Evidence and gap map of studies assessing the effectiveness of interventions for people with disabilities in low‐and middle‐income countries

Correspondence Ashrita Saran, Evidence Synthesis Specialist, Campbell Collaboration, Global Development Network, ISID Complex, Vasant Kunj Institutional Area, New Delhi 110070, India. Email: asaran@campbellcollaboration.org Abstract Background: There are approximately 1 billion people in the world with some form of disability. This corresponds to approximately 15% of the world’s population (World Report on Disability, 2011). The majority of people with disabilities (80%) live in low‐ and middle‐income countries (LMICs), where disability has been shown to disproportionately affect the most disadvantaged sector of the population. Decision makers need to know what works, and what does not, to best invest limited resources aimed at improving the well‐being of people with disabilities in LMICs. Systematic reviews and impact evaluations help answer this question. Improving the availability of existing evidence will help stakeholders to draw on current knowledge and to understand where new research investments can guide decision‐making on appropriate use of resources. Evidence and gap maps (EGMs) contribute by showing what evidence there is, and supporting the prioritization of global evidence synthesis needs and primary data collection. Objectives: The aim of this EGM is to identify, map and describe existing evidence of effectiveness studies and highlight gaps in evidence base for people with disabilities in LMICs. The map helps identify priority evidence gaps for systematic reviews and impact evaluations. Methods: The EGM included impact evaluation and systematic reviews assessing the effect of interventions for people with disabilities and their families/carers. These interventions were categorized across the five components of community‐based rehabilitation matrix; health, education, livelihood, social and empowerment. Included studies looked at outcomes such as, health, education, livelihoods, social inclusion and empowerment, and were published for LMICs from 2000 onwards until January 2018. The searches were conducted between February and March 2018. The EGM is presented as a matrix in which the rows are intervention categories (e.g., health) and

subcategories (e.g., rehabilitation) and the column outcome domains (e.g., health) and subdomains (e.g., immunization). Each cell lists the studies for that intervention for those outcomes, with links to the available studies. Included studies were therefore mapped according to intervention and outcomes assessed and additional filters as region, population and study design were also coded. Critical appraisal of included systematic review was done using A Measurement Tool to Assess Systematic Reviews' rating scale. We also quality-rated the impact evaluation using a quality assessment tool based on various approaches to risk of bias assessment. concern health interventions. Education is next most heavily populated with 40 studies in the education intervention/outcome sector. There are relatively few studies for livelihoods and social, and virtually none for empowerment. The most frequent outcome measures are health-related, including mental health and cognitive development (n = 93), rehabilitation (n = 32), mortality and morbidity (n = 23) and health check-up (n = 15). Very few studies measured access to assistive devices, nutrition and immunization. Over half (n = 49) the impact evaluation come from upper-middle income countries. There are also geographic gaps, most notably for low income countries (n = 9) and lower-middle income countries (n = 34). There is a fair amount of evidence from South Asia (n = 73) and

Results
Sub-Saharan Africa (n = 51). There is a significant gap with respect to study quality, especially with respect to impact evaluation. There appears to be a gap between the framing of the research, which is mostly within the medical model and not using the social model of disability.
Conclusion: Investing in interventions to improve well-being of people with disabilities will be critical to achieving the 2030 agenda for sustainable development goals. The EGM summarized here provides a starting point for researchers, decision makers and programme managers to access the available research evidence on the effectiveness of interventions for people with disabilities in LMICs in order to guide policy and programme activity, and encourage a more strategic, policy-oriented approach to setting the future research agenda. 2.6 | Screening, data extraction and quality appraisal Title and abstract screening and the evidence classification were undertaken by two independent reviewers, and any discrepancies were resolved by a third reviewer. The studies that passed on to full text were screened against the eligibility criteria by two independent reviewers, and conflicts resolved by third reviewer. After screening, all studies were coded for a wide array of information and populated into the map.

