Connecting perspectives on stroke disability: The measurement and the classification approach

1. Background Stroke is a frequently occurring condition and a
common cause of death and disability. Many stroke survivors are
facing long-term disability. The consequences of stroke on
patients’ functioning are usually complex and heterogeneous.
Precise knowledge of patients’ stroke related disability is
necessary in health services provision and research. Clinical
stroke management, but also epidemiological and clinical research,
depend on the careful detection of functioning problems, as well as
resources, in patients with stroke. Two conceptual approaches to
describe patients’ disability can be distinguished: the health
status measurement and the classification approach. Health status
measures, like standardized performance tests, rating scales, and
questionnaires are used to operationalize and to assess patients’
burden of disease, functioning and health. The classification
approach towards the description of patients’ health state is
represented by the World Health Organization’s International
Classification of Functioning, Disability and Health (ICF). The ICF
provides a comprehensive conceptual framework and a unified
standardized language to describe health and health related states,
both at the individual, as well as at population levels. To enhance
the applicability of the classification, ICF Core Sets for specific
health conditions have been developed in an evidence based and
consensus based process. The ICF Core Sets for Stroke are
selections of salient ICF categories out of the whole
classification, which describe the spectrum of problems in stroke
patients’ functioning based on the universal language of the ICF.
The ICF Core Sets for Stroke represent the practical implementation
of the classification approach in clinical practice and research.
The two approaches to represent stroke related disability, the
health status measurement and the classification approach, can be
regarded as complementary principles. From the classification
perspective, the ICF and the ICF Core Sets can serve as standards
to define what to measure. From the perspective of health status
measurement the question how to measure can be answered. An
explicit connection between the two approaches can be established
by the so-called linking method. Thereby, using the ICF’s category
system the contents of measures can be mapped, explored and
compared in a standardized, transparent and straightforward way.
The linking method can be useful for various purposes. The
application of the linking method along with the ICF Core Sets
constitutes a new approach for examining health status measures’
content validity. However, beyond content validity, meaningful
measurement essentially depends on the psychometric quality of the
applied instruments. Techniques based on modern test theory,
especially Rasch analysis, are increasingly adopted to ensure
instruments’ psychometric properties. 2. Objectives In the
following, the doctoral thesis is subdivided into four parts. The
first three parts present different studies performed to pursue the
objectives named below. Each of the three studies contains a
respective discussion section referring to the results of the
study. The fourth part of the doctoral thesis refers to aim four,
namely the discussion of the relationship between the methods
presented in the previous three parts. The current doctoral thesis
aims (1) to illustrate, how the connection between the health
status measurement approach and the classification approach can be
established by the application of the linking method, (2) to
demonstrate, how this approach can be used to select health status
measures based on their content validity, (3) to show, how the
psychometric features of health status measures can be examined
based on Rasch analyses, and (4) to discuss the relationship
between the demonstrated methods in the context of the connection
of the health status measurement and the classification approach.
3. Applying the linking method: Content comparison of
patient-centered health status measures used in stroke based on the
International Classification of Functioning, Disability and Health
(ICF) The first study, “Content comparison of patient-centered
health status measures used in stroke based on the International
Classification of Functioning, Disability and Health (ICF)”
illustrates the application of the linking method in stroke
measurement. The objective of this study was to examine and to
compare the contents of patient-centred health status measures used
in stroke. The specific aims of the study included the
identification of generic and condition-specific patient-centred
health status measures applied in stroke patients, the examination
of the contents of the single measures based on their linking to
the ICF, and the comparison of the contents of generic and
stroke-specific measures. A systematic literature review was
conducted to identify current generic and condition-specific
patient-centred health status measures applied in stroke. The most
frequently used instruments were selected. The contents of the
selected measures were examined by linking the concepts within the
instruments’ items to the ICF. Six generic and seven
stroke-specific health status measures were selected. Within the
selected instruments 979 concepts were identified. 200 different
ICF categories were used to map these concepts. No single ICF
category is contained in all instruments. Out of the total 200
different ICF categories used, 77 (40%) applied to only one of the
13 selected measures. Overall, the most frequently used category is
b152 emotional functions’ contained in 53 items from 10
instruments. Stroke-specific measures more often address mental
functions, while the selected generic instruments more often
include Environmental Factors. The study provides an overview on
current patient-centered health status measures in stroke and their
covered contents. The results of the content comparison provide
valuable information to facilitate and to account for the selection
of appropriate instruments for specific purposes in clinical as
well as research settings. 4. Selecting health status measures
based on content validity: Comparison of stroke-specific health
status measures with the Comprehensive ICF Core Set for Stroke The
