Systematic review of interventions for reducing stigma experienced by children with disabilities and their families in low‐ and middle‐income countries: state of the evidence

To identify and assess the evidence for interventions to reduce stigma experienced by children with disabilities and their families in low‐ and middle‐income settings.


Introduction
It is estimated that 150 million children globally live with a disability [1], of whom the majority (80%) live in low-and middle-income countries (LMIC). Fifty million children aged under five years are estimated to have developmental disabilities [2] and are likely to experience complex intellectual, physical and sensory impairments over their lifetime. There is evidence that experiences of stigma and discrimination are common for children with disabilities and their families [3][4][5][6] and this experience may vary by type and severity of disability [7][8][9]. However, information on approaches to, and impact of, interventions that address stigma in the context of children with disability and their families in LMIC is generally lacking.
Stigma is a complex psychosocial concept that lacks a universally agreed theoretical approach or definition.
Conceptualisations of stigma have increasingly drawn on human rights frameworks and recognise stigma as a form of social oppression. Link and Phelan [10] define stigma as the recognition and labelling of differences between people that connect to negative stereotypes, and therefore result in separation, status loss or discrimination. Healthrelated stigma has been defined by Weiss (2008) as 'a social process, experienced or anticipated, characterised by exclusion, rejection, blame or devaluation that results from experience, perception or reasonable anticipation of an adverse social judgement about a particular group' [11]. For this paper, we will draw on Van Brackel's recent conceptual model [12], which builds on definitions by Weiss [11] and Scambler [13], and differentiates between the internal perspective of 'people who are stigmatised' and the 'sources of stigma'. Considering people who are stigmatised, stigma is further categorised into 'anticipated stigma' (the expectation of encountering stigma), 'internalised (or self) stigma' (a sense of shame, guilt and fear) and 'experienced stigma' (discrimination). Sources of stigma can include the community, health staff, teachers, laws and policies, and this includes 'enacted stigma' (which refers to discrimination) and 'negative attitudes and prejudice' perpetuated by others, social processes or structures.
Stigma, prejudice and negative attitudes lead to discrimination and the social and economic exclusion [9,14] of children with disabilities and their families, increasing their vulnerability. There is evidence from LMIC that stigma is associated with poor physical and mental health outcomes, social isolation [3,4], limited access to health and education services [5][6]15] and increased financial and emotional strain [16][17][18][19][20]. Stigma may also be a contributor to children with disabilities being at increased risk of abuse, premature death and infanticide, compared with children without disabilities [21,22]. The widespread detrimental consequences of stigma related to child disability highlight the need for interventions aimed at reducing this stigma. Although stigma related to disability is not restricted to lower resourced settings, Kemp et al. (2019) suggest stigma may be a greater impediment to accessing services in these settings and that the same cultural and structural factors that influence stigmatising attitudes may also limit the acceptability and uptake of the interventions themselves [23].
There is growing evidence related to some health conditions (e.g. HIV, mental disorders and leprosy) that stigma-reduction interventions can be effective. For example, contact interventions (involving interactions between the public and affected persons with the aim of improving attitudes and reducing discrimination and exclusion) have been found to improve community attitudes about mental health [24] and leprosy [25]. Rights-based peer counselling was found to be effective at reducing internalised stigma and promoting social inclusion among adults with leprosy [26]. The use of 'change agents' or popular opinion leaders to display positive attitudes has shown promising results in the spread of non-stigmatising messages through the modelling of a new behaviour related to HIV and sexually transmitted infection interventions [27,28]. A systematic review of interventions aimed at addressing stigma for children with epilepsy identified different education and counselling programmes, which had variable benefit for the well-being of children with epilepsy [29].
This systematic review aimed to identify and assess the effectiveness of interventions to address stigma experienced by children with disabilities and their families in LMICs.

Search strategy
The systematic review was undertaken in accordance with PRISMA guidelines [30]. The protocol was registered with PROSPERO International Prospective Register of systematic reviews CRD42018102811. The following seven databases were searched in May 2018 to identify interventions published from January 2000 to April 2018: MEDLINE, EMBASE, Global Health, PsycINFO, Social Policy and Practice, CINAHL, International Bibliography of the Social Sciences. A search was carried out using terms for both 'child with disability' and 'stigma and discrimination', with LMIC keywords (according to the World Bank definition July 2017). Boolean, truncation and proximity operators were used to construct and combine searches for the key concepts as required for individual databases, and an example is available as Appendix S1.

