Severe disability and its prevalence and causes in northwestern Ethiopia: evidence from Dabat district of Amhara National Regional State. A community based cross-sectional study

Background: Disability is the social outcome of unfavorable interactions between individuals’ impairments, on the one hand, and inaccessible physical spaces, disenabling cultural environment and negative attitudes towards disability and persons with disabilities, on the other. Despite the fact that it directly affects 15% of Ethiopians as well as families and communities, policy-relevant studies on disability and the conditions of persons with disabilities are lacking. The Dabat Demographic Health Surveillance System (DHSS) was established to collect vital statistics in Dabat district, Amhara region. But it did not collect data on disability types, causes and community responses. With this drawback in DHSS, this study aimed at assessing the prevalence, types and major causes of severe disability in Dabat district. Method: A community-based cross-sectional study design was employed and covered 17,000 households residing in 13 Kebeles of Dabat district. The study was conducted in 2018 utilizing the modied 12-item World Health Organization’s Disability Assessment Schedule (WHODAS 2.0) and 7-item WHO Minimal Generic Set of Domains of Functioning and Health to collect data on disability in the district. The data were organized and presented as frequencies and percentages in table and gures. Relevant variables were used to construct a logit model that predicts the likelihood of disability whereby P-value < 0.05 was considered as statistically signicant. Results: This study found that the overall prevalence of disability in Dabat district was 2.14% [95% CI: 2.03, 2.24]. About 8.3% of households reported at least one member with disability. Visual impairment was the most commonly reported impairment (51%) while 8.19% had multiple impairments. Eighty-three percent of immediate the causes of disability were modiable – specically, illness (36.93%), injury (17.81%), and congenital (10.86%). The aged [AOR=1.0; 95% CI: 1.03, 1.04], those unable to read or write illiterate [AOR=1.57; 1.15, 2.14], the unmarried/single [AOR=1.39; 95% CI: 185,2.47] and the separated [AOR=2.78; 95% CI:


