannot do" in at least one of the following domains: seeing, hearing, walking/climbing, communicating (understanding/being understood), remembering/concentrating, self-care, upper-body strength, fine dexterity. • Reported experiencing symptoms of anxiety or depression "daily", at a level described as "a lot".
• Children (5-15): • Caregiver reported that, compared with other children of the same age, the child experienced "a lot of difficulty" or "cannot do" in at least one of the following domains: seeing, hearing, walking, self-care, understanding, being understood, learning, remembering. Child was worried/sad "a lot more" often than other children.
Interviewers were instructed to ask each household member about his/her functioning directly, if the person was over the age of consent (16+ years) and present at the time of the household visit. Household heads/caregivers reported on children's and absent members' functioning. The cut-offs used in this study to define disability are in line with the Washington Group guidelines (Washington Group on Disability Statistics 2010), and are comparable to eligibility criteria for disability-targeted social protection in Nepal. Further, a disability severity score was created by summing the level of difficulty across all activities (no difficulty = 0, some difficulty = 1, a lot of difficulty = 2, cannot do = 3; anxiety/depression = 3) and dividing it over the maximum possible score for the two question sets (27 for adults, 21 for children) to obtain a score from 0 to 100%. This method for constructing disability severity has been used in other research (Mitra 2017).
All individuals identified during the household survey as having a disability were recruited into the case-control study. The case-control questionnaire explored in greater depth need for, awareness of and participation in disability-targeted and non-targeted social protection programmes. Each person with a disability (case) was then matched to a person without a disability (control), who was of similar age (± 5 years), the same sex and resided in the same cluster. Controls could not be from households with members with disabilities. Additionally, 92 people with disabilities were recruited from registers of the Disability Allowance. To select participants from registers, 2-3 people were randomly selected from non-selected segments of included clusters. These participants received the same questionnaires as cases recruited in the population-based survey, but they were not matched to controls. All register-recruited and population-based cases who received the Disability Allowance received a third questionnaire, which included questions about the application experience.
Trained data collectors administered questionnaires, which had been translated into Nepali and created using Open Data Kit (ODK), using computer tablets. Multivariate regression using STATA 15 was undertaken to compare participation in various schemes between respondents with and without disabilities, controlling for age, sex and location.

Qualitative Research
In-depth, semi-structured interviews were carried out with people with disabilities who were and were not recipients of social protection (namely disability-targeted programmes), as well as stakeholders in disability and/or social protection in the district. People with disabilities were interviewed about their awareness of disabilitytargeted programmes. Social protection recipients were asked about their experience applying for and participating in different schemes. For stakeholders, interviews focussed on social protection policies and programmes, strengths and challenges of existing programmes, including factors affecting access for people with disabilities.
A purposive sample of participants with disabilities was recruited using data collected through the population-based survey. Participants were selected to reflect variation in terms of sex, age (children, working-age and older adults), impairment type and geographic distribution (rural versus urban). Adults with disabilities were interviewed directly; however, in some instances, proxies were used if the individual's impairment severely affected his/her ability to understand or communicate (e.g. people with profound hearing impairments but who had never learnt sign language, severe cognitive/intellectual impairments), even with available supports (e.g. sign language, visual aids). For children, interviews were predominately with caregivers, as young children were unlikely to be directly involved in decisions to apply or manage the application process.
In addition, national-, district-and community-level stakeholders were selected through recommendations from in-country advisors (e.g. National Federation of Disabled, Nepal) and snowball sampling. Stakeholders included representatives from relevant government agencies, United Nations agencies, non-governmental organisations (NGOs), DPOs, disability service providers and administrators responsible for social protection implementation.
Interviews with all participants were transcribed in Nepali, and a thematic approach was used to analyse findings.

