FIM™ internal construct validity revisited
taken to consider whether only motor items should be
assessed. In the US model system for burn injury, for
example, only the FIM™ motor subscale is assessed
(38). However, with regards to the ageing population
and related comorbidity (39), both subscales may be
of interest in musculoskeletal patients.
This study has the limitations of secondary data
analysis. For example, there is a lack of information on
the accuracy and consistency of the data-entry process,
the selection of DIF factors was limited to the variables
of the dataset and the use of non-validated translations.
However, this design enabled a well-tailored calibration
sample to be obtained from a large sample size. Another
limitation is in the 2-testlet approach, which provides
the basis for the transformation table. On the one hand,
this approach was successful in attaining model fit. On
the other hand, the approach does not allow a statement
to be made about the hierarchy and difficulty of single
items or a conceptually related group of items, since it
focuses on the whole construct being measured through
the assessment tool. However, while, for the purpose
of quality or outcome reports, the FIM™ is based on
the total score or change scores, data collection is still
conducted on an item level, which allows clinicians to
gain insight into the development of a single patient in
a certain item or group of items, or to conduct a quality
check of scores at the item level if the FIM™ was, for
example, applied within a payment system.
The analysis of threshold disordering is also not pos-
sible with the testlet approach. Indeed, there is some
evidence that disordered thresholds can themselves be
caused by local dependency (20, 40). For example, if
items are analysed within their subscales, threshold or-
dering may appear correct, but become disordered when
subscales are summated together. Thus, it becomes
impossible to determine if disordered thresholds are a
consequence of local dependency, as the solution for
local dependency renders interpretation of traditional
thresholds invalid (20). Since the 2-testlet approach is
a relatively new one for health assessment tools, further
investigations are needed to confirm the influence of
local dependency on such matters. However, this ap-
proach has the advantage that the total scores of a well-
established and widely used assessment tool, such as
the FIM™, can be converted on an interval-scale level,
without deleting or re-scoring items.
We recommend the use of the interval transforma-
tion table provided in this study for neurological and
musculoskeletal patients for national rehabilitation
quality monitoring, in order to be able to calculate
interval-scaled patient change scores for the FIM™,
compared with its original ordinal scoring system (6).
If the total scores are available in a digital format, as in
the ANQ datasets, transformation can be implemented
199
easily in an electronic information system, by simply
re-coding the total scores according to the table pro-
vided in the results. This interval scoring system has
the advantage that it provides an important basis for
the application of a standardized reporting system for
functioning information (2, 41) in which the FIM™
could be integrated as a widely used instrument in
rehabilitation. This is beneficial, as the standardized
reporting of functioning information enables clinicians
to continue using currently implemented assessment
tools while also being able to compare and aggregate
the information within and across tools, institutions or
even countries. One caveat to this is that the interval-
scale transformation is actually measured with error,
as can be seen in its logit form in Appendix S4 1 .
In conclusion, the results of this study support the
internal construct validity of the FIM™ 18-item ver-
sion and, consequently, the reporting of its total score,
by applying the interval-scaled transformation table
provided in this study. The fact that all the variance
could be accommodated in the final estimate suggests
that previous reports of multidimensionality may have
been driven by a breach of the local independence as-
sumption. This supports the intention of its developers
and the way the FIM™ scores are used in clinical prac-
tice and in institutional and national monitoring. It is
recommended to use the interval-scale transformation
of the FIM™ total score for national quality monitoring
for neurological and musculoskeletal patients, in order
to adequately report change scores in patients’ functio-
ning. Furthermore, interval transformation provides
a basis for integrating the FIM™ into a standardized
reporting system for functioning information.
ACKNOWLEDGEMENTS
The authors thank Dr L. Menzi, Head of Rehabilitation ANQ,
and K. Schmitt, Corporate Development Director of the Swiss
Paraplegic Centre, for their good advice, provision of valuable
information and fruitful discussions about the project.
This project is part of the cumulative Dissertation of Rox-
anne Maritz, which is funded by the Swiss National Science
Foundation´s National Research Programme “Smarter Health
Care” (NRP 74) within the NRP74 StARS Project ”Enhancing
continuous quality improvement and supported clinical decision
making by standardized reporting of functioning.
The authors have no conflicts of interests to declare.
REFERENCES
1. Stucki G, Bickenbach J. Functioning: the third health
indicator in the health system and the key indicator for
rehabilitation. Eur J Phys Rehabil Med 2017; 53: 134–138.
2. Stucki G, Prodinger B, Bickenbach J. Four steps to follow
when documenting functioning with the International
Classification of Functioning, Disability and Health. Eur J
Phys Rehabil Med 2017; 53: 144–149.
J Rehabil Med 51, 2019