Frameworks
For this paper, we drew on the review by Heijinders and Meij [31], which differentiates between the following five intervention/strategy implementation levels: intrapersonal, interpersonal, organisational/institutional, community and governmental/structural level. Recognising that stigma is a complex social process, we also aimed to identify the 'type' of stigma targeted by the interventions. We included four broad types of stigma characterised by Weiss [11], adapted by Van Brackel [12] and extended here to include caregivers/family as well as the affected child: 1 Negative attitudes and prejudice towards the child/family perpetrated by others, social processes or structures; 2 Discrimination or social exclusion 'enacted' by the community, health staff, structures, laws or policies (the 'sources of stigma') towards the child/family or by family members towards the child; 3 Internalised (or self) stigma including internalised negative stereotypes or negative attitudes, feelings of shame or guilt, low self-esteem, withdrawal from social participation by the child and/or by family members; and 4 Anticipated stigma: the perception or fear by the individual that stigmatisation is likely to occur.

Inclusion and exclusion criteria
We used deliberately broad inclusion criteria as we expected limited research in the area and wanted to capture different types of interventions that have been evaluated. There were therefore no restrictions on study design or language. We included studies of stigma-reduction interventions, for example quantitative studies including RCTs, controlled and uncontrolled pre-post studies, cross-over studies and longitudinal panel studies. Qualitative or mixed-method studies were also included. Participant inclusion criteria were as follows: (i) child with impairment or disability and (ii) family of a child with impairment or disability. We broadly included children with disabilities, as well as specific impairment types, such as physical and sensory impairment, mental illness, cognitive impairment, epilepsy, fits and seizures. We excluded studies that focussed on participants with (i) conditions that constituted a very specific field of research and intervention, such as chronic illnesses and diseases (cancer, heart disease, diabetes, etc.), communicable diseases including HIV/AIDS, drug and alcohol-related issues and short-term disabling conditions and (ii) participants with disabilities or impairments over the age of 18.

Search strategy
Article citations were uploaded and organised for title and abstract review using the reference manager programme Endnote X5. Titles were screened by two reviewers (TS and SP) to determine whether they included relevant information. If the article was deemed relevant by at least one reviewer, the abstract was retrieved. Two reviewers (TS and JA) screened the abstracts for relevant information. If at least one reviewer deemed the abstract relevant, or if the full text had to be obtained to determine if the abstract was relevant, the full text was reviewed. Discrepancies were discussed with a third reviewer (SP) and consensus was reached as to whether or not to include the article. We undertook double data extraction using a standardised form. The data extraction form was piloted with four studies and included information about the WHO region in which the study was undertaken, study design and participants, intervention type and outcomes related to stigma. We also recorded results on 'knowledge/understanding' about the condition/disability under study if this was assessed alongside another stigma related outcome (e.g. attitudes) because improved knowledge may challenge myths, beliefs and/or stereotypes and therefore contribute to improved attitudes or self-perception [32]. In classifying the intervention, effectiveness results from quantitative studies were summarised as being 'positive' (evidence of statistically significant improvement in the stigma related outcome measure), negative (evidence of statistically significant decrease), 'null' (no statistically significant change) or mixed (findings were a mix of 'positive' and 'negative'/'null').

Quality assessment
The full texts of all eligible studies were assessed against quality assessment criteria adapted from Lund et al. [33] and independently assessed by two reviewers (TS and SP; Table 1 shows quality assessment criteria). Differences between the reviewers were discussed, and consensus was reached on all papers.

