PhD Thesis
Translation and validation of ‘the STarT Back Tool’
– a clinical screening tool for predicting outcome and guiding
targeted treatment for patients with low back pain
Lars Morsø, PT, MPH
Research Department, Spine Centre of Southern Denmark,
Hospital Lillebaelt, Middelfart
Institute of Regional Health Services Research, University of Southern Denmark
Faculty of Health Sciences
University of Southern Denmark
2013
Table of contents
Preface
1
Supervisors
1
List of publications
2
Acknowledgement
3
Summary in Danish
6
Summary in English
8
Framework of the thesis
10
Background
12
Objective and aims
16
Overall methods
17
Designs
17
Materials
18
Analyses
19
Ethics
19
Study 1: ‘Translation and discriminative validation of the SBT’
Questions and considerations of the process, part I
Study 2: ‘The predictive and external validity of the SBT in Danish primary care’
Questions and considerations of the process, part II
21
24
26
29
Study 3: ‘Is the psychosocial profile of people with low back pain seeking care in
Danish primary care different from those in secondary care?’
Questions and considerations of the process, part III
32
35
I
Study 4: ‘The predictive ability of the SBT in a Danish secondary care setting’
36
Overall Discussion
40
Perspectives
46
Overall Conclusions
48
References
49
Papers
56
Manuscript 1
Manuscript 2
Manuscript 3
Manuscript 4
Appendix
i
II
Preface
________________________
Preface
Supervisors
Claus Manniche, Professor, D.M.Sc (main supervisor)
Research Department, Spine Centre of Southern Denmark, Hospital Lillebaelt, Middelfart, Institute
of Regional Health Services Research, University of Southern Denmark
Hanne B. Albert, MPH, PhD (project supervisor)
Research Department, Spine Centre of Southern Denmark, Hospital Lillebaelt, Middelfart, Institute
of Regional Health Services Research, University of Southern Denmark
Peter Kent, PhD (project supervisor)
Research Department, Spine Centre of Southern Denmark, Hospital Lillebaelt, Middelfart, Institute
of Regional Health Services Research, University of Southern Denmark
1
Preface
________________________
List of publications
1. Translation and discriminative validation of the STarT Back Screening Tool into Danish.
Morsø L, Albert H, Kent P, Manniche C, Hill J. Eur Spine J. 2011 Dec;20(12):2166-73.
2. The predictive and external validity of the STarT Back Tool in Danish primary care.
Morsø L, Kent P, Albert HB, Hill JC, Kongsted A, Manniche C. Eur Spine J. 2013 Feb 10.
[Epub ahead of print]
3. Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?
Morsø L, Kent P, Albert HB, Manniche C. Man Ther. 2013 Feb;18(1):54-9.
4. The predictive ability of the STarT Back Screening tool in a Danish secondary care setting.
Morsø L, Kent P, Albert HB, Manniche C. (Submitted to Eur Spine J. 18. February 2013)
2
Preface
________________________
Acknowledgement
This thesis is based on four cohort studies carried out from 2010 to 2013. The project could not
have been undertaken without the political willingness and vision from the Region of Southern
Denmark. In 2010, a decision was made to dedicate money to the research of various chronic
conditions, including low back pain. Substantial financial support was provided from this fund in
the Region of Southern Denmark. Financial support was also generously given by the University
of Southern Denmark, the Danish Physiotherapy Association, the Danish Rheumatism
Association and the Danish Society for General Practice.
Many people have willingly (and unwillingly) been involved in the process of this project. I would
like to express my heartfelt thanks and gratitude to everyone for their support during the completion
of my thesis, and I apologise that I cannot thank every one of you by name.
I wish to thank;
Claus Manniche, Professor, D.M.Sc, for his open-minded way of thinking in initiating this project
and for his support during the process to develop my academic skills and my physiotherapy career
at the Spine Centre.
Hanne B. Albert, Ph.D., for her never-ending optimism, friendship, kind nature and faith in me, for
critical and constructive input throughout the last 3 years and for sharing thoughts and ideas in
times of difficulty. Thanks for never doubting me.
Peter Kent, Ph.D., for his extensive and highly skilled supervision, for always having time for ‘just
one more question’, for having the ability to make me feel competent, even when I was asking the
most stupid questions, and for his kindness and support. It has been a privilege to work with you.
3
Preface
________________________
Jonathan Hill, Ph.D., for letting me have access to STarT-data and for inviting me to visit Keele
University, for his hospitality and kind nature during the stay, for his constructive critical advice,
always wanting the best for me.
Lene Ververs, Anne Marie Rosager, Heidi L. Rasmussen and Pernille K. Clausen, the
secretaries in the Research Department, for their willingness to help me, no matter how much my
requests interfered with their work.
My colleagues in the Research Department, Bodil Arnbak, Rikke K. Jensen, Tue S. Jensen, Per
Kjær, Søren O’Niell, Thomas B. Nielsen and Alice Kongsted (the perfect ‘office buddy’) for
interest, support and collaboration.
Suzanne Capell, for linguistic editing, revision and constructive feedback.
Lene Marie Isaksen and Dorte Lemvigh from the management of the Spine Centre for their
support.
Kirsten Kyvik, head of Institute of Regional Health Services for her sincere interest in the PhD
project.
The clinicians at the Spine Centre for their interest through the project and for their patience
during my ‘appraisal’ of the SBT.
The Spine Centre ‘home boys’ who always had time for a beer on Friday after work.
John Banke, Flemming Pedersen, the primary care physiotherapy clinics, the GP research
unit, Henrik Schroll and DAK-E, for involvement and enthusiasm in collecting the primary care
data.
Researchers from the Keele University, Staffordshire, England for kindly providing data from
English primary care.
4
Preface
________________________
Finally, I would like to thank my friends and family, and most of all the three most important
people in my life, my fiancée Camilla and my children Sven and Thea, for their love, support and
endless patience.
This thesis is dedicated to my sister Carina.
Lars Morsø, 1 March, 2013
5
Summary
_________________________
Summary in Danish
Baggrund
STarT Back Screening Tool (SBT) er et 9-punkts patientspørgeskema til inddeling af patienter med
uspecifikke lænderyg smerter. Dette korte multidimensionelle spørgeskema er udviklet til primær
sektor og kan identificere modificérbare risiko faktorer som udbredt smerte, funktionsnedsættelse
og psykosociale elementer. Patienterne inddeles i 3 subgrupper (lav, mellem eller høj) alt efter
risiko for dårlig prognose. Det er vist, at SBT har både prognostisk og behandlingsmæssig
implikation. Det aktuelle Ph.d. projekt havde til mål at: (i) Oversætte SBT til dansk, (ii) Teste den
interne validitet, (iii) Teste den prædiktive validitet i dansk primær sektor, (iv) Undersøge forskelle
i den psykosociale patient profil mellem dansk primær- og sekundærsektor, (v) Undersøge den
prædiktive evne i dansk sekundær sektor.
Metode
Oversættelsen af SBT blev udført efter metoder anbefalet i internationale guidelines, og den
diskriminative validering af SBT blev foretaget, og sammenlignet med tidligere engelske resultater.
SBT’s prædiktive værdi i primær sektoren i Danmark blev undersøgt og sammenlignet med
resultater fra engelsk primær sektor. Forskelle i den psykosociale profil hos rygpatienter i primærog sekundærsektor, samt den prædiktive værdi af SBT i sekundær sektoren i Danmark blev udført
ved at anvende og sammenligne data fra primær- og sekundær sektor.
Resultater
Den danske oversættelse af SBT var sproglig præcis og kunne anvendes af patienter, til trods for
forskelle fundet ved validering af den psykosocial subskala. SBT blev fundet brugbart, med
tilstrækkelig diskriminativ evne til at kunne anvendes i primær sektor i Danmark. Den prædiktive
6
Summary
_________________________
evne i lav- og mellemrisiko grupperne var i overensstemmelse med fund fra England, hvorimod
SBT viste reduceret evne til at forudsige prognose i højrisikogruppen.
Sammenligning af den psykosociale patientprofil hos patienter fra dansk primær- og
sekundærsektor viste signifikant højere grad af bevægeangst og katastroferingsadfærd hos patienter
fra sekundærsektor, derimod var
Hvad var kendt inden dette Ph.d. projekt?
de
mindre
’frygtsomme’
end
patienter fra primærsektor. På
trods af signifikante forskelle på
disse
parametre,
vurderedes
•
SBT kan identificere modificerbare risiko faktorer i primær sektor.
•
SBT kan klassificere patienter i relevante subgrupper
•
SBT har prognostisk og behandlingsmæssig implikation.
•
Målrettet SBT behandling har vist klinisk og økonomisk effekt.
forskellene til at være af en
Hvad har dette Ph.d. projekt bidraget med?
størrelsesorden, som ikke gjorde
•
Oversættelsen af SBT til dansk er forståelig og anvendelig.
dem klinisk relevante. Test af den
•
Den diskriminative validitet af SBT acceptabel
prædiktive evne i sekundær sektor
•
Den prædiktive værdi i primær sektor er acceptabel.
•
Prædiktion af prognose i sekundær sektor reduceret.
viste, at SBT i mindre grad kunne
forudsige prognose ved 6 måneders opfølgning i sekundær sektor sammenlignet med primær sektor.
Konklusion
Samlet set viste resultaterne fra oversættelsen, valideringen og test af den prædiktiv evne, at SBT er
et anvendeligt og brugbart klassifiseringsredskab i dansk primær sektor. På trods af
sammenlignelige psykosociale patient profiler på tvær af sektorer, så var SBT’s evne til at forudsige
prognose i sekundær sektor ikke så stærk som i primær sektor.
7
Summary
_________________________
Summary in English
Introduction
The STarT Back screening Tool (SBT) is a nine-item patient self-report questionnaire for triage of
non-specific low back pain patients in primary care. This short multidimensional questionnaire
identifies modifiable risk factors such as pain, activity limitation and psychosocial constructs, and
its three-level classification (low, medium, high risk of poor outcome) has prognostic and treatment
implications. This project: (i) translated the SBT into Danish, (ii) tested its concurrent validity, (iii)
quantified its predictive validity in Danish primary care, (iv) investigated differences in
psychosocial characteristics between Danish primary and secondary care settings, and (v) quantified
its predictive validity in a Danish secondary care setting.
