The effectiveness of community based rehabilitation programs: An impact evaluation of a quasi-randomised trial

Abstract

Background Community-based rehabilitation (CBR) programmes have been described as highly effective means of promoting the rights and opportunities of persons with disabilities (PwD). Although CBR is often the main way in which PwD in low-income and middle-income countries access rehabilitation services, there is little literature providing rigorous evaluation of their impact on people's well-being.
Methods Data were collected in the Mandya and Ramanagar districts (Karnataka state, India), between December 2009 and May 2010. In total 2540 PwD were interviewed using stratified random sampling: 1919 CBR beneficiaries (who joined the programme between 1997 and 2009) and 621 persons who were living in villages not covered by the programme. We controlled for the systematic differences between people joining and not joining the programme using the propensity score matching method controlling for covariates at individual and village level. We evaluated the impact of the programme on the subgroups of PwD who are disadvantaged on the dimensions of interest: access to pensions, use of aid appliances, access to paid jobs and improvement in personal-practical autonomy after 4 and 7 years of joining the CBR.
Results We observed a positive and significant impact of the programme on access to services, rights and opportunities of PwD. The results indicate that compared with the control group access to pensions and allowances, aid appliances, access to paid jobs and personal-practical autonomy increased by 29.7%, 9.4%, 12.3% and 36.2%, respectively, after 7 years.
Conclusions The CBR programme analysed has a positive impact on access to services and the well-being of PwD who are particularly deprived on outcomes of interest.

Full text

1
The Effectiveness of Community based rehabilitation programs: An impact
evaluation of a quasi-randomised trial
Vincenzo Mauro*, Mario Biggeri*, Sunil Deepak**, Jean-Francois Trani°
*Dept. of Economics, University of Florence, **Associazione Italiana Amici di Raoul Follereau
(AIFO), °Brown School of Social Work and Institute of Public Health, Washington University in St
Louis.
Corresponding author: Jean-Francois Trani Assistant professor, Brown School and Institute of
Public Health, Washington University in St. Louis, Campus Box 1196, Goldfarb Hall, Room 243,
One Brookings Drive, St. Louis, MO 63130 [o] 314.935.9277 [c] 314.412.0077 [e]
jtrani@wustl.edu
Keywords: Capability Approach, Community Based Rehabilitation, Disability, Quasi-experiment.
Word count: 3223
ABSTRACT
Background Community based rehabilitation (CBR) programs have been described as highly
effective means of promoting the rights and opportunities of persons with disabilities (PwD).
Although CBR are often the main way in which PwD in low and middle income countries access
rehabilitation services, there is little literature providing rigorous evaluation of their impact on
people’s well-being.
Methods Data was collected in the Mandya and Ramanagar Districts (Karnataka State, India),
between December 2009 and May 2010. 2,540 PwD were interviewed using stratified random
sampling: 1,919 CBR beneficiaries (who joined the program between 1997 and 2009) and 621
persons who were living in villages not covered by the program. We controlled for the systematic
differences between people joining and not joining the program using the propensity score matching
(PSM) method controlling for covariates both at individual and village level. We evaluated the
impact of the program on the subgroups of PwD who are disadvantaged on the dimensions of

2
interest: access to pensions, use of aid appliances, access to paid jobs and improvement in personal-
practical autonomy after four and seven years of joining the CBR.
Results We observed a positive and significant impact of the program on access to services, rights
and opportunities of PwD. The results indicate that compared to the control group access to
pensions and allowances, aid appliances, access to paid jobs and personal-practical autonomy
increased by 29.7%, 9.4%, 12.3% and 36.2% respectively after seven years.
Conclusions The CBR program analysed has a positive impact on access to services and the well-
being of PwD that are particularly deprived on outcomes of interest.

