Interventions to Improve the Labour Market Situation of Adults with Physical and/or Sensory Disabilities in Low- and Middle-Income Countries: A Systematic Review

1.1.1 The Problem

Disability is a key development issue. Recent estimates suggest that more than one billion people (or about 15% of the world's population) are living with some form of disability---80% of whom live in low- and middle-income countries (World Health Organization [WHO], 2011). ¹ People with disabilities are over-represented among the world's poor, and many experience multiple deprivations at higher rates and in higher breadth, depth, and severity than people without disabilities (Mitra, Posarac, & Vick, 2013; Samman & Rodriguez-Takeuchi, 2013). The lack of access to paid work and/or wider economic activity is a significant social disadvantage and helps maintain the link between poverty and disability in many country contexts (Braitwaite & Mont, 2009; Haveman & Wolfe, 1990; Hoogeveen, 2005; Peiyun & Livermore, 2008; WHO, 2011; Zaidi & Burchardt, 2005). The Millennium Development Goals (MDGs) on eliminating poverty, launched by the United Nations in 2000, are unlikely to be achieved unless explicit and specific efforts are undertaken to support disabled people's participation in labour market activities (Department for International Development [DFID], 2000; Groce & Trani, 2009).

There is no single, universally accepted definition of disability and defining it remains complex and controversial. In this study, disability is understood following the International Classification of Functioning, Disability and Health (ICF) which was developed by the WHO in 2001 in a long process involving academics, clinicians, and---importantly---people with disabilities (WHO, 2001). Representing a workable compromise between medical and social models, the ICF understands disability as arising from the interaction of health conditions with contextual factors (both environmental and personal). Disability is thus viewed not as a static feature of an individual, but rather as a complex, multi-dimensional, and changing experience for the individual (Schneider & Hartley, 2006). An implication of the ICF model of disability is that by removing barriers, persons with health conditions can be enabled to function and participate.

In many countries data on the employment of people with disabilities are not systematically available. Yet, where data exist, individuals with disabilities are found to be disadvantaged in both accessing and sustaining work and other forms of economic activity. Employment participation rates for people with disabilities are below that of the overall population; and, when disabled people do work, they generally do so for longer hours and lower incomes, have fewer chances of promotion, are more likely to work in the informal labour market, and are at greater risk of becoming unemployed for longer periods (Coleridge, 2005; Contreras, Ruiz-Tagle, Garces, & Azocar, 2006; Houtenville, Stapleton, Weathers, & Burkhauser, 2009; Mete, 2008; Mitra, 2008; Mitra et al., 2013; Mitra & Sambamoorthi, 2006; Mizunoya & Mitra, 2012; Roulstone, 2012; Roulstone, Gradwell, Price, & Child, 2003). Using data from 27 countries, a recent study from the Organization for Economic Co-operation and Development (OECD) found that people with disabilities are half as likely to be in paid work as their non-disabled counterparts (44% compared with 75%) (OECD, 2010).

Across the globe, people with disabilities find it difficult overcoming barriers to the workplace. Although this is not simply an issue in certain countries or even continents, the costs of disability are particularly acute in LMICs, where up to 80% of people with disabilities of working age can be unemployed, around twice that for their counterparts in industrialised countries (Contreras et al., 2006; Groce, Kembhavi, Wirz, Lang, Trani, & Kett, 2011; Houtenville et al., 2009; International Disability Rights Monitor, 2004; Mete, 2008; Mitra, 2009; Mitra et al., 2013; OECD, 2010). Rates of employment vary widely from country to country, from lows of 30% in South Africa to highs of 92% in Malawi (Loeb & Eide, 2004; 2004; Mitra, 2008). In many developing countries, a significant proportion of people work in the informal economy, and so are further disadvantaged. In India, for example, 87% of people with disabilities who work are in the informal sector (Mitra & Sambamoorthi, 2006a). It is less clear, however, whether the wage gap between disabled and non-disabled persons is as marked in developing counties as it is in industrialised countries (Mitra & Sambamoorthi, 2006b; OECD, 2003; WHO, 2011). Recent studies in India, for example, have produced mixed results (Mitra & Sambamoorthi, 2008, 2009).

Not all people with disabilities are equally disadvantaged. In practice, the extent of the negative effect of disability on employment is likely to vary depending on a variety of factors (Goertz, van Lierop, Houkes, & Nijhuis, 2010; Ingstad & Reynolds-Whyte, 1995; Kidd, Sloane, & Ferko, 2000; Mitra et al., 2013; OECD, 2010; Sena-Martins, 2010; World Bank, 2009). These include personal factors such as age, sex, level of education, motivation to work, and lack of financial resources. Women with disabilities, for instance, are recognised to be multiply disadvantaged, experiencing exclusion on account of their gender and their disability. There is evidence that disabled women tend to have less access to jobs, lower employment rates, and considerably lower earnings than male peers in similar jobs (Emmett, 2006; Mitra & Sambamoorthi, 2006; Mitra et al., 2013). Data for 51 countries from the World Health Survey, for example, show that employment rates are 19.6% for women with disability and 52.8% for men with disability (WHO, 2011). The links between disability and employment also vary considerably across impairment categories, severity and duration (WHO, 2011; World Blind Union, 2004). There is some evidence, for example, that individuals with mental health difficulties or intellectual impairments experience the lowest employment rates (Thornicroft, 2006) and those with more significant impairments are least likely to obtain work.

