Childhood disability and socio-economic circumstances in low and middle income countries: systematic review

Background The majority of children with disability live in low and middle income (LAMI) countries. Although a number of important reviews of childhood disability in LAMI countries have been published, these have not, to our knowledge, addressed the association between childhood disability and the home socio-economic circumstances (SEC). The objective of this study is to establish the current state of knowledge on the SECs of children with disability and their households in LAMI countries through a systematic review and quality assessment of existing research. Methods Electronic databases (MEDLINE; EMBASE; PUBMED; Web of Knowledge; PsycInfo; ASSIA; Virtual Health Library; POPLINE; Google scholar) were searched using terms specific to childhood disability and SECs in LAMI countries. Publications from organisations including the World Bank, UNICEF, International Monetary Fund were searched for. Primary studies and reviews from 1990 onwards were included. Studies were assessed for inclusion, categorisation and quality by 2 researchers. Results 24 primary studies and 13 reviews were identified. Evidence from the available literature on the association between childhood disability and SECs was inconsistent and inconclusive. Potential mechanisms by which poverty and low household SEC may be both a cause and consequence of disability are outlined in the reviews and the qualitative studies. The association of poor SECs with learning disability and behaviour problems was the most consistent finding and these studies had low/medium risk of bias. Where overall disability was the outcome of interest, findings were divergent and many studies had a high/medium risk of bias. Qualitative studies were methodologically weak. Conclusions This review indicates that, despite socially and biologically plausible mechanisms underlying the association of low household SEC with childhood disability in LAMI countries, the empirical evidence from quantitative studies is inconsistent and contradictory. There is evidence for a bidirectional association of low household SEC and disability and longitudinal data is needed to clarify the nature of this association.


Background
In 2004 the Global Burden of Disease report estimated that over 100 million children under the age of 15 years had a moderate or severe disability, the majority of whom live in low and middle income countries (LAMI) [1,2]. Research on these children in LAMI countries however, has been described as 'woefully inadequate' [3]. Reviews of research have found that the majority of studies have focused on cross-sectional, communitybased epidemiologic studies aimed at finding the prevalence or aetiology of certain conditions or impairments [3]. In addition, the quality of many studies has been judged inadequate [3][4][5]. It has been argued that it is important to move beyond prevalence studies to generate informative data on the circumstances of people with disability, including how these compare with those of their non-disabled peers [5]. There is now widespread acceptance that disability definitions and measures should no longer focus solely on impairments and other individual characteristics and conditions, important though those undoubtedly are. Definitions and measures of disability should also incorporate contextual dimensions which may enable or act as barriers to disabled children's participation and human rights [6][7][8][9]. A paradigm of disability, which incorporates both individual characteristics and social circumstances, is reflected in the recently developed International Classification of Functioning Disability and Health -Children and Youth (ICF-CY) [10] and the United Nations Convention on the Rights of Persons with Disabilities (UNCRPD) [11]. These factors make it an imperative to remedy the significant knowledge gap about the lives of children with disability in LAMI countries [1,2,12]. The World Disability Report [1] places emphasis on the role of environmental factors in disability; however, it identifies the absence of clarity on the relationship between household socio-economic circumstances and disability among both children and adults in developing countries. An important starting point is to scope existing studies that address this relationship.
Objective A number of important reviews of childhood disability in LAMI countries have already been undertaken and point to significant gaps in knowledge about this important group of children and their families [3,5,[13][14][15]. The purpose of this paper is to establish the current state of knowledge on the socio-economic circumstances (SEC) of disabled children and their households in LAMI countries through a systematic review and quality assessment of existing research. We are taking a broad definition of disability consistent current thinking in the World Report on Disability and for the definition of SEC, we include the following variables; income/asset measures, parental education, poverty, area-based measures of socio-economic factors, occupation-based measures and housing-based measures. To our knowledge, this is the first review specifically reporting the association between childhood disability and the home socioeconomic circumstances in LAMI countries.