Background
With multiple types and causes, impairment or disability can affect any person at some point in life, temporarily or permanently. Though impairment and disability had been used interchangeably well into the 1970s, disability movement espoused a social model that distinguished between the two i.e., impairment as a function of the condition the body or mind, and disability as an undesirable "outcome of an oppressive relationship between people with impairments and the rest of society" (31). This powerful distinction between the body and the society/culture, impairment and disability has helped the disability movement towards the advancement of the rights of persons with disabilities (PwDs).
But these achievements of the social model came at the cost of, some writers (e.g., Hughes and Paterson 1997) argued, emptying disability from the rich personal experiences of impairment, which made disability inadequate as a concept in postmodern identity politics centered on the body as a contested cultural terrain (32). Hughes and Paterson (1997) proposed not only to bring back the body/impairment experience into the cultural, symbolic space but also account for the multiple, individualized representations of pain and the oppressive societal/environmental factors. If the body enters the cultural realm in disability studies and movement, they argued, it can become a powerful 'emancipatory' concept.
This study drew on the WHO-ICF de nition of disability whereby a person is considered disabled if and when his/her activity performances in his/her usual environment are limited in nature, function, or quality. This de nition shifts the focus from individual factors measured in terms of medical or health indicators to broader factors that include health, education, social relationships and cultural participation. It treated disability in contrast to full functionality and both are "conceived as a dynamic interaction between health conditions (diseases, disorders, injuries, traumas, etc.) and contextual factors." (33) In a way, such a de nition draws on the complementary conceptualization of disability -with their own variations, of course -in the social model and the capabilities framework i.e., not only in understanding disability in terms of social/environmental barriers to equal opportunities and resources but also in identifying discrimination and oppression as central to its characterization (31). As such, this study recognizes all physical or learning and intellectual impairments that cause limited social functionality, opportunities and autonomy that are inconsistent with human dignity, equality and inclusion as disability.
Disability statistics, services and barriers. Using the de nition of disability adopted in this study, the WHO and the World Bank (WB) estimated that about 80% of the global 1 billion persons with disabilities (PwDs) currently live in low and middle-income countries (1). In most of these countries, especially on the African continent, community attitude towards disability and persons with disabilities are less than penchant and socioeconomic support and rehabilitation services are poor or non-existent (40,41). These challenges persisted, and the numbers of PwDs are increasing -both globally and in developing countries -due to population growth, man-made and natural disasters, war, accidents and aging.
In Ethiopia, data on the prevalence of disability is highly fragmented, or sometimes misleading and contradictory. For example, the 2007 Population and Housing Census reported the prevalence of disability to be 1.2%. Conversely, WHO estimates the national prevalence of disability in Ethiopia around 17% (1) The in-congruence between the two disability statistics could be due to several factors. The Census report was based on self-identi cation of household heads and members as being disabled or not, which generally underestimates prevalence rates due to stereotypes and stigma attached to disability (33). Furthermore, statistics on disability varies "according to the purpose and application of the data, the conception of disability, the aspects of disability examined, the de nitions, question design, reporting sources, data collection methods, and expectations of functioning" (1). Many (34,35) have reported skepticism about the Census statistics on disability in Ethiopia that could have been affected by "under-reporting, the way the Census questions were formulated, lack of disability awareness among enumerators and data managers and use of a narrow de nition of disability". In fact, there are reports that PwDs tend to avoid reporting their conditions due to negative community attitude towards disability and PwDs, which are prevalent in Ethiopia (34,35,39). In similar contexts, studies had identi ed institutional and cultural factors such as cost of treatment and stigma where there is signi cant underreporting (38). These reports show how Census reports actually underline the unreliability of statistics on PwDs in Ethiopia.
The signi cance of reliable disability statistics cannot be overstated. Having reliable statistics on disability informs better, policy making, planning and programming centered on inclusive and sustainable development. Data on the level of functioning in the population and the need for social and rehabilitation services will guide interventions to promote the equalization of opportunity and access to social resources such as education and health for PwDs. In recent years, national and global actors have used disability statistics to push the issue of disability to the center and front of policies, programming and service provision to ensure the effective inclusion of PwDs in development.
Empirical evidence on causes and impacts of disability on people are also relevant to prevention and rehabilitation policy making and programming. In Africa, studies identi ed the major causes of disability as communicable diseases, war, accidents, and inadequate prenatal and neonatal health-care services (36,37). It has been reiterated that disability is strongly linked to underlying and basic problems of poverty, poor nutrition and restricted access to basic services such as health and schooling. These factors are also relevant to the Ethiopian context where the list of causes/factors include low standard of living, malnutrition, natural and man-made disasters, accidents, infectious and non-infectious diseases, intermittent wars and violence.
The challenges PwDs face -as we have noted above -emanate from unfavorable relationships between their impairments and their social environment. The outcomes of disability have more to do with the way disability is perceived and treated by communities than the degree and type of physical impairments. African communities harbor misconceptions that deter the inclusion of PwDs and expose them to debilitating stigma. For instance, traditional/rural African communities attribute impairments to such factors as supernatural forces -curse, demonic possession or wrath of god for a family's mischiefs (42,43). This scenario has, in turn, led to PwDs poor mental health and well-being.
But PwDs do not form a homogenous group. Despite inconsistencies, the literature and statistical reports on disability in Ethiopia identify mobility, visual and hearing impairments as the top-three most prevalent forms of disabilities, in that order. The prevalence and types have also been reported to be similar in North Gondar Administrative Zone of the Amhara National Regional State (ANRS), Ethiopia.
In addition to difference in the type of impairments (e.g., physical, psycho-social, sensory, intellectual, neurological), PwDs live in a variety of contexts with different intersecting identities. This includes gender, age, education status, employment and economic status, and, geographical location. How these contexts and different characteristics intersect lead to very different experiences of disability [15]. In Ethiopia, for instance, about 46% of PwDs are women who carry the double burden of stigma, discrimination and other challenges due to patriarchy and misgivings about disability. Age is also a factor whereby children with disabilities (CwDs) are less likely to attend schooling or access healthcare services leaving them vulnerable to poverty and poor health that substantially reduce their quality of life [17,18]. As a group, without underplaying these variations within however, PwDs are still among the most disadvantaged segment of the society.
With the understanding that disability statistics and surveillance are necessary to assess and examine trends and conditions of PwDs to inform policies, strategies and programming, this study aims at lling the existing knowledge gap on disability, mainly severe disability, its prevalence, types and causes in Ethiopia, speci cally northwestern Ethiopia, and promote the effectiveness of services and interventions to meet the challenges of PwDs. It builds on the Dabat Health Surveillance System (DHSS) that the University of Gondar (UoG) established in 1995 to collect, organized and disseminate regular and up-to-date information on vital statistics -birth, death, migration and pregnancy -in Dabat district, northwestern Ethiopia. The justi cation for the DHSS lay in the relevance of up-to-date information on population to policy, planning and programming in all levels and sectors. But it included only one question on disability -"Are there members in the household with disability?" -which did not produce detailed information on degree, types and causes of disability.. It is against this backdrop that this study aims at assessing the prevalence, types and causes of severe disability in Dabat district employing a disability measurement tool adapted from WHO Disability Assessment Schedule (WHODAS