Description of the Study Samples
For the population-based survey, 6000 household members were included and 5692 screened for disability (response rate 94.9%). Overall, 214 individuals were identified as having a disability [prevalence 3.8%, 95% confidence interval (CI) 3.3-4.3%]. Prevalence of disability was slightly higher in men compared with women [adjusted odds ratio (aOR) 1.4, 1.0-1.7], and increased substantially with age (from 1.5% in children aged 5-18 to 19.6% in adults aged 76+ , p < 0.001).
Overall, 418 people from the population-based survey (209 cases and 209 controls) took part in the case-control study (response rate 97.9%). Cases and controls were well matched by age, gender and location, as there were no significant differences between groups in these characteristics. An additional 92 people with disabilities were recruited from Disability Allowance registers, who were not matched to controls, but received similar questionnaires.
For the qualitative research, 35 people with disabilities were recruited (response rate 100%). Of the 35 people, 9 were caregivers of children and 14 involved direct interviews with adults with disabilities. The remaining 12 adults involved people with severe intellectual/cognitive and/or communication impairments. In these instances, proxies either assisted in providing additional information (n = 9) or answered fully on behalf of the participant (n = 3). Within this group, 28 participants had a disability card. By impairment type, the following breakdown was observed: physical/mobility (n = 17), communication (n = 10), vision (n = 5), hearing (n = 5), psychosocial (n = 5), intellectual/cognitive (n = 5); 14 respondents had multiple impairments. Respondents ranged in age from 5 to 86 years (10-17 years: n = 9; 18-64 years: n = 22; 65+ years: n = 4), and there was a near-equal mix by gender (female, n = 19). For key informants, 13 district-and community-level and 15 national-level stakeholders were interviewed.

Enrolment in Social Protection
Overall, 65 (31.1%) people with disabilities identified during the population-based survey had a disability card (Table 2). This included 34 (52%) people who were eligible for the Disability Allowance (red card, n = 13; blue card, n = 21). The remainder (47%) had lower-level disability cards (yellow card, n = 18; white card, n = 13). Over one-third of all people with disabilities identified in the survey received some form of social assistance, which was significantly higher in comparison with people without disabilities (aOR = 3.0, 95% CI 1.6-5.3). The Old Age Allowance was the predominant form of social assistance received, among both people with and without disabilities. The Old Age Allowance had universally high coverage, as over three-quarters of eligible older adults were enrolled. Coverage of the Disability Allowance was lower at 13.4%. No one who did not meet the study's definition of disability was receiving the Disability Allowance, indicating low inclusion errors. However, six people (7.7%) who had received a disability card of an eligible classification were not receiving the Allowance. Disability cardholders are entitled to a range of benefits other than the Disability Allowance. However, uptake of most of these benefits was low (Table 3). Transportation discounts and educational scholarships had the highest uptake (25.5% and 13.0%, respectively). Uptake of these benefits varied by disability card level, with people with the lowest assessed severity more likely to use benefits such as transportation discounts and scholarships. Table 4 compares characteristics of people with disabilities who were and were not disability cardholders. Overall, coverage was highest for children and adults under 50. Coverage was lowest for people with sensory limitations as well as anxiety and depression. People with multiple functional limitations and limitations in selfcare were particularly likely to have a disability card.
While older adults with disabilities were less likely to have a disability card, this age group had the highest coverage of social assistance overall, in large part due to the high coverage of the Old Age Allowance (Table 5). There was no clear association between functional limitations and receipt of social assistance, although recipients had slightly higher severity scores. Similarly, no differences in the likelihood of receiving social assistance were observed by gender or location.

Factors Influencing Enrolment in and Uptake of Social Protection Among People with Disabilities
Across study methods, several factors emerged which impacted enrolment in and uptake of social protection among people with disabilities. These factors concerned: (1) geographic accessibility, (2) financial accessibility, (3) determining eligibility, (4) understanding the application process, (5) awareness and perceived utility of programmes and (6) compliance among service providers. These are described in detail below. While the focus was predominantly on disability-targeted schemes, other programmes are also discussed where relevant.