| Selection criteria
The studies were coded by the intervention category and subcategory. The intervention categories are those from the WHO community-based rehabilitation (CBR) guidelines (WHO, 2010): health, education, livelihood, social and empowerment. Advocacy and governance was added as a sixth category, given its importance to the DFID approach.
The coded information includes: bibliographic details for the study, the interventions from the framework that the study evaluates, the outcomes from the framework that the study measures and other relevant aspects such as population, region and countries. This coding was done by two independent reviewers and conflicts reconciled by a third reviewer. The quality of the included systematic reviews was assessed using A Measurement Tool to Assess Systematic Reviews (AMSTAR 2) and done independently by two reviewers. We also quality rated the impact evaluation The studies are unevenly distributed across intervention areas.
Health is the most heavily populated area of the map. A total of 118 studies of the 166 studies concern health interventions. Education is next most heavily populated with 40 studies in the education intervention/outcome sector. There are relatively few studies for livelihoods and social, and virtually none for empowerment.
The most frequent outcome measures are health-related, including mental health and cognitive development (n = 93), rehabilitation (n = 32), mortality and morbidity (n = 23) and health check-up (n = 15).
Very few studies measured access to assistive devices, nutrition or immunization.
Over half (n = 49) the impact evaluation come from upper-middle income countries. There are important geographic gaps, most notably for low income countries (n = 9) and lower-middle income countries (n = 34). There is a fair amount of evidence from South Asia (n = 73) and Sub-Saharan Africa (n = 51). There is a significant gap with respect to study quality, especially with respect to impact evaluation.
There appears to be a gap between the framing of the research, which appears to be mostly within the medical model (i.e., change at individual level) and not on the social model of disability (i.e., change at service or system level).
The majority of studies focus on people with physical impairments. There is a significant lack in studies focusing on people with visual or hearing impairment.
There is an important gap with respect to study quality, especially with respect to impact evaluation. Many of the included systematic reviews were assessed to have methodological limitations.
The findings from this EGM highlights a number of gaps, as mentioned above. Due to the strong concerns on the quality of reviews and impact evaluation, the evidence base needs to be strengthened on what works to improve the well-being of people with disabilities and their families in LMICs. We identify the following implications for research: 1. More studies are needed to fill an important gap in measuring intervention for people with disability and incorporating considerations for equity, with increased focus on low income settings.
2. Ensuring that the funding and research agencies adopt best practice approach for conducting and reporting research to raise the quality of available data.

| The problem, condition or issue
Disability is an umbrella term, covering impairments, activity limitations and participation restrictions. The Preamble to the UNCRPD acknowledges that disability is "an evolving concept", but also stresses that "disability results from the interaction between persons with impairments and attitudinal and environmental barriers that hinder their full and effective participation in society on an equal basis with others" (United Nations General Assembly, 2006 This approach draws on the earlier, medical model focussed more on the importance of impairments, as well as the social model which concentrates on the role of society in the exclusion of people with impairments. More than 1 billion people in the world have some form of disability. This corresponds to approximately 15% of the world's population (World Health Organisation, 2011). The majority of people with disabilities (80%) live in LMICs and disability disproportionately affects the most disadvantaged . People with disabilities are more likely to experience a range of exclusions, including from employment, education, healthcare access and social participation (World Health Organisation, 2011). As a consequence, people with disabilities are more likely to live in poverty, both because disability causes poverty, but also because people who are poor are more likely to become disabled (World Health Organisation, 2011). In addition to economic impact, employment serves many nonfinancial functions. For example, at the individual level, work provides a sense of purpose and belonging in society, leading to improved self-esteem, greater autonomy and an enriched quality of life (Walsh & Tickle, 2013). More broadly, disability is linked to social exclusion and low levels of autonomy and sense of empowerment.
The link of disability and poverty is also borne at a global level, as evidenced by a large systematic review . In 2004, the World Bank estimated the global GDP loss due to disability to be between $1.71 trillion and $2.23 trillion annually (Metts & Mondiale, 2004 People with disabilities are not a homogenous group, and include people with different ages, genders, impairment types and living in different settings, and this may influence the impact of disability. The systematic review showed that the link between poverty and disability is apparent for both males and females and regardless of poverty measure used . Poverty and disability were linked across impairment types, although a clearer link may have existed for people with mental conditions. Similarly, a study on link between poverty and disability found that people with mental illness face higher levels and intensity of poverty, partly as a result of stigma and prejudice (Trani & Loeb, 2012). There is some evidence that the relationship is strongest in countries with higher income level, that is, in upper-middle versus lower income countries. This means that as countries move out of poverty the people with disabilities are increasingly left behind. The review showed that the association of poverty and disability may be strongest in the working population age group. Similarly, another study that used internationally comparable data of 15 developing countries, found that people with disabilities aged 40 and above and people with multiple disabilities were more likely to be multidimensionally poor' (Mitra, 2013).
A key argument in attaining welfare for people with disabilities is to equalize social and economic opportunities from both humanitarian and economic perspectives. From a humanitarian perspective, interventions are implemented to secure basic human rights for people with disabilities. From an economic perspective, programmes SARAN ET AL.