Study Area
This paper discusses the prevalence, types and major causes of severe disability in Dabat district, northwestern Ethiopia -the district wherefrom DHSS collects longitudinal study. Dabat town is the administrative center of Dabat district and it is located 60kms northwest of Gondar city. The Dabat district covers 1,199.15km 2 area and has an estimated population of 168,331 (male/female: 50%) (CSA 2013). It is administratively organized under 5 urban and 27 rural Kebeles -the smallest administrative unit in Ethiopiawith altitudes ranging between 1000 and 3000 meters above sea level (Dabat Rural Project Statistics, 2015).
The district has 29 health posts, 3 health stations and 2 health centers. UoG established DHSS in 1995 to collect demographic, social and health data in 13 Kebeles (9 rural and 4 urban) of 32 Kebeles of the district. According to the 2014 Re-census Baseline Survey, DHSS collects data on 17,000 households, with 72,000 inhabitants, semiannually.

Study Design
To identify the prevalence, types and causes of severe disability at Dabat district, the study used a communitybased cross-sectional design. Data were collected by enumerators and supervisors who visited all households in the DHSS catchment area and interviewed household heads between January and June 2018. This survey makes part of the add-on cross-sectional survey launched in 2016 to speci cally address the gap in the DHSS data on the prevalence, types and causes of disability.

Study Population
This study covered members and permanent residents of 17,000 households in 13 Kebeles of Dabat district. Thirtythree trained and experienced enumerators, overseen by nine supervisors, interviewed household heads to collect data on incidences, types and causes of disability among all household members and permanent residents.
Though this study collected data on the same population as the DHSS, it uniquely focused on collecting data on disability through an adapted measurement tool from WHODAS 2.0 and ICF.

Sample size and sampling procedures
Theoretically, this study included all members and permanent residents of 17,000 households as well as those on the streets in the 13 Kebeles within the Dabat district. Actually, there were no street dwellers in the district and the study excluded those who were not permanent residents (less than 6 months, as per the CSA de nition of 'household members') at the time of data collection.