Methods
The translation was performed using methods recommended by international guidelines, and the
concurrent validity of the questionnaire was performed cross-culturally using Danish and UK
datasets. The predictive validity of the SBT in primary care was described and compared crossculturally also using data from Danish and UK primary care. Differences in psychosocial
characteristics and secondary care
What was known prior to this PhD project?
predictive validity were studies
using data from Danish primary and
•
The SBT can identify modifiable risk factors in primary care.
secondary care.
•
The SBT can classify patients into relevant subgroups.
•
The SBT has prognostic and treatment implications.
•
The targeting of treatment has been shown to have positive
Results
clinical effectiveness and economic impact.
The Danish SBT translation was
linguistically accurate and, despite
differences
found
in
What does this PhD project add?
the
performance of the psychosocial
•
sub-scale, the resultant version of
The Danish translation of SBT was linguistically accurate and
acceptable to patients.
the SBT had sufficient patient
•
The discriminative validity of SBT was acceptable.
acceptability
discriminative
•
The predictive ability in primary care was acceptable.
validity to be used in Denmark. The
•
The predictive ability of SBT in secondary care was less.
and
predictive ability of the low- and
8
Summary
_________________________
medium-risk SBT subgroups in Danish primary care was similar to that in UK primary care but was
slightly reduced in the high-risk group in DK primary care.
The comparison of patient psychosocial profiles across Danish primary and secondary care settings
showed significantly higher movement-related fear and catastrophisation in secondary care but
lower anxiety. However, the size of these differences was unlikely to be clinically important.
Testing of the SBT predictive validity in secondary care showed it was less able to predict poor
outcome at 6-month follow-up in a Danish secondary care setting than in a Danish primary care
setting.
Conclusion
Collectively, the results from these studies on the translation, discriminative validity and predictive
validity of the Danish SBT indicate that it is suitable as a triage tool in primary care. Although there
were no clinically important differences in the psychosocial profile of patients between primary and
secondary care, the predictive ability of the SBT classification subgroups was weaker in Danish
secondary care which there may be many reasons for.
9
Framework
________________________
Framework of the thesis
Overall, the PhD project consisted of four studies that each addressed components of the overall
objective. From these studies, four papers emerged that describe the creation and validation process
of the Danish version of the SBT.
The first component of the objective was to investigate whether the SBT was able to identify
subgroups of patients with different risks of poor outcome in Danish primary care. This component
was addressed by ‘Translation and discriminative validation of the STarT Back Screening Tool into
Danish’ (Paper 1), and ‘The predictive and external validity of the STarT Back Tool in Danish
primary care’ (Paper 2).
The second component of the objective was to explore whether the SBT might have some
applicability in Danish secondary care, which was addressed by ‘Is the psychosocial profile of
people with low back pain seeking care in Danish primary care different from those in secondary
care?’ (Paper 3), and ‘The predictive ability of the STarT Back Screening tool in a Danish
secondary care setting’ (Paper 4).
The content of this thesis summarises and expands selected background, methods, results and
discussion points from the PhD project. Each of the four studies will be summarised and some of
the dynamics in the validation process of the Danish SBT will be described. Discussions, questions
and considerations that emerged in the process will be addressed in bridging sections of ‘Questions
and considerations of the process’, between each of the four studies. This flowchart (Figure 1) will
be used to guide the reader through the thesis:
10
Framework
_________________________
Preface - Summary
Framework
Background
Objective
Overall methods
Study 1
Study 2
Study 3
Study 4
Considerations
Considerations
Considerations
Considerations
Overall Discussion
Perspective
Overall Conclusions
Figure 1. Flowchart of the thesis
11
Background
________________________
Background
Low Back Pain (LBP) is a common condition with a lifetime prevalence as high as 80% in adults
[1] and a one year prevalence of 55% in a Danish cohort [2]. A 2010 report from the National
Institute of Public Health in Denmark based on 173,129 respondents showed that 51.3% had been
bothered by back pain within the previous 14 days, including 14.0% who had been ‘very bothered’
[3]. The economic burden of LBP for society is high, and includes the costs of treatment,
rehabilitation, days off work, loss of earnings and early retirement. Collectively, this Danish
economic burden has been estimated to be EUR 1.73 billion per year, with expenses for treatment
alone accounting for EUR .75 billion annually [4]. Besides the economic consequences, LBP also
has great consequences for individuals in terms of pain, disability, days off work, social relations,
perception of overall health and co-morbidity [3-5].
Over the last few decades, research in the field of LBP has increased [6] with an emphasis on the
epidemiology and diagnostics of LBP. However, this effort has been challenged by the
heterogeneity of LBP and by difficulty in reaching definitive tissue-specific causes for pain in most
individual patients [7-9]. These challenges are reflected in several international guidelines which
recommend triaging LBP patients into three broad groups: specific LBP, radiculopathy, and nonspecific LBP (NSLBP) [10-16]. In primary care, this method of triage still leaves approximately
80% of patients classified in the group having NSLBP [8], and this group still contains people with
highly diverse clinical presentations. This diversity has been shown to influence outcome [17, 18].
In an attempt to address this heterogeneity and to test whether approaches other than the ‘one size
fits all’ model result in better patient outcomes, there has been an increased focus in recent years on
classification subgroups and various classification models have been suggested [19-21]. Previously,
the underlying clinical approach towards NSLBP was found to be focused on structural and
12
Background
_________________________
physical models [22] and it has been argued that, beyond the exclusion of specific serious causes of
LBP, no evidence-based agreement exists for the classification and prioritisation of NSLBP patients
in primary care. Therefore it has been suggested that alternative methods of classification be
considered [23].
Recently, there has been an acceptance of LBP as a multidimensional condition that is highly
influenced by psychosocial components [22, 24]. Guidelines also recommend assessment of
psychosocial characteristics [16], and several studies have highlighted psychosocial components as
being risk factors for poor outcome [12, 24-28]. However, the recognition of psychosocial factors in
the daily clinic can be challenging. Firstly, many clinicians feel inadequately trained to assess these
characteristics and there is evidence that clinician intuition is not very accurate [29]. Secondly,
many clinicians remain uncertain of the impact of these factors on outcome and of how to
appropriately manage these factors [30, 31]. Consequently, an increased focus has been on
questionnaires validated for detecting psychosocial constructs [26]. Unfortunately, many of these
questionnaires are very comprehensive and time-consuming [32]. Therefore, there is renewed
interest in questionnaires that screen LBP patients for psychosocial factors in practical ways in daily
care settings, especially with the purpose of stratifying prognostic risk in LBP.
The STarT Back Screening Tool (SBT) is a nine-item patient self-report questionnaire for triage of
NSLBP patients in primary care [33]. This short multidimensional questionnaire identifies
modifiable prognostic factors such as co-morbid pain, activity limitation and several psychosocial
constructs - all factors known to be risk factors for the persistence of NSLBP [34]. Answers to the
SBT questions create a total-score and sub-score, which classify patients into subgroups of people
with low, medium, or high risk of poor prognosis based on symptom complexity. This complexity
13
Background
_________________________
is based on the sum of physical and psychosocial characteristics of the individual patient (Appendix
1 – the score-flow) [33]. During the development phase that took place from 2006 to 2008 in the
United Kingdom (UK), the SBT underwent a thorough validation process that involved testing its
measurement properties, deriving the subgroup cut-points, and describing the predictive and
criterion validity of the subscale [35]. This evidence showed that, in the UK, the SBT is a reliable
and valid screening tool with adequate discriminative ability to classify patients into relevant
subgroups based on poor risk of outcome [35]. Besides being predictive of outcome, the SBT also
has treatment implications via subgroup-targeted treatment pathways [36]. The results from a highquality randomised controlled trial (RCT) in UK primary care showed that subgroup-targeted
treatment was more clinically effective and more cost-effective than usual care [37].
In the year 2009, on the basis of the UK validation work and preliminary results from the RCT, the
SBT looked appealing as a screening questionnaire for Danish primary care. In parallel with the UK
development work, the Region of Southern Denmark had an increasing focus on the assessment of
LBP patients in the Region, partly driven by a desire to improve the management of LBP patients in
the Region, but also driven by a focus on governmental expenses in the musculoskeletal field. The
Region started the development of assessment guidelines for primary care and wanted the SBT to
be included in the guidelines to assist GPs in recognising risk of poor outcome and in decisionmaking about appropriate referral and treatment pathways [38]. Although the SBT had face validity
for Danish primary care, at that time-point it had only been validated in UK primary care. No
Danish translation of the SBT was available and no definitive results of the effectiveness of targeted
treatment in any setting had been published. Many questions about its appropriateness in the Danish
context needed to be addressed and there was awareness that the sequence in which these questions
were addressed was very important. For example, would a Danish translation of the SBT retain the
14
Background
_________________________
discriminative and predictive validity of the original? To ensure an adequate foundation for the
Danish version of the SBT, a thorough translation and validation process had to be performed.
Parallel to these fundamental considerations about the applicability of the SBT for Danish primary
care, there was also a desire to explore the opportunities of expanding the SBT into other care
settings, patient populations and time-points of measurement.
15
Objective
________________________
Objective and aims
The overall objective of the thesis was to investigate whether the SBT could identify subgroups of
patients and predict risk of poor outcome in Danish primary and secondary care settings.
Therefore, the project had the following four aims:
1. To translate the English version of the SBT into Danish and to test its discriminative validity.
2. To test the predictive and external validity of the Danish version of the SBT in Danish
primary care and compare it with the English version of the SBT in UK primary care.
3. To investigate whether the psychosocial profile of patients in Danish primary and secondary
care settings were different.
4. To compare the predictive ability of the SBT in a Danish secondary care setting and a
Danish primary care setting.