| 5 of 34
are expected to increase the human capital of people with disabilities, and thus enable them to reduce their dependence on income transfers and other forms of public support. This economic expectation addresses disability as a development issue. Research is now required to determine the most cost-effective ways to overcome the above obstacles and develop policies and strategies that increase the economic contributions of people with disabilities (Metts & Mondiale, 2004).
Disability is also a human rights issue, as well as a development issue, and this is highlighted in a range of international documents, including the World Programme of Action Concerning Disabled People (WPA, 1982), the Convention on the Rights of the Child (CRC, 1989), the Standard Rules on the Equalisation of Opportunities for People with Disabilities (United Nation, 1994) and most importantly The United Nations Convention on the Rights of Persons with Disabilities (UNCRPD, 2006). The UNCRPD aims to "promote, protect and ensure the full and equal enjoyment of all human rights and fundamental freedoms by all persons with disabilities, and to promote respect for their inherent dignity". It reflects the major shift in global understanding and responses towards disability, and emphasises that people with disabilities have the right for full inclusion.
Inclusive development includes and involves everyone, especially those who are marginalized and often discriminated against (UNDP, 2010). The justification for disability inclusive development is that unless people with disabilities are brought into mainstream it is impossible to break the cycle of poverty and discrimination.

| The intervention
The "Twin-Track approach" promotes integration of disabilitysensitive measures into the design, implementation, monitoring and evaluation of all development policies and programmes, called as "mainstreaming disability", while simultaneously undertaking "targeted measures" such as disability-specific policies, programmes and initiatives to ensure the inclusion and full enjoyment of human rights by persons with disabilities (UNDP, 2010). A twin-track approach may be required to enable people with disabilities to contribute to creating opportunities, share in the benefits of development, and participate in decision-making (DFID, 2000). The twin-track approach aims to break this cycle between disability, poverty and exclusion, by both empowerment of individuals/families/organisations and by breaking down barriers in society, and is advocated for by many international donors (e.g., the World Bank, DFID, German Cooperation; the European Community [EC], the Finnish Cooperation) and non-governmental organisation (NGOs).
The World Health Organisation's (WHO) CBR guidelines is based on this approach. CBR is a multisectoral, bottom-up strategy which operates at the community level. While the UNCRPD provides the philosophy and policy of disability-inclusive development, CBR is a practical strategy for its implementation (Mendis, Gunnel, Ann, & Einar, 1989). CBR activities are designed to meet the basic needs of people with disabilities, reduce poverty, and enable access to health, education, livelihood and social opportunities-all these activities fulfil the aims of the UNCRPD. Therefore, the CBR will serve as a guiding framework for the EGM and the five pillars of CBR: health, education, livelihood, social and empowerment will form the intervention and outcome categories.

| Why it is important to develop the EGM
Over the past decade, the academic literature on disability outcomes and effectiveness has grown substantially (Andresen, Lollar, & Meyers, 2000;Ramey et al., 2016). However, several important questions have not been adequately addressed.
For example, what type of evidence is needed, and what are realistic expectations, for disability inclusive interventions? A lack of rigorous and comparable data on disability and evidence on programmes that work can impede understanding and action.
Understanding the numbers of people with disabilities and their circumstances can improve efforts to remove disabling barriers and provide services to allow people with disabilities to participate on an equal basis with others. Many efforts are currently underway to fill these knowledge gaps and generate internationally comparable data on the living situation and needs of people with disabilities. This is an important step towards persuading policy and programme decision makers that disability is an issue that needs urgent attention. However, it does not help them in determining which actions are required.
Knowledge production to influence policy and programme action takes place across several sectors (health, social welfare and education), focuses on various populations (different ages, ethnicities or with different needs), and involves rather diverse methodical approaches (e.g., systematic reviews, impact evaluation of different designs etc.). A mapping of the existing knowledge base is, therefore, required to provide a comprehensive overview of existing knowledge in this area, to improve the discoverability, and thereby the use, of that evidence. Furthermore, an EGM can show implementing agencies where there is no relevant information for their programmes and enable the purposeful and targeted commissioning of future research, tailored to the most eminent needs for knowledge and guidance. The EGM can also help to identify gaps to be filled by evidence synthesis where sufficient information is available within one subject area. This overview of the existing evidence is provided by the EGM presented in this study. ii. Map available systematic reviews and impact evaluation on the effectiveness of disability interventions in LMICs in this framework, with an overview provided in a summary report.
iii. Provide database entries of included studies which summarize the intervention, context, study design and main findings. The map has additional dimensions which capture study or intervention characteristics, such as study design, region, countries and population subgroup (which includes type of disability).
F I G U R E 1 Snapshot of disability evidence and gap map SARAN ET AL.

