The Measurement and Prediction of Physical Functioning after Trauma
de Graaf, Max Willem
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
de Graaf, M. W. (2019). The Measurement and Prediction of Physical Functioning after Trauma. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Chapter 8
General discussion
In the field of trauma surgery, outcome of treatment used to be traditio-nally evaluated by means of “objective” measures, such as range of motion of an affected joint, vital signs, radiographic imaging or mortality. These tradi-tional outcome measures have shown to correlate poorly with how patients
experience their functional outcome.1 Although these measures are relevant
for a better understanding of complicated contexts, such as hemodynamic status, they do not provide the information required to understand complex contexts such as (recovery of) physical functioning after trauma. Problems within a complex context are not to be measured and analyzed, but should be probed and explored in order to be able to get a better understanding of the
problem at hand (Chapter 1). Nonetheless, measurement instruments may
help to explore and probe important determinants within the context. In the past decade, physicians and researchers have become increasingly aware of the importance of exploring how patients experience their physical func-tioning after trauma. Hence there was an increasing need for high-quality measurement instruments that can be used to assess physical functioning after trauma. Patient-reported outcome measures (PROMs) can be used to this end. The Short Musculoskeletal Function Assessment (SMFA) was developed as an instrument to assess physical functioning in patients with a broad range of musculoskeletal conditions. However, it required further evaluation of its properties in order to appraise its quality.
The main aim of this thesis was to evaluate the quality of the Dutch SMFA questionnaire (SMFA-NL) with regard to its ability to assess physical func-tioning in patients that sustained a broad range of acute traumatic injuries. This thesis was divided into two parts: the first part focused on evaluating the clinimetric properties of the SMFA-NL and the second part aimed to inves-tigate the interpretability and prognostic performance of the SMFA-NL with regard to (recovery of) physical functioning after trauma. The current chapter provides an overview and discussion of the main findings that were presented in this thesis, followed by clinical implications and placement of the SMFA-NL within the context it is intended to be used in. The chapter closes with a general conclusion and perspectives for future research.
Part I – Clinimetric properties of the SMFA-NL
Measurement of functional outcome after trauma is essential in clinical
practice and research.2 While the SMFA questionnaire has been designed as an
instrument that can be used to assess physical functioning in patients with a broad range of musculoskeletal conditions, research was needed to determine whether the SMFA-NL can be used to assess the physical functioning of trauma patients. According to the Patient Reported Outcome Measure Information System (PROMIS), physical functioning has been defined as: “self-reported capability rather than actual performance of physical activities. This includes the functioning of the upper extremities, lower extremities and central regions, as well as instrumental activities of daily living such as running errands”.3 The
fact that the SMFA can be used to assess a construct that cannot be observed directly, stresses the vital importance of proving that the instrument actually measures what is intended to be measured. The clinimetric properties of the SMFA-NL were studied to determine whether it can be used to measure physi-cal functioning in a broad range of trauma patients.
In Chapter 2, the structural validity of the SMFA-NL was assessed using a
confirmatory factor analysis. Structural validity has been defined as “the degree to which the scores of a PROM are an adequate reflection of the dimensionality of the construct that is measured”.4 The structure of a PROM can be regarded as
the foundation of the questionnaire. It determines how many subscales should be used and which subscale contains which items. In the Consensus-based Standards for the Selection of health Measurement INstruments (COSMIN) checklist, a confirmatory factor analysis is considered to provide the strongest evidence of structural validity.5,6
A remarkable finding was the insufficient structural validity of the Func-tion and Bother Indices. The indices were originally proposed upon the
deve-lopment of the SMFA.7 Swiontkowski et al. considered that the two indices
measured two different constructs.7 The Function Index was meant to be a
relatively strict measure of functional limitations. However, even with few functional limitations, patients can still be extremely bothered by these limi-tations. The bother due to the functional limitations could therefore be evalu-ated with the Bother Index. It is worth noting that the choice of the Function and Bother Indices was based on a theoretical hypothesis, but was not verified using a factor analysis.8,9 This left uncertainty regarding the structural validity
factor analyses but never reported the two-index structure as the most
appro-priate structure.10-14 All these studies implicitly questioned the structural
validity of the two-index structure. Nonetheless, the indices continued to
be widely used.8,9 The confirmatory factor analysis presented in this thesis
confirmed that the SMFA-NL does not measure physical functioning as two separate unidimensional constructs of functioning and bother, hence use of the Function and Bother Indices is discouraged in Dutch patients with trau-matic injuries.
These findings may have an important consequence. The Function and Bother Indices were originally reported to have sufficient construct validity, reliability and responsiveness7, yet the findings presented in Chapter 2
indica-ted that there was little evidence supporting the conceptual basis of the Func-tion and Bother Indices. Hence these findings put into quesFunc-tion the validity of previous research that used these indices. This also stresses the importance of structural validity of a PROM, as it determines how many subscales should be used and which subscale contains which items. The other clinimetric proper-ties (reliability, construct validity, responsiveness) are evaluated separately for the identified subscales. Evidence of structural validity should therefore precede any evaluation of the other clinimetric properties. Poorly evaluated structural validity may lead to false assumptions of validity and to the use of inappropriate subscales. Structural validity is still a frequently overlooked
clinimetric property that should receive more attention.15
The research presented in Chapter 2 showed sufficient structural validity
and good internal consistency for a four-factor structure of the SMFA-NL that was originally proposed by Reininga et al.12 The four factors are Upper
Extrem-ity Dysfunction (6 items), Lower ExtremExtrem-ity Dysfunction (12 items), Problems with Daily Activities (20 items) and Mental and Emotional Problems (8 items).