16
Overall methods
________________________
Overall methods
Designs
Research of prognostic factors aims to identify factors associated with clinical outcomes. This
identification might reveal factors that are useful as modifiable targets for intervention [39].
However, often prognostic studies do not reach the high research standards in other research
designs [40]. Recently a series of papers proposed four themes as a framework for understanding
and improving prognosis research (PROGnosis RESearch Strategy or PROGRESS) [39-42]. This
research strategy was proposed to address the gap between the potential of prognostic research and
the actual impact, challenges and quality of prognostic research [40]. These challenges and
methodological flaws in prognostic modelling have also been previously described in the
investigation of LBP and other health conditions [43-45]. Despite these challenges, the potential
and opportunities of prognostic models are outlined by the PROGRESS group along with
recommendations on how to improve prognosis research [40].
In this context, different layers of prognostic questionnaire validation have been suggested in the
literature [45-47]. A fundamental assumption of this PhD project was that its purpose was not to
create a new SBT tool in the Danish context but to test the classification validity of a Danishtranslated version of the current English language SBT. Therefore, the design of this validation
process did not retest the question/factor structure or construct validity of the SBT. Instead, we built
on the construct validity work already conducted by Hill et al in the UK [33, 35]. The validation
pathway that was chosen in this PhD project was; (i) a cross-cultural validation comparing the
discriminative validity of the Danish-translated and original UK versions; (ii) a comparison of the
SBT predictive validity for poor outcome at 3 months in Danish and UK primary care cohorts; and
17
Overall methods
_________________________
(iii) a comparison of the predictive validity of poor outcome at 6 months in Danish primary and UK
secondary care cohorts.
Materials
This PhD project was based in the Medical Department of the Spine Centre of Southern Denmark.
Primary care collaborators (the Danish Quality Unit of General Practice, GPs, Physiotherapists and
the DAK-E research unit) were involved in the recruitment of patients for three of the four studies.
Cohorts from several settings were recruited to broaden the external validation. As the research
questions were different in each study, the data used in each study also varied. Two of the studies
used cross-sectional data (Studies 1 & 3) and two studies used longitudinal data (Studies 2 & 4).
The collection of data in each of the Danish cohorts occurred during the period from March 2010 to
October 2012 and some cohorts varied in their outcome time-points. Table 1 gives a visual
overview of the Danish and UK cohorts used in each study.
Table 1. Overview of cohorts used for each study
Study 1
Danish
translation cohort
from secondary
care
Baseline only
Study 2
Study 3
Study 4
Baseline only
Danish primary
cohort (GP, PT)
Danish
secondary
predictive cohort
UK primary care
development
cohort
UK primary care
BeBack cohort
Baseline only
Baseline & 3month outcomes
Baseline only
Baseline & 6month outcomes
Baseline & 3month outcomes
Baseline & 6month outcomes
Patients from Danish primary care, who were 18-65 years old, were recruited at baseline in two
ways: (1) From GPs on the basis of relevant diagnostic coding (L03= Back pain, L84=
Degenerative changes, L86= Back pain with leg pain), or (2) From physiotherapists using the
18
Overall methods
_________________________
criteria suggested in the European NSLBP guidelines [16]. Three and 6-month follow-up data were
collected for both primary care settings by way of postal questionnaires. Data from Danish
secondary care patients were collected using paper-based baseline and follow-up questionnaires for
one study and collected electronically for another. The paper-based questionnaires were
consecutively posted until 300 completed baseline questionnaires were returned. Patients recruited
electronically were included on the basis of a fully completed baseline SBT. For both the Danish
primary and secondary care cohorts, we intentionally did not require restrictive inclusion and
exclusion criteria, as we wanted the cohorts to reflect broad clinical practice. The data collection
methods and inclusion criteria of the UK primary care cohorts have been described in detail in
published studies [33, 48]. More detail on all of the cohorts is described in each of the four papers
contained in this thesis.
Analyses
In prognostic research, predictive validity has been tested using a variety of methods [49, 50]. For
the studies in this PhD project, we chose to mirror the statistical approaches taken in the original
UK development studies, as this allowed comparison of results across cohorts, settings and studies.
Dichotomized distribution-based outcome measures used in the UK study, additional relative risk of
poor outcome when classified by subgroup and ability of discrimination by using the Area Under
the Curve (AUC) statistic was applied in the analyses. Additional regression models were also built
for further exploration of predictive differences found between the cohorts.
Ethics
This PhD project was approved by the Scientific Ethics Committee of the Region of Southern
Denmark (S-20100036) and all patients gave written informed consent for the use of their data for
19
Overall methods
_________________________
research. Permission for collection and storage of data in concordance with the rules by Hospital
Lillebaelt was given by the Danish Data Protection Agency (2011-41-6286).
20
Study 1
________________________
Study 1: ‘Translation and discriminative validation of the SBT’
Aim
The aims of this study were to translate the English version of SBT into Danish and to test its
discriminative validity.
Methods
There were two phases in this study: (1) a linguistic and cultural translation phase; and (2) a crosssectional validation phase of the discriminative ability of the SBT. The first phase was conducted
using a convenience sample from secondary care, as we believed it unlikely that the concurrent
validity / discriminative ability would be affected by episode duration or care setting. The second
phase also included data from the original UK primary care cohort (Table 1a).
Table 1a: Cohorts used for the Danish SBT translation and discriminative validation
Danish translation
cohort from
secondary care
Baseline only
Danish primary
cohort (GP, PT)
Danish secondary
predictive cohort
UK primary care
development cohort
UK primary care
BeBack cohort
Baseline only
This study followed the translation method recommendations of international guidelines [51-53],
which resulted in the following translation process being used (Table 2).
Table 2. Phases in the linguistic and cultural translation
Phases
Tasks of the translation process
1
Liaison with SBT developers
2
Translation from English to Danish
3
Back translation from Danish to English
4
Synthesis
5
Translation committee consensus?
6
Pilot testing
7
Testing of final version
21
Study 1
_________________________
Before initiating the translation process, contact was made with the research team at the Arthritis
Research Primary Care Centre at Keele University in Staffordshire, England who developed the
SBT. This collaborative link between Keele University and our Danish research group was
established to determine whether any other researchers had expressed an intention to undertake the
Danish translation work, whether the UK developers were aware of investigations into the
appropriateness of the SBT in non-primary care settings, and to request access to their original
validation data so that we could undertake comparative studies. They granted us access to data from
the original validation sample, which enabled us in this study to compare the cross-sectional
concurrent validity across Danish and UK cohorts.
Data that needed to be collected for the Danish cohort consisted of the SBT scores and all the
reference standard questionnaires used for the SBT constructs. The reference standards were the
Roland Morris Disability Questionnaire (RMDQ) for activity limitation [54, 55], the Tampa Scale
for Kinesiophobia (TSK) for fear of movement [56, 57], the Hospital Anxiety and Depression Scale
(HADS) for anxiety and depression [58, 59] and the Coping Strategies Questionnaire (CSQ) for
catastrophisation [60, 61]. These data allowed us to compare the Danish and the UK cohorts on
seven of the nine items included in the SBT. For two items, comparable data were not available co-morbid pain and bothersomeness - as reference standard questionnaires for these constructs were
not readily available. The comparison of the discriminative validity was performed using the Area
Under the Curve (AUC) statistic derived from Receiver Operating Curves [62].
Results
After a thorough translation process that included minor linguistic adjustments being made during
phases one to five of the translation process, the Danish version of the SBT was pilot-tested. During
each pilot-test, uncertainty and hesitation were noted by a researcher and the findings were
22
Study 1
_________________________
discussed and adjusted in the plenary group. Pilot-testing was repeated until no further uncertainty
was observed. The final Danish version of the SBT was then complete (Appendix 2).
For the concurrent validation process, data from 311 secondary care patients were available. There
were minor differences in baseline characteristics across the Danish and the UK cohorts with a
higher proportion in the Danish secondary care cohort reporting leg and shoulder/neck pain.
The discriminative ability using AUC was analysed for both cohorts. Overall, the AUC point
estimates calculated were similar for five items, but there were differences on three psychosocial
sub-score items. Table 3 shows the results for the three sub-score items that differed and the full
results are shown in Paper 1.
Table 3. Area under Curve for each SBT question compared with its reference standard
HADS=Hospital Anxiety and Depression Scale, CSQ= Coping Strategy Questionnaire (Full model in Paper 1)
Question in SBT
6.
Worrying thoughts have been
going through my mind a lot of the
time
7.
I feel that my back pain is terrible
and it’s never going to get any
better
8.
In general I have not enjoyed all
the things I used to enjoy
Danish
English
Reference Standard
Point Estimate
(CI95%)
HADS ANX
.837
(CI95% .792 to .882)
Reference Standard
Point Estimate
(CI95%)
HADS ANX
.918
(.894 to .942)
CSQ
.779
(CI95% .726 to .832)
CSQ
.925
(.902 to .948)
HADS DEP
.735
(CI95% .678 to .792)
HADS DEP
.902
(.876 to .929)
Discussion
The results of the translation and discriminative validation processes showed that the SBT had
sufficient patient acceptability and discriminative validity to be used in Danish primary care.
Divergence was observed on three psychosocial constructs which could have occurred for a number
of reasons: the SBT containing inappropriate screening questions for the Danish context, linguistic
inaccuracies, cultural differences, differences in association between screening item and reference
23
Study 1
_________________________
standard, inaccuracies in translation of the reference standard questionnaires, the Danish sample
being from secondary care (as opposed to primary care) or just simple sampling variability across
samples.
Questions and considerations of the process, Part I
The translation and discriminative study reassured us of the patient acceptability of the Danish
version and indicated that the SBT had sufficient discriminative validity to be applicable in Danish
clinical practice. The discriminative validity study had tested and confirmed the first component of
the ‘foundation’ of a Danish version of the SBT. However, the study also showed a weaker
discriminative ability in the Danish SBT psychosocial sub-scale, but as there were many potential
sources of that finding, we believed that this should not inhibit us from proceeding with
investigating other aspects of the validity of the Danish SBT.