aim of the second study is to demonstrate how the ICF as a
fundamental reference can be used to select health status measures
according to their content validity. The specific aims are (1) to
examine the content validity of the selected stroke-specific health
status measures by comparing them with the Comprehensive ICF Core
Set for Stroke, and (2) to discuss the selection of measures based
on their coverage of the ICF Core Set. Taking the results from the
previous study, the seven stroke-specific patient-centered health
status measures are involved in the current analyses and compared
to the Comprehensive ICF Core Set for Stroke. Descriptive frequency
analyses are conducted to indicate the instruments’ bandwidth and
specificity of content coverage regarding the ICF Core Set. 67
(52%) out of the 130 categories of the ICF Core Set are covered by
at least one of the examined instruments. The single measures cover
in total between 29% and 14% of the Comprehensive ICF Core Set for
Stroke. Overall, 31 categories of the ICF Core Set are measured at
the more specific 3rd and 4th levels by at least one of the seven
instruments. All instruments cover Activity and Participation and
Body Functions, but only two instruments address Environmental
Factors. No categories of the ICF component Body Structures are
contained in the examined instruments. In contrast the
Comprehensive ICF Core Set for Stroke contains categories from all
ICF components. The Quality of Life Instrument for Young
Haemorrhagic Stroke Patients (HSQuale), the Stroke-specific Quality
of Life Scale (SSQoL), and the Stroke Impact Scale(SIS) represent
the top three instruments according to bandwidth as well as
specificity of content coverage. The comparison of instruments
against the Comprehensive ICF Core Set for Stroke can be used to
characterize and compare measures’ content validity. The
examination and comparison of patient-centered health status
measures’ content validity accounting for the bandwidth and the
specificity of content coverage can serve as a first step of
selecting a measure. However, further features of the measures have
to be considered. Especially, their psychometric properties have to
be carefully examined to accomplish the well-founded choice of
appropriate measures to assess stroke related health status. 5.
Applying the Rasch method: Evaluation of the Stroke Impact Scale
using Rasch Analyses The third study, the “Evaluation of the Stroke
Impact Scale using Rasch Analyses” undertakes the psychometric
evaluation of the Stroke Impact Scale 2.0 (SIS), in a German sample
adopting Rasch based techniques. The specific aims of the study
were to examine (1) the unidimensionality of the SIS domains and
item fit, (2) the structure of the response scales, (3) the
targeting of the domains, (4) reliability, (5) differential item
functioning (DIF) or item bias for relevant patient groups, and (6)
to compare the fit results of this study with the Rasch analysis
results of the SIS 2.0 in a North American sample which led to the
creation of the most current version of the SIS, the SIS 3.0. The
Rasch analyses based on Master’s Partial Credit model has been
carried out using data collected from stroke patients in Germany
within an ongoing multicentric international study. 196 stroke
patients from 16 study centers participated in the study and
completed the Stroke Impact Scale. Unidimensionality of the eight
SIS domains was confirmed according to the mean infit statistics
(.97 to 1.02). 7 items displayed model misfit. Response categories
of 25 items showed threshold disordering. For the domains
Communication and Memory/Thinking ceiling effects (>3 logits)
became apparent. Reliability values lay above .80 in six domains.
No DIF was found as to age, gender, disability severity, and
rehabilitation setting. Item fit results in the current study
differed from those in the reference study of the SIS 2.0 in a
North American sample. The SIS is according to its psychometric
qualities a sufficiently robust, valid and reliable measure of
stroke-specific quality of life. It seems suitable to capture
consequences of stroke in patients’ with different levels of
disability severity, within an inpatient as well as an outpatient
setting, across age groups and genders. However, the response
categories currently used with the SIS should be object to further
study and revision. The fit results of the reference study of the
SIS 2.0 could not be replicated and therefore, an item reduced
version of the German SIS equivalent to the most current SIS 3.0
can not be created. Thus, the SIS needs to prove its cross-cultural
validity in future. 6. Discussion: Towards a unified measurement
approach in stroke Two complementary principles towards the
description of disability have been introduced: the health status
measurement and the classification approach. Connected to these
conceptual approaches two methodological procedures have been
regarded: the linking method and the Rasch method. The application
of both methods has been illustrated. The connection of the
classification approach with the health status measurement approach
entails advantages reaching beyond the applications presented here.
Rasch analysis and the linking method, representing qualitative and
quantitative methods, may shed light on different facets of stroke
measurement, which combined increase information value and lead to
a complete picture of functioning and health. Advances of the
classification approach, like the development of the ICF Core Sets
for Stroke, and advances in health status measurement, like the
application of Rasch analyses can be concatenated by the linking
method. From this concatenation of different conceptual and
methodological approaches, unified and comparable, conceptually
sound, high quality measurement of functioning can emerge. The
integration of a common reference framework with the merits of
objective measurement within the proceedings of item banking and
adaptive testing can contribute to compass a common standard and
agreement on what and how to measure. A unified measurement
approach could thereby be achieved in stroke. Advanced measurement
can serve to promote precise, comprehensive, and efficient
knowledge of stroke disability at the individual and at population
levels, to enable better decisions for treatment and action, in the
long run improving stroke care and relieving the burden to the
patients.