Methods
We carried out a review of childhood disability in LAMI countries focusing on four areas: methodological issues in studying childhood disability; household SEC of children with disabilities; the rights of children with disabilities; the use of the International Classification of Functioning, Disability and Health -Children and Young people (ICF-CY) in relation in LAMI countries. This paper is concerned only with the household SEC of children with disability in LAMI countries. Further papers will address the literature related to the rights of children with disability and the use of the ICF-CY. The review by Maulik and Darmstadt [3] covers the methodological issues in detail and our review does not add significantly to their conclusions.

Search Strategy Databases searched
The search for evidence explored six electronic databases, including medical literature (MEDLINE, EMBASE, and PUBMED), Web of Knowledge and CSA databases (PsycInfo, ASSIA). Broad searches were also conducted in Virtual Health Library, and POPLINE databases, as well as in Google scholar http://scholar. google.co.uk. Publications from organisations such as World Health Organisation (WHO), The World Bank and International Monetary Fund (IMF), United Nations Children's Fund (UNICEF); UN Statistics Division; Statistical Information and Monitoring Programme on Child Labour (SIMPOC); Development Banks (Inter-American; African; Asian); Save the Children; Washington Group; Paris21 consortium (Partnership in Statistics for Development in the 21st. century) and The Foundation for Scientific and Industrial Research (SINTEF) were screened. Bibliographies of selected articles were also searched. We set out to identify all relevant publications regardless of language. Only primary studies and reviews from LAMI countries evaluating SEC of disabled children, methodological issues in childhood disability data collection, the use of ICF-CY, and the rights of children with disability were considered eligible for review. Only literature published between 1990 and June 2009 were included because earlier papers may not now be relevant and the United Nations Convention on the Rights of the Child, the basis of children's rights, was published in 1989.

Search terms used
The search terms and strategies used for the searched databases are summarized in Table 1.

Inclusion criteria
Both primary studies and reviews relating to household SEC of children with disability in LAMI countries were included. Only quantitative studies with data on the association of childhood disabling conditions and household SEC were included. Seven studies that reported on children and adults combined with no separate analysis of childhood disability and household SEC were excluded. Qualitative studies and reviews were included if they reported on childhood and adult disability and household SEC.

Instrumentation
The studies were independently assessed for inclusion and categorization by two researchers, DS and NS. There was a very high level of 96% agreement overall and 92% for included studies, final decisions were made by agreement. All included studies were entered into a spreadsheet and then categorised using variables that included details of the research design, screening tools/ method used, primary population studied, specific health condition and broad aim. The key focuses of the papers were categorized according to the broad themes including methodology, household socio-economic circumstances, rights, and ICF/ICF-CY. This paper is concerned only with those papers focusing on the household socioeconomic circumstances of children with disabilities.

Quality assessment
Papers were assessed for quality using the STROBE criteria http://www.strobe-statement.org modified for the specific requirements of the review (Table 2). Studies were judged to have low risk of bias if they had optimum quality in at least 6 out of the eight quality domains and no domains with least valuable quality. Studies with medium risk of bias were those with less than six optimum quality domains with only one least valuable quality domain or six optimum but with one least valuable domain. High risk of bias was defined as more than one least valuable domain or case series with no controls or comparison group.

Analysis
Summary findings related to the household SEC of disabled children in developing countries were extracted from primary studies and reviews identified by the search. The findings of qualitative studies were summarised. The association of household SEC with childhood disability was expressed as odds ratios with 95% confidence intervals where available or p values where odds ratios were not stated by the authors. Meta-analysis of the findings of the empirical studies was not attempted due to heterogeneity of the studies.