A range of environmental and personal factors can present barriers for persons with health conditions to function and participate in economic life. For example, the physical accessibility of local workplaces and transport facilities, available accommodations, and social attitudes can restrict participation in the labour market (Baldwin & Johnson, 2006; Bound & Burkhauser, 1999; Mitra & Sambamoorthi 2008). Lack of access to education and training or to financial resources can result in exclusion from the labour market, and social protection systems may create disincentives for people with disabilities to enter the labour market. There is also some evidence that people with disabilities seeking to access and sustain employment in competitive, tight labour markets are especially disadvantaged (Mitra, 2009). The policy context is relevant, too. The particular educational facilities, employment supports, health services, disability benefit systems, and other interventions that are available in a given context can influence whether, and to what extent, disability has employment consequences.

1.1.2 Why it is Important to do this Review

The barriers faced by people with disabilities globally in accessing and sustaining paid work is a profound social challenge. Many reasons are provided as to why these barriers exist and new policy visions are frequently offered up. Yet, efforts to promote development and poverty reduction have not always adequately included disability; for example, people with disabilities are not explicitly included in any of the MDG targets and indicators (WHO, 2011). Disability issues are, however, slowly being brought into the mainstream of development policy and practice, and over the past two decades there has been a noticeable change in the legal and policy responses of many governments and bilateral and multilateral donor agencies (DFID, 2000, 2007; Thomas, 2005). In 2002, for example, the World Bank embarked on mainstreaming disability into Bank operations and analysis (Mont, 2007). A major catalyst has been the Convention on the Rights of Persons with Disabilities (CRPD) adopted by the United Nations in 2006, which marked a significant advance in the recognition of the rights of disabled persons, including the right to work, on an equal basis with others (United Nations [UN], 2006). With increasing recognition of employment as a key factor in the process of empowerment and inclusion into society of people with disabilities, a shift to a broader framework for action has occurred, and the role of interventions to improve labour market outcomes is receiving increased international attention (DFID, 2000; International Labour Organization [ILO], 2008; WHO, 2004). Nevertheless, translating policy commitments into better lives for people with disabilities remains a profound social challenge. Establishing a firm evidence base to support the implementation of the CRPD is therefore a priority. Building a clearer understanding of which measures are effective at improving employment outcomes, and under which circumstances, can provide such an evidence base for policy development and contribute to the development of practical suggestions for meeting this challenge.

The existence of a growing body of evidence on interventions to increase the labour market participation of people with disabilities is highlighted in a recent comprehensive review of the literature in this area (Waddell, Burton, & Kendall, 2008). Taking a broad definition of vocational rehabilitation, and focusing on the conditions that account for two-thirds of long-term sickness absence in developed countries---mild/moderate musculoskeletal, mental health, and cardio-respiratory conditions---the study reviews the data from a large number of scientific reports and literature reviews, covering a wide range of intervention strategies. While the authors conducted a systematic search, assessed the strength of the evidence, and included data in evidence tables, they did not report effect sizes or perform a meta-analysis, making it difficult to judge and compare the effectiveness of the interventions. Other systematic reviews are more limited in scope, focusing on (a) specific countries (e.g., Bambra, Whithead, & Hamilton, 2004; Clayton et al., 2011); (b) single aspects of disability/illness, such as autism (e.g., Westbrook et al., 2012), mental illness (e.g., Crowther, Marshall, Bond, & Huxley, 2001; Underwood, Thomas, Williams, & Thieba, 2006), multiple sclerosis (e.g., Khan, Ng, & Turner-Stokes, 2009), traumatic brain injury (e.g., Graham & West, 2012), low back pain (e.g., Tveito, Hysing, & Eriksen, 2004) or spinal cord injury (e.g., Lidal, Huynh, & Biering-Sørensen, 2007); or (c) particular intervention types, such as interventions based on an empowerment perspective (e.g., Varekamp, Verbeek, & Dijk, 2006), workplace disability management programmes (e.g., Gensby et al., 2012) or workplace-based return-to-work interventions (e.g., Franche et al., 2005). Some of these reviews use meta-analytic synthesis methods, but several are quite dated and none explicitly focus on programmes conducted in LMICs.

There are a small number of recently published reviews in this area that focus on developing countries. The literature on assistive technology in LMICs is examined in two non-systematic reviews (Andrysek, 2010; Borg, Lindstrom, & Larsson, 2011). Some evaluative activities were identified, none of which measured employment outcomes. Another recent LMIC-focused non-systematic review (Velema, Ebenso, & Fuzikawa, 2008) examines evidence for the effectiveness of community-based rehabilitation (CBR) programmes for people with disabilities on a range of outcomes, including employment. A descriptive overview of the literature is presented, with no pooling of data. A non-systematic literature review by Mitra and Sambamoorthi (2006a) focused on impact evaluations conducted in India of the

People with Disabilities (PWD) Act and government programmes designed to promote employment among people with disabilities. More recently, a protocol was submitted for a joint Campbell/Cochrane systematic review of CBR for people with physical and mental disabilities in LMICs (Iemmi et al., 2012). Data will be collected on a number of functional outcomes (including employment) for different types of interventions, and variation of effects for different subject populations will be examined.

In sum, whilst existing reviews provide some evidence about the effectiveness of programmes to support the inclusion of persons with disabilities in the labour market, a number of them are now quite dated and specific gaps in the evidence base remain. Taking into account current policymaker priorities, this suggests a need to comprehensively assess the full evidence base relating to low- and middle-income countries, using appropriate methods to evaluate the impact of a range of different intervention types supporting the employment of adults with physical and/or sensory disabilities.