Although linguistically accurate and discriminatively acceptable, the predictive validity of the SBT
in any Danish care setting had not yet been established. We initially investigated the predictive
validity of the Danish SBT in primary care [47, 63]. For this purpose we needed longitudinal data
from a Danish primary care cohort that was comparable to data from UK primary care. In terms of
the overall SBT validation process, our measuring of its predictive validity in another cohort
(Danish) and at another time-point (the original UK studies used 6-month outcomes but we chose to
study 3-month outcomes), also complied with recommendations that suggest broad validation
criteria should include validation at time-points not previously studied [47, 63]. Creation of a
comparable Danish primary cohort required contact with Danish primary care researchers and the
involvement of GPs and physiotherapy primary care clinics. That collaboration resulted in an
electronic form of the SBT for use in GP practices. This electronic questionnaire was triggered by a
pre-defined diagnostic code when entered into the Danish national medical system by the GP. The
24
Study 1
_________________________
questionnaire was completed during the consultation and a sub-grouping classification was instantly
calculated for the GP to use in his/her clinical decision-making. The development of the electronic
format and the collection of data occurred within the framework of an audit in general practice in
the Region of Southern Denmark. It was not possible to translate the use of this electronic format of
the SBT into the physiotherapy setting and so data collection there was by patient self-completion
in a paper format.
25
Study 2
________________________
Study 2: ‘The predictive and external validity of the SBT in Danish
primary care’
Aim
The aims of this study were to test the predictive and external validity of the Danish version of the
SBT in Danish primary care and compare that with the English version of the SBT in UK primary
care.
Methods
Investigation of the predictive ability of the SBT in the context of Danish primary care would
clarify whether its ability to predict outcome, based on potentially modifiable prognostic factors,
was similar in the UK and Denmark. The predictive validity of the UK SBT had originally been
established using 6-month outcomes [33], but 3-month UK data were also available and use of that
time-point allowed an opportunity for broader external validation. As 3-month outcomes have been
shown to be the most important in the clinical course of LBP in primary care [64, 65] and most
Danish primary care patients are seen in that period, this was also a reason for our choosing to
investigate the SBT predictive validity for outcomes at that time-point. A Danish primary care
cohort consisting of patients from GPs and physiotherapists was recruited. This cohort (n=344) was
compared with an existing UK primary care cohort (n=856) from the BeBack study [48] (Table 1b).
Descriptive information and standardised questionnaires were extracted from both cohorts at
baseline and at 3-month follow-up and entered into a database.
Table 1b. Cohorts used for the predictive validity in primary care
Danish translation
cohort from
secondary care
Danish primary
cohort (GP, PT)
Baseline & 3 monthoutcomes
Danish secondary
predictive cohort
UK primary care
development cohort
UK primary care
BeBack cohort
Baseline & 3 monthoutcomes
26
Study 2
_________________________
Various methods have been used in testing the predictive validity of patient-reported health
questionnaires [49, 50]. We chose to mirror the three statistical methods used in the UK
development study [33]. This standardisation allowed us to compare our results with those from
previous studies and also facilitates comparison in future studies. Comparison was made between
proportions of patients with poor clinical outcome at 3 months stratified by SBT, additional risk of
poor outcome by being in a higher risk SBT subgroup was estimated and AUC statistics described
the ability to discriminate between people with poor outcome at 3 months on three outcomes.
Results
The results from both the Danish and the UK cohorts followed the pattern seen in previous studies
[33, 66], with the lowest proportion of patients with poor outcome in the low-risk subgroup and the
highest proportion in the high-risk subgroup (Figure 2).
Figure 2. Proportions of patients within each SBT subgroup who had a poor
clinical outcome on activity limitation at 3 months and their relative risk.
100%
Relative Risk
4.5 [3.6; 5.6]
90%
Relative Risk
2.7 [1.8; 3.8]
80%
Proportions
70%
Relative Risk
2.4 [1.7; 3.4]
Relative Risk
3.1 [2.5; 3.9]
60%
50%
40%
Relative Risk
1*
Relative Risk
1*
30%
20%
10%
0%
DK
UK
These unadjusted results also indicated that the predictive ability in DK primary care equalled that
of the UK for the low- and medium-risk subgroups. However, they also suggested that the
predictive ability in the Danish cohort did not have the same magnitude of step increase from
medium-risk to high-risk subgroups when compared with the UK cohort. This divergence of results
27
Study 2
_________________________
seemed to be centred round the psychosocial subscale of the SBT and it was our impression that this
might be a product of the very different treatment exposure that had occurred between the cohorts.
DAK-E who administered the electronic registry extracted data showing that approximately 60% of
the Danish cohort had been exposed to physiotherapy treatment, compared with approximately 18%
in the UK cohort. Adjusting for changes in the psychosocial factors over the 3 months in the Danish
cohort resulted in the adjusted predictive ability of the high-risk subgroup being almost identical to
the unadjusted predictive ability observed in the UK cohort (Table 4). Unfortunately, as change data
were not available in the UK data, adjusted results could not be calculated in both cohorts.
Table 4. The odds of having poor clinical outcome on activity limitation at 3 months by SBT subgroup in the Danish
and UK cohorts, estimated using logistic regression. (Full model in Paper 2)
Danish cohort (n=322)
UK cohort (n=845)
Odds ratio
p-value
Odds ratio
p-value
[CI 95%]
[CI 95%]
Unadjusted model
SBT low-risk subgroup2
1.00
1.00
SBT medium-risk subgroup
4.24 [2.45; 7.32]
p<.001
5.56 [3.99; 7.76]
p<.001
SBT high-risk subgroup
5.57 [2.97; 10.47]
p<.001
16.88 [9.71; 29.34]
p<.001
Constant
.32 [.21; .48]
p<.001
.22 [.16; .26]
p<.001
Parsimonious model adjusted for care setting
and change on SBT psychosocial constructs
(n=296), using manual backwards step-wise
procedure
SBT low-risk subgroup2
1.00
SBT medium-risk subgroup
7.89 [3.87; 16.11]
p<.001
SBT high-risk subgroup
15.73 [6.60; 37.47]
p<.001
Care setting3
0.31 [0.13; .71]
p=.006
Change in anxiety6
0.81 [0.73; .89]
p<.001
Change in pain bothersomeness7
0.27 [0.17; .43]
p<.001
Interaction between care setting and change in
2.48 [1.41; 4.34]
p=.002
pain bothersomeness
Constant
1.02 [0.49; 2.15]
p=.951
Discussion
The results of the predictive study in primary care indicated that the ability to predict increased risk
of poor outcome at 3 months in Danish primary care was similar to that seen in UK primary care for
the low-risk and medium-risk subgroups, and after adjusting for change in the psychosocial factors,
28
Study 2
_________________________
the predictive ability of the Danish high-risk subgroup was almost identical to unadjusted estimates
from the UK cohort. Divergence in predictive ability between the cohorts was centred on the highrisk subgroup, which is based on the psychosocial subscale of the SBT. As seen in Study 1, which
had examined the discriminative validity of the SBT, a number of reasons could account for this
divergence. Those reasons could include differences in treatment exposure or treatment
effectiveness, but could also be due to cultural differences in the influence of psychosocial factors
on outcome.
Questions and considerations of the process, part II
Overall, the two studies conducted to translate and test the discriminative and predictive validity of
the SBT in Danish primary care concluded that the Danish version was a useful prognostic
stratification tool. Despite minor differences in discriminative and predictive validity compared
with other primary care settings [33, 66, 67], the perception was that SBT had potential to support
and guide clinicians in their daily clinical decision-making, although the targeted treatment
implications of the SBT subgroups remained untested.
However, the question as to whether the SBT had applicability in other care settings remained
unaddressed, although others had speculated on this in the literature [68]. Although investigation of
the SBT’s applicability in secondary care was appealing, several considerations were raised within
the PhD project group. Firstly, the trajectories of recovery might be different in primary and
secondary care in ways beyond that simply attributable to episode duration (Figure 3).
29
Study 2
_________________________
Recovery %
Figure 3. The trajectory of recovery and care setting of intervention. Inspired by data from Pengel [65]
Time
Primary care intervention
Secondary care intervention
Alternative trajectory
In the secondary care setting of the Spine Centre of Southern Denmark, patients are referred by GPs,
chiropractors or medical specialists due to sub-optimal improvement in primary care and patient
data indicate that they are more complex and have poorer recovery rates. Given that, we wondered
whether screening these patients for poor outcome would also be more complex than in primary
care, and whether secondary care screening would require the inclusion of additional or alternative
constructs than those contained in the SBT.
The notion of a classifying model based on the prediction of poor outcome makes intuitive sense in
a recent-onset episode of LBP, but perhaps it was not as applicable for secondary care patients who
were already experiencing persistent pain and who had a higher proportion of leg pain and specific
LBP (radiculopathy and central stenosis) [69]. Does it make sense to classify some secondary care
patients as being at low risk of poor outcome when they are referred on the basis of experiencing a
poor outcome in primary care in the first place?
Secondly, the question emerged as to whether the psychosocial profile of patients differed across
these care settings and therefore whether the SBT psychosocial subscale was suitable in secondary
care. Prior research suggests that the clinical course of patients is different in primary and secondary
care [70] but, although it has been shown that psychosocial factors impact on prognosis and
outcome [25, 27, 70], there were very limited data available about whether these psychosocial risk
profiles differed across primary and secondary care settings. Similarly, differences in the
psychosocial profile of people from primary and secondary care classified by the SBT had not been
30
Study 2
_________________________
previously reported. Therefore, we believed it to be important to investigate potential differences in
the psychosocial profile of primary and secondary patients before performing further testing of the
predictive ability of the SBT in secondary care.
31
Study 3
________________________
Study 3: ‘Is the psychosocial profile of people with low back pain
seeking care in Danish primary care different from those in secondary
care?’
Aim
The aim of this study was to investigate whether the psychosocial profile of patients in Danish
primary and secondary care settings was different.