Data collection: tools and procedure
The survey employed a structured and pre-tested questionnaire adapted from WHODAS-2.0and ICF. WHODAS-2.0 is a standardized and validated tool to assess health and disability status both at individual and community levels, and across cultures. As a generic tool, it assesses disability induced limitations on activity and participation both in community and clinical set-up. Furthermore, it has been fruitfully deployed to measure the effectiveness of disability interventions. The study applied WHODAS 2.0 when interviewing respondents above the age of 18. The survey questionnaire speci cally adapted items from WHODAS-2.0 that measure the six domains of disability: learning and intellectual, mobility, self-care, getting along, life activities and participation.
The ICF -children and youth version (ICF-CY) -was used to assess disability, activity limitations and other outcomes among population under 18 years of age.
The questionnaire was rst designed in English and then translated to Amharic -the o cial language of Ethiopia and spoken by residents of the Dabat district. The adaptation and validation of the questionnaire to the local context involved piloting, which was conducted in one Kebele -later excluded from the survey -and provided information to assess the relevance and appropriateness of its items and their wordings and sequencing. The pilot involved all enumerators, supervisors -who were trained for 5 days on research methods, interviewing and the questionnaire -and researchers who later engaged in critical re ection and integration of the piloting results to make the questionnaire, and the items wording, sequencing, etc., as relevant and appropriate as possible.
Furthermore, the research team drew on their research and programing experiences on disability, community-based rehabilitation and inclusion in Dabat district to ensure the culture-appropriateness and scienti c validity of the tool.
Enumerators, after contacting household heads and informing them above the purpose of the study, asked them to provide information on household members with regard to impairment and/or disability by reading out a list of possible cases including hard of hearing or total loss, visual impairment or blind, speech impairment, loss of senses or limbs, paralysis, diagnoses insanity, etc. Due to stereotypes and stigma attached to disability, enumerators asked household heads to identify member with 'functional limitations' rather than directly asking them the question, "Is there anyone with disability in the household?" Enumerators applied items from WHODAS 2.0 or ICF-CY depending on the age of the individual disability information were provided for. The information gathered on disability through the community survey did not necessarily account for mild or minor types of impairments that household heads were not aware of and/or decided to be insigni cant. As a result, the data collected and reported in this study were on complete/severe visual, auditory, motor or intellectual impairments sustained by household members.
The use of self-reporting rather than clinical diagnosis of household members to collect data on disability is identi ed as the limitation of this study. Whenever new cases of disability were identi ed during data collection, they were referred and/or linked to local Community-based Rehabilitation (CBR) eldworkers.

Data analysis
Data were entered into the Household Registration System (v-2.1) and the survey data was entered into and analyzed using STATA (v.12) software. During data cleaning and organization, few items that were adapted from WHODAS 2.0 or ICF-CY but returned with questionable and unreliable responses were excluded to ensure the quality of the report. Descriptive statistics -means, percentages, standard deviation, etc. -were employed to describe the characteristics of the study population. Table and gures were used to present aggregated and disaggregated, as appropriate, data. Binary logistic regression was tted to test and identify factors signi cantly associated with disability. The bivariate analysis was carried out, and variables with p-values of <0.2 were entered into the multivariable logistic regression model. The crude odds ratio (COR) and nally, a Logistic regression models were constructed on predictors of disability among the study population. The adjusted odds ratios with the corresponding 95% Con dence Interval (CI) were used to show the strength of association between independent variables (age, sex, place of residence, religious a liation, educational attainment, occupation marital status) and disability. A P value of <0.05 was considered statistically signi cant.

Results
Respondents' Socio-demographic characteristics: As (  Table 2 and Figure 1 reveal, the prevalence of disability increased with increase in age, and reached the highest level among persons of 65 and above years old (10.18%). Conversely, without alluding causal attribution, the prevalence of disability declined with increase in educational status. For instance, the highest proportion of disability -971 (63.2%%)was recorded among those who cannot read and write, and the difference is stark when we compare it to those with Grade 11/12 (0.72%) and Grade 12-plus (0.98%) education (Table 2 and 3) ( gure 1).
More than 8% of households had a member with disability (1411 in 17,000 households). Table 1 also reports that the prevalence of disability was high among those with 'separated' (63 (10.21%)) and widowed (176 (9.33%)) marital statuses as well as the unemployed (5.42%). There was marginal variation between urban (2.08%) and rural (2.15%) Kebeles of Dabat district in terms of the number of persons affected by disability (Table 2).
Eight percent of the PwDs had more than one type of disability. In terms of single-disabilities, 51% had visual impairment, while 22.3% had mobility impairment and 22.3% were hard of hearing. there was a marked increase in hearing and visual impairments among ageing respondents (Table 3 & Figure 2, 3).
On the other hand, only 253 (16.5%) PwDs attended any form of formal education. In fact, as educational level increases, the proportion of PwDs who attended higher grade-levels fell signi cantly with less than 1% completing secondary education (Table 4).
The survey identi ed the most frequent causes of disability as illness (36.93%), followed by accidents (17.81%), unknown (11.57%), congenital (10.86%) and ageing (5.2%). In other words, more than 83% of the reported immediate causes of impairments could have been prevented from causing functional limitations if they were identi ed and treated early (Table 6 and Figure 4). This argument is congruent to studies in other African and Latin American settings where families and persons with impairments are more likely to avoid early treatment (44), seek the services of traditional healers than modern medicine (40) or suffer from the inadequacy of primary health care services for screening and treatment (30).
Factors associated with disability: Multivariate logistic regression model was constructed to predict the occurrence of disability in the study population from data on respondents' age, educational attainment and marital status.