Geographic Accessibility
In Nepal, applications for disability cards, a precondition to receiving the Disability Allowance and other disability-targeted social protection entitlements, are processed in district headquarters. In contrast, applications for non-disability-targeted social protection benefits are conducted at the more local VDCs, which are geographically nearer. As VDCs cover a relatively small catchment area, few people reported problems travelling to these offices. Getting to district headquarters, however, was cumbersome for many people, particularly for people with mobility limitations or who lived in remote areas. Tanahun covers an area of over 1500 km 2 , much of which is rural with limited roads and transportation links. The lack of accessible transportation in many parts of Tanahun   Table 5 Characteristics of recipients of social assistance (disability-targeted and non-targeted) compared with non-recipients with disabilities (population-based survey and register-recruited participants) Note: includes people with disabilities from population-based survey (n = 209) and registers (n = 92) a Adjusted by age, sex and location b Not mutually exclusive (i.e. sum > 100%). Domains derived from Washington Group questions as follows: physical (difficulties walking, with upper-body function or fine dexterity), sensory (hearing/seeing), communication, cognitive (remembering, learning and understanding) c Total across Washington Group domains (0 = no difficulty, 1 = some, 2 = a lot, 3 = cannot do for each domain; 3 = anxiety/depression), divided by maximum score (21 for  and the rest of the country was noted by key informants, people with disabilities and their caregivers alike as compounding difficulties getting to application points. Mirroring these responses, in the quantitative survey of Disability Allowance recipients, over 60% of people who had successfully completed the application process for the disability card reported difficulties getting to the application point. Furthermore, upon reaching the application office, many people were asked to provide medical documentation before their application for the disability card could be processed. Gathering the necessary documentation may involve travel to cities in other districts; For example, there are no ear, nose and throat (ENT) specialists in Tanahun, so people with hearing impairments were required to travel to Pokhara, in the neighbouring district (50 km from the district headquarters of Tanahun) for assessment. In fact, many respondents reported travelling to Pokhara and other cities to get medical documents for a range of different impairments for their applications. These difficulties were also reinforced in the survey of Disability Allowance recipients, as 85% reported needing a medical evaluation, which for 11% required travel outside the district.
In recognition of these and other barriers, outreach camps are organised by the WCDO, the District Health Office and community-based organisations. These camps are held in various locations throughout the district so that people can submit applications and undergo an assessment of disability without travel to the district headquarters. Outreach camps were started in Tanahun but have since been rolled out throughout Nepal, and are now mandated in the Disability Identification Card Distribution Guidelines 2008. Still, key informants involved in the process note that, while initially effective, they are looking to "cut off" outreach camps as they believe most people have been reached and anyone remaining "should come on their own as it's too expensive to organise outreach programmes for a few people".

Financial Accessibility
Officially, there are no direct costs (i.e. application fees) for submitting applications for any social protection programmes. However, 20% of surveyed Disability Allowance recipients reported paying nominal fees at application points for their application to be processed (mean NPR 266 [US $2.50]). Additionally, indirect costs such as for travel to application points are often incurred. For applications conducted at the VDC, these costs are typically minimal. However, to obtain a disability card, applicants must travel to the district headquarters, which can involve substantially higher transport costs. In the survey, 49% of Disability Allowance recipients noted transportation costs as a challenge. Accommodation costs are also common, as several applicants in the qualitative research reported waiting for several days in the district headquarters to meet with the required officials, for the Disability Identification Committee to assemble or to gather additional documentation.
Furthermore, disability card applicants and anyone accompanying them often must forgo time spent on productive activities such as work or schooling; For example, the mother of a young woman with an intellectual impairment highlighted the financial challenges associated with applying for her daughter's disability card. Her costs included travel to the district headquarters and fees at a government hospital for medical documentation. She was fortunate to have family members to stay with while the application process was being completed, otherwise she would have had to pay for room and board. When the application was delayed, she had to leave her daughter with relatives to complete the process, as she had to return to home: "I requested [to the programme official], 'Don't do such a thing [delay assessment], sir. I am alone, no one at home. I have left cattle at home, sir. In this planting season of Jestha (May/June), don't do this, please make it [the disability card]." While fortunate to have relatives near the headquarters who could assist with the application when she had to return, others do not have this support.