Methods
This cross-sectional study was conducted to determine whether patient profiles on the psychosocial
constructs (movement-related fear, catastrophisation, anxiety and depression) included in the SBT
where different across primary and secondary care settings. For this study, baseline values from the
Danish secondary care cohort in Study 1 and from the Danish primary care cohort in Study 2 were
used (Table 1c). Therefore, the study was a secondary analysis of the SBT scores and the full
psychosocial construct scores for the five SBT items on the psychosocial subscale.
Table 1c. Cohorts used for the comparison of the psychosocial profile across primary and secondary care
Danish translation
cohort from
secondary care
Baseline only
Danish primary
cohort (GP, PT)
Danish secondary
predictive cohort
UK primary care
development cohort
UK primary care
BeBack cohort
Baseline only
A comparison of psychosocial scores was made overall across cohorts and across SBT classification
subgroups. Linear regression models were also created for each of the psychosocial constructs to
adjust for potentially influential covariates (age, gender, work participation, episode duration, pain
intensity and activity limitation).
32
Study 3
_________________________
Results
Although there were no significant differences across care settings in the distribution of patients
across the SBT subgroups, a slightly higher proportion of patients were classified in the mediumrisk subgroup in primary care (42% vs. 32%). An unexpectedly large proportion of patients from
secondary care were classified in the low-risk subgroup (39.2%). Overall, there were significantly
higher scores in secondary care on movement-related fear and catastrophisation, but lower scores on
anxiety. The observed differences across care settings on the psychosocial constructs were also
retained when patients were stratified by SBT subgroup (Figure 4).
Figures 4. Differences between care settings on psychosocial construct scores, when stratified by SBT
classification groups.
Movement-related Fear
Catastrophisation
67
35
57
37,6
35,3
37
32,2
*
25
40,6
Primary care
Secondary care
32,3
CSQ score
TSK score
30
43,2*
47
*
19
20
14,5
15
*
15
Primary care
Secondary care
10
10
27
7
5
5
17
low risk
medium risk
0
high risk
low risk
STarT group
medium risk
high risk
STarT group
4a
4b
Anxiety
Depression
30
20
25
Primary Care
10
5
Secondary care
6
*
7
*
6
H A D S sco re
HADS score
15
20
Primary care
15
Secondary care
10
*
6
4
4
5
2
3
1
1
4
2
0
0
low risk
medium risk
STarT group
4c
high risk
low risk
medium risk
high risk
STarT group
4d
* Significant differences
The trend towards increased psychosocial reference standard scores by increased risk SBT
classification subgroup that has been reported by others [33] was also found in this study. To further
33
Study 3
_________________________
test for care setting as a dominant variable on these psychosocial constructs, separate linear
regression models were made for each construct. When simultaneously entering the independent
variables of care setting, age, gender, employment status, pain intensity, activity limitation and
duration of episode, the variable of care setting was not significant for any of the psychosocial
constructs. When tested in a forward stepwise model, care setting was always entered late in the
model. Both these analyses indicate that care setting was not a dominant variable on any of the
psychosocial constructs.
Discussion
The results from this study indicated small but statistically significant differences on four of five
psychosocial constructs across Danish primary and secondary health care settings. Although these
differences were also broadly retained when stratified by SBT subgroup, our interpretation was that
they were so small in magnitude that they were unlikely to be clinically relevant from a patient
perspective. This interpretation was based on previous estimates of minimally important clinical
differences [71, 72]. Overall, the trend of increased scores on the psychosocial constructs in higher
risk SBT subgroups was similar for both settings and reinforced the construct validity of the SBT.
Although the distribution of patients across the three SBT subgroups in primary care was very
similar to that reported previously in primary care [33], we noted the surprisingly high proportion of
patients allocated to the low-risk subgroup in secondary care. This seems inconsistent with the
expectation of these patients recovering well in primary care. There could be several reasons for this
high proportion of low-risk patients in secondary care: lack of improvement of low-risk patients in
primary care due to inadequate reassurance and information on self-management or over-treatment,
the SBT not being able to detect clinical characteristics that are important for the different phase of
34
Study 3
_________________________
LBP in secondary care, and different stages of psychosocial response through the clinical course of
LBP.
Questions and considerations of the process, part III
Although some new questions did emerge in this study, the comparison of psychosocial patient
profiles across care settings encouraged us to further explore the applicability of the SBT in
secondary care. It was reassuring that differences in the psychosocial profiles of patients from these
care settings were not large and not all in the same direction. Less clear were the implications for
the SBT predictive ability in secondary care, for a large proportion of secondary care patients being
classified in the low-risk SBT subgroup.
Therefore, multiple potential aspects influencing the predictive ability of the SBT in secondary care
were considered. In Denmark, secondary care is defined as government-funded, specialised care
requiring specific referral1 and we knew from Study 3 that patients from secondary care settings had
longer episode duration, more frequent leg pain and greater pain intensity. An earlier study had also
shown that there were differences in patient case-mix with an increased proportion of patients at the
Spine Centre having specific LBP (radiculopathy and central stenosis) [69]. In addition, we were
also thoughtful about any potential influence of the differences in the concurrent validity and
predictive ability of the Danish SBT psychosocial subscale noted in Study 1 and Study 2. On the
other hand, this would be the first study to contribute knowledge about the predictive ability of the
SBT in secondary care and thereby to initiate the first step of testing the SBT in this care. setting
1
As defined in the Great Danish Encyclopaedia (Published 2005.Gyldendals Forlag)
35
Study 4
________________________
Study 4: ‘The predictive ability of the SBT in a Danish secondary care
setting’
Aim
The aim of this study was to compare the predictive ability of SBT in a Danish secondary care
setting with a Danish primary care setting.
Methods
In this study, the secondary care component was conducted using a new cohort (n=960) from the
Medical Department of the Spine Centre of Southern Denmark. Patients are referred there for
evaluation after sub-optimal improvement in primary care. As 6-month outcomes were believed to
be more clinically meaningful for secondary care patients, the secondary and primary care cohorts
were designed to contain comparable data at baseline and at 6-month follow-up. The primary care
component of the study was a secondary analysis of the physiotherapy subsample (n=172) of the
primary care cohort collected for Study 2. Only the physiotherapy subsample was used, as 6-month
outcome data were only available for this subsample (Table 1d).
Table 1d. Cohorts used for the predictive validity study in secondary care
Danish translation
cohort from
secondary care
Danish primary
cohort (PT only)
Danish secondary
predictive cohort
Baseline & 6-month
outcomes
Baseline & 6-month
outcomes
UK primary care
development cohort
UK primary care
BeBack cohort
This study also mirrored the statistical methods used in the original development study conducted in
the UK [33], as this allowed us to contextualise the results relative to our previous primary care
study and those from the UK. In addition, to explore explanations for the results, we used logistic
regression to calculate odds ratios for poor outcome adjusted for baseline differences between the
36
Study 4
_________________________
cohorts and also calculated the relative risk of poor outcome using baseline pain intensity and
baseline activity limitation as predictors.
Results
Overall, there were significant differences at baseline across the two cohorts on duration of episode
and pain intensity. In concordance with earlier findings [33, 66], the pattern of increased score on
pain intensity and activity limitation across the SBT subgroups from low risk to high risk was also
found in both cohorts. At 6-month follow-up, there were differences between the two cohorts with
patients in secondary care having higher pain intensity and activity limitation, and these differences
between cohorts were retained when stratified by SBT subgroup (Table 5).
Table 5. Outcome at 6-month follow-up for the Danish secondary and primary care cohorts.
Secondary care cohort
Pain intensity (0-10 scale)
• Low back pain, median, (IQR)
•
Leg pain, median, (IQR)
Activity limitation (0-100 scale)
• Median, (IQR)
•
Proportion of patients > 30
Primary care cohort
Pain intensity (0-10 scale)
• Low back pain, median, (IQR)
•
Leg pain, median, (IQR)
Activity limitation (0-100 scale)
• Median, (IQR)
•
Proportion of patients > 30
Total cohort
n=960
SBT
Low risk
n=252
(27.7%)
SBT
Medium risk
n=296
(32.5%)
SBT
High risk
n=363
(39.9%)
4.7 (2-6)
3.0 (2-5)
4.3 (2-6)
5.7 (3-7)
2.7 (0-5)
1.3 (0-4)
2.7 (0-5)
3.7 (1-6)
47.8 (28-70)
26.1 (13-48)
47.8 (22-70)
65.2 (35-83)
656 (69.0%)
120 (47.8%)
209 (71.3%)
292 (81.3%)
Total cohort
n=144
SBT
Low risk
n=48
(36.9%)
SBT
Medium risk
n=52
(40.0%)
SBT
High risk
n=30
(23.1%)
3.0 (2-5)
3.0 (2-4)
2.0 (0-4)
4.5 (2-5)
1.3 (0-4)
0.5 (0-3)
2.0 (1-5)
1.0 (0-5)
26.0 (9-57)
17.0 (4-26)
30.0 (9-55)
65.0 (25-75)
51 (40.2%)
9 (20.0%)
21 (46.7%)
18 (69.2%)
37
Study 4
_________________________
The proportion of patients with poor outcome on activity limitation at the 6–month follow-up
increased from low risk to high risk in both cohorts. However, a larger proportion in secondary care
had poor outcome in all the SBT subgroups, most notably in the low risk, where almost half of the
secondary care patients still had an RMDQ score > 30 points (on a 0 to 100 scale). The results also
showed that, although statistically significant, the gradient of relative risk in secondary care was not
as steep as in primary care (Figure 5), which indicated that the predictive ability of the SBT was
weaker in secondary care than in primary care.
Figure 5. Relative risk of poor clinical outcome on activity limitation at 6-month follow-up by SBT
subgroup in the Danish secondary and primary care cohorts.