Discussion
This study found that though the general prevalence of disability was 2.14%, it signi cantly affects older persons in the community. The proportion of households with members with disability i.e., 8.3% was high when compared to previous studies on the study area. As questionable they were in reliability, representativeness or timeliness, CSA and Chala reported 1.82% and 1.09%, respectively, disability prevalence for Dabat district, which were signi cantly lower than the 2.14% reported in this study. However, 2.14% is still signi cantly lower than an estimated 17.6% national disability prevalence rate in Ethiopia that WHO and WB (2011) reported.
There are several possible explanations why different sources provide different reports on the prevalence of disability. For one, WHO conceptualize disability broadly and include various factors and types, such chronic illnesses as diabetics as well as moderate and minor impairments, within the rubric of 'disability.' However, limited resources for rehabilitation services and supporting PwDs in developing countries may have dictated a narrower operationalization of disability. In other words, pragmatic conceptualization of disability would set realistic expectations and recommendations for policy formulation and social work practice. For another, this study relied on responses household heads provided to estimate the prevalence of disability in the district. By its design, the quality of data could be affected by respondents' lack of knowledge on the nature and degree of one's and household members' disability, and negative attitude towards disability and PwDs, among other. Hence, lack of understanding on disability, stigma attached to disability in the district and respondents' inability to identify all types and degrees of disability may have lowered reports on disability in the study area.
In terms of type of disability, this study reported similar results as the same team's previous study in Dabat district.
Both studies found that visual impairment affects the highest proportion of PwDs; but while the current study found that mobility and hearing impairments are second and third in the number of persons they affect, their relative proportion was reversed three years earlier.
A small fraction of PwDs attended formal education; and more importantly, with advance in educational level, their progression to higher grades decreases signi cantly. This result was consistent with the Handicap International report that indicated a small fraction (3%) of the estimated 2.4 to 4.8 million school-age CwDs went to school.
CwDs' low school attendance and participation are due to underlaying reasons that include stigma against CwDs, school inaccessibly, rigid teaching practice, poorly trained teachers to accommodate children with special needs and lack of adaptive hearing resources. Another study in South Africa documented how PwDs and their families live with stigma and social exclusion that affect their personal development and community participation. Other studies also report CwDs are less likely to attend schooling and access health care, and they are more vulnerable to poverty which may result in substantial decline in their quality of life.
This study reported that the majority of disabilities were due to modi able illnesses, injuries, etc.; and the causes and impacts of disability are multifaceted. Concurrent to these ndings, studies have documented that the majority of disabilities in Africa result from illness, injury, and accidents -causes that are preventable (5,45,46). In Ethiopia, diseases like measles, poliomyelitis, etc. as well as accidents and injuries cause illnesses to many people that -if identi ed and treated early or on time -may not necessarily amount to serious functional and activity limitations. This does not necessary amount to disability entirely avoided for individuals concerned -but that if trachoma, measles, illnesses, etc. could be treated early and prevented from causing serious damage to the body organism thereby reducing the severity impairment and/or enabling functionality. This scenario is not unique to Ethiopia as, for instance, a study on nine Latin American countries categorized 80% of visual impairments as avoidable with early detect and treatment [30]. Most disabilities and their causes can be linked to poverty, and restricted access to basic services. As this study reported, most PwDs were poor -both in urban and rural areasand a signi cant majority of them were involved in small-scale farming with strained livelihoods. Besides, inadequate health care services, poor health literacy and health seeking behavior could were also contribute to the disability problem on the continent, which is the same in the Dabat district.
Disability causes serious bad health outcomes, and the provision of long-term care for PwDs in resource limited settings is a major public health problem. Disability is especially a concern for developing countries where prevention approach and health service services are inadequate or very scarce (40,47). Moreover, stigma and stereotypes limit educational and employment opportunities for PwDs thereby creating dependency and exclusion for PwDs and this study underlines these facts.
When it comes to women with disabilities, they carry the double burden of stigma and discrimination and economic problem due to their gender-roles and disability. Disability and poverty exacerbate their socioeconomic wellbeing and their general quality of life.