1.2.1 The Intervention

The scope of this review is not limited to one type of intervention. Rather, it extends to any intervention likely to help adults with disabilities in LMICs enter, re-enter, or maintain employment. Such interventions may take the form of a device, policy, programme, strategy, or other type of action. For the purposes of this review, a typology of intervention types was developed prior to undertaking the review, with a view that we would refine it on the basis of the review findings, if appropriate. The broad groupings are presented in Table 1.1. The characteristics of eligible interventions are broad. They (i) encompass complex, specialised, multidimensional programmes that implement multiple strategies as well as much simpler interventions based on a single strategy; (ii) may be implemented in any setting, including the workplace, health care facility, home, or community; (iii) include both routine and structured/tailored interventions; (iv) can vary not only by type but also by intensity; (v) can be delivered at various stages of the employment process (pre-employment, transition to employment, and post-employment); and (vi) need not have the core objective of restoring capacity for work.

1.2.2 How the Intervention Might Work

Conceptual understanding of the causal pathways through which available interventions may influence the employment prospects of people with disabilities in developing country contexts is under-developed. It was necessary, therefore, to develop a logic model specifically for this review. Originating from the field of programme evaluation, logic models (also known as theoretical, conceptual, or impact models) are typically diagrams or flow charts that illustrate pathways between inputs, strategies, outputs, and short-term, intermediate and longer-term outcomes (Anderson et al., 2011; Joly et al., 2007). Designed to read from left to right, they provide a valuable road map that spells out how, and for whom, a programme is meant to produce the desired outcomes. We hypothesised that the types of interventions detailed in Table 1.1 affect a range of different labour market outcomes for people with disabilities through various mechanisms. The model shown in Figure 1.1 illustrates both intermediary factors through which the intervention may exert its impact, and additional personal and contextual factors that may modify or inhibit the desired effect.

OPEN IN VIEWER

1.1:

Logic Model

3.2.1 Types of Participants

Study participant characteristics:

Geographical location: Low- or middle-income country, as classified by the World Bank for the fiscal year 2013-2014, ending on June 30 2014 (see Table 9.2, Appendix I). The World Bank's country classifications are based on estimates of gross national income (GNI) per capita for the previous year. For the fiscal year starting 1 July 2013, these are: low income ($1,035 or less), middle income ($1,036 to $12,615).

Widely used by researchers and policy makers when addressing disability issues in the global development literature, the ICF is adopted as the conceptual framework for this systematic review. Disability is therefore understood as an umbrella term embracing impairments, activity limitations, and participation restrictions (WHO, 2001). The term impairment implies specific problems in body functions and structures, often identified as symptoms or signs of health conditions (i.e., diseases, injuries, and disorders). ² The following additional definitions and restrictions also apply.

Definitions. Physical impairment is defined as problems with the structure, development, or function of the bones, muscles, joints, and/or central nervous system. Physical characteristics may include paralysis; altered muscle tone (ranging from loss of muscle mass to uncontrolled muscle contraction); an unsteady gait; loss of, or inability to use, one or more limbs; difficulty with gross-motor skills (such as walking); and/or difficulty with fine-motor skills (such as writing). Sensory impairment is defined as full or partial loss of one or more senses (e.g., sight, hearing, smell, touch, taste, and/or spatial awareness), causing difficulty with communication, gross-motor skills, fine-motor skills, and/or access to information.

Eligible studies. The focus is on impairments that meet customary and/or statutory definitions of disability. These are usually long-standing, for example, lasting at least one year, and have a substantial impact on a person's ability to do normal daily activities, such as getting dressed. The impairment/health condition may be acquired or congenital. It may be acute, chronic, progressive, or intermittent, and may or may not need ongoing medical intervention.

Studies of participants with the following types of health condition/physical impairments were eligible for the review: communicable diseases (e.g., leprosy); metabolism disorders (e.g., diabetes); respiratory conditions (e.g., asthma); neurological impairments (e.g., multiple sclerosis, epilepsy, those associated with brain injury); musculoskeletal conditions (e.g., arthritis, amputations); cardiovascular diseases; and body disfigurements (e.g., burn injuries).

Where study participants were described as multiply disabled, the study was included if physical and/or sensory impairment was the primary diagnosis. Where study samples were comprised of people with different disabilities, we included the study if: (a) the majority of the sample was physically and/or sensory disabled; or (b) the authors reported disaggregated results according to type of disability.

Studies focused on work-related and non-work related health conditions were both eligible for inclusion in the review. Finally, eligibility for the review was extended to both primary studies that incorporated the ICF diagnostic framework in identifying and selecting its subjects and studies that did not use this framework.

Non-eligible studies. Studies focused solely on (i) people with mental health conditions and/or intellectual impairments, (ii) those with chronic illnesses that predominate in later life (e.g., chronic obstructive pulmonary disease (COPD), cancer, stroke, and renal disease), or (iii) HIV/AIDS were not eligible for this review, on the grounds that these groups have different rehabilitation needs.3 The review focuses on long-term disability, rather than persons with minor health problems, such as fractured bones or allergic rhinitis (hay fever). Therefore, evaluations of return-to-work (RTW) interventions for employees on short-term sick leave were outside the scope of this review.