4
3.5*
3,5
3
2.3*
2,5
2
1,5
1
1.5*
1.7*
1
Low risk
Medium risk
High risk
1
0,5
0
Secondary care
Primary care
*Significant differences
As significant baseline differences were found across cohorts on episode duration and pain intensity,
estimates of SBT predictive ability were adjusted for these differences. Although in these regression
models, episode duration and LBP intensity were significantly associated with outcome, the odds
ratios for the SBT subgroups predicting outcome changed only marginally. In contrast, this analysis
showed that episode duration was predictive in both cohorts and had an influence that was
independent of the predictive ability of the SBT subgroups.
Although it was apparent that the predictive ability of SBT subgroups was less in secondary care
than in primary care, we lacked any secondary care reference standards to which the predictive
38
Study 4
_________________________
ability of SBT subgroups could be compared. Therefore, we performed a post hoc analysis using
alternative reference standard predictors (categorised baseline pain intensity and categorised
baseline activity limitation) and found their predictive ability to be nearly identical to that of the
SBT subgroups.
Discussion
The SBT subgroups were less predictive in Danish secondary care than in primary care but were as
predictive as similarly categorised baseline scores on pain intensity or activity limitation. The later
finding is notable, as baseline pain intensity and activity limitation are known to be strong
predictors of outcome [73, 74] and the baseline values of the outcome being investigated (in this
case, activity limitation) are usually the strongest predictor [67]. Although there were similar
proportions across the cohorts with poor activity limitation at baseline, these proportions were
different at 6 months, especially in the low-risk subgroup. This indicates less favourable recovery
trajectories in secondary care, as has been shown previously [70]. Based on the predictive ability of
the SBT subgroups and the predictive ability of baseline pain intensity and activity limitation, it
seems that predicting outcome overall in secondary care is challenging, perhaps due to a
combination of the secondary care group’s generally having a less favourable clinical course and
perhaps due to greater heterogeneity in the outcomes within the subgroups. In the case of the SBT,
it may also be that the identification of increased risk of poor outcome due to psychosocial
components is more complex in secondary care. Other explanations could also include differences
in treatment exposure or differences in case-mix.
39
Overall discussion
________________________
Overall discussion
When this PhD project was initiated during 2009-2010, the work describing the development of the
SBT in the UK had just been published [33] and only preliminary results from the ‘STarT Back
Trial’ [37] existed. Since then, the SBT has gained much attention and at the XII LBP Forum for
Research in Primary Care in 2012, the SBT was described as having larger potential in the field of
LBP than any other prognostic research for the last 10 years. Since 2009, the SBT has been
translated into several languages [75, 76] and tested in a number of studies [66, 67, 77].
Methodological developments in prognostic research have also occurred in this period. Proposals
for improved quality criteria for the measurement properties of patient-reported outcomes have been
suggested by the COSMIN group [78] [47]. Furthermore, a framework for prognostic research has
been recommended by the PROGRESS group [39-42].
The aim of the current project was to investigate whether the SBT was able to identify subgroups of
patients predictive of risk of poor outcome in Danish primary and secondary care settings. For that
purpose, we chose a project design of performing (i) a cross-cultural validation comparing the
discriminative validity of the Danish-translated and original UK versions; (ii) a comparison of the
SBT predictive validity for poor outcome at 3 months in Danish and UK primary care cohorts; and
(iii) a comparison of the predictive validity of poor outcome at 6 months in Danish primary and UK
secondary care cohorts. Prior to this work, an extensive development phase had been completed in
the UK [35]. Results from that work indicated that the fundamental stages of developing the SBT
had been thoroughly investigated, including the selection of specific modifiable prognostic factors,
the testing of measurement properties, the formation of subgroup allocation rules, and the
description of the predictive validity of the SBT subscale relative to another psychosocial screening
questionnaire [32]. In the context of the PROGRESS framework, the initial UK development work
40
Overall discussion
_________________________
represents activity in the first three phases of the framework: fundamental prognostic research,
prognostic factor research and prognostic model research [40]. Collectively, the validation process
of the Danish SBT represents research occurring within the third phase of the PROGRESS
framework (prognostic model research) [42].
In this PhD project, we initially built on the construct validity work already conducted by Hill et al
in their development studies in the UK [33, 35]. Though it could be theoretically argued that
different items might have been appropriate in the Danish SBT, our results from Study 1 did not
support that notion and it was not our intention to re-examine the content validity of the SBT. Our
strategy of testing the external and predictive validity of an existing prognostic model is strongly
concordant with the recommendations of the PROGRESS group [39, 40].
Throughout the process of validity testing, the methods in the original development studies were
replicated and the results compared across several Danish and UK cohorts at 3- or 6-month followup time-points. This approach allowed us to validate the SBT in different cohorts, in different
settings, at different time-points and across national health care systems. This method of validation
is recommended [47, 63] and external validation is considered highly important [42]. The choice of
the same methods and outcome parameters also creates results more suitable for future systematic
review purposes.
The predictive ability of the SBT has been investigated in other studies using alternative statistical
approaches [66, 67]. In those studies, continuous measures (SBT sum scores instead of SBT
subgroups classification) were used to predict continuous outcome measures (such as RMDQ raw
scores rather than distribution-based dichotomised scores). It has been argued that the use of
41
Overall discussion
_________________________
continuous outcome measures avoids the use of arbitrary cut-off estimates that could lead to
misclassification and preserves all the potential information contained within continuous variables
[39, 40]. In our opinion, the SBT was designed for the stratification of patients into three subgroups
of risk and clinicians use the SBT raw scores only as a means to calculate subgroup membership.
As such, the SBT is a predictive prognostic tool, not an explanatory prognostic model [79].
The SBT was explicitly developed as an easy applicable screening tool for daily use in the clinic
[35], and its capacity to indicate increased risk has therefore not been reported by regression
coefficients or regression line slopes but mostly in the more clinically interpretable terms of relative
risk. Therefore, while being fully aware that dichotomised outcomes remove potential information,
we believed that modelling the SBT subgroups was more interpretable and relevant for clinicians
than modelling risk on continuous scales [43].
This PhD project found that classification into the high-risk subgroup using the Danish SBT was
less predictive of poor outcome than in the UK. The unadjusted results in Danish primary care were
not as strong as in UK primary care (Study 2) and in Danish secondary care not as strong as in
Danish primary care (Study 4). While adjustment for covariates in primary care (Study 2) suggested
that care setting (GP/physiotherapy) and change over time in psychosocial factors confounded the
unadjusted estimates of SBT predictive ability in Danish primary care, adjustment for selected
covariates did not alter the predictive ability in Danish secondary care (Study 4).
These results focus upon the psychosocial subscale and question whether the items included in the
psychosocial subscale are sufficient in a Danish context. In Study 1, we also examined whether
alternative questions from the reference standard questionnaires might have shown stronger
concurrent and discriminative validity than those chosen originally for the SBT. We did not find
evidence that alternative questions better suited these constructs for Danish patients. It remains
42
Overall discussion
_________________________
possible, however, that alternative or additional psychosocial constructs might improve the
performance of the psychosocial subscale of the Danish version of the SBT. There also may be
broader public health and social issues that are not represented in the psychosocial subscale that
could be considered as additional prognostic factors in secondary care [64, 80, 81]. For example, in
pregnancy-related pelvic pain, it has been shown that socio-demographics are influential on
outcome [82]. An over-representation of lower socio-demographics in the secondary care cohort
could possibly have influenced the SBT predictive ability in that care setting. So, maybe for the
SBT to have better predictive ability in secondary care, broader prognostic factors would need to be
included.
Overall, the Danish SBT was not as predictive in secondary care as it was in primary care. However,
describing additional risk might not be very useful in a setting where more than 50% of the patients
in the reference subgroup (low-risk) do not improve. This reference category leaves little room for
the increased risk of poor outcome in the medium- or high-risk subgroups to be relatively large.
This was also shown in this setting by the predictive ability for constructs usually considered as
strong prognostic factors (categorised baseline activity limitation and pain intensity) to be
equivalent to that of the SBT subgroups [73, 74]. This predictive difficulty seems broader than the
considerations about the Danish SBT psychosocial subscale, as it applies to all the SBT subgroups.
Strengths and weaknesses of the PhD project
The strengths of the project were (i) the collaborative relationship with the developers of the SBT at
Keele University in the UK, which ensured a contemporary understanding of the SBT and allowed
the project to have access to data from the original UK validation cohorts, (ii) the mirroring of
statistical approaches used in the UK studies, as this allowed comparison of results and will
43
Overall discussion
_________________________
facilitate future synthesis of these results with others, (iii) the use of a translation method
recommended by international guidelines, (iv) the use of cross-cultural and cross-care setting
comparisons to broaden validity testing, (v) the use of different outcome time-points to broaden
external validity testing, and (vi) the use of regression to explore and explain variability in findings.
However, the project also has a number of potential weaknesses. Aspects of this project were
conducted in close collaboration with the original developers of the SBT and this potentially could
have biased our judgement about the applicability of the SBT in Danish health care. However, the
collective oversight of the PhD team, the conducting of the project using internationally
recommended methods and the convergence of results across countries are likely to have minimised
this potential bias.
A secondary care cohort was used for testing the Danish translation (Study 1) but subsequent results
(Study 3) showed there were statistically significant but not clinically important differences in the
psychosocial profiles of patients in Danish primary and secondary care settings. It was the view of
the project team, that this and other differences between care settings, such as pain intensity and
episode duration, were unlikely to impact the concurrent validity / discriminative ability of the SBT
in a cross-sectional study design. However, we do not have empirical data to test whether that view
is correct.
Some of the project data were collected in paper format and some were collected electronically. The
electronic format had not yet been validated and therefore data collected in that format could have
contained potential bias. Preliminary results from a new study conducted at the Spine Centre
44
Overall discussion
_________________________
indicate that SBT data collected in electronic and paper formats are equally valid, but these results
are yet to be published.
When collecting data electronically, participating GPs had immediate access to the SBT score and
subgroup classification. This could potentially have affected their patient management and thereby
have affected treatment exposure and predictive validity. We have data suggesting that at least 60%
of the GP patients were referred for physiotherapy and, due to incomplete registrations at the GP
practices, this number might have been even higher. However, this electronic SBT scoring is readily
available to GPs that opt to use it and we therefore think that the results obtained in Study 2 reflect
contemporary Danish primary care.