Strengths and Limitations
The study aimed at exploring the prevalence, types and causes of severe disability in low resource setting i.e., Dabat district, Northwestern Ethiopia. Literature review revealed that there is dearth of scienti c, empirical information to understand the trend and changes on the prevalence and magnitude of disability and its effects in Ethiopia. It has also established the evident fact that there is low community and governmental attention on disability and the empowerment of PwDs in developing countries such as Ethiopia. Therefore, by identifying contributing factors to disability and its prevalence and impacts on peoples' lives in northwest Ethiopia, this study will promote disability awareness and inclusive society as well as inform the agenda for disability research, policy and practice in the district and beyond. To this end, large sample was drawn to collect representative empirical data about the population.
The two main limitations of this study emanate from drawing on the reports of household heads to identify PwDs at the household level, and the focus on severe/extreme functional limitations due to impairments rather than the whole range of impairments/disabilities. The researchers attempted to address the rst limitation to a certain extent by listing possible types of impairments/disabilities to help household heads frame and/or prompt their responses. Due to the large sample size i.e., 17,000 households the researchers could not employ disability screening as a method of data collection. Consequently, the reported disability prevalence in the district was signi cantly lower than the WHO estimation for low-and middle-income countries. But we have to take into account the fact that household heads reported visible and severe impairment.

Conclusion
This study found the proportion of reported disability per household to be high. Vision and mobility disability were the most frequently observed types of disability, and most immediate causes of disability were modi able. This conclusion, however, does not apply to underlying structural factors to disability such as poverty, aging, inaccessibility of health services, etc. that require programmatic interventions at national, regional and local levels.
On the other hand, this study indicated that the prevalence of disability increased among the ageing-groups. As educational level increase, the proportion of PwDs attending school decreases, and most PwDs do not complete secondary education. Similarly, a small proportion of PwDs were government employees, and the majority were engaged in farming.
Early intervention and appropriate medication will prevent the burden of disability. To this end, appropriate attention should be given to improve quality of healthcare facilities and services to reduce the burden of disability on families, communities and the health care system. Moreover, community awareness creation and enhancing inclusive education will improve education opportunity for PwDs. With improve in educational level and skills, PwDs will be able to nd and/or create jobs thereby promoting their independence and better quality of life. Services and the College of Medicine and Health Sciences -were submitted to Dabat District Administrator. Letter of information describing the purpose and objectives of the study, and written informed consent was obtained from the study participant (household head), and consent forms were given or read to study participants prior to the commencement of the interview. Participants were granted the right to terminate their participation in the study at any point if and when they chose to.
Once in the eld, researchers and DHSS staff put their maximum efforts in assuring all ethical and scienti c standards are maintained. Participants' privacy and con dentiality were respected during and after interviews.
Once data was collected, each questionnaire was coded, and personal identi ers were removed to maintain anonymity and con dentiality of participants. All data was entered and saved on a computer with a strong password which was only accessible by the research team. Participants who reported disability or were found ill during data collection were either linked to UoG-CBR eldworkers or health centers for better care.
Consent for publication: "Not applicable" Availability of data and material: Since the data presented in this report are the part of the large DHSS survey data, All authors prepared the draft manuscript, read and approved the nal manuscript.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.