The four factors of the SMFA-NL closely follow the dimensionality of this defi-nition, and conceptualize physical functioning as the ability to use the four extremities, ability to perform daily activities, and psychological problems due to possible functional limitations. Other structures that had been proposed in previous research did not show sufficient structural validity and were therefore considered unsuitable as factor structure of the SMFA-NL. Three previously proposed structures of the SMFA-NL could not be tested due to changes that
were made to the questionnaire by splitting some double-barreled items.16,17
were considered impractical for clinical use. Each of these untested structures structure is unique for patients with a specific condition: isolated upper extre-mity fractures, lower extreextre-mity fractures, and patients with an injury severity score (ISS) above 15.16,17 In addition to the impracticality, separate structures
for separate injuries implies that the concept of physical functioning itself is different between types of injuries, which is unlikely. In summary, in patients with acute traumatic injuries the SMFA-NL may be best scored and interpreted using the four subscales as proposed by Reininga et al.12
Structural validity and internal consistency alone do not provide suffi-cient evidence whether the SMFA-NL actually measures what is intended to be measured (physical functioning). Hence additional clinimetric proper-ties such as test-retest reliability, construct validity and responsiveness were
evaluated in Chapter 3 of this thesis. Reliability and construct validity were
studied to determine the extent to which the measurements made with the SMFA-NL were free of measurement error, and the SMFA-NL measured the constructs that were intended to be measured. Responsiveness was studied to evaluate whether the SMFA-NL can be used to assess change in physical functioning over time. The four subscales of the SMFA-NL showed good to excellent test-retest reliability (intraclass correlation coefficients ranging from 0.80 to 0.98). Construct validity and responsiveness were sufficient (86% and 79% of all predefined hypotheses were confirmed, respectively). The SMFA-NL can therefore be used to make a valid and reliable assessment of physical functioning, as well as to make valid measurements of change in physical functioning over time.
An important aspect of reliability is measurement error, which can be expressed as smallest detectable change (SDC). The SDC is the smallest change
in score that can be considered free of measurement error.18 A small SDC
denotes little measurement error, which is preferred to ensure straightforward interpretability of a PROM. The SDC of the SMFA-NL ranged from 11.0 to 17.4
points (on a 0 to 100 scale, Chapter 3) and was considered moderate. Reininga
et al. reported standard error of measurement values of the SMFA-NL, indi-cating an SDC ranging from 23.3 to 31.3 points. The measurement error in their sample was larger, which was most likely due to a more heterogeneous sample that consisted of both orthopedic and trauma patients with an injury that was at least three months old and the absence of a criterion to identify unchanged patients.
Whether the SMFA-NL is a high-quality measurement instrument cannot be concluded from just one of the studies presented in this thesis. Rather, its quality and ability to measure physical functioning should be appraised in a greater context. By aggregating all available research on the measurement properties of the SMFA-NL, its measurement quality can be appreciated. At present all important clinimetric properties have been evaluated, and
show sufficient content7,19 structural (Chapter 2), construct (Chapter 3)
and cross-cultural12 validity. Various aspects of reliability (internal
consis-tency (Chapter 2), reliability and measurement error (Chapter 3)) have been
investigated, showing that the SMFA-NL is a reliable instrument. Lastly, the
SMFA-NL has shown to be responsive (Chapter 3). According to the COSMIN
checklist of bias, the employed methodology of the studies that evaluated these properties was “very good” for the majority of the tested aspects.5 Based
on all this information together, it can be concluded that the SMFA-NL is an instrument that can be used to make high-quality measurements of physical functioning in patients with a broad range of traumatic injuries. The SMFA-NL does have some limitations, which are discussed below.
Limitations
The research presented in Chapter 3 exposed that the performance of the
Upper Extremity Dysfunction subscale has some limitations. Due to the obser-ved floor effects, the instrument showed to be less suitable for the evaluation of minor arm injuries. The susceptibility to floor effect also limits the detection of improvement in upper extremity functioning over time. These problems may be solved by adding upper extremity-related items to the SMFA-NL, but that was beyond the scope of this thesis. Modification of the questionnaire would require a re-analysis of the measurement properties of the question-naire. Alternatively, other outcome instruments such as the Disabilities of Arm Shoulder and Hand (DASH) may be used to evaluate upper extremity functioning.20-23
The clinimetric properties of the SMFA-NL have been assessed in
working-age patients (Chapters 2 and 3). The validity and reliability of measurements
taken among the elderly and patients with neurological deficits is not warran-ted. Because of possible cognitive impairments, different limitations in physi-cal functioning and injury characteristics, these patients should be considered a “different population”. Additional studies would be needed to justify use of
Part II – Interpretability and clinical application of the SMFA-NL
Strong clinimetric properties are a prerequisite for the use of a measu-rement instrument. Proper interpretability however is equally important to justify routine or occasional use of a measurement instrument in clinical practice. This specifically applies to a standardized multi-item instrument, since the meaning of its scores is not immediately clear. Interpretability has been defined as “the degree to which one can assign qualitative meaning (i.e. clinical or commonly understood connotations) to an instrument’s quantita-tive score or change in score”.4 The challenge for clinicians is to interpret the
meaningfulness of the scores derived from PROMs. Various pieces of infor-mation can contribute to the interpretability of instruments, as investigated in Part II of this thesis.