3
INTRODUCTION
Community based Rehabilitation (CBR) is defined in the 2004 position paper - adopted by World
health Organisation (WHO), International Labour Organisation (ILO) and the United Nations
Educational, Scientific and Cultural Organization (UNESCO)- as “a strategy within general
community development for the rehabilitation, equalization of opportunities and social inclusion of
all people with disabilities” based on a participatory process involving people with disabilities
(PwD), their families, organizations and communities [1]. The new CBR guidelines [2], are based
on the principles of the UN Convention on the Rights of Persons with Disabilities (CRPD) [3], as
well as on empowerment, self-advocacy and sustainability.
The CBR matrix consists of five components: health, education, work, empowerment and social
participation. Each component includes five elements [2]. Each CBR program is supposed to select
components fitting local needs and priorities and considering available resources.
Issues relating to CBR feature twice among the top ten most relevant and challenging priorities (out
of 83) for future research on disability [4]. However, although CBR is considered the most cost-
effective approach for improving the wellbeing of PwD [2-5], and for fostering their participation in
the community and society at large [6, 7], we lack evidence of the impact of CBR programs. This
jeopardises the development and implementation of effective rehabilitation policies and
programmes [7, 8, 9, 10-12]. Sharma (2007) carried out an analysis of evaluation reports of 22 CBR
programmes in 14 countries: she found that studies focused mainly on accessibility, program reach,
identification of needs and specific outcome.
CBR research in low-income countries has recently increased [9]: Alavi and Kuper (2010)
identified 51 studies evaluating the impact of rehabilitation programs in Africa, Asia and Latin
America. But few studies were focusing on the program impact. The type of study design was
quasi-experimental in 19 studies, randomised controlled trials (16) and case-series (14). Only 2
studies had a comparison group, one being the present study.

4
Our study contributes to filling this gap by measuring the impact of CBR programs on PwDs that
are particularly deprived on four core dimensions of interest (access to pensions, use of aid
appliances, access to paid jobs and improvement in personal-practical autonomy) after four and
seven years of joining the CBR. The field study is built on a potential outcomes framework [13, 14]
using a large-scale household case-control study.
Our theoretical framework and the outcomes of interest are based on the CBR guidelines [2], the
CRPD [3], and the capability approach [15, 16]. CBR’s strategy “promotes the rights of people with
disabilities to live as equal citizens within the community, to enjoy health and well-being, to
participate fully in educational, social, cultural, religious, economic and political activities” [2, p.4].
As a result, CBR programs aim at expanding capabilities of PwD (capacities and opportunities) and
agency to live a life they value. [17, 18]
METHODS
Study location and population
Our study analyses a CBR program implemented in the Mandya and Ramanagar districts of
Karnataka State, India, with a total population of over 2.2 million people (figure 1). From 1997
onwards, this program reached 2,045 villages and 22,000 PwD. We defined disability following the
form 1 instructions of the WHO CBR Manual [19].
Figure 1. Map of the districts under research (approximately here)
The CBR program is managed by two non-governmental organisations (MOB, Maria Olivia
Bonaldo and SRMAB, Sri Raman Maharishi Academy for Blind) and reached PwD from different

5
age groups and with different types of disabilities in rural, semi urban and urban areas. The
implementation process is carried out by trained community CBR workers supported by a
supervisor and a project coordinator with a high involvement of PwD themselves through Self-Help
Groups (SHGs).
Randomisation
We measured the impact of CBR by comparing PwD in three neighbouring districts (figure 1). The
first two, Mandya and Ramanagar, were covered by the intervention, while the third –Mysore-
housed the control sample. The choice of the Mysore area was due to its extreme similarity with the
intervention areas. Furthermore, the villages in the control area were selected for their proximity
with Mandya district. Two big cities, Bangalore and Mysore, are close to both the treatment and the
control areas. As a result, a similar range of referral services and specialised institutions in health,
education, social welfare and occupation are accessible to participants and controls. In particular,
the two cities provide access to vocational training, specialized health service and technical
appliances. The three districts also have few local services, such as special schools for children with
hearing and speech impairments and a vocational training centre for children with visual
impairment.
Villages and PwD were sampled in all areas following a one-stage cluster sample design, using the
villages as first-stage units. Three variables were chosen for stratification: the geographical area
(using sub-districts as strata), the size of the village and the year the CBR program was introduced
(for areas covered by the program only). In selected villages of the intervention areas, all PwD
participating in the CBR program were interviewed. In control areas, all PwD were interviewed.
PwD were identified using existing data in participating villages and a 9 items screening tool [20]
tool in control areas. Our final dataset consisted of 2,540 respondents including 1,919 participants
in 237 villages and 621 controls in 28 villages not yet covered. The average response rate was
91.8%.