Determining Eligibility
In comparison with disability-targeted schemes, determining eligibility for nondisability-targeted programmes is relatively straightforward. All social assistance programmes require a citizenship card for people over 16 or a birth certificate for children. This documentation is sufficient to prove eligibility for the Old Age Allowance, Child's Grant and allowances for endangered indigenous groups. For the Single Women (age 60+)/Widows Allowance, a death or divorce certificate is also needed if the woman was ever married. While several participants reported difficulties gathering these documents, once obtained, assessments of eligibility are mostly clear-cut. Age, caste, ethnicity and marital status are all relatively objective criteria that can be determined directly from these documents.
In contrast, determining if a person meets disability eligibility criteria is more complex. Disability assessments for disability cards are based on an applicant's level of difficulty in performing daily activities. The language used in policies and guidelines for defining and categorising disability is broadly in-line with the UNCRPD and the International Classification of Functioning, Disability and Health (World Health Organization 2001); For example, categorisations focus on difficulties performing daily activities, rather than the presence of medical impairments. Further, assessors are instructed to consider the role of individual and environmental characteristics (e.g. availability of support) when assessing functional status.
However, guidance documents on how to classify individuals into the four categories-namely the Disability Identification Card Distribution Guideline 2008were perceived as vague. Training of assessors was also reported by key informants to be limited. Consequently, in practice, there is a large degree of subjectivity to the assessment. Key informants familiar with the process reported that, for observable disabilities such as physical impairments, blindness or severe intellectual impairments, assessments were straightforward. However, for mental health conditions or mild to moderate communication and developmental impairments, classifications were more challenging; For example, mental health providers reported that few of their patients had upper-level disability cards even if they had a severe mental health condition. Additionally, functional decline due to ageing is often not considered to be a disability. These attitudes on disability may lead to placement in lower card levels or the denial of a card altogether.
The lack of clarity on assessment guidelines combined with a low understanding on the impact of certain impairments among assessors may lead to an overreliance on medical documentation in determining eligibility. When the committee cannot decide on a classification, they will request that the applicant seek a medical assessment by a specialist before making a decision on their application. In the quantitative survey, 85% of disability card holders had been asked to provide medical documentation, which was primarily obtained from the district hospital. This additional requirement can be cumbersome for applicants and leads to a more medical-modelbased assessment rather than one in line with the UNCRPD.
In Tanahun, mental health providers and DPOs reported collaborating with the Disability Identification Committee to improve their understanding on the disabling impact of certain impairments and conditions. Across Nepal, registered DPOs are often part of the Disability Identification Committee and can provide input into the disability assessment. In Tanahun, the involvement of DPOs was seen as beneficial by other key informants, and significant weight was given to their recommendations. However, not all districts in Nepal have a registered DPO, and the capacity of some DPOs is limited.
A final improvement to determining eligibility was the removal of quotas on the number of blue disability cardholders who could receive the Disability Allowance. Previously, these quotas led to arbitrary rationing decisions, excluding many eligible people with disabilities from the cash transfer. After the removal of the quota, the number of Disability Allowance recipients nearly doubled between 2014/15 and 2015/16 (from 33,578 to 62,320).