45
Perspectives
________________________
Perspectives
The SBT is a classification tool based on subgroups of increased risk of poor outcome/baseline
symptom complexity. The SBT subgroups have prognostic and treatment implications in primary
care. This project investigated the discriminative and predictive validity of a Danish-translated
version of the SBT. Investigation of the targeted treatment implications of the SBT was beyond the
scope of the work undertaken in this PhD. SBT-targeted treatment has been investigated in a large
scale, high quality RCT in the UK and was more clinically effective and cost-effective than usual
care [37]. Therefore, it would be relevant to investigate the SBT-targeted treatment implications in
Danish primary care. Such investigation would also be in concordance with the next phase of the
PROGRESS framework that encourages research of stratified medicine [41].
Broader than the SBT approach, previously tested types of targeted treatment have been based on
best evidence from the literature and international guidelines [83], but there is minimal evidence
that any one type of targeted treatment is more effective than alternatives. Therefore, research that
combines other types of targeted treatment with the SBT subgroup classification may be of interest,
especially for the high-risk subgroup.
Our data do not support the use of the SBT as a prognostic screening tool in Danish secondary care
and suggest that predicting subgroups of poor outcome in this setting is challenging. The
complexity of components influencing prognosis might be different from those in primary care and
require a more complex screening tool. Investigation of the association between other forms of
predictive/prognostic models [20, 84] and the SBT classification might be relevant. In addition, the
integration of non-patient self-reported measurement pathways could provide further useful
information, such as clinical signs measured by clinicians like neurological signs or movement
46
Perspectives
_________________________
patterns and imaging findings (for example selected MRI findings). It may also be the case that
predictive models in secondary care need to be specific for different types of case-mix (non-specific
pain, radiculopathy, central stenosis). The exploration of more complex predictive models in
longitudinal studies might also help in the understanding of trajectories of LBP across care settings
and at different time-points - from onset of a LBP episode until the cessation of health care-seeking.
The implementation of the SBT as a guidance tool for decision-making in primary care in the
Region of Southern Denmark has been initiated and the SBT has been introduced as a mandatory
component of the Region’s treatment guidelines. GPs can choose to use the electronic format of the
SBT in their clinics, with automatic subgroup allocation and additional decision guidance during the
patient consultation. Further development of the electronic format of SBT is in progress, as is an
investigation of any potential impact of using the SBT in different formats (electronic versus paper,
self-administered versus clinician-administered).
47
Overall conclusions
_________________________
Overall Conclusions of the PhD project
•
The Danish translation of the SBT questionnaire was linguistically accurate.
•
The discriminative validity of the Danish SBT was comparable with the English version,
though lower discriminative validity was found on three psychosocial questions.
•
The SBT had a 3-month predictive ability in Danish primary care that was similar to that in
UK primary care (a test of the external validity) but the predictive ability of the high-risk
subgroup in Danish primary care was reduced.
•
The SBT had sufficient patient acceptability, discriminative and predictive validity to be a
suitable prognostic triage tool for LBP patients in Danish primary care.
•
There were statistically significant, but probably not clinically important, differences in the
psychosocial profile of patients in Danish primary and secondary care settings.
•
The SBT was not as strong in predicting outcome at 6 months in a Danish secondary care as
compared with a cohort from primary care.
48
References
(1) Walker BF, Muller R, Grant WD. Low back pain in Australian adults: prevalence and
associated disability. J Manipulative Physiol Ther 2004 May;27(4):238-44.
(2) Leboeuf-Yde C, Nielsen J, Kyvik KO, Fejer R, Hartvigsen J. Pain in the lumbar, thoracic
or cervical regions: do age and gender matter? A population-based study of 34,902 Danish
twins 20-71 years of age. BMC Musculoskelet Disord 2009;10:39.
(3) Illemann Chrinstensen A, Ekholm O, Davidsen M, Juel K. Sundhed og sygelighed i
Danmark 2010 & udviklingen siden 1987. The National Institute of Public Health,
University of Southern Denmark; 2012.
(4) Bjerrum Koch M, Davidsen M, Juel K. De samfundsmæssige omkostninger ved
rygsygdomme og rygsmerter i Danmark. The National Institute of Pulic Health,
University of Southern Denmark; 2011.
(5) Hestbaek L, Leboeuf-Yde C, Manniche C. Is low back pain part of a general health pattern
or is it a separate and distinctive entity? A critical literature review of comorbidity with
low back pain. J Manipulative Physiol Ther 2003 May;26(4):243-52.
(6) Hoy D, Bain C, Williams G, March L, Brooks P, Blyth F, et al. A systematic review of the
global prevalence of low back pain. Arthritis Rheum 2012 Jun;64(6):2028-37.
(7) Delitto A. Research in low back pain: time to stop seeking the elusive "magic bullet". Phys
Ther 2005 Mar;85(3):206-8.
(8) Deyo RA. Diagnostic evaluation of LBP: reaching a specific diagnosis is often impossible.
Arch Intern Med 2002 Jul 8;162(13):1444-7.
(9) Fourney DR, Andersson G, Arnold PM, Dettori J, Cahana A, Fehlings MG, et al. Chronic
low back pain: a heterogeneous condition with challenges for an evidence-based approach.
Spine (Phila Pa 1976 ) 2011 Oct 1;36(21 Suppl):S1-S9.
(10) Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-Moffett J, Kovacs F, et al.
Chapter 4. European guidelines for the management of chronic nonspecific low back pain.
Eur Spine J 2006 Mar;15 Suppl 2:S192-S300.
(11) Dagenais S, Tricco AC, Haldeman S. Synthesis of recommendations for the assessment
and management of low back pain from recent clinical practice guidelines. Spine J 2010
Jun;10(6):514-29.
(12) Krismer M, van TM. Strategies for prevention and management of musculoskeletal
conditions. Low back pain (non-specific). Best Pract Res Clin Rheumatol 2007
Feb;21(1):77-91.
(13) Laerum E, Storheim K, Brox JI. [New clinical guidelines for low back pain]. Tidsskr Nor
Laegeforen 2007 Oct 18;127(20):2706.
49
(14) Liddle SD, Gracey JH, Baxter GD. Advice for the management of low back pain: a
systematic review of randomised controlled trials. Man Ther 2007 Nov;12(4):310-27.
(15) Quebec Task Force. Scientific approach to the assessment and management of activityrelated spinal disorders. A monograph for clinicians. Report of the Quebec Task Force on
Spinal Disorders. Spine (Phila Pa 1976 ) 1987 Sep;12(7 Suppl):S1-59.
(16) van Tulder M, Becker A, Bekkering T, Breen A, del Real MT, Hutchinson A, et al.
Chapter 3. European guidelines for the management of acute nonspecific low back pain in
primary care. Eur Spine J 2006 Mar;15 Suppl 2:S169-S191.
(17) Dunn KM, Jordan KP, Croft PR. Contributions of prognostic factors for poor outcome in
primary care low back pain patients. Eur J Pain 2011 Mar;15(3):313-9.
(18) Hill JC, Fritz JM. Psychosocial influences on low back pain, disability, and response to
treatment. Phys Ther 2011 May;91(5):712-21.
(19) Fairbank J, Gwilym SE, France JC, Daffner SD, Dettori J, Hermsmeyer J, et al. The role
of classification of chronic low back pain. Spine (Phila Pa 1976 ) 2011 Oct 1;36(21
Suppl):S19-S42.
(20) Freynhagen R, Baron R, Gockel U, Tolle TR. painDETECT: a new screening
questionnaire to identify neuropathic components in patients with back pain. Curr Med
Res Opin 2006 Oct;22(10):1911-20.
(21) Petersen T, Olsen S, Laslett M, Thorsen H, Manniche C, Ekdahl C, et al. Inter-tester
reliability of a new diagnostic classification system for patients with non-specific low back
pain. Aust J Physiother 2004;50(2):85-94.
(22) O'Sullivan P. It's time for change with the management of non-specific chronic low back
pain. Br J Sports Med 2012 Mar;46(4):224-7.
(23) Dunn KM, Croft PR. Classification of low back pain in primary care: using
"bothersomeness" to identify the most severe cases. Spine (Phila Pa 1976 ) 2005 Aug
15;30(16):1887-92.
(24) Chou R, Qaseem A, Snow V, Casey D, Cross JT, Jr., Shekelle P, et al. Diagnosis and
treatment of low back pain: a joint clinical practice guideline from the American College
of Physicians and the American Pain Society. Ann Intern Med 2007 Oct 2;147(7):478-91.
(25) Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of outcome in acute
and subchronic low back trouble. Spine (Phila Pa 1976 ) 1995 Mar 15;20(6):722-8.
(26) Linton SJ, Boersma K. Early identification of patients at risk of developing a persistent
back problem: the predictive validity of the Orebro Musculoskeletal Pain Questionnaire.
Clin J Pain 2003 Mar;19(2):80-6.
(27) Main CJ, Foster N, Buchbinder R. How important are back pain beliefs and expectations
for satisfactory recovery from back pain? Best Pract Res Clin Rheumatol 2010
Apr;24(2):205-17.
50
(28) Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal
pain: a state of the art. Pain 2000 Apr;85(3):317-32.
(29) Haggman S, Maher CG, Refshauge KM. Screening for symptoms of depression by
physical therapists managing low back pain. Phys Ther 2004 Dec;84(12):1157-66.
(30) Hill JC, Vohora K, Dunn KM, Main CJ, Hay EM. Comparing the STarT back screening
tool's subgroup allocation of individual patients with that of independent clinical experts.
Clin J Pain 2010 Nov;26(9):783-7.
(31) Kent PM, Keating JL, Taylor NF. Primary care clinicians use variable methods to assess
acute nonspecific low back pain and usually focus on impairments. Man Ther 2009
Feb;14(1):88-100.