Results
24 primary studies of the relationship of childhood disability with family and household social circumstances in LAMI countries were identified [2,4,16-38] (see Table  3). We found no reviews focusing specifically on the relationship of childhood disability with family and household social circumstances in LAMI countries. Three reviews [39][40][41] set out to address the association of both adult and child disability with poverty in LAMI countries; a further 10 reviews [3,5,[13][14][15][42][43][44][45][46] refer to  CSA (PsycInfo, ASSIA) KW = ((disabled children) or ICF-CY or (disability rights)) or KW = ((sensation disorders) or (sensory impairment) or (mental retardation)) KW = ((developing countries) or (low -income countries) or (poor countries)) or KW = ((latin american countries) or (south east asian countries) or (south american countries)) or KW = ((african south of sahara) or (subsaharan african countries) or (low middle -income countries)) 1 AND 2 the association within the context of reviewing other aspects of disability in LAMI countries.
Of the identified primary studies, 5 reported qualitative data within country level reports and 19 were quantitative studies conducted in more than 50 LAMI countries (Table 3). Five studies reported multi-country data, one of which [33] included developed countries as well as developing countries. The age range of children included in the quantitative studies varied from 0-20 years; eight studies reported on children under the age of 10 years, 2 on adolescents only and the remaining studies included young children as well as adolescents. The qualitative data included both adults and children; however, the ages of the children were not specified.
Two of the quantitative studies were based on prospective cohort studies [16,17] and one (reported in two papers) on a randomised control trial [27,28]; the remaining 16 were cross-sectional surveys or case-control studies (Table 3). A range of disabilities were studied in the quantitative studies; total disability rates were reported in 8 studies, emotional & behavioural disorders and mental retardation in 6, hearing loss in 2, visual impairment in 1, and neurological, neurodevelopment and musculoskeletal impairment in 4. The most commonly used instrument to measure disability in the quantitative surveys was the Ten Questions Questionnaire (TQQ); this was used in 8 studies, four as a screening tool in an initial phase of the study and four as the main measure of disability. The qualitative data were based on interviews and focus group including adults and children (or their parents) with a range of disabilities including hearing, visual, intellectual and motor impairment.
Eleven quantitative studies used multiple measures of household SEC. Parental education (mostly maternal education) and income/asset-based measures were the most commonly used measures. Six studies used areabased measures, two occupation-based measures and two housing-related measures. The poverty measures used in the reports containing qualitative data were not specified.
Although the reviews that primarily address disability and poverty [39][40][41] categorically state that poverty is both a cause and consequence of disability, they review very limited empirical data on the relationship and none related specifically to the household SEC of children     with disability in LAMI countries. As shown in Table 3 the qualitative data describe a close association of disability (both adult and child) with poverty; however, the association is not reported consistently in the quantitative studies. Potential mechanisms by which poverty and low household SEC may be both a cause and consequence of disability are outlined in the reviews and the qualitative studies. The reviews identified the living conditions of poor people in LAMI countries as a primary causal mechanism. Preventable impairments associated with communicable, maternal and perinatal disease and injuries [40], and malnutrition in childhood [39] were identified as key mechanisms by which poverty caused disability. The reviews (see particularly Mitra [41]) presented socially plausible mechanisms by which disability itself both in adults and children causes poverty and exacerbates existing poverty. Baylies [43] and Dudzik et al [44] identified social structures, war and social unrest which have the greatest impact on the poor as mechanisms linking poverty and disability. Smith [46] reviewing hearing impairment in developing countries, identified poor preventive health services as a factor in high prevalence of deafness and hearing impairment among children due to perinatal factors, chronic otitis media and ototoxic drugs.