Results
The database search generated 2860 records, from which 907 duplicates were removed. When screened by abstract, 397 records did not fulfil the necessary criteria. The full texts of 72 papers were then assessed, of which 20 were eligible for inclusion. Data were provided from 16 Data represented fits the views of the participants studied (credibility) Analysis is grounded in the data (confirmability) Risk of bias Low All or almost of the above criteria were fulfilled and those that were not fulfilled were thought unlikely to alter the conclusions of the study Medium Some of the above criteria were fulfilled, and those not fulfilled were thought unlikely to alter the conclusions of the study High Few or no criteria were fulfilled, and the conclusions of the study were thought likely or very likely to alter with their inclusion. *Not required for qualitative studies.
countries. Reasons for excluding the full text articles can be found in Figure 1. Table 2 summarises the characteristics of the studies eligible for inclusion. The 20 included studies provided data from 26 different study settings. The most common WHO study region was Europe (n = 8, 31%), followed by the Americas (n = 7, 27%) and the Africa Region (n = 6, 23%). The majority (n = 24, 92%) of interventions targeted sources of stigma: negative attitudes (n = 19, 73%) and exclusion (n = 5, 19%), while only two (8%) studies targeted people who are stigmatised (internalised stigma) and no studies explicitly assessed anticipated or experienced stigma. Most interventions targeted a single social level, most commonly organisational/institutional (n = 13, 54%) followed by community (n = 6, 25%) and intrapersonal (n = 3, 13%). No interventions were delivered at government/structural level. Twentyfour stigma-reduction strategies were included in the 20 studies, and the majority used education (n = 15, 63%), followed by four studies of contact (n = 4, 17%) interventions. The interventions targeted children with a limited range of impairments types; the most common was epilepsy (n = 10, 50%) followed by intellectual impairment (n = 4, 20%). Table 3 summarises the designs of the included studies. The majority of studies were quantitative in nature (n = 15), two were qualitative, and three used mixed methods (both qualitative and quantitative). Fourteen studies had before-after study design; however, the majority had no control group (n = 10), only one study used random assignment to intervention or control, and only five described a follow-up period, which varied from 4 weeks to 2 years. The remaining studies only collected data post-intervention. There were two multi-country studies, both of which used phenomenological qualitative methods. Study participants (the group targeted in the intervention) were most commonly primary school teachers (n = 5; 25%), followed by parents (n = 4; 20%). In terms of method of outcome assessment of the quantitative studies, one used a previously validated questionnaire [34] the 'Opinions Relative to Mainstreaming' [35], which assessed teachers' opinions and attitudes related to mainstreaming special needs students in regular education environments. Three studies used questionnaires from previous studies [36][37][38]: Elafros et al. [36] used a threeitem assessment to assess felt stigma in Zambia [39], Eze et al. [37] used a questionnaire adapted from a previous study of teachers' perception of epilepsy in Nigeria [40] to assess the trainee teachers' knowledge, attitudes and first aid management of epilepsy, and Tilahun et al. [38] used a questionnaire assessing beliefs and social distance towards children with autism, adapted from the World Psychiatric Association's programme to reduce stigma and discrimination because of schizophrenia [41]. Eleven studies developed bespoke self-reported tools [42][43][44][45][46][47][48][49][50][51][52].

Study characteristics
Three studies that used qualitative methods undertook interviews with a topic guide [53][54][55], and the data collection approach in the remaining two studies was unclear [56,57].

Risk of bias in included studies
The quality of the studies was generally relatively poor; two (10%) were assessed to have a medium risk of bias, and 18 (90%) had high risk of bias. No studies were deemed to have a low risk of bias. Common methodological limitations included lack of control groups (n = 15), clearly defined, valid stigma assessment measures and non-representative samples that result in limited generalisability. Studies predominantly measured aspects of stigma (e.g. negative attitudes) through self-report questionnaires but evidence was lacking on the validity or reliability of the questionnaires used in the study setting. Few studies included control groups (n = 4), and lack of adequate adjustment for confounding was also a concern; whilst some distributions of principle confounders were partially described (n = 9), few studies accounted for confounding in the study design or analysis. Loss to followup was reported in fewer than half of the studies (n = 8), and characteristics of losses of participant follow-up were inconsistently taken into account and reported in eight (40%) studies. No studies demonstrated a comprehensive attempt to measure adverse effects. Power calculations were only provided in two studies and although some studies assessed for significant difference through before/ after designs, no studies calculated effect sizes.

Type of interventions
We present the results of the 20 included studies according to level at which the intervention was delivered: organisational/institutional, community, intrapersonal and interpersonal, and multiple levels (Tables 4-7).  Tropical Medicine and International Health volume 00 no 00

Interventions at organisational/institutional level
The majority of interventions were delivered at organisational/institutional level (n = 9) and aimed to reduce negative attitudes towards children with disability, most commonly epilepsy (enacted stigma). Training programmes were the most commonly delivered interventions (n = 8), and different approaches were used including didactic and interactive teaching sessions, videos, theatre and small group discussions. The programmes targeted teachers (n = 7) and school pupils (n = 2) ( Table 4). Seven studies reported positive results, with significant improvement in knowledge and reduction in negative attitudes. The remaining two studies reported mixed results, with improvement in knowledge but limited change in attitudes towards children with epilepsy post-intervention [42,45]. However, the majority (n = 7) of studies were assessed to have a high risk of bias, with two [43,48] assessed to have a medium risk of bias.