3.2.2 Types of Interventions

The scope of this review extends to any intervention with the means to help adults with disabilities in LMICs gain or maintain employment. Such interventions may take the form of a device, policy, programme, strategy, or other type of action. Examples of relevant interventions were detailed in Section 1.2.

3.2.3 Types of Outcome Measures

To be eligible for this review, studies must have measured/reported at least one quantitative labour market outcome. Eligible outcomes include primary outcomes and intermediate outcomes.

Primary outcomes: People with disabilities in LMICs are often prevented from work, constrained in the type and amount of work that they do, and/or have difficulty sustaining work. As a consequence, they are predominantly employed in the informal sector, which is characterised by low pay. In consideration of this, the primary labour market outcomes of interest are those relating to the general constructs (a) employment participation (b) work productivity/performance, and (c) income.

For employment participation, examples of relevant indicators are: gaining initial employment; return-to-work (e.g., from non-employment, or from long-term sick leave); gaining formal employment (i.e., a ‘better’ job in that it has written contract, etc.); job retention; promotion (i.e., vertical job mobility); change in job role/function (i.e., horizontal job mobility).

For work productivity/performance, examples of relevant indicators are: number of weekly hours worked; number of days worked per month.

For income, examples of relevant indicators are: overall annual income; monthly earnings; weekly wages; average hourly rate of pay; and profits or income from self-employment.

Unless otherwise stated, employment refers to paid employment and self-employment. The following definitions of paid employment and self-employment apply to this review.

Paid employment: defined as jobs involving some form of contractual relationship between the individual worker and an employer over time for remuneration. Employment contracts may be explicit (written or oral) or implicit. Remuneration is typically in the form of wages and salaries, but people may also be paid by commission from sales, from piece-rates, bonuses, or in-kind payments such as food (ILO, 1993, para. 6). Those workers employed in the informal economy, over which there is little or no official control, are likely to be paid in cash. Within the definition of ‘paid employment’, the review includes both (a) competitive paid employment, broadly defined as jobs that are available on the open market and open to anyone who applies, and that offer payments and benefits that are comparable to industry/sector standards, and (b) jobs in an integrated work setting for individuals with disabilities who are working toward competitive employment with ongoing support services.

Self-employment: defined as jobs where ‘the remuneration is directly dependent upon the profits (or the potential for profits) derived from the goods or services produced … The incumbents make the operational decisions affecting the enterprise, or delegate such decisions while retaining responsibility for the welfare of the enterprise. In this context “enterprise” includes one-person operations’ (ILO, 1993, para. 7). Self-employment may take place anywhere: in the worker's home, fields, or any public place. Within the definition of self-employment, the review includes hawking, vending, and other street entrepreneurial activities (such as rickshaw pulling), but excludes other forms of making a living, such as begging, foraging, and scavenging.

Intermediate outcomes: Intermediate outcomes reflect the pathways through which the primary outcomes may be influenced. Studies that measured an intermediate job-related outcome (i.e., where individuals are still in the process of preparing for, and gradually moving closer to, work) were eligible for inclusion in the review. Studies that only reported non-work-related intermediate outcomes were not eligible.

Work-related intermediate outcomes may include, but are not limited to: attitudes to work, job search skills, job-related self-efficacy/confidence, career management skills, work readiness, job applications, and job interviews.

Other (non-work related) outcomes may include, but are not limited to: educational outcomes (e.g., attainment and attendance), health outcomes (e.g., intensity/severity of pain), functional limitations (e.g., range of movement), health care resource utilisation, and quality of life.

If any of the included studies measured outcomes for employers or other relevant stakeholders (e.g., co-workers, supervisors), in addition to outcomes for people with disabilities, we collected this outcome data.

Notes: The focus of this review is on economically productive ‘work’; therefore, it is not concerned with unpaid productivity, such as voluntary work, internships, household work, and family responsibilities/caring. It is also important to note that participation in education and training (including job training) is not defined as an employment outcome in this review. This takes into consideration growing evidence that many people with disabilities, particularly the young, are trapped in a ‘revolving door’ of training and vocational preparation abstracted from any real job opportunities (Corrigan & McCraken, 2005).

3.2.4 Types of Study Designs

As the literature was expected to be scarce, eligibility was extended to (a) randomised experiments, (b) rigorous quasi-experimental designs (QEDs) that used some method for removing biases due to non-random assignment of treatment (e.g., regression adjustment, difference-in-differences estimation, instrumental variables regression, fixed effects regression, regression discontinuity, matching, or inverse-propensity-weighted estimation), and (c) QEDs those that used less credible methods for constructing the counterfactual (including those that rely exclusively on before-and-after comparisons). ⁴

Eligible designs include those in which one of the following is true:

Experimental designs

participants are randomly assigned to the treatment and control groups by the investigator, using a process of random allocation, such as a random number generation (randomised controlled trial);

Quasi-experimental designs

decisions about which individuals receive the intervention and which serve as the controls are not in the hands of the investigator; instead, this is decided by the individuals themselves or by other circumstances (includes, for example, designs commonly referred to as controlled before-and-after studies and natural experiments)

Studies collecting data at baseline and endline, and those collecting only endline data, were eligible for inclusion in the review (conditional on meeting all other criteria). Individually-allocated and cluster-allocated studies were also both eligible. The review included studies that adjust for confounders at either the design or analysis stage (e.g., studies using propensity score matching or regression analysis) and studies that have made no attempt to account for differences between the groups. No restriction was placed on the timing of outcome measurements (i.e., duration of follow up).