| 7 of 34
The online version of the map (Figure 1) is interactive so that users may click on entries to see a list of studies for any cell in the map. The map is constructed using software prepared by the EPPI Centre. The cells of the table contain a bubble whose size is proportional to the number of studies reporting that outcome for that intervention. There are separate bubbles for impact evaluation and systematic reviews, with the reviews further divided by study quality. The map includes a set of filters allowing evidence to be shown just for certain sub-populations, such as specific regions or countries.

| Type of population
The target populations for this EGM are people with disabilities living in LMICs, based on the World Bank Classification (World Bank, 2017). We also included studies targeting parents/caregivers of people with disabilities. Other populations (e.g., teachers) may be targeted as a means for improving circumstances for people living with disabilities. For this map, we do not focus on the prevention of impairments.
People with disabilities include those 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 (UNCRPD, 2006).
For this map we will include following type of disabilities: 1. Physical: A physical impairment is the long-term loss or impairment of part of a person's body function, resulting in a limitation of physical functioning, mobility, dexterity or stamina. It will include conditions as cerebral palsy, Spina Bifida, poliomyelitis, spinal cord injuries.
2. Visual: Visual impairment, also known as vision impairment or vision loss, is a decreased ability to see to a degree that causes problems in daily life. Conditions may include complete or partial loss of vision, due to conditions such as macular degeneration, retinal detachment and so on.
3. Hearing: Hearing impairment refers to partial or total inability to hear. 4. Intellectual: Also known as learning disability. This condition is characterized by significantly impaired intellectual and adaptive functioning which arises before the age of 18. This involves a permanent limitation in a person's ability to learn. 5. Mental: This category includes conditions such as Schizophrenia, Alzheimer's, bipolar disorders, psychosis.
If a paper includes mixture of disability types, that paper was coded for all types of disabilities as included. Similarly if the study included individual participant with multiple disabilities, again the study was coded for all those disabilities.
In recent years, the inclusion of traditionally underrepresented groups in research has received increasing attention, including racial and ethnic minorities, women, elderly individuals and children (Glickman et al., 2008). These groupings are relevant with respect to disability, as these characteristics may heighten vulnerability in the face of disability, and may also relate to a higher prevalence of disability.
Hence, the population sub-groups of interest for this EGM include: women, vulnerable children (particularly children in care), conflict (conflict and postconflict settings), migrants and ethnic minority groups.

| Types of interventions
The SDG guidelines highlight that implementing the SDGs should build upon existing international and national commitments and mechanisms, in order to generate an inclusive and global dialogue.
The WHO's CBR recognizes CBR as a comprehensive and multisectoral strategy to equalize opportunities and include people with disabilities in all aspects of community life. Therefore, the CBR will serve as a guiding framework for the intervention and outcome categories as listed below, in order to realize the full inclusion and empowerment of persons with disabilities. We have added "Advocacy and Governance" as one of the components, as strong advocacy may be required to prevent and/or address abuse, neglect and exploitation that people with disabilities may experience (CBM, 2012).
The six main intervention categories are: 1. Health.

| Types of outcome measures
The five main outcome categories are as mentioned below and they are plotted against the WHO's CBR indicators: 1. Health.

| Types of settings
The EGM includes studies from LMICs. Studies with multiple populations are included in the map as long as they have a LMIC focus. Reviews with a global focus are included as eligible if they do not exclude countries from LMICs.
The World Bank region classification will be used as filters. There is also a filter for studies in conflict and postconflict settings.

| Search methods and sources
The EGM is based on comprehensive search for impact evaluation and systematic reviews based on the framework of interventions and outcomes as outlined above.
The Campbell Collaboration policy brief for searching studies and information retrieval, informed the search strategy as presented below (Hammerstrøm, Wade, Jørgensen, & Hammerstrøm, 2010). In addition, information retrieval specialist John Eyers was consulted during the preparation of search strings, while several search retrieval specialists provided recommendations during the peerreviewing process of the study protocol. The lead author conducted the searches once the protocol had been peer-reviewed and approved by Campbell Collaboration. The searches were conducted during the period February 19, 2018 to March 9, 2018.
At the end of the screening process, key journals were searched using key-terms up to the end of January 2018. Studies in any language and from any country were included, provided the abstract was in English.
Searches were completed, as per protocol with a number of minor additions. In some cases the search string could be copied and pasted directly from the protocol, whilst other databases required the search string to be manually populated as recommended by Higgins and Green (2011), the search strategy is reported in Appendix B. Details of additional grey literature databases are included as recommended by Campbell Collaboration information retrieval specialists.