Interventions at community level
At community level, three contact-based interventions aimed to address negative attitudes and exclusion. Two of these involved direct contact: (i) a film screening in Brazil, Colombia and Japan about lives of children with disabilities and their caregivers followed by community debates [57] and (ii) an inclusive sports programme in Germany, Hungary, Poland, Serbia, Ukraine, including people with and without intellectual disabilities in sports teams [53]. One study in Ethiopia used indirect contact through an educational comic entitled 'We'll make it', which included traditional views of epilepsy and introduced the concept of inclusion and football [52] ( Table 5). All studies demonstrated a positive effect; qualitative evidence from the film screening and the sports programme suggested a change in enacted stigma including a decrease in negative attitudes and social exclusion by community members and sports participants. Knowledge and attitude scores significantly improved among children who participated in/received the educational comic book intervention; however, all studies were assessed to have a high risk of bias.

Interventions at the intrapersonal and interpersonal level
Three studies targeted the intrapersonal level [36,44,49], and one study was conducted at the interpersonal level [54]. The strategies to address stigma at the intrapersonal level included support groups. In one study, peer support groups, where content was chosen by the participants who had epilepsy, aimed to target internalised stigma and Tropical Medicine and International Health volume 00 no 00  Tropical Medicine and International Health volume 00 no 00 Positive High *Results from quantitative studies (only) categorised as 'positive' (evidence of statistically significant improvement in the stigma related outcome measure), negative (evidence of statistically significant decrease), 'null' (no statistically significant change) or mixed (findings were a mix of 'positive' and 'negative'/'null'). *Results from quantitative studies (only) categorised as 'positive' (evidence of statistically significant improvement in the stigma related outcome measure), negative (evidence of statistically significant decrease), 'null' (no statistically significant change) or mixed (findings were a mix of 'positive' and 'negative'/'null'). non-disclosure [36]. Two studies investigated the effect of parent support groups. One study investigated parent to parent counselling for caregivers of children with cerebral palsy, which took place for 90 min in weekly sessions, and aimed to facilitate exchange of knowledge and experience [49], and the other study combined parent support groups for caregivers of children with epilepsy with an educational component [44]. The interpersonal level intervention [54] consisted of home visits and community-based rehabilitation by community health workers to assist trained professionals in supporting parents in their home environment. The study aimed to reduce negative attitudes and exclusion. Two of the four studies targeted internalised stigma (e.g. shame and guilt) of the child [36] and parent [49] and two addressed negative attitudes (among caregivers/ family members about the child with a disability [44,54]. While one study of support groups found reduction in internalised stigma [36], the effect was either mixed or unclear for the remaining studies [44,49,54] (Table 6).

Interventions targeting multiple levels
The most commonly combined intervention levels were organisational/institutional and community. The studies included schools and healthcare settings and tended to combine individual-level information provision and/or skills building through training, with community-level activities, such as theatre. All studies targeted enacted negative attitudes. One study by Dalal et al. [56] intervened at the interpersonal level with organisational/institutional and community, combining community-based rehabilitation, education and contact. Activities included medical checks to enable access to disability certificates, children with disabilities collecting donations for flood victims (door to door and procession), community discussions around abilities of youth with disability and establishing an integrated school. This study demonstrated qualitative evidence of change in positive attitudes, community and social inclusion ( Table 6). Three studies included a training programme intervention; Tilahun et al. [55] assessed the effect Health Education and Training (HEAT) Mental Health Training on exclusion of children with autistic spectrum disorder as well as negative attitudes of community health workers; Tilahun et al. [38] assessed the effect of delivering 10 sessions of classroom-style training to community health workers on awareness-raising efforts in community; and Somoza 2013 [50] used theatre in primary schools for school children, parents and teachers and seminars in hospital for paediatric staff, to address negative attitudes of epilepsy. Results were predominantly positive; however, negative attitudes remained a barrier to training for some health extension workers (HEW) [38] (Table 7).