Although they fail to protect against most threats to internal validity (Shadish, Cook, & Campbell, 2002), studies using historical control and single-group pre-test/post-test (SGPPT) designs were included in the review as prior reviews and our preliminary scoping exercise suggested a scarcity of randomised controlled trials (RCTs) and robust quasi-experimental designs in this area. This decision was made prior to conducting the review. The inclusion of studies using weaker designs can help provide a fuller picture of strategies that are being utilised in the field and to

determine if the research base adequately represents the range of programmes currently in operation.

3.2.5 Date, Language and Form of Publication

For this review, eligibility extended to studies published or reported within the period 1 January 1990 to 31 December 2013. Studies published in any language were eligible, provided they met all other criteria. Studies were included regardless of their publication type (i.e., we did not exclude specific forms of publication, such as unpublished working papers, theses or dissertations).

3.5.1 Criteria for Determination of Independent Findings

Efforts were made to identify all affiliations between studies/reports before coding commenced. Information on study sample sizes, intervention details, grant numbers, and so on were used to identify multiple reports from single studies. In cases where several different reports relating to a single study exist, reviewers classified the publication containing the most complete data set as the main report; when extracting data, the full set of relevant reports was used.

3.5.2 Data Extraction and Quality Appraisal

A coding tool was developed to capture study design characteristics, participant characteristics, intervention characteristics, outcome characteristics, data for effect size calculation (and/or other outcome data) and other substantive and descriptive study features.

Each study was appraised to determine its internal validity---that is, make sure that the study has been designed and conducted in such a way as to minimise the risk of bias, and that the outcomes are likely to be attributable to the intervention being assessed, rather than some other factor.

Included studies were coded on the following domains:

Potential for selection bias/confounding due to non-random assignment, no adjustment for differences in baseline measurements, etc.

For each of these domains, we coded the paper as ‘Yes’ if it addressed the issue, ‘No’ if it did not, ‘Unclear’ if it was unclear, and ‘Not relevant’ if the issue was not applicable for that particular study. For those domains with two parts (‘a’ and ‘b’), both parts had to be answered ‘Yes’ for the study to score ‘Yes’ overall for that domain. We then aggregated to an overall risk of bias as follows:

Low Risk of Bias (bias, if present, is unlikely to alter the results of the research): ‘Yes’ for four or five categories.

The coding tool was pilot tested and modified as necessary. Members of the review team worked independently on a purposive sample of eligible studies, which were selected to test the tool on the full range of relevant study designs, before meeting to compare their decisions. Reviewers were retrained on any coding items that showed discrepancies during this process and the coding manual was adapted accordingly. This process was repeated until a very high level of consistency in reviewers' application of the codes was achieved, at which point the tool was finalised. A draft version of the codebook for data collection is presented in the study protocol. The risk of bias section of the tool is detailed in Table 10.10 (Appendix I).

The EPPI-Reviewer software was used to collect data necessary for the description, analysis and quality appraisal of studies. All studies included in the review were independently evaluated by two reviewers who came together to compare their decisions. Any uncertainties and discrepancies were resolved by discussion, further review of the respective study reports and, where necessary, consultations with a third reviewer.

Data were collected for all labour market outcomes reported (both positive and negative), relevant sub-groups and where studies include multiple follow-up time periods. For the data analysis, where studies used several indicators for a particular outcome variable, our approach was to drop indicators. This involved selecting the indicator most similar to those used by other studies in that category and retaining only the data for that outcome indicator in the analysis. The reviewers attempted to contact the authors of study reports that were missing data that would allow the computation of effect sizes.

Efforts were made to identify all affiliations between studies/reports before coding commenced, using information on study sample sizes, intervention details, grant numbers, and so on. In cases where a single report described more than one study, each study was coded separately (i.e., as if they had been published separately).

5.2.1 Treatment/Therapeutic Interventions

Four treatment/therapeutic interventions were assessed in three of the included studies (see Table 5.5). One study was set in Brazil and evaluated a small-scale intervention for people with physical impairments (Pereira-Guizzo et al., 2012). The remaining three interventions in this category were designed for the visually impaired: two small-scale interventions in Nigeria (Eniola & Adebiyi, 2007) and a regional programme operating in India (Finger et al., 2012).

Interventions targeted at people with physical impairments . One study evaluated a multi-component professional social skills programme for unemployed people with a range of physical disabilities.

Pereira-Guizzo et al. (2012) evaluated the Program for the Development of Social Skills for the Work Environment, which aimed to develop disabled people's work-related social skills and include them into the labour market. The intervention was delivered at institutions in two cities in the state of São Paulo, Brazil - the Municipal Bureau of social welfare and a philanthropic association. It was composed of 16 group sessions, carried out twice a week, lasting approximately 90 minutes each. The overall structure of the programme was based on the experiential method associated with cognitive-behavioural techniques. It appeared to have been designed by the authors of the study.

Interventions targeted at people with visual impairments . The two studies in this category evaluated three interventions. One study evaluated two different therapeutic techniques based on cognitive behavioural therapy (CBT) principles, and the third study assessed the impact of a cataract outreach programme.

Eniola and Adebiyi (2007) assessed the impact of two interventions (labelled as ‘emotional intelligence’ and ‘goal setting’ techniques) upon the motivation to work among visually impaired students attending the School for Handicapped Children, Ibadan and Osogbo, Nigeria. Both interventions were delivered over six weeks (two sessions per week) and entailed a number components, including lectures, home assignments, and other activities. The authors of this pilot study appear to have designed the interventions, and do not clearly report how the two techniques differed.