Understanding the Application Process
Across social assistance programmes, a common challenge was the yearly application cycle. Applications not submitted before the one annual deadline must wait an additional year before receiving any allotments. Amongst Disability Allowance recipients in the quantitative survey, almost half (47%) reported waiting over a year between submitting their application to the VDC and receiving their first payment. Similarly, for recipients of the Old Age Allowance, many are not aware that they need to apply the year before they meet the minimum starting age.
Applications for disability cards presented further challenges. Many people with disabilities and/or their caregivers assisting with the application process reported being unclear on which documents were needed. Even if they had all required documents, many were asked to provide extra medical documentation to complete the assessment, as described above. Furthermore, since the Disability Identification Committee meets infrequently, applicants may face long waiting periods if they are not aware of the schedule. In the quantitative survey, people who had successfully gotten a disability card reported an average of three visits to application points to complete the process.
Lack of clarity on application procedures led to delays or frustration; For example, a father faced many difficulties getting a disability card for his daughter, who is blind and has a hearing impairment. He explained that he needed to go to Pokhara three times and Damauli (district headquarters) four times, as he was told "this thing or that thing was missing or would not do." In certain cases, the process appeared so daunting that it deterred starting or continuing an application altogether.
Finally, people with disabilities, including disability cardholders, and their caregivers were often not clear on how assessments, including card categorisations, were decided. Most people equate the disability card with the Disability Allowance, so frequently expected to get a cash transfer if they are applying for a card. Lack of awareness about the assessment criteria and the benefits attached to each category can lead to resentment and distrust of the process; For example, several respondents felt assessment decisions came down to political connections, such as the father of a man, aged 49, who has mobility and communication impairments. He did not apply for a card as he believes that lack of afno manchhe (one's acquaintance in power or position) would be a barrier. This perspective was also mirrored by some of the key informants; For example, some felt that those who are "clever" and literate go to the district and get the type of card they wish, but those who have greater needs often do not get it.
Several provisions were reported to have improved the ease of the application process, particularly for disability-targeted programmes; For example, for all social assistance programmes, policies are being changed to increase the number of annual deadlines. Additionally, in Tanahun and other districts, the involvement of Disability Identification Committees in disability card assessments is limited to complex cases. Since the Disability Identification Committee meets infrequently and irregularly, key informants noted that having the WCDO, the office where applications are first submitted, complete most assessments streamlines the process: wait times are decreased and more predictable, which also improves financial accessibility given that many people travel and reside in the district headquarters while their application is processed. Finally, in Tanahun, key informants reported strong involvement of local DPOs, including in guiding applicants through the application process.

Awareness and Perceived Utility of Benefits
While awareness of the Disability Allowance and other social assistance programmes was generally high, many people with disabilities and their caregivers did not know about benefits such as transportation and healthcare discounts that are available for lower-level cardholders.
Additionally, even if people with disabilities or their caregivers were aware of these programmes, many did not perceive them to be useful. This perception was linked to concerns about the quality and availability of services; For example, public transportation is not available in many parts of Nepal, and vocational training programmes may not offer relevant skills. Similarly, schools might not have adequate resources to support the learning of children with disabilities. Of note, while scholarships are available to any level cardholder, national-level key informants stated that scholarships were not intended for children with the most severe disabilities (red cardholders) as "they will not be going to school".

Compliance Among Linked Service Providers
Particularly for healthcare and transportation discounts, several respondents in the qualitative research reported difficulties accessing benefits at point of use; For example, a 49-year-old woman related that bus drivers scold her when she presents her card for a discount. Similarly, the mother of a man with an intellectual impairment went to the district hospital after hearing about healthcare discounts associated with the disability card. She reported that "not a drop [of medicine] was given for free" and was instead told by a staff member that a "disability card won't do anything in the hospital".
Key informants attributed poor compliance amongst service providers to lack of awareness of disability card benefits, as well as motivation to maximise profits. To improve awareness, the WCDO runs information sessions with transportation owners' organisations and other service providers. However, monitoring and enforcement mechanisms to ensure compliance are admitted to be weak.