Interpretability of single scores
Proper interpretability starts with a thorough description of the study
population and the distribution (mean and SD) of the scores on the scale.24
By observing the distribution of scores of known groups, clinicians and rese-archers may get a feel for what is a usual score for the group of interest. In the
research presented in this thesis (Chapter 3), for example, patients with an
upper extremity injury scored 25 points on the Upper Extremity Dysfunc-tion subscale versus 5 points for patients without an upper extremity injury. Patients with a lower extremity injury scored 46 points on the Lower Extremity Dysfunction scale versus 14 points for patients without a lower extremity injury. In the literature, Barei et al. reported SMFA scores for various conditions,
but that was only for the Function and Bother Indices.9 Due to the novelty of
the four subscales of the SMFA-NL, only one previous study reported scores using these subscales; these were scores of orthopedic and trauma patients
who had been treated three months earlier.12 Patients scored 36 points on the
upper extremity dysfunction subscale and 30 points on the lower extremity dysfunction subscale. These values provide a reference of the scores that may be expected in patients with traumatic injuries. Two other studies reported SMFA-NL scores in patients with isolated injuries of the extremities and after
major trauma.16,17 However these studies used a modified SMFA-NL
question-naire with alternative subscales. The scores are therefore not directly compara-ble to the four-subscale structure of the SMFA-NL that was recommended in this thesis. The interpretability of single scores of the SMFA-NL may be further
improved by knowledge of the scores of patients with specific injuries, such as pelvic or proximal humeral fractures, using the four-subscale configuration.
The interpretability of a measurement instrument may be further enhan-ced by normative data.5,24 In medicine, normative data are of vital importance
to interpret findings of any measurement instrument, whether it is the range of motion of a joint, a laboratory test, or a score of a PROM. Absence of popu-lation-based norm values throws a barrier into the routine use of PROMs, because of the difficulty of interpreting scores at the individual patient level and at group level. Normative data of the American population have been reported for the SMFA Function and Bother Indices, yet did not contain data
of the four subscales of the SMFA-NL.25 For this reason, age- and
gender-spe-cific normative data of the Dutch population were reported in Chapter 4.
Significant differences between men and women and between different age groups were found. These findings stress the importance of presenting age- and gender-specific normative data.
Normative data that have been obtained from a general population gene-rally provide a reference of what can commonly be considered a “healthy”
score.24,26 The data presented in Chapter 4 can be used to assess whether
patients who have sustained trauma have returned to a normal level of physical
functioning. This has been applied in Chapter 7 of this thesis, where a
prog-nostic model was developed that can be used to predict whether patients will return to a normal level of functioning. The normative data of the SMFA-NL may also provide a standard independent reference group to which treatment effectiveness can be compared in clinical research.
Interpretability of a change in score
Repetitive measurement of physical functioning is a core aspect of the follow-up of trauma patients. It is therefore important to know which
chan-ges in score reflect a meaningful change in physical functioning. In Chapter
6 anchor-based minimal important change (MIC) values of the SMFA-NL
were determined for each subscale. The MIC values of the SMFA-NL ranged from 5 to 26 points, and denote the smallest longitudinal change in physical functioning that can be considered important to patients at the individual level. An extensive literature search showed that MIC values have never been
reported for the SMFA.8 The research provided in Chapter 6 is an important
understanding may be further improved by evaluating MIC values in patients with other musculoskeletal conditions.
Several connotations have to be made to the MIC values obtained. The MIC values of the Problems with Daily Activities and Lower Extremity Dysfunction subscale were larger than the smallest detectable change (SDC) reported in
Chapter 3. For those subscales, any change that is important can be
consi-dered true and not due to measurement error.27 This was not the case for the
Upper Extremity and Mental and Emotional Problems subscale. Both subsca-les showed MIC values that were smaller than the SDC of the subscasubsca-les. This means that a change in score that falls between the MIC value and the SDC
may be important, but cannot be distinguished from measurement error.27
This limits the interpretability of these two subscales.
In terms of accuracy, the ability of the MIC values of the SMFA-NL to discri-minate between improved and non-improved patients ranged from a “below acceptable” to a “well acceptable” level. Depending on the subscales, the Area Under the Curve (AUC) ranged from 0.65 to 0.78. Studies that evaluated MIC values of similar instruments in trauma patients reported AUC values ranging from 0.65 to 0.70. Though the SMFA-NL performed slightly better compared to these studies, a better discriminative ability of the SMFA-NL would have been preferable for a clearer interpretation of changes in scores.
The level of inaccuracy (e.g. amount of misclassification) of the MIC values may be due to a limitation of the anchor-based method that was used. The single item that is used to determine whether the patient has changed (i.e. the anchor) has been criticized.28 It is assumed that a single item is less
reliable and less valid than a multi-item instrument, therefore the single-item criterion may be a profound source of inaccuracy. A second explanation suggested by Terwee et al. is that MIC values may be dependent on groups or received treatment.28 If the MIC value is specific for each injury type and
injury severity, calculating MIC values in a heterogeneous sample – such as the one studied in this thesis – could lead to profound misclassification. Some presumptive evidence for the dependency of injury severity on the MIC values was found in this thesis. MIC values of patients with at least one injury to the upper or lower extremity were higher on the extremity-specific subscales than for the complete sample. A definitive cause for the inaccuracy could not be determined though. Further research is needed to disentangle the source of
misclassification in anchor-based MIC values and the dependency on popu-lation characteristics.