The qualitative data provided further support for the role of poverty as both cause and consequence of disability. Ingstad and Grut [23], based on case studies derived from their fieldwork in Kenya, identified congenital conditions, conditions occurring in pregnancy and childbirth, malaria and epilepsy as mechanisms through which poverty leads to disability. All five reports contained rich data from their case studies illustrating how disability exacerbates and precipitates poverty.
Of the 8 studies reporting total childhood disability rates by household SEC, three were single country studies [24,30,38]. In Kandamuthan's case-control study [24] disabled children aged 0-14 years were more than twice as likely to have a mother with low education as their non-disabled peers. Natale et al's study [30] reported a more than twofold increase in prevalence of disability (17.4%) among children aged 2-9 years among those living in the lowest socioeconomic area compared with children living in a slightly more advantaged but still poor area (8.2%). Although VanLeit et al [38] report 49% of the children aged 0-18 years in their study living in poverty, their study has no control group so the significance of this finding is difficult to interpret. By contrast, there is no consistent association of total childhood disability with household SEC reported by the five multi-country studies [2,4,30,26,33]. Six out of the nine countries included in Filmer's study [20] show significant difference in prevalence of childhood disability by household SEC although only one, Indonesia, shows a clear social gradient. However, in the same author's 2008 study [4], in only two out of 13 countries, Indonesia and India, were significant differences between rich and poor households found. The studies by Loaiza & Cappa [26] and UNICEF [2], based on the Multi-Indicator Cluster Surveys 2 (MICS2) and 3 (MICS3) respectively, report variable findings with only a few countries showing significant relationships of total disability among 2-9 year old children with poorer household SEC. In their study, Suris and Blum [33] use United Nations disability data from both high and low/middle income countries to examine the association of countrylevel prevalence of disability among adolescents aged 15-19 years. They found no correlation of country level female illiteracy with prevalence of disability in this age group.
The six studies that examined the association of child mental illness, behavioural problems and mental retardation with household SEC [16,17,19,22,32,34] all reported significant relationships with poor household SEC. Bashir et al [17], Durkin et al [19] and Shawky [32] reported a strong association of mild mental retardation with household SEC. Durkin et al [19] reported a strong association of severe mental retardation (IQ < 50) with no maternal education and rural living. Anselmi et al [16], in a prospective cohort study, found a significant association of externalising and internalising behaviours, and attention problems at age 12 years with low family income at 4 years of age. Hackett et al [22] reported similar findings in a cross-sectional survey. Izutsu et al [34] found a significant association of residence in a slum area with conduct problems but not with other behavioural difficulties.
None of the studies reporting on the association of neurological, neurodevelopmental and musculoskeletal problems with household SEC [25,[27][28][29]31] found any significant association. Shawky [32] in a case-control study, found a very high odds ratio (13.3(95% CI 7.2,27.8)) of children and young people aged 6-20 years with hearing loss having a mother with no education. Bastos et al [18] reported a two-fold increase in prevalence of hearing loss within the speech frequency range in urban areas (37%) compared with rural areas (18%) of northern Tanzania. The only study reporting on visual impairment [32] found a significant association with no maternal education. The quality of included studies was variable. Table 4 summarises the quality assessment of the quantitative studies using the criteria listed in Table 2. Seventeen of the 19 quantitative studies had a medium or low risk of bias. The main sources of potential bias were single cross-sectional survey design and high or unreported non-response rates. Two studies had high risk of bias: Suris and Blum [33] as the study was based on national disability rates with no information on definitions, no information of how data were collected and only national level female illiteracy rates as a measure of socioeconomic status; VanLeit et al [38] as the study included only adolescents with disabilities with no comparison or control group.
The methodologies used in the country-based reports to collect qualitative data, while enabling rich data to be collected, had weaknesses related to inadequately described sampling methods and limited explanation of how themes and case studies were identified, selected and analysed. The samples included both adults and children limiting the validity of the findings for studying childhood disability.