| Electronic searches
The search was as comprehensive as possible, using (but not limited to) relevant systematic review database for first stage along with bibliographic databases (Appendix D), EGM databases, web-based search engines, websites of specialist organisations, bibliographies of relevant reviews, and targeted calls for evidence using professional networks or public calls for submission of articles. Database for EGMs was also searched to identify any map and relevant populated studies.
In addition, reference lists of the included reviews were reviewed and the authors contacted for information on other relevant sources.
Citation searches were also performed (see Appendix B).

| Searching other sources
We searched the following databases to identify unpublished reviews studies: Dissertation Abstracts, Conference Proceedings and Open Grey. We also searched a number of agency websites.
To identify ongoing studies, we searched ClinicalTrials.gov and WHO International Clinical Trials Registry Platform and CENTRAL Trials Register within the Cochrane Library for published trials. The WHO's CBR programme recognizes CBR as a comprehensive and multisectoral strategy to equalize opportunities and include people with disabilities in all aspects of community life. CBR activities are designed to meet the basic needs of people with disabilities, reduce poverty, and enable access to health, education, livelihood and social opportunities-all these activities fulfil the aims of the UNCRPD. Therefore, the CBR will serve as a guiding framework and the five pillars of CBR: health, education, livelihood, social and empowerment will form the intervention and outcome categories.

| Stakeholder engagement
Clinical/pharmacological interventions to prevent or treat the primary impairment/health condition are beyond the scope of the map and hence such studies are excluded. We will be including studies that focus specifically on people with disabilities, as well as studies referring to interventions for families of people with disabilities.

| Conceptual framework
The matrix (Figure 1) illustrates the different sectors, which can make up a CBR strategy for the welfare of people with disabilities. It consists of five key components, each divided into five key elements.
The elements are subdivided into content headings. The matrix should not be seen as sequential, and all components will not be needed by every person with disabilities Figure 2.

| Description of intervention/problem categories
The included interventions cover all main strategies to reduce disability related outcome as described in the CBR. The six main intervention categories are:

| Description of population/geographic location
The EGM has two primary dimensions: interventions (rows) and outcomes (columns). The screening of studies in relation to inclusion/exclusion was undertaken in two stages. The first stage involved title and abstract; the second involved full text documents.
Three independent researchers were involved at each stage. The screening was carried out based on predefined eligibility criteria (Appendix A) by two independent reviewers and the third screener resolved the conflicts. Prior to data extraction and coding, the three independent reviewers met to discuss and pilot the extraction and coding procedures on a sample of abstracts.
Stage 1: Screen on Title and abstract The screening was carried out based on predefined eligibility criteria (Appendix A) by two independent reviewers and the third screener resolved the conflicts. The conflict was resolved by third reviewer through group discussion with team. Title and abstracts which passed the first stage were retrieved in full text for a more comprehensive review.
Stage 2: Screen on full text Full text documents were retrieved for all documents that passed stage one. Two reviewers independently evaluated all studies.
Studies had to meet all of the inclusion/exclusion criteria set out previously in order to advance to full review.

| Data extraction and management
Each included study was coded independently by two coders using the coding tool covering study characteristics, population, intervention, outcomes, region, countries and type of disability. The coding tool is added in the Appendix C.

| Tools for assessing risk of bias/study quality of included reviews
Each study in the map has a rating for the quality of evidence.
Items 2, 4, 7, 9, 11, 13 and 15 are termed "critical". Study quality is rated high if there is no more than one noncritical weakness, and medium if there is no critical weakness but more than one non critical weakness. Studies with one or more critical weaknesses are rated low quality. SARAN ET AL.