In clinical practice, the MIC values of the SMFA-NL may be specifically helpful to evaluate whether improvement in functioning of individual patients has occurred. However, physicians should always be aware of the inaccuracy that is associated to a MIC value. A substantial inaccuracy increases the risk of misclassification, which may lead to false conclusions as to whether the patient has improved importantly. In research settings, knowledge of which change in score reflects an important improvement may be used in clinical trials. In such case the MIC values may be used to discriminate between patients who
did and did not show improvement in each arm of a trial. The proportion of
improved patients can be subsequently compared between treatments.29 MIC
values may also be used to calculate required sample sizes in clinical studies.27
In Chapter 5 further efforts were made to improve the interpretability of
the SMFA-NL. To study treatment effect after trauma, information of patients’ pre- and post-injury functioning is important. Due to the acute character of traumatic injuries, true pre-injury assessments are usually not available. To interpret change in physical functioning after trauma, two different methods are frequently applied: 1) normative data are used as a reference of pre-injury functioning, and 2) recalled pre-injury functioning reported shortly after
sustaining the injury can be used as proxy for pre-injury functioning.30 When
directly compared, recalled pre-injury health status and physical functioning
were 2 to 9 points higher (better) than the general population (Chapter 5).
An extensive review of the literature shows that pre-injury health status in a broad range of trauma patients has consistently been reported to be higher than the general population’s health status (EQ-5D: 0.06 to 0.08 points, SF-36: 0 to 6 points).30 The differences reported in the literature are comparable to
those found in Chapter 5.
Throughout the literature, interpretation of this difference in pre-injury health status and health status of the general population has been found problematic. On the one hand, it has been hypothesized that the difference is true, and that trauma patients may be a specific sub-sample of the general population.31,32 On the other hand, concerns have been raised that the
differ-ences in pre-injury health status may be caused by recall bias and response
shift, rather than being a true difference between the samples.30-32 Scholten
needed to conclude whether the difference is “true”. In such a study, trauma patients could be assessed before the injury, thereby bypassing recall bias and response shift. We could not provide a study that met the suggested design.
The research in Chapter 5 however did show that adjustment for general
characteristics reduced the difference in pre-injury health status of trauma patients and the general population to 1 to 4 points for most scales. These findings indicate that the original difference was largely explained by the differences in general characteristics of the samples. The remaining difference in health status (regardless of being a true effect, response shift or recall bias) was considered to be too small to represent a meaningful difference. It was considered that there was little evidence that trauma patients had a different pre-injury health status compared to the Dutch population. As a consequence, recalled pre-injury and normative data may both be used to compare groups of patients that sustained a broad range of traumatic injuries.
Concluding remark on interpretability
Whether the SMFA-NL is a well-interpretable instrument cannot be directly concluded from just one of the studies presented in this thesis. Rather, all studies that evaluated aspects of interpretability should be reviewed from a broader perspective. Single scores can be interpreted using the distributions of
scores of groups of patients (Chapter 3). One may get an idea of which scores
are “normal” for trauma patients. Single scores may further be interpreted by means of pre-injury scores and the normative data of the Dutch population (Chapters 4 and 5). Changes in scores may be interpreted by means of the
MIC values (Chapter 6). The interpretability of change in physical
functio-ning of the upper extremity and mental and emotional problems at the indi-vidual level is limited though. Altogether, the SMFA-NL was considered to be a reasonably good interpretable instrument. Additional knowledge of the distribution of scores in patients with specific injuries may further improve the interpretability of the SMFA-NL.
Clinical implications
In complex contexts, PROMs can be used to probe domains of interest. Recovery of physical functioning after trauma is a complex context that requi-res exploration of the domains that relate to it. Domains that are considered important to patients with acute musculoskeletal conditions have been defined in a core set of the International Classification of Functioning, Disability and
Health (ICF).33 The items of the SMFA-NL closely match with the items in the
“Activity Limitations” domain of the core set. Hence the SMFA-NL may be used to investigate most of the relevant aspects of the “Activity Limitations” domain of the ICF model.
At the individual patient level, repeated evaluation of physical function-ing is one of the cornerstones of post-trauma follow-up. Trauma may cause serious disabilities, impacting lives of patients and their relatives, so recov-ery of physical functioning should be strictly monitored. When patients are evaluated by a different physician each time, it is difficult to get a clear idea of the recovery process. Based on individual patient records it can be difficult to determine how much a patient has improved and whether it is sufficient. The same applies when the intervals between assessments are long, and both physician and patient may have forgotten about the previous level of physical functioning. In such cases a high-quality standardized assessment tool may be useful as an additional instrument to quantify physical functioning in a follow-up trajectory. The SMFA-NL may be used as a standardized measure-ment instrumeasure-ment in clinical care, to quantify (the improvemeasure-ment in) physical functioning of trauma patients.
At the administrative level, improving quality of healthcare services is a priority of healthcare providers and decision-makers. Quality of care can be divided into clinical effectiveness, safety and patient experience.34 PROMs and
patient-reported experience measures (PREMs) are increasingly recognized as quality indicators.35 Quality registries such as the Dutch Institute for
Clini-cal Auditing grant PROMs and PREMs an important role in the evaluation of quality of care. Considering its properties, the SMFA-NL may be used to eval-uate physical functioning after trauma, which can also serve as a benchmark to evaluate treatment results between healthcare professionals or between hospitals.