Discussion
To our knowledge, this is the first review specifically reporting on the available literature on the association of childhood disability with home socio-economic circumstances in LAMI countries. We identified primary quantitative and qualitative studies and reviews that specifically addressed the household SEC of disabled children in LAMI countries. As this is a narrative review, we have not carried out meta-analysis of results of empirical quantitative studies.

Main findings
This review has shown that evidence from available literature on the association of childhood disability and household SEC in LAMI countries is inconsistent and inconclusive. This is likely to be due to differences in: measures of household SEC used in the studies; definitions of disability; outcomes studied; study design and methods. Some measures of household SEC, for example, residence in urban versus rural areas and female illiteracy rates are likely to be less precise than those based on individual household SEC measures. Studies using more than one measure of household SEC are more likely to identify associations with childhood disability than those using single measures as different measures may capture different pathways and mechanisms associated with the outcomes. For example, mild learning disability (termed mental retardation in the included studies) is likely to be associated with household education level while hearing impairment is likely to be associated with poor perinatal care and lack of preventive health care secondary to family poverty.
The definition of disability has developed in recent decades and the International Classification of Functioning, Disability and Health promotes a bio-psycho-social model that incorporates components of the medical and social models [10]. Disability arises out of the interaction between health conditions with contextual factors including the environment (SEC is part of this) and personal factors. With the United Nations Convention on the Rights of Persons with Disability, disability is becoming a rights based issue and its definition is changing [11]. However the literature described in this paper defines disability in terms of impairments.
The association of poor household SEC with learning disability and behaviour problems was the most consistent finding of the review. All six studies reported a positive association of poor household SEC with these outcomes and all these studies had a low or medium risk of bias. By contrast, studies with neurodevelopmental and neuromuscular problems as the outcome reported no association with poor household SEC. These studies also carried a low or medium risk of bias. One possible explanation for this differential association is that children with physical or neurological problems die prematurely and so are not available for counting whereas children with learning disability or behavioural problems survive; another explanation is that the neurodevelopmental and neuromuscular conditions may be more likely to have a genetic origin not associated with SEC.
Only two studies reported on hearing loss as the outcome of interest. They reported strikingly different associations with SEC. These divergent findings may be partly explained by the different measures of household SEC used.
The studies in which overall disability was the outcome of interest also reported divergent findings. The multi-country studies, based on secondary analysis of the UNICEF Multiple Indicator Cluster Surveys (MICS), Demographic and Health Surveys and other survey datasets, reported a positive association of childhood disability rates with poor household SEC in a few countries but no association in the majority of countries studied. Children with disability were identified using the TTQ in several of these studies. This instrument was designed with two stages; the questionnaire and then a clinical assessment of children who scored positively on the questionnaire. Unfortunately the clinical assessment is often not performed, making the TQQ invalid. One of these studies [33] had a high risk of bias; the other four studies had a medium risk of bias. A positive association of childhood disability rates with poor household SEC was reported by the three studies that collected and analysed primary data; however, one of these studies [38] had a high risk of bias as it was based on a case series with no control or comparison group. The reviews identified by the search strategy reported socially and biologically plausible mechanisms by which poverty might be both a cause and consequence of disability. They identified the living conditions of poor people in LAMI countries as a primary causal mechanism; as Elwan [40] states "disability in developing countries stems largely from preventable impairments associated with communicable, maternal and perinatal disease and injuries". Malnutrition was identified as a specific cause of disability in childhood [39]. The reviews (see particularly Mitra [41]) presented socially plausible mechanisms by which disability causes poverty and exacerbates existing poverty.
The qualitative data from country-based reports had methodological weaknesses that limited the validity of their findings in relation to the association of poor household SEC with childhood disability. However, the studies provided further support for the role of poverty as both cause and consequence of disability. Ingstad and Grut [23], based on case studies derived from their fieldwork in Kenya, identified congenital conditions, conditions occurring in pregnancy and childbirth, malaria and epilepsy as mechanisms through which poverty leads to disability. All five studies reported rich data from their case studies illustrating how disability exacerbates and precipitates poverty.