| 11 of 34
Impact evaluation: The quality assessment for the impact evaluations is based on existing approaches to risk of bias assessment. Many of the items in this assessment, such as study design and baseline balance, relate to possible sources of bias. Other items relate to clarity of reporting, especially of the intervention and outcomes. The assessment used the following criteria (see Tables 1-3): 1. Study design (potential confounders taken into account): Impact evaluations need either a well-designed control group, preferably based on random assignment, or an estimation technique which controls for confounding and the associated possibility of selection bias.
2. Adequate sample size: Small samples generally mean that a study in underpowered, that is, there is a high risk of not finding an effect even if the intervention works.
3. Attrition (or loss to follow-up) can be a major source of bias in studies, especially if these is differential attrition between the treatment and comparison group so that the two may no Overall study quality is the lowest rating awarded any one of the above six criteria. As a result of this process, a total of 166 studies were included for coding. Of these, 59 are systematic reviews (see references for a list of included studies), and 107 impact evaluations. We then screened the included studies in the 59 systematic reviews to assess their eligibility for inclusion in the map.

| Synthesis of included studies
Studies in the map, especially systematic reviews, may be coded under more than one intervention category or subcategory. This means that there are many more entries in the map then there are studies. The number of studies contained in the map is stated clearly at the top of the map.   Policies and resources are conducive to quality education for people with disabilities and ensure smooth transitions through different stages of learning People with disabilities participate in and complete quality primary education in an enabling and supportive environment People with disabilities have resources and support to enrol and complete quality secondary and higher education in an enabling and supportive environment People with disabilities experience post school options on an equal basis with their peers Attendance People with disabilities have resources and support to enrol and complete quality secondary and higher education in an enabling and supportive environment Education in mainstream education facilities/ inclusive education Percentage of people with disabilities who acquire education in mainstream education facilities Social and life skill development People with disabilities make use of youth or adult centered learning opportunities to improve their life skills and living conditions Learning and achievement People with disabilities experience equal opportunities to participate in learning opportunities that meet their needs and respect their rights Access to educational services People with disabilities participate in a variety of nonformal learning opportunities based on their needs and desires People with disabilities actively participate in early childhood developmental activities and play, either in a formal or informal environment Livelihood Employment in formal and informal sector People with disabilities have paid and decent work in the formal and informal sector on equal bases with others People with disabilities earn income through their own chosen economic activities People with disabilities acquire marketable skills on an equal basis with others through a range of inclusive training opportunities Access to job market People with disabilities have access to job markets on equal basis as others Control over own money People with disabilities have control over the money they earn Access to financial services such as grants and loans

| Risk of bias in included reviews
People with disabilities have access to grants, loans and other financial services on an equal basis with others People with disabilities participate in local saving and credit schemes Poverty and out-of-pocket payment Percentage of people with disabilities who are covered by social protection programs Access to social protection programs People with disabilities access formal and informal social protection measures they need Participation in development of inclusive policies Inclusive policies, practices and appropriate resources, defined with people with disabilities enable equal participation of women and men with disability in livelihood (training, finance, work opportunities and social protection)

Social Stigma and discrimination
Communities have increased awareness about disability, with a reduction in stigma and discrimination towards people with disabilities Safety People with disabilities feel safe in their family and community Participation in mainstream recreational, leisure and sports activity People with disabilities participate in inclusive or specific recreation, leisure and sports activities (Continues) High confidence: outcome measure clearly and fully described, preferably with reference to validation Medium confidence: brief description of outcome Low confidence: outcome named but not described 6 Baseline balance (not applicable for before versus after) High confidence: RCT or baseline balance report and satisfactory (imbalance on 5 or less than 5 percent) Medium confidence: Imbalance between 5-10 percent Low confidence: Baseline balance not reported, or reported and lack of balance on 10 or more than 10 percent Overall confidence in study findings End of intervention Lowest rating across items 1a, 4a, 6 and 7 a https://homvee.acf.hhs.gov/HomVEE-Attrition-White_Paper-7-2015.pdf.

SARAN ET AL.
| 15 of 34 Figure 6 shows the quality trends of both impact evaluations and systematic reviews over the years. The proportional number of low quality impact evaluations has increased over the years as compared to medium/high quality impact evaluations. A high proportion of systematic reviews identified had methodological limitations and were of low quality.

| By intervention categories
Systematic reviews are concentrated in the health sector: 45 (80%) of reviews report effects of health interventions (Figure 7). Randomised controlled trials (RCTs) account for close to half of the impact evaluations (44 RCTs out of 107) being particularly prominent in health and education (30 and 13 studies, respectively), where some studies cover both sectors.
As mentioned above a single study may appear in more than one category. For example,  review of rehabilitation programmes states that the interventions covered include "home visits by trained community workers who taught disabled persons skills to carry out activities of daily living, encouraged disabled children to go to school, helped find employment or an income generating activity, often involving vocational training and/or microcredit. Many programmes had a component of influencing community attitudes towards disabled persons". This study is coded under each of health, education, livelihoods, empowerment and advocacy and governance intervention type.

