evaluating research in developmental disabilities: a conceptual framework for reviewing treatment...
TRANSCRIPT
Evaluating research in developmentaldisabilities: a conceptualframework forreviewing treatmentoutcomes
Charlene Butler* EdD;
Henry Chambers MD;
Murray Goldstein DO MPH;
Susan Harris PhD;
Judy Leach RPT;
Suzann Campbell PhD;
Richard Adams MD;
Johanna Darrah PhD; Treatment Outcomes Committee,
American Academy for Cerebral Palsy and Developmental
Medicine, Rosemont, IL, USA.
*Correspondence to first author at 1818 Westlake Avenue
North, Suite 106, Seattle, Washington, 98109-2707, USA.
E-mail: [email protected]
The study of developmental disabilities, not being confined to
one medical field, poses a challenge in evaluating outcomes
research. It is a multidisciplinary area of study which encom-
passes health-care, rehabilitation, psychosocial, educational,
and biotechnology specialties and involves biological, social,
and behavioral effects of intervention. Consequently, there is a
lack of consistency in what has been studied, how the out-
comes have been measured, and where these results have been
recorded. Naturally, these disparate outcomes data need to be
consolidated in such a way that comparison of treatments can
be made, within and across professional disciplines.
In an attempt to facilitate this, the American Academy for
Cerebral Palsy and Developmental Medicine (AACPDM) is
committed to providing critical and useful appraisal of the sci-
entific literature to help clinicians keep abreast in their own as
well as other relevant disciplines1. The Academy has explored
several classification systems to consolidate and interpret data,
and has established a two-part conceptual framework (1) to
aggregate treatment outcomes and construct evidence tables
based on a model of disablement which classifies treatment
outcomes by the dimension in which they have an effect; and
(2) to determine the degree of confidence that can be placed in
the scientific evidence available in support of an intervention.
Classification systems and models of disablement Before selecting a classification system, the AACPDM exam-
ined and carefully considered the merits of the most promi-
nent classification in disablement models in rehabilitation as
well as a system currently used to classify effects of orthope-
dic surgery. The rehabilitation models were found to be
more applicable to the multidisciplinary nature of develop-
mental medicine.
In 1980, the World Health Organization (WHO) developed
a model of disablement; this model has had the most univer-
sal influence on rehabilitation. The model is described in two
companion publications: the International Classification of
Disease (ICD-10)2 which classifies diseases, disorders or
injuries; and the International Classification of Impairment,
Disability and Handicap (ICIDH)3 which describes the con-
sequences of health conditions. The WHO model describes a
key concept which is the basis of its classification system, i.e.
that the consequences of disease occur at different levels:
disease at the level of molecules and cells; impairment at the
level of organs; disability at the level of individuals; and
handicap at the level of society. The WHO’s use of the term
handicap for one of these levels became the subject of vigor-
ous international debate. Consequently, the US Institute of
Medicine (IOM)4 and the US National Center for Medical
Rehabilitation Research (NCMRR)5 chose not to use the lan-
guage of the ICIDH. Instead, they adopted the classification
language of a conceptually similar model which was pro-
posed by Nagi in 19696.
Table I shows that this concept of reference levels is central
to all the rehabilitation models, but that the number and nam-
ing of levels is not consistent7. For example, disability refers to
the person level in the 1980 WHO model, but in the IOM
model, disability refers to the societal level. The NCMRR
model added a fifth level which addresses how society may
impose disablement on a person with a disability. Although
disability and other related terms have different meanings
within these existing classification systems, a common con-
ceptual framework of levels of reference underlies all of them.
In addition to the controversy over the term handicap and
the confusion of terminology which resulted from efforts to
improve on the WHO classification system, conceptual think-
ing about disablement has occurred over the years. Current
thinking by the WHO8, the IOM9, and the NCMRR (personal
communication, Marcus Fuhrer, Director of NCMRR, 1998)
has: (1) refocused the classification terminology toward neu-
tral rather than negative attributes (i.e. function instead of
functional limitation) to reflect a more positive approach or
enablement model; (2) acknowledged that the responsibility
for the disadvantages of disablement may be assigned to the
individual, to the environment, or both; and (3) recognized
that the consequences of disease are not necessarily uni-
directional and linear.