Comparison with literature from high income countries and other literature from LAMI countries
To our knowledge, no equivalent review of the association of childhood disability with poor home circumstances/low SES in high income countries has been published. There is a substantial literature from the UK (for example, Gordon et al [47]; Blackburn et al 2010 [48], Emerson et al [49]), the USA (for example, Newacheck [50]; Newacheck and Halfon [51]) Australia (for example, Bor et al [52]; Leonard et al [53]) and Scandinavia (for example, Hjern et al [54]; Berntsson and Kohler [55]) showing that childhood disability is associated with higher risk of living in a poor/low income household. Gordon et al [47] assert that poverty in high income countries is both cause and consequence of childhood disability but we are not aware of specific studies confirming that poverty is causally related to disability in childhood.
The findings of this review are consistent with the literature in high income countries in that poverty is viewed as both cause and consequence of childhood disability; however, although there is strong evidence in both high income and LAMI countries for poverty as a consequence of disability, there are no empirical studies demonstrating the causal relationship of poverty to disability. Whereas studies from high income countries show a consistent association of childhood disability with poor home circumstances/low SES, the quantitative studies reviewed here show a much less consistent association. This is likely to result from differences in definition of disability, in data quality and sampling methods.
Three quantitative studies [56][57][58] not included in this review as they did not analyse child and adult disability separately, reported lower incomes and fewer assets among households with disabled members (adults and/ or children) and those with no disabled members in three sub-Saharan African countries (Namibia, Malawi and Zimbabwe). Their findings indicate that the use of rigorous research methodology with attention to sampling, disability definition, and detailed measurement of household SEC can result in very similar findings in different countries.

Limitations
Publication bias is a possible threat to the conclusions of this review. Although we aimed to identify a broad range of journal papers and reports on childhood disability, we did not include a search of grey literature, hand-searching of key journals or advice from key informants. As a consequence, the review is unlikely to include all available studies of the links between household SEC and childhood disability in LAMI countries. We used general search terms for disability, did not include specific diagnoses and chose to limit the search to publications from 1990 onwards in order to exclude studies with outdated information and to limit the volume of literature identified. This may have led to significant omissions from the review.

Implications for further study
The reviews and qualitative studies included in this review suggest that poverty is both a cause and consequence of disability in childhood. They outline biologically and socially plausible mechanisms by which poverty may exert its influence. By contrast, the empirical studies reviewed here report an inconsistent association of household SEC with childhood disability. Further research in this area is essential to clarify the relationship. The findings of this review, and those of Maulik and Darmstadt [3] indicate the need to address the following issues in future research into the association of childhood disability with household SEC: Definitions of disability -although many of the studies reviewed here clearly stated the definitions used, there is a need, as Maulik and Darmstadt [3] point out, for consistency and standardisation of the definitions used and awareness of the problems of definition that impose limitations on conclusions that can be drawn Types of disability -this review is consistent with Maulik and Darmstadt's [3] observation that the majority of studies focus on intellectual impairment and overall disability rates. Future studies could focus on other relatively common disabling conditions such as cerebral palsy. We found no studies using functional definitions of disability, such as defined in the ICF-CY -these would provide further valuable insights into the association with household SEC Measures of household SEC -the review demonstrates that more than one measure of household SEC is necessary to study comprehensively the association with childhood disability as different measures such as maternal education and household wealth may have different associations with disability.
Study design -cross-sectional surveys, the most common design among the quantitative studies reviewed here, are relatively cheap to organise and are valuable for descriptive epidemiology. Further study to describe the relationship of childhood disability and household SEC will be valuable and cross-sectional surveys will continue to have a role. However, such surveys are only able to show associations; longitudinal prospective studies are necessary to comment on causal mechanisms particularly as relates to poverty as a cause rather than consequence of disability.
Mixed methods studies -the review shows that the qualitative studies provide valuable insights into possible social and biological mechanisms by which poverty might impact on disability. Future research would be strengthened by a mixed methods design that combined rigorous qualitative research preferably nested within a well-designed prospective quantitative study.

Conclusion
This review indicates that, despite socially and biologically plausible mechanisms underlying the association of low household SEC with childhood disability in LAMI countries, the empirical evidence from quantitative studies is inconsistent and contradictory. There is an urgent need for a more robust evidence base to inform the development of effective health and social policies aimed at reducing the burden of childhood disability in LAMI countries.