Discussion
This systematic review identified 20 studies of interventions aimed at reducing aspects of stigma experienced by children with disabilities and their families in LMIC. In terms of type of intervention, the majority of interventions targeted a single social level only (most commonly organisation/institutional) and there was limited evidence for multi-level interventions. Most interventions targeted a single domain of stigma; predominantly, negative attitudes with few studies focussing on other aspects of the stigma process, including internalised stigma. The most common disability type targeted was epilepsy, followed by intellectual disability while physical and sensory impairments were relatively neglected, limiting any comparison of intervention impact by disability type. The most common stigma-reduction strategy utilised was education (n = 15, 63%), followed by 'contact' interventions (n = 4, 17%). The majority of the studies found either a positive or a 'mixed' impact of the intervention on an aspect of stigma. However, caution in the interpretation of findings is warranted because the studies were characterised by a high risk of bias.
There are no previous reviews of stigma-reduction interventions focussed specifically on children with disabilities with which to compare this review. However, our review has some findings in common with previous reviews (which included all-ages) on health-related stigma reduction. Our finding that 'sources of stigma' (negative attitudes and discrimination/exclusion) were most commonly addressed and that education/training was the most common intervention approach aligns with reviews of health-related stigma-reduction interventions in LMIC [23] and multi-level interventions globally [58]. Although the quality of evidence was relatively poor, this review suggested some encouraging trends for education and contact-based interventions in terms of improving attitudes. This aligns with findings of Heijnders and Van Der Meij [31] who suggested that education and contact interventions show promising results in the field of HIV/ AIDS, mental illness, leprosy, TB and epilepsy, and Mehta et al. [59] who reported that social contact reduced mental-health-related stigma.
Our review also highlighted concerns about the quality of existing studies assessing effectiveness of stigma-reduction interventions related to disability. These concerns align with findings from previous reviews of Heijnders and Van Der Meij [31] and Mehta et al. [59], underscoring a need for well-designed research in this area. This included identifying a need for more rigorous assessment of intervention effect, a concern that was also raised in a review by Kemp et al. [23]. In particular, studies lacked control groups, validated measures of stigma [23] and reported statistical significance but not effect sizes [58].
A critical assessment of the studies included in this systematic review suggests key gaps in the literature. The majority of studies evaluated short-term outcomes but lacked evidence of long-term impact, and no studies included measures of change in behaviour. Stigma-reduction interventions focussed on a narrow range of impairments, primarily on children with epilepsy or intellectual impairment and typically focussed on single levels. Considering the qualitative evidence that experiences of stigma vary by type and severity of disability [8,9], this deserves further attention. Few studies appeared to involve people with disabilities in the design and implementation of stigma-reduction strategies. Active involvement of people with disabilities is important for maximising the feasibility, acceptability, sustainability and impact of interventions. Heijnders & van der Meij (2006) argue the need for multi-level interventions that aim to change negative attitudes and discrimination alongside empowerment of affected individuals by ensuring that they take an active role as in the design and implementation of stigma-reduction strategies [31].
Given the poor quality of studies assessed in this review, it is important that results are interpreted with caution. Future research directions should include multilevel interventions that address and/or assess internalised stigma as well as negative attitudes and discrimination/exclusion perpetrated by the 'sources of stigma'. Comprehensive intervention descriptions are necessary to replicate interventions in different contexts and to evaluate the conditions under which stigma may be optimally reduced. In addition, a wider range of disabilities evaluated with these interventions should be included in design and implementation of future studies. A lack of available validated tools for assessing stigma experienced by children and their families is an important area that warrants attention.
The purpose of this review was to describe the evidence on interventions to reduce stigma experienced by children with disabilities and their families in LMIC and inform potential future research studies. We used a comprehensive search strategy that followed PRISMA guidelines, and robust methods that included double data extraction and review to produce an accurate, comprehensive state of the evidence composition. This review has several limitations. Our study did not limit inclusion of articles through methodological appraisal. While we include information on intervention effectiveness, the lack of rigour in these studies may have led to non-generalisable conclusions.
Studies undertaken in high-income countries were excluded to focus on the unique challenge of addressing stigma in LMIC in contexts with limited financial and logistic resources and unmet need. Inclusion of studies from high-income settings in future reviews may inform additional learning. The assessment of outcomes that lacked uniformity and validity made both interpretation and comparison of study results difficult.

Conclusions
This systematic review highlights key gaps in the evidence around effective stigma-reduction strategies for children with disabilities and their families in LMIC. There are some promising findings around education and contact interventions to reduce negative attitudes. However, given the methodological limitations we found, these findings have to be interpreted with caution. The validation and consistent use of contextually relevant quantitative measures of stigma may advance this field of research.