Discussion
This research measured coverage and uptake of disability-targeted and non-targeted social protection programmes among people with disabilities in Nepal, and explored factors that encourage or impede participation. Evidence in this area has thus far been relatively limited (Devereux and Sabates-Wheeler 2004), so this research can help inform planning and delivery of social protection systems to ensure equitable access for people with disabilities.
Few studies have measured social protection enrolment among people with disabilities, in either disability-targeted or non-targeted schemes. Concerning disability-targeted schemes, slightly less than one-third of people with disabilities had a disability card, and only 13.4% were receiving the Disability Allowance. A small portion (7.7%) of eligible cardholders were not receiving the Disability Allowance, which is significantly lower than the 58% exclusion reported in other areas of Nepal in a survey conducted before the removal of quotas for blue card holders (Holmes et al. 2018). Overall, coverage of the Disability Allowance is slightly higher than the modelled regional estimate for disability-targeted cash benefits, which is estimated at 9.4% for Asia and the Pacific (International Labour Organization 2017).
Over one-third of people with disabilities were receiving some type of social assistance, which was much higher compared with people without disabilities. The Old Age Allowance was the main cash transfer accessed amongst both people with and without disabilities, and had universally high coverage. High coverage of the Old Age Allowance amongst older adults as a group has been reported in other research from other areas of Nepal (Johnson and Subedi 2017;HelpAge International 2009). This research indicates that older adults with disabilities are accessing this cash transfer in equal proportion to older adults without disabilities, indicating equity in access. Among older adults eligible for both the Disability Allowance and the Old Age Allowance, there was a clear preference for the Old Age Allowance. The Old Age Allowance currently provides the same amount as the Disability Allowance for red cardholders, although disability cards also provide access to other benefits (e.g. transportation discounts). However, the application process is much more straightforward for the Old Age Allowance: eligibility is relatively easy to assess, applications are conducted in the local VDC and application procedures are straightforward.
In general, disability-targeted schemes in Nepal appear much more challenging for eligible individuals to enrol in than non-targeted schemes. No other schemes require travel to the district headquarters as part of the application process, a requirement that may be particularly onerous for people with disabilities. As people with disabilities are more likely to be living in poverty, meeting the financial costs associated with travel will be more difficult (Banks et al. 2017). Similarly, the lack of accessible transportation is particularly disadvantageous to people with mobility limitations or in remote areas. Other research, including in other areas of Nepal (Johnson and Subedi 2017;Drucza 2016), has highlighted that geographic and financial factors can be a barrier to accessing social protection among people with disabilities (Schneider et al. 2011a;Gooding and Marriot 2009;Devandas Aguilar 2017;Kidd 2017). As many areas of Nepal have even more inaccessible topography compared with Tanahun, such as across the Mountain Region, geographic and financial challenges associated with travelling to district headquarters for disability cards will likely be even greater in these regions.
Further, challenges in establishing and applying disability assessment criteria mirror research in other contexts, indicating a widespread challenge in the design and implementation of disability-targeted programmes (Graham et al. 2013;Devandas Aguilar 2017;Schneider et al. 2011a, b). At the policy level, Nepal's disability assessment guidelines focus on functioning, which is in line with international conceptualisations of disability (United Nations Office of the High Commissioner on Human Rights 2006; World Health Organization 2001). Functioning-based assessments can also be more practical to implement than medical-based protocols due to a lower reliance on specialised resources and expertise (Schneider et al. 2011b;Mitra 2005;Gooding and Marriot 2009;Devandas Aguilar 2017;Schneider et al. 2011a;Mont et al. 2016). However, evidence from Tanahun and other areas of Nepal suggests implementation does not always follow official guidelines (Johnson and Subedi 2017). The majority of applicants reported requiring medical documentation to complete their application, which was both cumbersome to obtain and moves away from the recommended functioning-based approach to disability assessment. While functioning-based tools for assessing mental health conditions and some other impairments are a global challenge ), many other disability types can be effectively captured without medical assessment.