A common and non-inflammatory classification system has
been needed to facilitate the measurement, management, and
research of rehabilitation outcomes and to help remove the
barriers between medical and social models of rehabilitation
service delivery10. In 1993, the WHO initiated a worldwide
effort to accomplish this. Involving multiple partners in a con-
sensus-building exercise guided both by current scientific
thinking and the practical needs of use in the field, the WHO
revised the ICIDH. The draft revision, known as the ICIDH-2
and published in 1997, reflects changes proposed by many
users, experts, WHO-collaborating centers, and international
task forces8. This draft version is being subjected to systematic
Developmental Medicine & Child Neurology 1999, 41: 55–59 55
field trials and further consultation until 1999 and will be final-
ized after the field-trial results. Ongoing revision will extend
the main document to address specific use of the model in
clinical applications and medical research.
Compatibility with the WHO revision in progressIn 1995, when the Academy needed to select a classification
system, the WHO revision was in its pre-draft stage. The
Academy elected to use the NCMRR classification and contin-
ue to monitor the WHO revision process. The US (represented
by the Public Health Service, the IOM, and the NCMRR as well
as other groups) has subsequently joined Canada and Europe
in supporting the current and ongoing WHO revision effort11.
Once the WHO revision is fully accomplished with con-
sensus from disability activists, medical rehabilitation service
delivery systems, and research communities, the AACPDM
will consider changing to the WHO classification schema.
Similarities between the conceptual levels of reference used
in the WHO draft and the NCRMM model have been investi-
gated to assure that only the names given to the levels would
change (Table II). Thus future comparisons between inter-
ventions using evidence tables developed under the former
and latter classifications would be possible.
It will take time for any new terms of disablement to take
hold in a widespread and uniform manner. However, a com-
mon language is crucial to the clarification of rehabilitation
concepts, outcome definitions, and development of mea-
surement tools.
The NCMRR modelThe NCMRR, established in 1990 by the US National Institute
of Health, required a model to guide their agenda for mea-
surement and research of rehabilitation outcomes. An advi-
sory board of scientists, clinicians, and individuals with
disabilities had to determine the best way to translate med-
ical findings into clinical benefits for individuals with disabil-
ities. It examined the usefulness of current disability models
and how well they reflected the multidisciplinary nature of
rehabilitation and the importance of environmental factors
in mediating disabilities. The NCMRR constructed a model in
1993 based on, but also elaborating, the earlier Nagi, WHO,
and IOM models. Given the model’s intended purpose and
that its classification included levels from both medical and
social models, the NCMRR classification system seemed well-
suited to the needs of the AACPDM.
The NCMRR model describes five dimensions in which
congenital abnormalities, developmental disorders, genetic
conditions, injury or disease and their consequences may
occur (see Table III). The term ‘dimensions’ replaced that of
‘levels’ to suggest the multidirectional nature of the interac-
tions proposed; this change in terminology has also been
adopted in the WHO revision. Table IV shows how the effects
of cerebral palsy can be described using this model.
Table V gives one example of how evidence tables based
on this model can consolidate data from existing research for
analysis.
Evidence tables: how much and what kind of research isavailable? Evidence tables are a convenient way to summarize research
findings to determine what is known, what is not yet known,
and what it would be helpful to know about an intervention.
Campbell et al.12 demonstrated the usefulness of the
NCMRR model for analyzing data from their review of the
effects of intrathecally-administered baclofen on adults and
children with spasticity. They determined which dimension
of the disabling process was measured or observed for each
outcome published and then constructed a detailed evi-
dence table organized by these dimensions. Table V is a sum-
mary of that evidence table; it demonstrates one of the
simplest ways in which disparate data can be presented to
show the kind of evidence, as well as how much, exists for an
intervention. There are studies on how baclofen works in
the pathophysiological dimension, but the authors did not
attempt to review these or any work on societal limitations,
choosing, instead, to concentrate on those dimensions of
primary interest to rehabilitation specialists, i.e. impair-
ment, functional limitation, and disability. Table V shows
56 Developmental Medicine & Child Neurology 1999, 41: 55–59
Table I: Comparison of classification terminology in models of disablement
Terms used by Levels of reference for impact of disabilityCells/tissue Organ Person Social External barriers
Nagi 1969 Pathophysiology Impairment Functional limitation Disability –
WHO 1980 – Disease Impairment Disability Handicap
IOM 1991 – Pathology Impairment Functional limitation Disability
NCMRR 1993 Pathophysiology Impairment Functional limitation Disability Societal limitation
Table II: Comparison of NCMRR and revised WHO classification
Terms used by Levels of reference for impact of disabilityCells/tissue Organ Person Social External barriers
NCMRR 1993 Pathophysiology Impairment Functional limitation Disability Societal limitation
WHO 1997 Health condition Impairment Activity Participation Contextual– environmental
(revised draft) factors
that our scientific knowledge, based on this review of
human studies, is restricted to effects in the dimension of
impairment and, to a lesser extent, in the dimension of func-
tional limitation. All 20 studies assessed one or more effects
on impairment, specifically: effects on abnormal postural
tone, reflexes, spasms, pain, skin integrity, bladder capacity,
and so on. Effects on functional limitations (e.g. sleep dis-
turbance, dressing, hygiene care, sitting, and mobility) were
reported in 15 of the studies, but in three of them only anec-
dotal findings are reported. Only eight of the studies provid-
ed any information on effects in the dimension of disability
(e.g. ability to drive own van, full-time employment, sexuali-
ty) and this was exclusively anecdotal.