| By type of impairment
Nearly two-thirds of the studies (60%) of the studies relate to interventions for people with mental or intellectual impairments, 27% to physical impairment, with a small number identified as relating to hearing and visual and hearing impairments (see Figure 8; recall that some studies are coded under more than one category).

| By outcome domain
Since the most common intervention category is health, it is unsurprising that the health-related outcomes are reported in the largest number of studies (114 studies); see Figure 9. This is followed by education (46), social (46) and livelihoods (24 studies). Only 3 included studies report empowerment-related outcomes. Systematic reviews are concentrated in the health sector: 46 (78%) of reviews report effects of health interventions (Figure 8). RCTs being particularly prominent in health sector (37) and considerably less than 10 RCTs in other sectors.
Within health, mental health and cognitive development account for the largest number of studies (93 studies) followed by rehabilitation (32) (Figure 10).

| By region
Impact evaluations are unevenly distributed across World Bank region and countries (Table 4). Over half the impact evaluation come from four LMICs. These are concentrated in four countries: India (23), China (11), Iran and Turkey (9) studies each. South Asia is relatively well covered with studies from India (23), Bangladesh (5) and Pakistan (4) and as is East Asia on account of China (Figures 11-13).
We included in the map all reviews in which studies from LMICs were eligible in searches. However, only 17 of the 59 included reviews actually include eligible studies. Of the other 44 reviews, 20 studies had only included studies from high-income countries, 16 had LMIC studies which were not eligible for reasons of date or study design, 5 had no included studies and 2 are ongoing with results not yet reported.
Thirty-eight impact evaluations concerned fragile and conflicted affected states (Table 5).

F I G U R E 3 PRISMA for disability evidence and gap map
F I G U R E 4 Number of impact evaluation by study quality 8.9 | By quality assessment The systematic reviews were assessed using the AMSTAR tool described elsewhere in the document. Of the 59 reviews, 22 were assessed as low quality, 16 medium and 18 high quality, with the remaining three studies ongoing and, therefore, not yet scorable.  (Table 6). The total in Table 6 exceeds 166 (the total number of studies) as many studies are coded under more than one intervention and outcome.  EGMs are important building blocks in the evidence architecture and help in the following three ways:

| By population
1. Guide users to high quality reviews.
2. Guide users to where there are no high quality reviews.
3. Guide users to evidence gaps to be filled by new reviews and impact evaluations.
9.1 | Areas of high quality reviews and impact evaluations 9.1.1 | Systematic reviews Out of the 18 high quality review identified for this map, 13 (73%) are in health sector. A large proportion of these focuses on rehabilitation and health promotion.
Some high quality reviews were also identified in education (5) and social (5) pillars and may have some policy implications. Within education, high quality reviews were identified on early childhood Education and Non-formal education. While only one high quality review was identified in the primary and secondary education.
A significant number of reviews were found to have methodological limitation, particularly in empowerment and livelihood sectors. There are also very few studies from low-income countries, reflecting the relative neglect of many parts of Sub-Saharan Africa such as Rwanda, South Sudan, Somalia, Congo, Burundi and so on.
Studies were scarce for people with disabilities in many of these conflict affected regions.
There were also gaps by impairment type, with limited studies identified on people with visual and hearing impairment.
There is mostly low or medium confidence in study findings, and so another gap is the absence of high quality studies in the field.
Reviews are of higher quality overall, though less than one-third qualified as high quality and the studies they draw on tend to be of low quality.
We can draw on two Rapid Evidence Assessments undertaken from the EGM, and thereby go beyond the bounds of what reading the map can tell us . It is apparent from the reviews that the focus of studies is on fixing individuals, that is, the medical approach, for instance, focussing on improving social or learning skills for people with disabilities. Fewer studies focus on improving infrastructure or institutions, and therefore address social barriers to inclusion. Development agencies, including DFID, are stressing the biopsychosocial approach in their work, so the absence of evidence on what works to promote disability inclusion is a very striking gap. Future systematic reviews from the EGM will be able to provide more guidance for action for policy and programme decision-makers.