In Chapter 7 an additional interesting application for PROMs was
investi-gated: whether recovery of physical functioning after trauma can be accurately predicted. Due to the traditional focus on mortality, trauma-related prognostic models also used to focus on the prediction of mortality. The Injury Severity Score and the Revised Trauma Score, for example, are still frequently used in research and clinical practice. Since the late 1960s and early 1970s, inju-ry-related mortality is decreasing at a global level, and the Netherlands is no exception.36,37 Nowadays the vast majority of patients survive trauma, and the
burden of injury-related disability is increasing. As a consequence, the need
for a model that can be used to predict functional outcome has emerged.1,38
To the best of my knowledge, a model that is capable of predicting func-tional recovery after trauma had not been previously reported. To address this gap in the literature, the PROgnosis of functional recovery after Trauma (PRO-Trauma) prediction model was developed, enabling physicians and researchers to make long-term prognoses of functional recovery of individual patients at an early stage of treatment. Predictors of functional recovery were the number of chronic health conditions, having a partner, length of hospital stay and the SMFA-NL Problems with Daily Activities score at 6 weeks post-in-jury. The model was internally validated and showed good calibration. The final model showed a reasonably strong discriminative performance to identify patients likely to be functionally recovered after one year.
In current clinical practice, prognosis of functional recovery and expected injury-related disability is based on the experience of the physician treating the patient. But even with tremendous clinical experience, it remains diffi-cult to quantify an individual’s chances of functional recovery. It is known
that trauma patients often have unrealistic expectations of their recovery.39
The PRO-Trauma prediction model may enable clinicians to provide patients with a quantified estimate of their chances of functional recovery. A quanti-fied chance of being functionally recovered after one year may lead to more realistic recovery expectations and health literacy of trauma patients.39,40 In
the end, shared decision-making may benefit from patients who have a better understanding of their recovery prospects.
The importance of having a partner for recovery of physical functioning was a notable finding. Patients without a partner were at risk of not attaining functional recovery one year post-injury. Having a partner may be a proxy that is reflective of a complex context, namely the importance of social support. Previous research has identified the positive relation between social support
and health, and the benefits of social support for rehabilitation.41-43 After
sustaining trauma, patients frequently face problems that are not confined to the physical, but also comprise emotional, psychological, occupational and social issues.44-48 Strong social support is likely to be beneficial for overcoming
these problems. Family structures and social support have been associated with better functional recovery, less experienced pain and a higher return-to-work rate among trauma patients.45-47 Although the importance of strong social
support for recovery after trauma appears straightforward, in clinical practice the role of social support receives relatively little attention. This may be parti-ally due to a lack of well-studied measurement instruments that are easy to use. Such instruments have been developed and validated in patients with e.g. chronic orthopedic conditions, but have not been evaluated in patients with acute traumatic injuries.49 It is currently not known how exactly social support
relates to recovery after trauma, or how it could be used to improve recovery after sustaining trauma. Hence social support itself, as well as the validation of instruments that can assess social support in trauma patients, could be the subject of additional research.
The PRO-Trauma prediction model was developed for trauma patients without traumatic brain injury or neurological deficits, which may be consi-dered a limitation. Traumatic brain injury and neurological deficits due to trauma have different recovery patterns than musculoskeletal injuries. The model may therefore not be generalizable to patients with such injuries. In addition, the PRO-Trauma prediction model is not externally validated. The validity of the model should be further evaluated in an external validation procedure, with data from a different hospital.
The SMFA-NL has to be placed in the context of problems it is intended to be used in. This may provide some extra guidance on its use. In the
intro-duction of this thesis (Chapter 1), various contexts of problems were defined:
simple, complicated and complex problems were each shown to require a diffe-rent approach. (Recovery of) physical functioning in patients that sustained acute traumatic injuries is a complex context. Recalling from the introduc-tion, analytic reduction of complex contexts can be considered an improper approach. For this reason, the SMFA-NL should not be regarded as an analy-tical tool that provides a definitive answer to a problem. To illustrate this, a patient with a nonunion of a fractured ankle should not be rejected for surgical treatment because the SMFA-NL scores are good. Rather, complex contexts require the underlying problem to be probed and explored in order improve
our understanding of it (Chapter 1). The SMFA-NL should be considered as
an instrument that may help explore and understand the complex context of an individual patient’s (recovery of) physical functioning of an individual patient – “providing a piece of the puzzle”. Similarly, good SMFA-NL scores may provide a part of the information required to decide whether the patient with the nonunion of a fractured ankle may be treated surgically or not.
Conclusion
Physical functioning is impaired by musculoskeletal injuries that may lead to long-term disability. Patient-reported outcome measures may aid in the exploration of problems with physical functioning and recovery after trauma. The research presented in this thesis showed that the SMFA-NL can be used to make high-quality measurements of physical functioning in patients that sustained physical trauma. The SMFA-NL may be used to assess physical functioning at a single moment, as well as changes in physical functioning over time. In a population of patients with a broad range of acute traumatic injuries, the SMFA-NL is a reasonably good interpretable instrument. Single scores can be interpreted by the distribution of scores of patients with acute traumatic injuries and with the normative data of the Dutch population. Changes in scores can be interpreted by means of minimal important change values. The PRO-Trauma prediction model can be used to predict functional recovery after one year.
Future perspectives
The research presented in this thesis provided insight into the measure-ment and prediction of physical functioning after trauma. Physical functio-ning is mostly reflected by the “Activity” domain of the ICF model, whereas traditional outcomes are reflective of the “Body functions and structures” domain.33,50 However, in order to get a complete picture of a patient’s
functi-oning and disability, the “Participation” domain and environmental factors
should also be taken into account.33,50 The research presented in Chapter 7
of this thesis already showed that constructs such as social support may have a profound effect on recovery after trauma. Currently the domain of partici-pation is poorly understood and frequently overlooked. This is partly due to a lack of knowledge about how this domain relates to outcome and quality of life after trauma, and how it should be used to improve outcome after trauma. Another reason may be a lack of well-evaluated measurement instruments. Future research may focus on developing a better understanding of the parti-cipation domain in relation to recovery after trauma.