Finger et al. (2012) examined the impact of a multi-component cataract outreach programme in South India. This regional programme was available in the rural area of Tamil Nadu state, and was operated by a community eyecare provider, Sankara Eye Care Services, Coimbatore. Patients underwent cataract surgery and received a one-month follow up at the outreach centre. Patients' transport, surgery and inpatient hospital stays were provided free by the hospital as patients all fell under the poverty threshold. The study was embedded within routine services provided by Sankara Eye Care.

5.2.2 Assistive Devices & Accommodations

Two included studies evaluated assistive devices designed to improve general well-being (see Table 5.6). Both interventions were targeted at individuals with mobility impairments (Guarino et al., 2007; Shore & Juillerat, 2012).

Interventions targeted at people with physical impairments . Of the two evaluations in this category, one examined whether prostheses helped the employment prospects of lower-limb amputees, and the other investigated the impact of providing a simple wheelchair to persons with mobility impairments resulting primarily from stroke, muscular dystrophy or cerebral palsy.

The study by Guarino et al. (2007) assessed the benefits for lower-limb amputees of using a prosthesis. Patients attended a university-based rehabilitation centre, the Lar Escola Sao Francisco Rehabilitation Centre, Universidade Federal de São Paulo (UNIFESP), in the city of São Paulo, Brazil. A charge was made for the prosthetics.

Shore and Juillerat (2012) evaluated the impact of wheelchairs distributed by an international NGO, the Free Wheelchair Mission (FWM). As of 2012, FWM has distributed over 600,000 wheelchairs in developing countries. Wheelchairs are provided free of charge to recipients, made possible through local and national fundraising efforts. The cost to produce, ship, assemble and deliver the chair to recipients was under US$60 worldwide.

5.2.3 Occupational Rehabilitation

Four studies evaluated multi-component occupational rehabilitation programmes (see Table 5.7). Three of the four programmes in this category were managed by the Centre for the Rehabilitation of the Paralysed (CRP), an NGO that has operated in several regions of Bangladesh since 1979---the only organisation of its kind in the country (Hansen et al., 2007; Momin, 2004; Nuri et al., 2012). The remaining programme was a national initiative in China. Three programmes were targeted at people with physical disabilities, whereas the fourth programme was open to adults with any type of disability.

Interventions targeted at people with physical impairments . Of the three programmes in this category, two were designed specifically for those with spinal cord injuries (Hansen et al., 2007; Momin, 2004). The third intervention in this group was targeted at injured workers (Tang et al., 2011).

Hansen et al. (2007) evaluated a tailored work rehabilitation programme based in Savar, an area about 25km from Dhaka, Bangladesh. The initiative was delivered by CRP. It provides specialised services for people with spinal cord injuries, and aims at returning participants to their previous employment or a suitable alternative. The study focus is on a sample of those who completed the programme in the three-year period 2002-2005.

Momin (2004) evaluated another CRP run rehabilitation programme for people with spinal cord injuries. In this study, participants lived in the Dhaka, Narayangonj, Gazipur, Manikgonj, Munshigonj and Narshingdi districts of Bangladesh, and had received services between 1994 and 1999. Services focused on the person's whole life. Vocational training was provided, support was extended once the person returned to the community, and families were encouraged to become active participants in the rehabilitation process

Tang et al. (2011) evaluated a multi-dimensional return-to-work (RTW) programme provided by the Guangdong Provincial Work Injury Rehabilitation Center, the first and leading institution of occupational rehabilitation in China. The demonstration centre is located in the southeast of the country and has offered training courses nationwide since 2004. It is available to persons with occupational injuries who are on long-term absence from work or short-term sick leave. It is based on a case management approach and includes both social and occupational rehabilitation as well as follow-up after return to work. Clients attend the centre for three months and then receive six months follow-up support.

Interventions targeted at people with any/multiple types of impairments . One study evaluated a multi-component occupational rehabilitation programme that was open to persons with any/multiple types of impairment.

Nuri et al. (2012) evaluated the effectiveness of the Madhab Memorial Vocational Training Institute (MMVTI) programme in Bangladesh, which is part of the Centre for the Rehabilitation of the Paralysed (CRP). Disabled participants were selected from five districts across central Bangladesh. Data was also collected from 10 non-disabled key informants (local leaders, employers, and project staff). This programme specialises in the occupational rehabilitation of persons with disabilities through specifically designed vocational training and job placements. Individuals are carefully matched to courses after a full assessment by a multidisciplinary team of doctors, social workers and other professionals.

5.2.4 Financial Services

One study evaluated the impact of a financial programme (see Table 5.8). It was available to adults with physical disabilities (Metts & Oleson, 1995).

Interventions targeted at people with physical impairments . One study evaluated a multi-component business training and credit guarantee scheme established to facilitate informal sector self-employment for disabled micro-entrepreneurs.

Metts and Oleson (1995) evaluated the initial phase of the Disabled persons Loan Scheme, a multi-component programme funded by the United National Development Programme (UNDP). At the time of the study, it was available in 20 rural districts of Kenya. The typical loan recipient had an orthopaedic disability. Potential beneficiaries received basic business training and assistance with the process of apply for loans through Barclay's Bank of Kenya. Clients were identified by Kenya's Ministry of Culture and Social Services and selected by District Loan Review Committees. Business training and assistance with loan applications were provided by three sub-contracted NGOs, and beneficiaries received them for the duration of the loan repayment period. Loans were provided at market rates of interest and guaranteed by a credit guarantee fund deposited with the bank.