| Limitations of the EGM
• The EGM provide a rich source of information on existing systematic reviews and impact evaluations relating to interventions to improve the lives of people with disabilities and their families in LMICs.
• The EGM followed comprehensive search using predefined eligibility criteria, yet inevitably there are limitations to our approach.
• Eligible studies were restricted to those published after 2000 up until the start of 2018, and published in English. Also searching the "grey" literature is challenging, and consequently some eligible studies may have been missed.
• Sometimes it was difficult for the reviewer team to categorize interventions, mainly between empowerment and livelihood, as there can be overlaps. The categorization for such interventions was done based on expert consultations and the information as available to mitigate this issue as far as possible.

| AUTHORS' CONCLUSIONS
The mapping exercise is a first step to identifying priority areas for systematic reviews and impact evaluations. We identify initial steps that can help advance research to promote the welfare and inclusion of people with disabilities. We strongly believe that the online interactive visualization, list of references, and summary of studies will facilitate access and use of research.
10.1 | Implications for research, practice and/or policy • The available high and medium quality systematic reviews in health sector may suggest some implications for policy.
However, few of the studies are recent, and so they may need to be updated.
• Efforts are also needed to reach a consensus to identify priority areas for research with weak evidence synthesis by key funders and researchers in the field.
• More studies should be carried out given the relative lack of impact evaluation in many areas such as empowerment and livelihood. Impact evaluations will be more useful if they focus on more diverse set of outcomes and thereby fill multiple evidence gaps.
• More studies are needed to fill important gaps in equity and measuring interventions for vulnerable populations. This includes areas of gender, ethnic minorities and low-income and conflict affected settings.
• The geographical base of evidence needs to be expanded as well. Most of the studies come from upper-middle income countries and there is limited evidence from low income countries. Evidence need to be expanded in these countries.
• More studies are needed to generate evidence on all types of impairment, including visual and hearing impairments.
• Future research should also follow the best practice and improve reporting of intervention implementation in order to improve the quality of studies.
• Consideration needs to be given to improve quality of systematic reviews in terms of reporting and inclusion criteria or scope by adherence to standard guidelines as PRISMA. •

| Systematic review method expertise
All authors are experienced systematic reviewers, which means they are proficient in carrying out the various processes in an EGM, such as eligibility screening, quality assessment and coding.

EGM methods expertise:
All team members have previous experience in systematic review methodology, including search, data collection, statistical analysis, theory-based synthesis, which mean they are proficient in carrying out the various processes in an EGM, such as search, eligibility screening, quality assessment and coding.

Information retrieval expertise:
All authors have previous experience in developing search strategies.

Selection criteria Inclusion Exclusion
Publication year After 2000 Before 2000 Publication status Completed and on-going None Study design The EGM will include systematic reviews of effects of interventions and effectiveness studies that used either: (a) randomised experimental design, or (b) rigorous quasiexperimental design, (c) natural experiments, (d) regression discontinuity, (e) propensity score matching, (f) difference in difference, (g) instrumental variables, (h) other matching design and (i) single subject design Literature reviews, noneffectiveness studies, case studies and qualitative studies Population People with disability, and/or their family, their caregivers, their community living in low-and middle-income countries People with disabilities and/or their family, their caregivers, their community living in high-income countries Disability is defined as impairments, activity limitations, and participation restrictions denoting the negative aspects of the interaction between an individual (with a health condition) and that individual's contextual factors (environmental and personal factors) (World Health Organisation, 2011) For Impact evaluation we will include participants from lowand middle-income countries only, as this was the original commitment of CBR (Helander, 1989) Interventions We will include effectiveness studies. All the clinical trials, interventions for reversible form of illness will be excluded. A CBR programme is formed by one or more activities in one or more of the five components (health, education, livelihood, social and empowerment). List of activities for each element of the five components are presented within the CBR guidelines under the section "Suggested activities". The following activities are here given as examples: • Health: training PWD in the use of assistive devices; providing information to PWD and their family or their caregivers about time and location of activities for screening health conditions and impairments associated with disabilities • Education: providing education and training for families or caregivers of PWD; installing ramps in schools to make them accessible to PWD using wheelchairs • Livelihood: linking the jobseeker with disability to existing support services; advocating before relevant public and private agencies to ensure accessible housing for PWD • Social: converting institutions for PWD in rehabilitation centres; providing information to PWD about the sports opportunities available within the community • Empowerment: helping PWD running meetings of new selfhelp group; involving disabled's people organizations in CBR planning, implementation, and monitoring Interventions not focused on people with disabilities. We will also exclude studies that deals temporary or reversible form of disability for examples, maternal depression or back pain Outcome We will use the CBR framework for outcomes. None

Quality
We will not restrict based on quality None SARAN ET AL.