6
A comparison between respondents and non-respondents showed no significant differences with
respect to the available characteristics. We therefore assumed that unobserved data are missing
completely at random and the estimated effects are unbiased [21]. The PSM methodology
corroborates this assumption: our results remained unbiased under the less restrictive hypothesis
that missingness and the unit’s characteristics are conditionally independent given a set of variables.
Figure 2. Randomisation process (approximately here)
Procedures
The present trial is the first component of a larger study composed of: (i) an emancipatory research
focusing on the mapping of barriers faced by PwD in the communities, their strategies for
overcoming these barriers and the role played by the CBR program; and (ii) a participatory research
to provide in-depth understandings of the key issues identified in the quantitative and emancipatory
researches and elaborate recommendation to improve the program. Therefore, this research program
is compliant with article 30 of the CRPD.
The study benefitted from the input of an Advisory and Scientific committee (composed of
academics, PwD representative of Disabled Peoples Organizations in Mandya and Ramanagara,
experts from AIFO, local NGOs and social worker representatives). The theoretical framework and
survey instruments were elaborated following eight months of desk research, several in depth
interviews and various focus group discussions. The study instrument is composed of four sections
including socio-demographic characteristics, activity limitations, a participation scale, and objective
and subjective dimensions of quality of life. Ethical clearances were obtained from both UCL and
AIFO.

7
A two weeks training programme for a team of 5 supervisors and 35 enumerators covered
theoretical, practical and ethical issues, including a brief pilot survey, during November 2009. The
survey was administered between December 2009 and May 2010.
Variables
Selected outcomes of interest were identified based on specific CBR activities made available to
participants.[2] We focused on three objective outcomes (i.e., access to pension and allowances,
access to mobility and assistive device, and access to paid job) and one subjective outcome based
on self-perceived personal practical autonomy.
These outcomes reflect two of the CBR matrix components: livelihoods, health and rehabilitation.
They constitute required additional resources linked to disability – what Sen called the “conversion
handicap” [17]. PwD need more resources but also face social and environmental barriers that can
directly or indirectly influence the conversion of personal resources into the freedom to leading
lives they value.[18, 20]
The outcomes selected explore the contribution of the CBR to expanded functionings and
capabilities for participants. The opportunity to have a paid job is central to PwD as it provides
financial autonomy as well as family and community recognition, enhancing social inclusion and
self-esteem [11, 23]. Access to pensions and allowances similarly advances autonomy and social
recognition. Access to aid appliances is central to increase autonomy for people with mobility
restriction. The perceived impact of the CBR on individual autonomy is a good indicator of self-
esteem and well-being.[11]
Statistical analysis

8
The framework to identify causal effects is based on potential outcomes. This framework is rooted
in the statistical work on randomized experiment by Fisher [21] and Neyman [22], later extended by
Rubin [14, 23] and others to quasi-randomized studies and other forms of inference. This approach
known as “Rubin’s Causal Model” considers causal inference as a problem of missing data: the
explicit mathematical modelling of the assignment mechanism aims at revealing the observed data
[24]. We can reasonably assume that participants and control villages have similar characteristics as
shown above. We tested both treatment and control groups for the various cofounding factors and
tests always show that we cannot reject the hypothesis of equality of mean between groups for any
of these variables. This confirms the similarity between the two groups for each covariate included
in the P-score model (data not shown). Yet, participants and controls were not assigned randomly to
both arms of the quasi-experiment and therefore straightforward comparison between units may
introduce a selection bias [25-27]. PSM was used to address this bias, building treatment and
control groups with balanced pre-treatment covariates [13].
The impact of the program on all outcome variables is calculated over two periods of time, after 4
and 7 years of treatment respectively, in order to capture both short and medium-term effects. The
covariates included in the models estimating the propensity score are both at individual and village
level. There are seven main individual covariates: age, gender, education, household size, type of
disability, level of disability and level of wealth. Since caste and religion often play a significant
role in determining access to different services in India, we have included them in our analysis but
with some limitations because of their sensitive nature. For ethical reasons, and to avoid non-
response bias, the scientific advisory committee decided not to ask this information to the control
group. Nonetheless, in order to better understand the dynamics underlying these variables, caste and
religion were introduced as covariates in a comparison among PwD joining the program at different
times, measuring a two-year effect of the CBR. These results didn’t seem to be affected by the