A methodological aspect that requires further attention relates to the inter-pretability of changes in scores. In the literature, minimally important change (MIC) values are calculated using several methods. Terwee et al. showed
that different methods yield different MIC values.28 At present there is no
performance measures that characterize quality of an MIC value. There is a need for (a consensus-based) standardization of calculating and reporting MIC values.
Due to the shift in thinking about what health is and how it should be measured, PROMs have become increasingly popular. As a consequence, the number of PROMs that have been developed has increased dramatically. For example, over 140 outcome instruments have been developed to evaluate (aspects of) physical functioning of the upper extremity alone.51 The use of a
large number of PROMs that measure the same construct has led to several problems. One of the most serious issues is that all these instruments have different metrics, so studies that use different PROMS are very hard, if not impossible, to compare. There is a need for standardization and internatio-nal consensus within the various clinical disciplines on which PROMs should be used. One initiative that encourages such standardization is the Patient
Reported Outcome Measurement Information System (PROMIS).3 PROMIS
provides large item banks that have one common metric and attempts to stan-dardize PROMs. Future studies may be used to (cross-culturally) validate and evaluate measurement properties of item banks, such as that developed by PROMIS, encouraging the development of a standardized outcome measure.
The vast majority of PROMs currently use classical test theory (CTT) as underlying theory to measure the construct of interest. In CTT, the items of the instrument are hypothesized to identically reflect the construct of interest, and can be averaged to eliminate the measurement error accompanying each item. Item response theory (IRT) is an alternative measurement theory used in the PROMIS system. An interesting aspect of IRT is that items are considered to not identically reflect the construct of interest. For example, an item that asks whether patients can run 200 meters is more difficult than an item that asks whether a patient can walk 50 meters. IRT accounts for this difference in difficulty between items, since IRT-based instruments use the (dis)ability of the items themselves as the scale of the instrument, rather than a calcu-lated average of multiple items as a scale, as with CTT. A direct consequence of testing at item level rather than at scale level is that it enables computer adaptive testing (CAT). This is one of the major advantages of IRT. With CAT, a computer algorithm dynamically selects the next item based on all previously administered items, so that the next item retrieves the most useful information from the patient. In the process, CAT interactively probes the (dis)ability of
patients with a minimal number of questions. IRT-CAT systems shorten the test length and limit the administrative burden to the patient without losing information of interest. IRT-CAT systems can be considered a “next step” in the use of PROMs and in embracing complexity. Still, intuitive and easily applicable CAT software needs to be further developed. In addition, IRT-based items and questionnaires should be developed and knowledge about IRT-ba-sed measurement instruments propagated to facilitate widespread adoption. Finally, fracture treatment can be considered the working horse of trauma surgeons. Nonetheless, even for the most common fractures there is insuffi-cient evidence to decide which of the different treatment options is preferred in terms of functional outcome.52-61 This is remarkable, as functional outcome
should be one of the most important aspects in the decision for a specific treatment. The lack of knowledge regarding the effects of specific treatments on functional outcome is illustrative of the uncharted territory that may be studied using PROMs. Increased general awareness of the possibilities of using PROMs among clinicians and researchers may be an important step toward further improving the treatment of trauma patients.
References
1. Polinder S, Haagsma JA, Belt E, et al. A systematic review of studies measuring health-related quality of life of general injury populations. BMC Public Health. 2010;10:783-2458-10-783.
2. Mokkink LB, Terwee CB, Patrick DL, et al. The COSMIN checklist for assessing the methodological quality of studies on measurement properties of health status measurement instruments: An international delphi study. Qual Life Res. 2010;19(4):539-549.
3. HealthMeasures.net. List of health measures: Available PROMIS measures for adults. List of Health Measures: Available PROMIS measures for adults Web site. http://www. healthmeasures.net/explore-measurement-systems/promis/intro-to-promis/list-of-adult-measures. Published December 2017. Updated 2017. Accessed July, 17, 2018.
4. Mokkink LB, Terwee CB, Patrick DL, et al. The COSMIN study reached internatio-nal consensus on taxonomy, terminology, and definitions of measurement properties for health-related patient-reported outcomes. J Clin Epidemiol. 2010;63(7):737-745.
5. Mokkink LB, Terwee CB, Knol DL, et al. The COSMIN checklist for evaluating the methodological quality of studies on measurement properties: A clarification of its content. BMC Med Res Methodol. 2010;10:22-2288-10-22.
6. Mokkink LB, De Vet HC, Prinsen CA, et al. COSMIN risk of bias checklist for systematic reviews of patient-reported outcome measures. Quality of Life Research. 2018;27(5):1171-1179.
7. Swiontkowski MF, Engelberg R, Martin DP, Agel J. Short musculoskeletal function assessment questionnaire: Validity, reliability, and responsiveness. J Bone Joint Surg Am. 1999;81(9):1245-1260.
8. Bouffard J, Bertrand-Charette M, Roy J. Psychometric properties of the musculoskele-tal function assessment and the short musculoskelemusculoskele-tal function assessment: A systematic review. Clinical Rehabilitation. 2015.
9. Barei DP, Agel J, Swiontkowski MF. Current utilization, interpretation, and recom-mendations: The musculoskeletal function assessments (MFA/SMFA). J Orthop Trauma. 2007;21(10):738-742.
10. Taylor MK, Pietrobon R, Menezes A, et al. Cross-cultural adaptation and validation of the brazilian portuguese version of the short musculoskeletal function assessment ques-tionnaire: The SMFA-BR. J Bone Joint Surg Am. 2005;87(4):788-794.