Additionally, universally low coverage of social insurance (e.g. pensions) and social protection benefits other than cash transfers has been reported in other research from LMICs (Barrientos 2011). Social insurance is often limited to the formal sector, which covers a minority of people in most LMICs. For Nepal, 90% of the labour force works in the informal sector and are thus ineligible for social insurance (Khadka 2017). There is some evidence from other countries that people with disabilities are even more likely to work in the informal sector compared with people without disabilities (Mizunoya and Mitra 2013). Further, women with disabilities are particularly likely to be excluded from social insurance, due to gender as a source of exclusion from employment and greater engagement in unpaid and domestic work (Devandas Aguilar 2017;United Nations 2015).
Moreover, the restriction to only receiving one type of social assistance does not account for intersecting dimensions of exclusion (Drucza 2016); For example, disability and poverty are both more common in older age (Barrientos et al. 2003;World Health Organization & World Bank 2011). Older adults with disabilities will thus have to contend with both disability-related costs as well as the loss of income from retirement from the same the allotment. Older adults with disabilities will thus have to cover both disability-related extra costs as well as the loss of income from retirement from one cash transfer, without adjustments in value to account for increased need (Barrientos et al. 2003). Similarly, single women or widows with disabilities may not be able to cover both daily living costs (if they are not engaged in paid work) on top of disability-related costs from the same allowance. Removing restrictions on the receipt of multiple benefits or varying benefit levels would therefore be more equitable, to account for the multiple sources of vulnerability.
Finally, although this study did not measure the impact of receiving social protection, it is unlikely that in its current state social protection will be sufficient in meeting its intended goals among recipients with disabilities in Nepal, particularly in promoting stronger livelihoods and social inclusion. Integrated poverty reduction programmes, which combine income transfers, investments in human capital and improved access to services, are more likely to address chronic poverty and social exclusion compared with pure income transfers (Barrientos 2018;United Nations 2015). The design of disability-targeted social protection in Nepal has taken this multidimensional approach, by combining the Disability Allowance with linked benefits to strengthen recipients' capabilities (e.g. vocational training, education scholarships) with interventions to increase access to services (e.g. transportation and healthcare discounts). However, low uptake of these linked benefits, due in part to concerns over their quality, availability and lack of compliance among services providers, hampers the effectiveness of these tools. Further, complementary interventions may be needed to address discrimination and marginalisation of disability, as well as of overlapping vulnerabilities (e.g. gender, caste, religion).
Still, despite challenges, it is important to note that several changes to policy and practice appear to have improved access to social protection among people with disabilities in Nepal; For example, funding for all social protection has increased, and quotas on the number of Disability Allowance recipients have been removed (Khadka 2017). Additionally, the involvement of DPOs has increased awareness of disability-targeted programmes, including how to apply. DPOs have been reported to improve understanding of disability among assessors, which may help lead to more equitable classifications. Similarly, outreach camps target financial and geographic barriers to access, as well as improve the ease of the application process. Plans to increase the number of annual deadlines for all social assistance programmes are likely to reduce wait times to receiving payments.
Additional research is needed to explore access to social protection among people with disabilities in other areas in Nepal, as well as other countries globally. In particular, assessing the effectiveness of different tools and procedures for functioning-based disability assessment would be useful for social protection planning and implementation. Additionally, more information is needed on the impact of social protection schemes, particularly against intended aims of poverty reduction, strengthening livelihoods, decreasing inequalities and improving social inclusion (International Labour Organization 2012a, b). Research in other areas of Nepal indicates a high need for social protection among people with disabilities, as they are more likely to be living in poverty and face inequalities in areas such as access to education and work (Eide et al. 2016;Plan International 2014), which follows international trends (Banks et al. 2017;World Health Organization & World Bank 2011). Research on the impact of social assistance for people living in poverty more broadly indicates a range of potential positive outcomes, including greater investment in human capital and productive assets, protecting minimum standards of living and shifts in the balance of power in household decision-making regarding resource allocation (Barrientos 2012;Attanasio and Lechene 2002;Case and Menendez 2007;Gertler et al. 2012). Whether people with disabilities receiving social assistance and other social protection entitlements share similar benefits is less clear.