9
introduction of these covariates, suggesting no caste or religion-based discrimination in accessing
CBR programs.
There are five main village-level covariates in the model: size of the village, presence of a hospital,
presence of a middle school, distance from a main road, and quality/type of the road that leads to
the village.
Moreover, in order to reduce information bias linked to self-reported measure of personal practical
autonomy and to take into account adaptive preferences, we added a covariate based on a personal
score using a vignette [28]
The PSM approach allowed the identification of a control group of PwD unexposed to the program
that matched the intervention group on any characteristics that would influence the outcomes of
interest to avoid the introduction of a selection bias into the resulting program effect estimate. The
main drawback of PSM is that it can only take into account observed covariates. Latent factors that
affect assignment to treatment cannot be accounted for in the matching procedure. The risk of
selection bias arises at individual level (i.e. PwD in villages covered by the program may decide not
to join it). We argue that the village-level assignment mechanism can be considered random. In
order to gather information about the individual-level assignment mechanism, we performed a
complete census of the PwD in 17 randomly selected villages in order to identify PwD who did not
join the program. A statistical analysis of this subgroup of “never-taker” units (i.e. units who do not
receive the treatment even if they can) showed significant differences between those and treated
units in three key variables: level of wealth, age and severity of disability. Results show that
wealthier, older people experiencing milder disability are less likely to join the program, while the
caste registered non-significant differences (data not shown). These three variables are included in
the model estimating the propensity score, guaranteeing that the treated and control groups are
balanced with respect to these key characteristics, and implicitly supporting the unconfoundedness
assumption, that remains not directly testable.

10
We restricted our analyses to the subgroup of deprived PwD to obtain even more robust results.
This approach has two important advantages. First, it allows a straightforward interpretation of the
results. Considering the outcome as a binary variable (e.g. 0=deprived, 1=not deprived) allows
interpreting the results as the share of both participants and controls meeting the outcome.
The second advantage is that before being treated the two groups are perfectly balanced with
respect of the outcome variable (that equals 0 for all the units). For example, when analysing the
effect of the CBR program on getting a pension, we compared groups of treated and control units
who did not have a pension at the starting point. If the propensity score unconfoundedness
assumption holds, this leads to a setting that can be considered experimental. This approach also
presents some drawbacks -the main one being the impossibility of measuring the effect of the
program on PwD’s capacity to keep a pension that they already secured.
The counterfactual outcome of each unit treated is calculated via a kernel-weighted average of the
outcome of all non-treated units, using the distance between units to estimate the weights. This
method has been found to produce more robust results compared to a one-to-one matching method,
especially when the number of potential controls is lower than the number of treated unit, as it is the
case in our study [29]. In order to test the sensitivity of the results to the matching algorithm, the
results were recalculated via a nearest-neighbour method obtaining similar estimated effects.
RESULTS
Table 1 provides the distribution of treated and untreated groups for outcomes of interest. Table 2
reports the average treatment effect (ATT).