5.2.5 Community-Based Rehabilitation

Four studies evaluated the impact of four community-based rehabilitation (CBR) programmes (see Table 5.9). Three programmes were open to people with any/multiple types of impairment (Biggeri et al., 2012; Lagerkvist, 1992a; Lagerkvist, 1992b) and one was designed specifically for people with leprosy (Gershon & Srinivasan, 1992).

Interventions targeted at people with any/multiple types of impairment . Three studies (reported in two papers) were included in this category.

In the study by Biggeri et al. (2012) approximately three-quarters of the sample were described as having a physical and/or sensory impairment. This study was conducted in neighbouring districts in South Karnataka State, southern India. The authors indicated that the programme covered all five areas of the CRB matrix (health, education, livelihood, social and empowerment) and included activities such as home visits, health awareness, therapy services, referral services, aids/appliance support, and legal support. It was implemented through a participatory development approach with active participation of persons with disabilities at all levels.

In Zimbabwe, Lagerkvist (1992a) evaluated a CBR programme run by the Red Cross in two districts since 1985. Two rehabilitation assistants with 1-2 years medical education were responsible for assessment of clients, analysing a rehabilitation plan for each client, and training local coordinators and volunteers. Local coordinators were committee workers with some medical training, and responsible for 300-400 disabled persons in an area.

Lagerkvist (1992b) assessed the impact of a CBR programme that had operated in the Philippines since 1981. This programme had a different style of management to the one in Zimbabwe. It was managed and supervised from a rehabilitation centre and was based on local supervisors who were community workers recruited from the villages. Each one was expected to work one to two days a week with four to eight disabled persons. World Health Organization (WHO) training packages were used.

Interventions targeted at people with physical impairments . One study evaluated a CBR programme for people with leprosy in the Greater Madras area of India (Gershon & Srinivasan, 1992). Interest-free loans were offered to patients to help them start new trades or occupations. Social workers also helped organise job training and placement activities. The scheme also undertook the education of the children of patients.

6.2.1 What effects do interventions to support adults with disabilities have on motivation to work?

One study addressed this sub-question (see Table 10.18, Appendix I). It focused on persons with visual impairments. The direction of effect was positive and statistically significant.

Visual impairments . In the pilot study by Eniola & Adebiyi (2007) two therapeutic programmes for visually impaired students in Nigeria were evaluated using a pre- and post-test experimental group design, with a total sample size of 32 (16 in each group). The group means suggest that the emotional intelligence intervention may have had a more positive impact than the goal setting intervention (mean change scores 12.2 and 2.9 respectively). However, a significant interaction was not found. Across the whole sample, a statistically significant increase in the level of motivation post-intervention compared pre-intervention was observed (mean change score 6.5; F=7.98; df=1,28; p<0.05).

6.2.2 What effects do interventions to support adults with disabilities have on professional social skills?

One study addressed this sub-question (see Table 10.19, Appendix I). It examined an intervention for people with physical impairments. The direction of effect was positive and statistically significant.

Physical impairments . Pereira-Guizzo et al. (2012) used a multi-probe design to evaluate a therapeutic skills development programme in Brazil. The sample consisted of 16 people with physical disabilities, forming two intervention groups who received the programme at different intervals. Three different indicators of professional skills were assessed: “facing a job interview”; “offering a colleague some help”; and “dealing with a superior's fair criticism”. For Group 1, the programme had a significant impact on all three measures at the two-month follow-up (p=0.001, p=0.032, and p=0.030, respectively). Similar benefits from participation in the programme were also observed in Group 2.

6.2.3 What effects do interventions to support adults with disabilities have on paid employment?

Twelve studies examining 12 different interventions addressed this sub-question (see Table 10.14, Appendix I). Different measures of paid employment were used, ranging from engagement in any type of income-generating activity to formal employment. Baseline samples ranged in size from 1 to 620. Seven studies evaluated interventions designed for persons with physical disabilities and one focused on an intervention for the visually impaired. An additional four studies evaluated interventions that were open to individuals with any/multiple impairments. The direction of effect in all 12 studies was positive. Three study reports (Biggeri et al., 2012; Finger et al., 2012; Shore & Juillerat, 2012) presented results of tests for statistical significance and indicated study findings were significant.

Physical impairments . Of the seven studies in this category, two used a non-equivalent groups pre-test/post-test design. Both found better outcomes for the treatment group compared to the comparison group. The Brazilian study by Guarino et al. (2007) reported results for a group of lower limb amputees who received a prosthesis and those who did not. The majority (98%) of the total sample was working at baseline. At follow-up, rates of employment had decreased to 16% for the treatment group, compared to 0% for the comparison group. Momin (2004) compared outcomes for persons receiving specialised occupational rehabilitation services in Bangladesh with a group in receipt of general hospital care. Although little change was observed overall, the treatment group had a slightly better outcome; their engagement in paid employment was 6% at both baseline and follow-up. In contrast, the rate for the comparison group decreased from 9% to 6%. The remaining five studies in this category were conducted in Bangladesh, Kenya, Vietnam, India (two studies) and China, and investigated occupational rehabilitation services (Hansen et al., 2007; Tang et al., 2012), a community-based rehabilitation programme (Gershon & Srinivasan, 1992), the Disabled persons Loan Scheme (Metts & Oleson, 1995) and wheelchair provision (Shore & Juillerat, 2012). All five studies used a single-group pre-test/post-test design, and found that paid employment rates were higher in the follow-up period.