Linkage of levelsAnother type of analysis to which such an evidence table
lends itself is an examination of the interactions across
dimensions, which may occur as a result of an intervention.
One is tempted to think of these as five levels of a hierarchical
causal pathway; for example, in cerebral palsy, periventricu-
lar leukomalacia causes spasticity which is responsible for
awkward gait, and so on. While this may be true, the corollary
that successful treatment at one ‘level’ (i.e. the pathophysio-
logical or impairment dimension) will automatically and
positively affect another ‘level’ (i.e. functional limitation or
disability) has been an untested assumption underlying
intervention and research.
As a result of this assumption, studies have rarely incorpo-
rated measures in multiple dimensions to test specifically for
associated effects of an intervention. However, frequently
reported clinical observations did lead to a 1990 study of
selective posterior rhizotomy for cerebral palsy in which out-
come measures were included for both impairment and
functional limitation dimensions13. The results showed that,
in spite of the reduction in spasticity (i.e. a measure of
impairment) achieved by the intervention, children in the
study did not begin to use normal movement patterns (i.e. a
measure of functional limitation). In other words, positive
effects of treatment in one dimension did not automatically
produce effects in another dimension.
Almeida et al.14 published a case study in 1997 which sup-
ports a multidimensional assessment of outcomes. These
authors specifically designed a study to investigate the link-
age of effects of a treatment across dimensions, i.e. the use of
intrathecal baclofen in a child with cerebral palsy. Outcomes
were assessed longitudinally and included measures of
impairment (spasticity, speed of movement, range of
motion, and strength), functional limitations (gross motor
activities and activities of daily living), and disability (client
and family reports of changes at home and at school).
Although use of abnormal movement patterns did not
decrease in the child, reduced spasticity and increased speed
of movement were associated with improved independence
in transfers, dressing, and toileting; and the child won an
award for improved physical fitness in school as well as hav-
ing improved ball control during wheelchair basketball.
The NCMRR model emphasizes the possibility that its
dimensions may prove to be multidirectional rather than
hierarchical. The following examples demonstrate the com-
plex interactions suggested by the NCMRR model, i.e. that
effects may be top-down, bottom-up, or lateral. Antispasticity
drugs may intervene in the pathophysiological dimension by
affecting neurotransmitters; orthopedic surgery may act in
Evaluating Research in Developmental Disabilities Charlene Butler et al. 57
Table III: NCMRR model of disablement: five dimensions ofhuman functioning
Dimension Description
Pathophysiology Interruption or interference of normal
physiology and developmental processes
or structures
Impairment Loss or abnormality of body structure or
body function
Functional limitation Restricted participation in typical
societal roles
Disability Inability to participate in typical societal
role functions
Societal limitation Barriers to full participation in society that
result from attitudes, architectural barriers
and social policies
Table IV: Examples of selected effects of cerebral palsy ondimensions of human functioning
Level Examples
Pathophysiology Cystic lesions and white matter loss as a
result of periventricular leukomalacia of the
premature infant’s brain
Impairment Spasticity, contractures, low endurance,
perceptual dysfunction
Functional limitation Awkward walking with fatigue; difficulty
dressing; poor concentration and sustained
listening; reading problems
Disability Learning delays; education in restricted
environment; limited sports activity;
interference with dating and sexuality; not
able to take communion at church; cannot
participate in family activity by doing chores
at home; unable to achieve independent
living
Societal limitation Exclusion from school/city team sports;
denial of medical treatment or equipment
by insurer; government action that blocks
the building of independent living units for
people with disabilities; failure of voters to
support funding of wheelchair lifts for
public buses
Table V: Summary of a review of 20 studies of effects ofintrathecally administered baclofen on adults and childrenwith spasticity of differing etiology (1986 to 1993)13
Impairment Functional Disabilitylimitation
Nr of studies reporting 20 12 0
measured outcomes
Nr of studies reporting 1 3 8
anecdotal outcomes
the impairment dimension by changing the musculoskeletal
system. There may be positive, negative, or no effects of these
drug or surgical interventions in the dimensions of functional
limitation and disability. Assistive technology, such as pow-
ered mobility, may compensate for functional limitations by
providing an alternative means of efficient locomotion. It may
decrease disability by allowing a student to be independent
and to move about school faster with less effort. Its use may or
may not have positive or negative effects in the impairment
dimension, such as improved head control(positive), skin
breakdown (negative), or increased knee contractures (nega-
tive). Use of assistive technology may have effects in the func-
tional-limitation dimension through loss of ability for
independent transfers, or conversely, increase of transfer abil-
ity due to high motivation and effort to access public trans-
portation. Training for the Wheelchair Olympics may increase
ability. It may or may not also decrease emotional impairment
for a person with depression or low self-esteem. The relations
among the five dimensions of functioning have not been sys-
tematically investigated and need attention.