11
Table 1. Randomisation process (approximately here)
!"#$%&'()
*+',#'-)
.)
/'0+12'-)
3%4#+%5)
.)
/'0+12'-)
67)#+',#'-)
&,#$8'-)
9:);',+()<**=)
67)
$%4#+%5()
>'4(1%4)?),55%@,4$'()
!
!
!
!
Year!2002!
52.14!
72.65!
314!
230!
A%B151#;),1-),4-),0051,4$')
!
!
!
!
Year!2002!
91.97!
97.93!
637!
379!
>,1-)C%BD@%+E)
!
!
!
!
Year!2002!
58.30!
68.85!
263!
148!
>'+(%4,5D0+,$#1$,5),"#%4%&;)
!
!
!
!
Year!2002!
39.88!
38.08!
265!
!!115!
Table 2. Impact evaluation: access to selected dimensions (approximately here)
Note: Kernel Matching method
Figure 3 shows the impact of the CBR program on outcomes of interest at time 0 (before the CBR
started) and after 4 and 7 years (Figure 3).
Figure 3. Results of the outcomes analysed after 4 and 7 years (approximately here)
Outcomes
Treated
%
Improv.
Control
%
Improv.
N.
treated
N.
controls
Effect
%
Confidence
Interval
95%
St. dev
p-value
Pension & allowances
After 4 years
42.80
10.33
314
230
35.2
26.8 - 43.6
0.035
<0.001
After 7 years
67.15
38.38
314
230
29.7
22.8 - 36.5
0.046
<0.001
Mobility aid and appliance
After 4 years
6.81
1.85
637
379
4.9
2.3 - 7.5
0.012
<0.001
After 7 years
9.43
5.80
637
379
4.2
1.4 - 7.1
0.015
<0.005
Paid job/work
After 4 years
9.15
0.00
263
148
10.7
6.8 - 14.6
0.018
<0.001
After 7 years
14.34
1.43
263
148
12.3
7.8 – 16.7
0.019
<0.001
Personal/practical
autonomy
After 4 years
40.29
7.83
265
115
35.4
27.1 –43.6
0.041
<0.001
After 7 years
43.40
11.30
265
115
36.2
27.9 - 44.5
0.043
<0.001

12
The CBR program had a positive impact on all outcomes of interest. The effect on access to
pension/allowances was already significant after 4 years (35.2%), and remained robust after 7 years.
We found that people in the control areas obtained pension or allowance after a longer time period.
The impact on getting a paid job was positive and significant both after 4 and 7 years. We observed
a greater effect after 7 years, suggesting a long lasting impact. We also found a small but significant
impact of the program on obtaining mobility and aid appliances after 4 years. This effect was twice
higher after 7 years. Results reported in table 1 also show that treated PwD, who initially had little
or no personal autonomy in daily life activities such as keeping oneself clean and tidy benefited
from an increased probability of being able to look after themselves (35% after 7 years)
DISCUSSION
Our study is the first attempt to evaluate the impact of CBR programs using a quasi-randomised
trial. Our results showed that low-cost community driven CBR programs can improve various
aspects of the livelihoods and well-being of PwD by having a positive impact on access to public
pension schemes and allowances, paid jobs, mobility aid and appliances and perceived levels of
autonomy in daily life activities. Results also established the long lasting effect of the CBR
program: it remained significant after 7 years.
The outcomes of interest were identified based on the CBR guidelines [2] the UN CRPD and the
capability approach literature [16-18]. Literature shows that access to pension and allowances is
essential to fight poverty and promote agency of PwD [30]. Having a job has been showed to foster
social inclusion and quality of life of PwD [31]. Assistive devices are instrumental in removing
barriers in the environment and expand the capabilities set [17]. Many PwD depend on disability

13
equipment such as wheelchairs and ramps to enable them to participate in daily life and contribute
to productive activities [2] .
The increase of personal practical autonomies of PwD is one of the main relevant action/objective
of CBR workers and has a deep impact on the self-esteem and relational capacities of PwD [2].
Our study shares limitations of the propensity score matching approach. This includes the
fundamental and non-directly testable assumption -often referred to as “unconfoundedness”-
according to which adjusting for differences in observable characteristics linked to the outcome
removes biases in the comparison between treated and control units, thus allowing for causal
interpretation of the results [13].
What is already known
Existing CBR studies have been focusing on accessibility, importance of the program, identification
of needs and specific outcomes using only post-test or pre-post tests evaluation designs [6, 7, 32].
Very few investigated the impact of CBR programs using randomised control trial or quasi-
experimental design and had a limited sample size. Most evaluations were even exclusively
descriptive. Others made inference from qualitative interviews of a small sample of PwD [7, 24].
None explored access to employment [7].
What this study adds
Our study shows the overall effect of a multipronged program for PwD in Karnataka State using the
CBR matrix which has been recently recommended as the reference framework for CBR evaluation
[2]. The CBR program evaluated here has many components and must be interpreted as an overall
effect of joining the program for a PwD. Nonetheless, the qualitative and emancipatory researches
that complement our study underline that the range of circumstances in which CBR has successfully
improved the quality of life of PwD indicates that CBR workers possess a range of skills and
expertise beyond medical proficiency [33]. The most significant impact was detected in increased