11. Guevara CJ, Cook C, Pietrobon R, et al. Validation of a spanish version of the short musculoskeletal function assessment questionnaire (SMFA). J Orthop Trauma. 2006;20(9):623-9; discussion 629-30; author reply 630.
12. Reininga IH, el Moumni M, Bulstra SK, Olthof MG, Wendt KW, Stevens M. Cross-cul-tural adaptation of the dutch short musculoskeletal function assessment questionnaire (SMFA-NL): Internal consistency, validity, repeatability and responsiveness. Injury. 2012;43(6):726-733.
13. Lindahl M, Andersen S, Joergensen A, Frandsen C, Jensen L, Benedikz E. Cross-cul-tural adaptation and validation of the danish version of the short musculoskeletal function assessment questionnaire (SMFA). Qual Life Res. 2017.
14. Wang Y, He Z, Lei L, et al. Reliability and validity of the chinese version of the short musculoskeletal function assessment questionnaire in patients with skeletal muscle injury of the upper or lower extremities. BMC Musculoskelet Disord. 2015;16:161-015-0617-z.
15. Suk M, Hanson B, Norvell D, Helfet D. Musculoskeletal outcomes measures and instruments. 2nd Edition ed. Stuttgart: Thieme; 2009.
16. van Delft-Schreurs CCHM, van Son MAC, de Jongh MAC, Gosens T, Verhofstad MHJ, de Vries J. Psychometric properties of the dutch short musculoskeletal function assess-ment (SMFA) questionnaire in severely injured patients. Injury. 2016;47(9):2034-2040. doi: https://doi.org/10.1016/j.injury.2016.03.006.
17. Van Son MA, Den Oudsten BL, Roukema JA, Gosens T, Verhofstad MH, De Vries J. Psychometric properties of the dutch short musculoskeletal function assessment (SMFA) questionnaire in patients with a fracture of the upper or lower extremity. Qual Life Res. 2014;23(3):917-926.
18. Beckerman H, Roebroeck ME, Lankhorst GJ, Becher JG, Bezemer PD, Verbeek ALM. Smallest real difference, a link between reproducibility and responsiveness. Qual Life Res. 2001;10.
19. Martin DP, Engelberg R, Agel J, Snapp D, Swiontkowski MF. Development of a muscu-loskeletal extremity health status instrument: The muscumuscu-loskeletal function assessment instrument. J Orthop Res. 1996;14(2):173-181.
20. Hudak PL, Amadio PC, Bombardier C. Development of an upper extremity outcome measure: The DASH (disabilities of the arm, shoulder and hand) [corrected]. the upper extremity collaborative group (UECG). Am J Ind Med. 1996;29(6):602-608.
21. Slobogean GP, Noonan VK, O’Brien PJ. The reliability and validity of the disabilities of arm, shoulder, and hand, EuroQol-5D, health utilities index, and short form-6D outcome instruments in patients with proximal humeral fractures. Journal of Shoulder and Elbow Surgery. 2010;19(3):342-348.
22. Westphal T, Piatek S, Schubert S, Schuschke T, Winckler S. Reliability and validity of the upper limb DASH questionnaire in patients with distal radius fractures. Z Orthop Ihre Grenzgeb. 2002;140(4):447-451.
23. Schonnemann JO, Larsen K, Hansen TB, Soballe K. Reliability and validity of the danish version of the disabilities of arm, shoulder, and hand questionnaire in patients with fractured wrists. J Plast Surg Hand Surg. 2011;45(1):35-39.
24. De Vet HCW, Terwee CB, Mokkink LB, Knol DL. Measurement in medicine. Cambridge University Press; 2011.
25. Hunsaker FG, Cioffi DA, Amadio PC, Wright JG, Caughlin B. The american academy of orthopaedic surgeons outcomes instruments: Normative values from the general popu-lation. J Bone Joint Surg Am. 2002;84-A(2):208-215.
26. Kendall PC, Sheldrick RC. Normative data for normative comparisons. J Consult Clin Psychol. 2000;68(5):767-773.
27. Terwee CB, Roorda LD, Knol DL, De Boer MR, De Vet HC. Linking measurement error to minimal important change of patient-reported outcomes. J Clin Epidemiol. 2009;62(10):1062-1067.
28. Terwee CB, Roorda LD, Dekker J, et al. Mind the MIC: Large variation among popu-lations and methods. J Clin Epidemiol. 2010;63(5):524-534.
29. Schünemann HJ, Akl EA, Guyatt GH. Interpreting the results of patient reported outcome measures in clinical trials: The clinician’s perspective. Health and quality of life outcomes. 2006;4(1):62.
30. Scholten AC, Haagsma JA, Steyerberg EW, van Beeck EF, Polinder S. Assessment of pre-injury health-related quality of life: A systematic review. Popul Health Metr. 2017;15(1):10-017-0127-3.
31. Wilson R, Derrett S, Hansen P, Langley J. Retrospective evaluation versus popula-tion norms for the measurement of baseline health status. Health Qual Life Outcomes. 2012;10:68-7525-10-68.
32. Gabbe BJ, Cameron PA, Graves SE, Williamson OD, Edwards ER, Victorian Orthopae-dic Trauma Outcomes Registry (VOTOR) Project Group. Preinjury status: Are orthopaeOrthopae-dic trauma patients different than the general population? J Orthop Trauma. 2007;21(4):223-228.
33. Scheuringer M, Stucki G, Huber EO, et al. ICF core set for patients with musculos-keletal conditions in early post-acute rehabilitation facilities. Disabil Rehabil. 2005;27(7-8):405-410.