Grading studies: how good is the evidence?The ideal method for determining efficacy of a treatment is
through randomized clinical trials, but such trials are often
difficult to pursue for a number of reasons. As a result,
many studies employ less well-controlled research
designs. The variety of research designs in studies man-
dates use of a method to help evaluate diverse studies and
give weight to their findings. To determine the degree of
confidence that can be placed in the evidence available
about an intervention, a grading system commonly known
as Sackett’s rules of evidence can be used. Sackett15 has
described a method for grading research based on five lev-
els of evidence (Table VI).
Grade A, B, and C recommendations in support of an
intervention can be generated based on these five levels of
evidence. Outcomes supported by at least one and prefer-
ably more Level-I studies yield grade A recommendations.
Outcomes supported by at least one Level-II study receive a B
grade. Grade C recommendations are given to outcomes
supported by Level-III, -IV, or -V studies. The grade indicates
the degree of certainty of a conclusion, generated by the
strength of the supporting research evidence. Grade A con-
clusions are generated from the strongest research evidence;
they are the most definitive. Grade B conclusions are based
on weaker evidence and are only tentative. Grade C conclu-
sions are based on the weakest, suggestive evidence only;
thus, they are the least reliable.
While this system has much to offer in summarizing the
quality of evidence, there are some limitations. The major
limitation is that this system depends on large group studies
because it grew out of clinical work in epidemiology and
internal medicine where large group studies, even random-
ized controlled trials, are not only possible but increasingly
common. Unlike reviewers of research in developmental dis-
abilities, where small group studies and single subject
designs prevail, Sackett simply excluded less rigorous stud-
ies and relied on ones with stronger designs. In addition, this
system can be misleading because it focuses on the type of
design rather than on the actual design (i.e. the extent to
which threats to validity have actually been controlled).
Finally, there is potential for confusion. For example, two
randomized controlled trials may show opposite results. In
this classification, these Level I studies produce grade A evi-
dence in support of an intervention.
Despite these limitations, the AACPDM believes that ‘lev-
els of evidence’ is a useful concept and has therefore consid-
ered where to incorporate the single subject designs found
more commonly in the developmental disabilities research
into the Sackett schema. In fact, Sackett has more recently
promoted the use of the N-of-1 randomized trial, with an
individual subject to establish the efficacy of a treatment trial.
Furthermore, Sackett points out that using a levels-of-evi-
dence classification does not mean that we discard the body
of uncontrolled evidence that is available to us. Instead,
when more rigorous studies are not available, we must ‘fol-
low the trail to the next best external evidence and work
from there’16 (p 4).
ConclusionThis two-part conceptual framework which provides the orga-
nizational structure for evaluation of interventions in develop-
mental disabilities will be used by the AACPDM. Evidence
tables and analyses of current interventions will appear in sub-
sequent issues of this journal. Those wishing to learn more
about developing reviews of research using this framework
should visit the AACPDM website at http://www.aacpdm.org
for a step-by-step process, including an elaboration of the
NCMRR classification and the Sackett grading systems with
operational definitions of terms and case reports. Here you
will find a guideline entitled ‘Methodology for Developing
Evidence Tables and Reviewing Treatment Outcomes
Research’ written by The Treatment Outcomes Committee of
the AACPDM.