14
autonomy in daily life activities, which is the primary tenet of CBR. This has implication for a
better participation in family and community matters. This contributes to reducing stigmatising
attitudes towards PwD and promoting empowerment in line with the social model of disability and
the human rights framework [34-36]. Similar findings apply to Botswana, Ghana, the Philippines,
Zimbabwe, South Africa and Vietnam [37-40]. Another major finding is that the CBR program has
been promoting access to employment, which is an essential aspect of social integration and
changing attitudes towards PwD, particularly those with mental illness related disabilities such as
schizophrenia, bipolar disorders and severe depression. [41]. In fact, research in India has shown
that employment level disparity between PwD and non-disabled people is not explained by a major
difference in productivity and human capital but most probably by negative attitudes resulting in
discrimination in access to jobs [42] particularly for women [43].
Policy implication
There are several policy implications of our results. First, our study, inclusive of all types of
disability, and using validated instruments can be applied in other low-income settings to measure
the impact of CBR programs in the various domains of the CBR matrix – health, livelihoods,
employment, education, social participation and empowerment. Second, results show that CBR
programs are particularly effective in a context such as India where laws, regulations and public
policies addressing disability and mental health exist. Previous research has shown the lack of
knowledge PwD have of their rights in India [44]. PwD, particularly those with mental disability,
are also reluctant to claim their rights as they might face public stigma linked to public disclosure of
their status [45]. CBR workers were trained to navigate the Indian legal and administrative system
to promote information and inclusion of CBR participants in public services (education,
rehabilitation and health notably) as well as defend their rights (for instance to free transportation)
and protect them against prejudice [31]. In summary, our paper shows that a CBR program can set
in motion virtuous interactions in the communities through social empowerment, fostering new
opportunities for those who are discriminated against and lobbying efficiently for PwD’ rights.

15
Acknowledgement
We wish to thank participants in this study and their family who took the time to respond to our
survey. We also thank the team of data collectors and supervisors for their patience and efforts in
learning how to carry out the survey and for their outstanding work in the field. We thank AIFO
India in Bangalore, and the two partner organizations (SRMAB and MOB) in Mandya district for
the high level of commitment to the research project and for their continuous logistical support. We
have been privileged to receive help, comments and suggestions from a large number of people. For
this reason we would like to express our gratitude to Parul Bakhshi, Deveraj Basavaraj, Paolo
Battistelli, Marco Bellucci, Cristina Devecchi, Srinivas Gowda, Giampiero Griffo, Nora Groce,
Renato Libanora, Aron Martin, Fabrizia Mealli, Francesca Ortali, Enrico Pupulin and Enrico Testi.
"Competing Interest: None to declare."
Contributorship statement
The study was elaborated by VM, JFT and MB. The analysis were carried out by VM. The first
draft has been written by VM, MB and JFT. All authors collaborated and agreed to the final draft.
Funding: This work was supported by AIFO/Italy, SMHF/Japan, FIRAH/France and
DAHW/Germany.
Ethical approval was obtained from both UCL and AIFO.
Licence for Publication
The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of
all authors, an exclusive licence (or non exclusive for government employees) on a worldwide basis
to the BMJ Publishing Group Ltd to permit this article (if accepted) to be published in JECH and
any other BMJPGL products and sublicences such use and exploit all subsidiary rights, as set out in
our licence (http://group.bmj.com/products/journals/instructions-for-authors/licence-forms).