Strengths and Limitations
In interpreting the results of this research, several limitations should be taken into account. Notably, Tanahun was selected to highlight best practices in Nepal's social protection system, and thus may not be reflective of the situation throughout the country. Additionally, the Washington Group questions used to measure disability in the quantitative research may underestimate the prevalence of disability, as they may not capture all forms of functional limitations; For example, certain mental health conditions may not be captured (e.g. bipolar disorder, schizophrenia). However, the experience of people with these types of disabilities was explored through qualitative research.
A major strength of this study is the use of mixed methods. Combining qualitative and quantitative research presents an opportunity to measure coverage and uptake, while also exploring the underlying factors affecting those figures. Further, use of mixed methods allowed for triangulation of findings across different respondents and methodologies, which strengthens the validity of key results. Another key strength of this study is that participants were recruited from the general population, which improves the generalisability of results.

Conclusions
Social assistance remains the dominant form of social protection accessed by people with and without disabilities alike in Tanahun District, with the Old Age Allowance demonstrating universally high coverage. Overall, 37% of people with disabilities were accessing social assistance, which was significantly higher compared with people without disabilities. Few people with or without disabilities were accessing social insurance tied to employment in the formal sector.
Many people with disabilities remain excluded from programmes for which they are eligible. Points of exclusion occurred at different stages of design and delivery of social protection, with some strategies proving effective at promoting greater access.

Aligning Programme Benefits to the Needs of People with Disabilities
In their conception, Nepal's disability-targeted programmes take an integrated approach to poverty reduction as they combine a cash transfer (for certain recipients) with other benefits that could help to strengthen livelihoods and well-being (Barrientos 2018). However, the quality and availability of linked services, as well as compliance in honouring them among service providers, led to perceptions that they were not worth applying for or using. Further, although the Disability Allowance and other cash transfers are set to double in value in the 2016/2017 fiscal year, it is unlikely that they will cover the range of expenses that people with disabilities and their households often incur, such as disabilityrelated costs and loss of income from household members involved in caregiving or from exclusion of people with disabilities from work Schneider et al. 2011b). Improving the quality and availability of linked services, as well as the value of the cash transfer, will likely improve the impact of these programmes, as well as encourage greater enrolment.
Other programmes take less consideration of the needs of people with disabilities in their design. All social assistance programmes offer a set rate for all recipients, and an individual can only participate in one programme. Removing this restriction to a single scheme or adapting benefit levels could better address poverty stemming from multiple vulnerabilities (Drucza 2016;Kidd 2017). Similarly, social insurance programmes should be broadened to include the informal sector, in which the majority of Nepali citizens with and without disabilities work.

Awareness of Programmes
Overall, awareness of the Disability Allowance and other cash transfers was high amongst people with disabilities and their caregivers in Tanahun. However, awareness of some of the non-cash benefits was low, as was a clear understanding of application procedures and eligibility requirements. Clear communication strategies on programme availability and eligibility has been highlighted as an important strategy for minimising exclusion, which will require adaptations to reach people with certain impairments or who are illiterate (Kidd 2017;Gooding and Marriot 2009). The active involvement of DPOs in Tanahun and other areas of Nepal has been a good strategy for spreading knowledge of disability-targeted programmes, including application procedures. However, adaptations to communication strategies for non-targeted programmes may be needed to ensure people with disabilities receive adequate information.

Determining Eligibility
Disability-targeted programmes have the most complex assessments of eligibility compared with other social assistance programmes in Nepal. While disability eligibility criteria are in-line with the UNCRPD, administrative capacity is lacking to carry out assessments effectively, which has been reported as a common challenge in other contexts (Kidd 2017;Goldblatt 2009;Devandas Aguilar 2017;Mont et al. 2016). The involvement of DPOs in assessment committees has been useful in improving understanding of disability, although more rigorous training of assessors is still needed, particularly in areas without a strong DPO presence. Finally, the removal of quotas on Disability Allowances for blue cardholderswhich led to a doubling in the number of Disability Allowance recipients-illustrates the high level of exclusion resulting from arbitrarily rationing access.

Application Procedures
Disability-targeted programmes had the most cumbersome application procedures compared with other forms of social assistance. A key challenge to enroling in disability-targeted programmes was the requirement to travel to application offices in district capitals, as well as to gather medical documentation. Further, the infrequency and irregularity of assessment board meetings led to further delays, additional costs and frustration amongst applicants. Mobile outreach camps are an important innovation in reducing these challenges, which have also been used successfully in other contexts (SASSA and UNICEF 2013). Additionally, across social assistance programmes, the once annual deadline to register leads to long delays-and potential worsening of poverty and exclusion in the interim (Kidd 2017). The proposal to increase the number of annual registration deadlines across programmes is a positive change, which will likely improve access as well as impact.

Compliance with Ethical Standards
Conflict of Interest On behalf of all authors, the corresponding author states that there are no conflicts of interest.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.