Research Papers

Sophie Mitra, Usha Sambamoorthi: Disability Estimates in India: What the Census and NSS tell us

In India, official disability prevalence rates are estimated at about 2 percent. The prevalence estimates vary significantly across sources. According to the 58th round of the NSS, there were 18.5 million persons with disabilities in 2002 compared with 21.9 million reported by the Census of 2001. This translates to a 20 percent difference in the prevalence estimates, which can be considered as significant.

In this context, it is important understand the sources of differences in prevalence estimates from the Census of India of 2001 and the 58th round of the NSS of 2001. We first present as background different ways to define and measure disability. We then review difference in the NSS and the census in disability type. Finally, we conclude and offer recommendations for the design of disability questions in the census and the NSS.

Conceptual Models of Disability

Disability is a multifaceted and complex concept to define. Different conceptual models have been created to define disability {Altman 2001}. Over the years, the definition of disability has been evolving and can be classified into three different perspectives: (i) the medical model, (ii) the social model, and (iii) the ICF model.

The medical model: The medical model looks at disability as a problem that is directly caused by a disease, an injury or other health condition, and requires medical care in the from of treatment and rehabilitation. Further, it considers disability a health problem or abnormality that is intrinsic to an individual’s body and mind. Any individual with impairment is considered disabled, whether or not the person experiences limitations in his or her life activities. Under this model, for example, individuals with any brain injury or condition such as multiple sclerosis are considered disabled.

The social model: the social model considers disability purely as a social construct and a human rights issue. Under this model, even though impairments are at the individual level, disability is the direct result of society’s failure to account for the needs of persons with impairments. Disability is not the attribute of the individual; rather it is created by the social environment and needs social change. In the social model, disability is generally understood as the result of social oppression, this oppression can start in the form of poverty and later on lead to disability.

The ICF model: The World Health Organisation (WHO) developed the International Classification of Impairments, Disabilities and Handicaps in the early 1980s, which was recently revised and renamed the ‘International Classification of Functioning, Disability and Health (ICF) [WHO 2001]’. Conceptually, ICF is presented as an integration of the medical and the social models (2001:20): “ICF attempts to achieve a synthesis, in order to provide a coherent view of different perspectives of health from a biological, individual and social perspective”. The ICF model is sometimes termed the “biopsychosocial” model of disability [Bickensack, Chatterji, Badley and Ustun 1999].

This model starts with a health condition that gives rise to impairments, and then activity limitations and participation restrictions. Impairments are problems in body function or structure as a significant deviation or loss. An activity is the execution of a task or action by an individual, while participation is the lived experience of people in the actual context in which they live. Activity and participation domains include among others, learning and applying knowledge, mobility, self-care, education, remunerative employment, economic self-sufficiency. Functioning and disability are two umbrella terms, one being the mirror image of the other. Functioning covers body functions and structures, activities and participation, while disability includes impairments, activity limitations and participation restrictions.

Measures of Disability

As can be anticipated, there are different ways to measure disability that correspond to different conceptual definitions of disability. We describe below three disability measures that have been commonly used in applied disability research: impairment, functional limitations and activity limitation measures.

Impairment measures of disability focus on the presence of impairment intrinsic to the individual. For example, individuals may be queried about impairments that might include blindness, deafness, mental retardation, stammering and stuttering, complete or partial paralysis.

Functional limitations refer to difficulties experienced with particular bodily functions such as seeing, walking, hearing, speaking, climbing stairs, lifting and carrying, irrespective of whether the individual has an impairment or not. The above two measures of disability, impairments and functional limitations are consistent with the medical model definition of disability.

Activity limitations are limitations in activities of daily living (ADL) such as bathing or dressing. Activity limitations may also include participation limitation in major life activities such as going outside the home, work or housework for working age persons, and school or play for children. This measure may be considered to capture disability as per the social model. Finally, it should be noted that, if disability is defined as per the ICF model where it is an umbrella term for impairments, activity limitations and participation restrictions, then any or all of the above measures may be used.

Developed countries typically used disability screens that assess activity limitations, whereas developing countries tend to use impairment screen [Mitra 2005]. This makes cross country comparison of disability prevalence nearly impossible. Activity limitation screens generally lead to higher rates of reported disability than impairment screens. Indeed, individuals are more likely to identify activity restrictions because they immediately connect with daily experience; whereas the terms used to identify impairments may not be easily understood, and their nomenclature may be unknown. For instance, in Chile disability prevalence was estimated at 2.2 percent based on an impairment measure in the Census of 2002, compared to 21.7 percent based on an activity limitation measure in the Quality of Life and Health Survey in 2000 [Mitra forthcoming].