16
Figure 1. Map of the districts under research (approximately here)
Figure 2. Randomisation process (approximately here)
Figure 3. Results of the outcomes analysed after 4 and 7 years (approximately here)

17
REFERENCES
1. ILO, UNESCO, and WHO, CBR: A Strategy for Rehabilitation, Equalization of
Opportunities, Poverty Reduction and Social Inclusion of People with Disabilities. Joint
Position Paper, 2004, United Nations: Geneva.
2. WHO, et al., Community-based Rehabilitation: CBR Guidelines, Towards Community-
based Inclusive Development, 2010, WHO: Geneva.
3. United Nations, Convention on the rights of persons with disabilities, 2006, United Nations:
New York.
4. Tomlinson, M., et al., Research priorities for health of people with disabilities: an expert
opinion exercise. Lancet, 2009. 374(9704): p. 1857-1862.
5. Mitchell, R., The research base of community-based rehabilitation. Disability &
rehabilitation, 1999. 21(10&11): p. 459-468.
6. Cornielje, H., The Role and Position of Disabled People’s Organisations in Community
Based Rehabilitation: Balancing Between Dividing Lines. Asia Pacific Disability
Rehabilitation Journal, 2009. 20(1): p. 3-14.
7. Sharma, S., Community participation in community-based rehabilitation programmes. Asia
Pacific Disability Rehabilitation Journal, 2007. 18(2): p. 146-157.
8. Finkenflügel, H., H. Cornielje, and J. Velema, The use of classification models in the
evaluation of CBR programmes. Disability and Rehabilitation, 2008. 30(5): p. 348-354.
9. Alavi, Y. and H. Kuper, Evaluating the impact of rehabilitation in the lives of people with
disabilities and their families in low and middle income countries. A review of tools, 2010,
London School of Hygiene and Tropical Medicine and CBM: London.
10. Johnston, S., S. Mendis, and C.D. Mathers, Global variation in stroke burden and mortality:
Estimates from monitoring, surveillance, and modelling. The Lancet Neurology, 2009. 8(4):
p. 345-354.
11. World Health Organization and World Bank, World report on disability, 2011, World
Health Organization: Geneva. p. 325.
12. Groah, S.L., et al., Beyond the evidence-based practice paradigm to achieve best practice in
rehabilitation medicine: a clinical review. PM & R : the journal of injury, function, and
rehabilitation, 2009. 1(10): p. 941-50.
13. Rosenbaum, P.R. and D.B. Rubin, The central role of the propensity score in observational
studies for causal effects. Biometrika, 1983. 70(1): p. 41-55.

18
14. Rubin, D.B., Estimating causal effects of treatments in randomized and nonrandomized
studies. Journal of Educational Psychology, 1974. 66(5): p. 688-701.
15. Sen, A.K., Development as Freedom1999, Oxford: Oxford University Press.
16. Trani, J., et al., Disabilities through the Capability Approach lens: Implications for public
policies. European Journal of Disability Research, 2011. 5(3): p. 143-157.
17. Sen, A.K., 2009. The Idea of Justice. The Belknap Press of Harvard University Press,
Cambridge, Massachusets.
18. Burchardt, T., Capabilities and disability: the capabilities framework and the social model
of disability, 2004, Routledge. p. 735-751.
19. Helander E, Mendis P, Nelson G, Goerdt A. 1989. Training in the Community for People
with Disabilities. Geneva, Switzerland: World Health Organization
20. Mitra, S. The Capability Approach and Disability. Journal of Disability Policy Studies,
2006. 16(4): p. 236-247.
21. Fisher, R.A., The causes of human variability. The Eugenics review, 1919. 10(4): p. 213-20.
22. Neyman, J., On the application of probability theory to agricultural experiments: essay on
principles. Translated in Statistical Science in 1990, 1923. 1993(5): p. 465-480.
23. Little, R.J.A. and D.B. Rubin, eds. Statistical analysis with missing data. 2002, Wiley: New
York.
24. Loeb, M.E., A.H. Eide, and D. Mont, Approaching the measurement of disability
prevalence: the case of Zambia. ALTER Revue européenne de recherche sur le handicap,
2008. 2(1): p. 32-43.
25. Holland, P.W., Statistics and causal inference. Journal of the American Statistical
Association, 1986. 81(396): p. 945-960.
26. Robins, J.M. and S. Greenland, Adjusting for differential rates of prophylaxis therapy for
pcp in high-dose versus low-dose azt treatment arms in an aids randomized trial. Journal of
the American Statistical Association, 1994. 89(427): p. 737-749.
27. Sheiner, L.B. and D.B. Rubin, Intention-to-treat analysis and the goals of clinical-trials.
Clinical Pharmacology & Therapeutics, 1995. 57(1): p. 6-15.
28. Barnard, J., et al., A broader template for analyzing broken randomized experiments.
Sociological Methods & Research, 1998. 27(2): p. 285-317.
29. Salomon, J.A., A. Tandon, and C.J.L. Murray, Comparability of Self-Rated Health: Cross
Sectional Multi-Country Survey Using Anchoring Vignettes. British Medical Journal, 2004.
328: p. 258-261.