58 Developmental Medicine & Child Neurology 1999, 41: 55–59
Table VI: Sackett’s16 method for grading research
Level Description
Level I Large randomized trials, producing results with
high probability of certainty. These include studies
with positive effects that show statistical
significance and studies demonstrating no effect
that are large enough to avoid missing a clinically
significant effect
Level II Small randomized trials, producing uncertain
results. These are studies which have a positive
trend that is not statistically significant to
demonstrate efficacy or studies showing a negative
effect that are not sufficiently large to rule out the
possibility of a clinically significant effect
Level III Non-randomized prospective studies of concurrent
treatment and control groups, i.e. cohort
comparisons between contemporaneous subjects
who did and did not receive the intervention
Level IV Non-randomized historical cohort comparisons
between subjects who did receive the intervention
and earlier subjects who did not
Level V Case series without controls. The clinical course of
a group of clients is described, but no control of
confounding variables is undertaken. This is a
descriptive study which can generate hypotheses
for future research but does not demonstrate
efficacy
The Academy expects that using this format for organizing
the information on research outcomes will have many bene-
fits for the field of developmental disabilities. With such a
tool, consensus can be reached about the levels for which
there is evidence of efficacy. Meaningful comparisons
between interventions can be made. The levels for which
adequate information is lacking can be visualized and future
research encouraged to address the gaps in our knowledge.
Use of the model prompts the research community to
include multiple outcome measures in study protocols, so
that effects of an intervention in one level can be traced
across other levels. It will help professionals and clients alike
to see that recommendations may not be seen as conflicting
but rather as complementary by showing how different inter-
ventions can have effects in different levels. This can help
clients make informed decisions about treatment options
based on client values and on what the treatment offers. It
can lead us to think about intervention differently, i.e.
whether an individual child is best served at a particular
point by altering the environment rather than attempting to
change the child’s impairment or by considering whether
the disability of social isolation may be caused by depression
rather than limited mobility. Finally, it will serve to remind
each of us that as individuals in our communities, we have
the opportunity to reduce further the burden of disability in
the dimension of societal limitation by efforts to initiate and
support actions to remove all barriers to full participation in
society of people with disabilities.
Accepted for publication 9th October 1998.
References1. Butler C. (1995) Outcomes that matter. Developmental Medicine
and Child Neurology 37: 753–4.2. World Health Organization. (1992–1994) International
Statistical Classification of Diseases and Related HealthProblems, 10th Revision familiarly abbreviated as ICD-10.Geneva, Switzerland: WHO
3. World Health Organization. (1980) International Classificationof Impairments, Disabilities and Handicaps.Geneva,Switzerland: WHO.
4. Pope AM ,Tarlov AR, editors. (1991) Disability in America:Toward a National Agenda for Prevention. Washington, DC:National Academy Press.
5. National Institute of Health. (1993) Research Plan for theNational Center for Medical Rehabilitation Research. NIHPublication no. 93-3509. Washington, DC: US Department ofHealth and Human Services.
6. Nagi SZ. (1965)Disability and Rehabilitation. Columbus, OH:Ohio State University Press.
7. Dittmar SS ,Gresham GE. (1997) Functional Assessment andOutcome Measures for the Rehabilitation HealthProfessional.Gaithersburg, MD: Aspen Publishers.
8. World Health Organization. (1997) ICIDH-2: InternationalClassification of Impairments, Activities and Participation.Geneva, Switzerland: WHO.
9. Brandt EN, Pope AM. editors (1997) Enabling America: Assessingthe Role of Rehabilitation Science and Engineering. Washington,DC: National Academy Press.
10. Wilkerson D. (1997) On the language and classification ofdisablement and a new ICIDH. Rehabilitation Outlook.Winter: 5–7.
11. Seltser R. (1992) PHS Task Force on Improving Medical Criteriafor Disability Determination: Summary Report. Washington,DC: Public Health Service.
12. Campbell SK, Almeida GL, Penn RD ,Corcos DM. (1995) Theeffects of intrathecally administered baclofen on function inpatients with spasticity. Physical Therapy 75: 352–62.
13. Cahan LD, Adams JM, Perry J, Beeler LM. (1990) Instrumentedgait analysis after selective dorsal rhizotomy. DevelopmentalMedicine and Child Neurology 32: 1037–43.
14. Almeida GL, Campbell SK, Girolami GL, Penn RD ,Corcos DM.(1997) Multidimensional assessment of motor function in achild with cerebral palsy following intrathecal administration ofbaclofen. Physical Therapy 77: 751–64.
15. Sackett DL. (1989) Rules of evidence and clinicalrecommendations on the use of antithrombotic agents. Chest95: (Suppl.) 2–4.
16. Sackett DL, Richardson WS, Rosenberg W ,Haynes RB. (1997)Evidence-based Medicine: How to Practice and Teach EBM.New York: Churchill Livingstone.
Evaluating Research in Developmental Disabilities Charlene Butler et al. 59