Indian Disability Estimates

The two main official source of nationwide disability statistics are the NSS and the census. In the NSS, a disability schedule is administrated every 11 years, with the most recent one fielded in 2002. In the census, a disability question was included in selected years, recently in 2001.

The census of India and the NSS have different sample design. The census is an enumeration of the entire population of India while the NSS uses a nationally representative stratified sample. Although some of the differences in prevalence estimates of disability could be due to the differences in study design, different definitions of disability may also contribute to difference in estimates. 

Definitions of Disability: The Census of 2001 does not have a general definition of disability. Instead, a question on disability type was included in the population enumeration section (question 15) as follows: “If the person is physically/ mentally disabled, give appropriate code numbers from the list below: in seeing, in speech, in hearing in movement, mental”. Each of these disability types is defined in detail in an instructions manual for census enumerators and will be reviewed later in this article. What is notable is that there is no general definition or screen in the census for disability: Census enumerators coded a person as disabled if the person had any of the listed disability types.

In schedule zero of the 58th round of the NSS, fielded in 2002 some broad information about the households (e.g. housing conditions, ability) was collected during the household listing. This information was required mainly to identify and develop a frame for selection of households with persons with disabilities. A person is considered disabled “if the persons has restrictions or lack of abilities to perform an activity in the manner or within the range considered normal for a human being”. This general definition of disability in the NSS acts as screen leading to disability type question. Disability is thus defined overall as an activity limitation in the NSS.

In general, both sources classified type of disability in the following areas locomotor, visual, hearing, speech and mental. Despite similar sequence in the identification of disability types, there are substantial variations in prevalence estimates of disability types across the two data sources (Table 1).

Source: Census of India 2001 and NSS 58th Round 2002 as reported in Bhanushali (2005)

As shown in the table, the prevalence estimates of mental disability among persons with disabilities across both the census (10.3 percent) and the NSS (11.3 percent) are quite similar. This result is surprising given that the NSS and the census use different definitions for mental disability. In both sources, comprehension appropriate to age is used to capture persons with mental retardation. The definition of mental disability is also based on a general activities of daily living limitation in the census (“depend on her/his family members for performing daily routine”) while it refers to several specific functional and activity of daily living limitations in the NSS (“activities of communication (speech), self-care (cleaning of teeth, wearing clothes, taking bath, taking food, personal hygiene, etc), home living (doing some household chores and social skills”).

We observed substantial differences in estimates for other types of disability. In general, the NSS rates of disability were higher for hearing, and locomotor, while the census disability rates were higher for visual disabilities. The largest difference in prevalence estimated between the two sources is for visual disability. Visual disability prevalence is estimated at 10.6 million in the census (48.6 percent) and 2.8 million in the NSS 15.2 percent). NSS rates were higher than the census rates for locomotor disability by about 30 percent (57.5 percent versus 27.9 percent). For speech and hearing, the rates were 22.7 percent versus 7.5 percent and 16.6 percent versus 5.8 percent respectively.

Definitional differences

Part of these differences in prevalence estimated by disability type may be due to differences in definitions between the two sources. Therefore, we review the definitions of disability types used in the census and the NSS. These definitions are presented in Table 2.

Source: K Bhanushali (2005) and Government of India (2003a)

Visual: According to the Census 2001 10.6 million persons are visually disabled compared to 2.8 million persons in the NSS. In the census, individuals are considered visually disabled, if they have blurred vision and have had no occasion to test their vision, or if they have vision in one eye. The NSS has a visual functional test by asking whether the person can perceive light using both eyes, and can count fingers at a distance of three meters using spectacles (if the person wears spectacles). The NSS is silent on how persons who never had their vision tested should be treated. For visual disability, the census has a wider definition of visual disability” this may have contributed to higher rates of visual disability in the Census of 2001.

Hearing: Among persons with disabilities, 1.3 million persons have hearing disabilities in the census compared with 3.06 in the NSS. Part if these differences could be due to the stark contrast between the census and the NSS definitions hearing disability. In the census, persons who can hear only with one ear are counted as disabled, while they are considered as non-disabled in the NSS- this would argue for lower estimates in the NSS. In the Census 2001, a person who is able to hear, using hearing aid will not be considered as disabled. However, the NSS refers to a person’s hearing ability without the use of hearing aid. This difference in how hearing aid use is treated in the definition may explain the higher hearing disability prevalence estimates in the NSS compared to the census. 