19
30. Berg, G.D., An application of kernel-based versus one-to-one propensity score matching for
a nonexperimental causal study: example from a disease management program evaluation.
Applied Economics Letters, 2011. 18(5): p. 439-447.
31. Zaidi & Tania Burchardt. "Comparing Incomes When Needs Differ: Equivalization For The
Extra Costs Of Disability In The U.K," Review of Income and Wealth, International
Association for Research in Income and Wealth, 2005, vol. 51(1), pages 89-114, 03.
32. Biggeri M, Deepak S, Mauro V, Trani JF, Kumar J, Ramasamy P, Bakhshi P, Giriyappa R.
Impact of CBR: Community-Based Rehabilitation programme in Mandya district
(Karnataka, India). Bologna: AIFO; 2012.
33. Thomas M. Reflections on community-based rehabilitation. Psychology and Developing
Societies 2011;23(2):277-291.
34. Grandisson M, Hébert M, Thibeault R. A systematic review on how to conduct evaluations
in community-based rehabilitation. Disability and Rehabilitation 2014;36(4):265-275.
35. Lang R. Community-based rehabilitation and health professional practice: Developmental
opportunities and challenges in the global North and South. Disability and Rehabilitation
2011;33(2):165-173.
36. Deepak S., Kumar J., Ramasamy P., Griffo G.. “An emancipatory research on CBR and the
barriers faced by persons with disabilities” Journal of Disability and Rehabilitation, 2013,
Jun 13.
37. Kuyini, A.B., A.-R.K. Alhassan, and F.K. Mahama, The ghana community-based
rehabilitation program for people with disabilities: what happened at the end of donor
support? Journal of Social Work in Disability & Rehabilitation, 2011. 10(4): p. 247-67.
38. Chappell, P. and C. Johannsmeier, The impact of community based rehabilitation as
implemented by community rehabilitation facilitators on people with disabilities, their
families and communities within South Africa. Disability and Rehabilitation, 2009. 31(1): p.
7-13.
39. Lundgren-Lindquist, B. and L. Nordholm, Community-based rehabilitation-a survey of
disabled in a village in Botswana. Disability and Rehabilitation 1993. 15(2): p. 83-89.
40. Sharma, M. and S. Deepak, A participatory evaluation of community-based rehabilitation
programme in North Central Vietnam. Disability and Rehabilitation, 2001. 23(8): p. 352-
358.
41. Drew, N., et al., Global Mental Health Human rights violations of people with mental and
psychosocial disabilities: an unresolved global crisis. Lancet, 2011. 378(9803): p. 1664-
1675.

20
42. Mitra S, Sambamoorthi U. Disability and the rural labor market in India: evidence for males
in Tamil Nadu. World Development 2008;36(5):934-952.
43. Thomas T, Rajaraman D, Kiruba Shankar KV, Vaz M. Social, economic, and demographic
factors affecting risk of severe disability and employability in India. International Journal
on Disability and Human Development 2013;12(1):45-51.
44. Pal HR, Saxena S, Chandrashekhar K, et al. Issues related to disability in India: A focus
group study. National Medical Journal of India 2000;13(5):237-241.
45. Mishra NN, Parker LS, Nimgaonkar VL, Deshpande SN. Disability certificates in India: a
challenge to health privacy. Indian journal of medical ethics 2012;9(1):43-45.