Visual impairments . The one study in this category (Finger et al., 2012) was located in India and used single difference regression estimation methods applied to pre-test/post-test data. The authors reported that, at one year, more of the participants were engaged in income generating activities (44% to 77%, p<0.001) and participants who had successful cataract surgery were more likely to be engaged in income earning activities one year after surgery (OR 3.28; 95% CI 1.40-7.82; p =0.006).

Any type of impairment . Three of the four impact assessments in this category investigated community-based rehabilitation (CBR) programmes. One CBR study was conducted in India and involved a cross-sectional propensity score analysis (Biggeri et al., 2012). It demonstrated a small but significant effect on employment after two years (2002-2004) implementation, for those previously unemployed (ATT=0.05, SD=0.014, t=3.714). The remaining two CBR studies (Lagerkvist, 1992a; Lagerkvist, 1992b) utilised a single-group pre-test/post-test design. Both evaluations (Philippines and Zimbabwe) found that paid employment rates were higher in the follow-up period. The final study in this group (Nuri et al., 2012) examined an occupational rehabilitation programme in Bangladesh. It also found increased employment over the course of the study.

6.2.4 What effects do interventions to support adults with disabilities have on self-employment?

Two studies addressed this sub-question (see Table 10.15, Appendix I). Both evaluated interventions for persons with physical impairments. The direction of effect in both studies was positive. Neither study reported results of tests for statistical significance.

Physical impairments . The Kenyan study by Metts & Oleson (1995) utilised a single group pre-test/post-test design. It found that five of the 55 recipients of loans from the Disabled persons Loan Scheme started new businesses as the result of participating in the programme. Momin (2004) used a non-equivalent groups, pre-test/post-test design, with a total sample size of 64, to compare an occupational rehabilitation programme in Bangladesh with general hospital care. The rate of self-employment for the treatment group rose from 12% to 19% over the study period; in contrast, it decreased from 19% to 12% for the comparison group.

6.2.5 What effects do interventions to support adults with disabilities have on income?

Four studies addressed this sub- question (see Table 10.16, Appendix I); all utilised a single group pre-test/post-test design. Of these, one study (Finger et al., 2012) used logistic regression. Baseline samples ranged in size from 55 to 620. Three studies evaluated interventions for persons with physical disabilities and one focused on an intervention for the visually impaired. The direction of effect in all four studies was positive. Two studies (Finger et al., 2012; Shore & Juillerat, 2012) reported results of tests for statistical significance and indicated study findings were significant.

Physical impairments . Shore & Juillerat (2012) assessed the impact of providing free wheelchairs. At 12 months, the proportion of the total sample (across three different countries) reporting adequate income had increased from 42% to 52% (χ²=19.741, p=0.000). The authors also indicate that this change was driven mainly by large changes in India where those with adequate income increased from 12.6% to 23.4%. Metts & Oleson (1995) examined the impact of the Disabled persons Loan Scheme on net business income. The authors found that, on average, net monthly income increased by 58.3% (from 2035 to 3222 Kenyan Shillings per month). In their evaluation of a community-based rehabilitation programme for leprosy patients in India, Gershon & Srinivasan (1992) measured the average monthly income of patients before and after rehabilitation. They found that the percentage of those earning less than 200 Indian Rs. per month fell from 67% to 23%.

Visual impairments . Finger et al. (2012) assessed the impact of a cataract outreach service on household income. At one year, the proportion of the study sample reporting a monthly income of 0-1000 Indian Rupees (Rs.) decreased from 49% to 20% (p<0.001). Participants who had successful cataract surgery were significantly more likely to report a higher monthly household income one year after cataract surgery. For example, compared to the highest income category (>3000 Rs./month), participants were about five times less likely to report a monthly household income of 0-1000 Rs. (OR 0.22, 95% CI 0.08-0.62; p = 0.004).

6.2.6 What effects do interventions to support adults with disabilities have on number of hours worked?

One study addressed this sub-question (see Table 10.17 Appendix I). It evaluated an intervention for persons with physical impairments. The direction of effect was positive. The study did not report results of tests for statistical significance.

Physical impairments . Metts and Oleson (1995) used a single-group pretest/post-test design to examine the Disabled persons Loan Scheme in Kenya. Based on a sample size of 55, the authors found an increase in the number of monthly hours worked by employees in the businesses owned by programme beneficiaries from 660 to 1700 hours.

6.4.1 What were participants' observations, experiences and views about why the intervention they received had worked for them?

Two studies answered this question (Hansen et al., 2007; Nuri et al., 2012). Both examined occupational rehabilitations programmes managed by the same NGO in Bangladesh. The following factors were cited:

General health & well-being (one study)

6.4.2 What were participants' observations, experiences and views about why the intervention they received had not worked for them?

Three studies answered this question (Hansen et al. 2007; Nuri et al., 2012; Shore & Juillerat, 2012). Two studies examined occupational rehabilitations programmes in Bangladesh and one study evaluated the impact of providing manual wheelchairs to persons with mobility impairments in India and Vietnam. The following barriers to the success of the interventions were cited:

Discriminatory attitudes of prospective employers (one study)