Speech: Prevalence of speech disability is somewhat lower in the census (1.6 million persons) than in the NSS (2.2 million persons). The NSS has indeed a broader definition of speech disability than the census. In the NSS, persons who cannot speak, speak limited words or with loss of voice, or with stammering voice are classified as being speech disabled. In the census, speech disability covers persons who are dumb, or who cannot be understood. A person who stammers but whose speech is comprehensible is not considered as speech disabled.

Locomotor: The NSS includes persons with paralysis, amputation, deformity, dysfunction of joints and dwarfism as having locomotor disability. Persons with dwarfism or deformity are considered disabled even if they are not limited in the movement of their body limbs. In the census, locomotor disability covers the absences of all toes, all fingers, deformity, and the inability to move without aid, the inability to lift and carry any small article. NSS has a broader definition of locomotor disability than the census because it lists more impairments (e.g. dwarfism, paralysis) than the census does. This difference in the definition of locomotor disability is consistent with the much higher rate of locomotor disability prevalence in the NSS (57.5 percent of persons with disabilities, 10.6 million) than in the census (27.9 percent of persons with disabilities, 6.1 million).

As noted above, three disability type definitions (hearing, speech and locomotor) are more inclusive in the NSS than in the census. The reverse is true for visual disability where the census definition is wider than that in the NSS. The vast differences in prevalence estimates between the census and the NSS illustrate how essential it is to document the disability definition used, sources and caveats and to exercise great caution while comparing prevalence estimates or using statistics related to disability.

Conclusion

The census and the NSS are two essential data sources that may be used in India to understand the lives of persons with disabilities. It appears that the overall disability prevalence estimates in the census and the NSS are clearly not comparable There are difficulties in comparing the estimates because the census does not have an overall definition of disability while the NSS does. The census overall disability prevalence is the sum of prevalence estimates for the five disability types.

In addition, in both sources, the current definitions of disability types seem to mix activity limitations, functional limitations and impairments. For instance, in the NSS activity limitation is used for the general definition of disability, and the definitions of visual and metal disabilities, functional limitation is used for hearing and speech disability, while impairments are used to define locomotor disability. In the census visual and hearing disabilities are defined as functional limitations, movement disability is a mix of functional limitation (eg, inability to mover or lift any small article) and impairment (eg, lack of limb), and mental disability is defined as a combination of functional limitation (inability to understand) and activity limitation (self-care limitation). As a result, it is unsure what aspects of disability are captured by the census and NSS current disability definitions.

There are also inconsistencies in disability types in the census and the NSS with regard to whether the limitation applies to a situation where and assistive device is used. For instance, in the census, for locomotor disability, it refers to a person’s limitation without using aid while for hearing disability; it refers to a limitation experienced despite the use of hearing aid. In the NSS, the definition of visual disability refers to a person using spectacles or contact lenses, while the definition of hearing disability considers a person’s ability without using a hearing aid. Such inconsistencies certainly make it difficult for field staff to collect the data and for researchers to interpret the results.

Recommendations

The measurement of disability within surveys and the census is a very complicated task and there is no simple recommendation on a best approach. Ideally, one would need to include in survey instrument several disability measures that referent disability definitions in order to give researches the ability to pick the measure that is best suited to their research questions. For instance, research focused on disability prevention would use a medical model of disability and would need measures of impairments, while research focused on social participation would use measures of activity limitations. We describe below recommendations for the agencies involved in the design of the NSS and the census to improve on the definitions used.

First, in both the census and the NSS there should be separate questions getting at impairments, functional and activity limitations, which would allow researchers to use the data based on different conceptual definitions of disability. While having consistent definitions in the census and the NSS would facilitate the comparison of prevalence estimates across the two sources, effort should be targeted toward improving the definitions of disability types within each data source.

In the census, the term “insane” is used to define mentally illness: this term is subjective and using it to classify persons seems inadequate. In additional, in the census, the definition of visual disability needs to be narrowed down so that persons who have mild visual limitations are not counted as disabled. More importantly, the five disability types that the census covers seem limited. Some conditions that may be disabling such as autism, epilepsy, dyslexia or depression may not be captured under the current disability type definitions. It is also not known where persons with multiple disabilities fit among existing disability types.

There is scope for improving data collection on disability in the NSS as well. Disability schedules are administrated every 10 years or so and do not allow for the analysis of trends, nor the evaluation of the impact in the lives of persons with disabilities of policies such as the 1995 Persons with Disability Act. NSS could include disability questions in non-disability schedules. For instance, in the round of the NSS that collect employment data for large samples (eg, 60th round), fewer than a handful of disability questions would facilitate the comparison of the labour market experiences of persons with and with out disabilities.

Source: Sophie Mitra, Usha Sambamoorthi, Disability Estimates in India: What the Census and NSS tell us, Economic and Political Weekly, September 23, 2006.