34. National Health Service. Our approach to patient safety - NHS improvement. . 2017;1(1):1-24.
35. Kingsley C, Patel S. Patient-reported outcome measures and patient-reported expe-rience measures. Bja Education. 2017;17(4):137-144.
36. Aarts J,H. Medische verbeteringen: Eenliteratuurstudie over de consequenties van de medische wetenschp en verbeterde hulpverlening op de aantallen geregistreerde verkeers-doden. . 1989;1:1-47.
37. Gunst M, Ghaemmaghami V, Gruszecki A, Urban J, Frankel H, Shafi S. Changing epidemiology of trauma deaths leads to a bimodal distribution. . 2010;23(4):349-354.
38. de Jongh MA, Kruithof N, Gosens T, et al. Prevalence, recovery patterns and predictors of quality of life and costs after non-fatal injury: The brabant injury outcome surveillance (BIOS) study. Inj Prev. 2017;23(1):59-2016-042032. Epub 2016 May 6.
39. Sleney J, Christie N, Earthy S, Lyons RA, Kendrick D, Towner E. Improving recove-ry-learning from patients’ experiences after injury: A qualitative study. Injury. 2014;45(1):312-319.
40. Kadakia RJ, Tsahakis JM, Issar NM, et al. Health literacy in an orthopedic trauma patient population: A cross-sectional survey of patient comprehension. J Orthop Trauma. 2013;27(8):467-471.
41. Hupcey JE. The meaning of social support for the critically ill patient. Intensive and Critical Care Nursing. 2001;17(4):206-212. doi: https://doi.org/10.1054/iccn.2000.1568.
42. Stevens M, van den Akker-Scheek I, Spriensma A, Boss NA, Diercks RL, van Horn JR. The groningen orthopedic exit strategy (GOES): A home-based support program for total hip and knee arthroplasty patients after shortened hospital stay. Patient Educ Couns. 2004;54(1):95-99.
43. Sarason IG, Levine HM, Basham RB, Sarason BR. Assessing social support: The social support questionnaire. J Pers Soc Psychol. 1983;44(1):127-139.
44. Gironda MW, Lui A. Social support and resource needs as mediators of recovery after facial injury. Oral Maxillofac Surg Clin North Am. 2010;22(2):251-259.
45. Prang KH, Berecki-Gisolf J, Newnam S. Recovery from musculoskeletal injury: The role of social support following a transport accident. Health Qual Life Outcomes. 2015;13:97-015-0291-8.
46. Prang KH, Berecki-Gisolf J, Newnam S. The influence of social support on health-care service use following transport-related musculoskeletal injury. BMC Health Serv Res. 2016;16:310-016-1582-4.
47. Prang KH, Newnam S, Berecki-Gisolf J. “That’s what you do for people you love”: A qualitative study of social support and recovery from a musculoskeletal injury. PLoS One. 2018;13(4):e0196337.
48. Pape HC, Probst C, Lohse R, et al. Predictors of late clinical outcome following ortho-pedic injuries after multiple trauma. J Trauma. 2010;69(5):1243-1251.
49. van den Akker-Scheek I, Stevens M, Spriensma A, van Horn JR. Groningen orthopae-dic social support scale: Validity and reliability. J Adv Nurs. 2004;47(1):57-63.
50. World Health Organization. International classification of functioning disability and health: ICF, geneva, switzerland. . 2001(World Health Assembly, Geneva, Switzerland).
51. Jayakumar P, Williams M, Ring D, Lamb S, Gwilym S. A systematic review of outcome measures assessing disability following upper extremity trauma. JAAOS Global Research & Reviews. 2017;1(4):e021.
52. Handoll HH, Madhok R. Conservative interventions for treating distal radial fractures in adults. Cochrane Database Syst Rev. 2000;(2):CD000314. doi(2):CD000314.
53. Handoll HH, Parker MJ. Conservative versus operative treatment for hip fractures in adults. Cochrane Database Syst Rev. 2008;(3):CD000337. doi(3):CD000337.
54. Handoll HH, Huntley JS, Madhok R. Different methods of external fixation for trea-ting distal radial fractures in adults. The Cochrane database of systematic reviews. 2008.
55. Sayum Filho J, Lenza M, Teixeira de Carvalho R, Pires OG, Cohen M, Belloti JC. Inter-ventions for treating fractures of the patella in adults. The Cochrane Library. 2015.
56. Handoll HH, Pearce P. Interventions for treating isolated diaphyseal fractures of the ulna in adults. Cochrane Database Syst Rev. 2012;(6):CD000523. doi(6):CD000523.
57. Handoll HH, Brorson S. Interventions for treating proximal humeral fractures in adults. Cochrane Database of Systematic Reviews. 2015.
58. McNamara IR, Smith TO, Shepperd KL, et al. Surgical fixation methods for tibial plateau fractures. Cochrane Database of Systematic Reviews. 2015(9).
59. Wang Y, Zhuo Q, Tang P, Yang W. Surgical interventions for treating distal humeral fractures in adults. Cochrane Database Syst Rev. 2013;(1):CD009890. doi(1):CD009890.
60. Bruce J, Sutherland A. Surgical versus conservative interventions for displaced intra-articular calcaneal fractures. Cochrane Database Syst Rev. 2013;(1):CD008628. doi(1):CD008628.
61. Gosler MW, Testroote M, Morrenhof JW, Janzing HM. Surgical versus non-surgical interventions for treating humeral shaft fractures in adults. Cochrane Database Syst Rev. 2012;1:CD008832.