(32) Hill JC, Dunn KM, Main CJ, Hay EM. Subgrouping low back pain: a comparison of the
STarT Back Tool with the Orebro Musculoskeletal Pain Screening Questionnaire. Eur J
Pain 2010 Jan;14(1):83-9.
(33) Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE, et al. A primary care back
pain screening tool: identifying patient subgroups for initial treatment. Arthritis Rheum
2008 May 15;59(5):632-41.
(34) Hilfiker R, Bachmann LM, Heitz CA, Lorenz T, Joronen H, Klipstein A. Value of
predictive instruments to determine persisting restriction of function in patients with
subacute non-specific low back pain. Systematic review. Eur Spine J 2007
Nov;16(11):1755-75.
(35) Hill J. Identifying subgroups among patients with low back pain in primary care:
Evaluating the STarT Back Tool. Primary Care and Health Sciences, Keele University;
2008.
(36) Hay EM, Dunn KM, Hill JC, Lewis M, Mason EE, Konstantinou K, et al. A randomised
clinical trial of subgrouping and targeted treatment for low back pain compared with best
current care. The STarT Back Trial Study Protocol. BMC Musculoskelet Disord
2008;9:58.
(37) Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of
stratified primary care management for low back pain with current best practice (STarT
Back): a randomised controlled trial. Lancet 2011 Oct 29;378(9802):1560-71.
(38) Styregruppe i Region Syddanmark. Patientforløbsprogram for Rygområdet i Region
Syddanmark. Region Syddanmark; 2010.
(39) Riley RD, Hayden JA, Steyerberg EW, Moons KG, Abrams K, Kyzas PA, et al. Prognosis
Research Strategy (PROGRESS) 2: Prognostic Factor Research. PLoS Med 2013
Feb;10(2):e1001380.
(40) Hemingway H, Croft P, Perel P, Hayden JA, Abrams K, Timmis A, et al. Prognosis
research strategy (PROGRESS) 1: A framework for researching clinical outcomes. BMJ
2013;346:e5595.
51
(41) Hingorani AD, Windt DA, Riley RD, Abrams K, Moons KG, Steyerberg EW, et al.
Prognosis research strategy (PROGRESS) 4: Stratified medicine research. BMJ
2013;346:e5793.
(42) Steyerberg EW, Moons KG, van der Windt DA, Hayden JA, Perel P, Schroter S, et al.
Prognosis Research Strategy (PROGRESS) 3: Prognostic Model Research. PLoS Med
2013 Feb;10(2):e1001381.
(43) Altman DG, Lyman GH. Methodological challenges in the evaluation of prognostic
factors in breast cancer. Breast Cancer Res Treat 1998;52(1-3):289-303.
(44) Altman DG. Systematic reviews of evaluations of prognostic variables. BMJ 2001 Jul
28;323(7306):224-8.
(45) Hayden JA, Dunn KM, van der Windt DA, Shaw WS. What is the prognosis of back pain?
Best Pract Res Clin Rheumatol 2010 Apr;24(2):167-79.
(46) Altman DG, Vergouwe Y, Royston P, Moons KG. Prognosis and prognostic research:
validating a prognostic model. BMJ 2009 May 28;338:b605.
(47) Henrica C.W.de Vet, Caroline B.Terwee, Lidwine B.Mokkink, Dirk L.Knol. Validity.
Measrement in medicine. First ed. New York: Cambridge Universiry Press; 2011. p. 150201.
(48) Foster NE, Bishop A, Thomas E, Main C, Horne R, Weinman J, et al. Illness perceptions
of low back pain patients in primary care: what are they, do they change and are they
associated with outcome? Pain 2008 May;136(1-2):177-87.
(49) Childs JD, Fritz JM, Flynn TW, Irrgang JJ, Johnson KK, Majkowski GR, et al. A clinical
prediction rule to identify patients with low back pain most likely to benefit from spinal
manipulation: a validation study. Ann Intern Med 2004 Dec 21;141(12):920-8.
(50) Cleland JA, Fritz JM, Brennan GP. Predictive validity of initial fear avoidance beliefs in
patients with low back pain receiving physical therapy: is the FABQ a useful screening
tool for identifying patients at risk for a poor recovery? Eur Spine J 2008 Jan;17(1):70-9.
(51) Beaton DE, Bombardier C, Guillemin F, Ferraz MB. Guidelines for the process of crosscultural adaptation of self-report measures. Spine (Phila Pa 1976 ) 2000 Dec
15;25(24):3186-91.
(52) Bullinger M, Alonso J, Apolone G, Leplege A, Sullivan M, Wood-Dauphinee S, et al.
Translating health status questionnaires and evaluating their quality: the IQOLA Project
approach. International Quality of Life Assessment. J Clin Epidemiol 1998
Nov;51(11):913-23.
(53) Ware JE, Jr., Keller SD, Gandek B, Brazier JE, Sullivan M. Evaluating translations of
health status questionnaires. Methods from the IQOLA project. International Quality of
Life Assessment. Int J Technol Assess Health Care 1995;11(3):525-51.
52
(54) Albert HB, Jensen AM, Dahl D, Rasmussen MN. [Criteria validation of the Roland Morris
questionnaire. A Danish translation of the international scale for the assessment of
functional level in patients with low back pain and sciatica]. Ugeskr Laeger 2003 Apr
28;165(18):1875-80.
(55) Roland M, Morris R. A study of the natural history of back pain. Part I: development of a
reliable and sensitive measure of disability in low-back pain. Spine (Phila Pa 1976 ) 1983
Mar;8(2):141-4.
(56) Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW, Oostendorp RA.
Psychometric properties of the Tampa Scale for kinesiophobia and the fear-avoidance
beliefs questionnaire in acute low back pain. Man Ther 2003 Feb;8(1):29-36.
(57) Vlaeyen JW, Kole-Snijders AM, Boeren RG, van EH. Fear of movement/(re)injury in
chronic low back pain and its relation to behavioral performance. Pain 1995
Sep;62(3):363-72.
(58) Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the Hospital Anxiety and
Depression Scale. An updated literature review. J Psychosom Res 2002 Feb;52(2):69-77.
(59) Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand
1983 Jun;67(6):361-70.
(60) Rosenstiel AK, Keefe FJ. The use of coping strategies in chronic low back pain patients:
relationship to patient characteristics and current adjustment. Pain 1983 Sep;17(1):33-44.
(61) Swartzman LC, Gwadry FG, Shapiro AP, Teasell RW. The factor structure of the Coping
Strategies Questionnaire. Pain 1994 Jun;57(3):311-6.
(62) Kirkwood BRSJAC. Measurement error: assessment and implications.
Essential
Medical Statistics. 2nd edition 2003 ed. Oxford: Blackwell Science Ltd.; 1988. p. 429-46.
(63) Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of prognostic
information. Ann Intern Med 1999 Mar 16;130(6):515-24.
(64) Costa LC, Maher CG, McAuley JH, Hancock MJ, Herbert RD, Refshauge KM, et al.
Prognosis for patients with chronic low back pain: inception cohort study. BMJ
2009;339:b3829.
(65) Pengel LH, Herbert RD, Maher CG, Refshauge KM. Acute low back pain: systematic
review of its prognosis. BMJ 2003 Aug 9;327(7410):323.
(66) Fritz JM, Beneciuk JM, George SZ. Relationship between categorization with the STarT
Back Screening Tool and prognosis for people receiving physical therapy for low back
pain. Phys Ther 2011 May;91(5):722-32.
(67) Beneciuk JM, Bishop MD, Fritz JM, Robinson ME, Asal NR, Nisenzon AN, et al. The
STarT Back Screening Tool and Individual Psychological Measures: Evaluation of
Prognostic Capabilities for Low Back Pain Clinical Outcomes in Outpatient Physical
Therapy Settings. Phys Ther 2012 Nov 2.
53
(68) Hill JC, Hay EM. Invited commentary. Phys Ther 2011 May;91(5):733-4.
(69) Albert HB, Briggs AM, Kent P, Byrhagen A, Hansen C, Kjaergaard K. The prevalence of
MRI-defined spinal pathoanatomies and their association with modic changes in
individuals seeking care for low back pain. Eur Spine J 2011 Aug;20(8):1355-62.
(70) Grotle M, Vollestad NK, Brox JI. Clinical course and impact of fear-avoidance beliefs in
low back pain: prospective cohort study of acute and chronic low back pain: II. Spine
(Phila Pa 1976 ) 2006 Apr 20;31(9):1038-46.
(71) Angst F, Verra ML, Lehmann S, Aeschlimann A. Responsiveness of five conditionspecific and generic outcome assessment instruments for chronic pain. BMC Med Res
Methodol 2008;8:26.
(72) Woby SR, Roach NK, Urmston M, Watson PJ. Psychometric properties of the TSK-11: a
shortened version of the Tampa Scale for Kinesiophobia. Pain 2005 Sep;117(1-2):137-44.
(73) Chou R, Shekelle P. Will this patient develop persistent disabling low back pain? JAMA
2010 Apr 7;303(13):1295-302.
(74) Nicholas MK, Linton SJ, Watson PJ, Main CJ. Early identification and management of
psychological risk factors ("yellow flags") in patients with low back pain: a reappraisal.
Phys Ther 2011 May;91(5):737-53.
(75) Bruyere O, Demoulin M, Brereton C, Humblet F, Flynn D, Hill JC, et al. Translation
validation of a new back pain screening questionnaire (the STarT Back Screening Tool) in
French. Arch Public Health 2012;70(1):12.
(76) Gusi N, Del Pozo-Cruz B, Olivares PR, Hernandez-Mocholi M, Hill JC. The Spanish
version of the "STarT Back Screening Tool" (SBST) in different subgroups. Aten Primaria
2011 Feb 4.
(77) Kongsted A, Johannesen E, Leboeuf-Yde C. Feasibility of the STarT back screening tool
in chiropractic clinics: a cross-sectional study of patients with low back pain. Chiropr Man
Therap 2011;19:10.
(78) Mokkink LB, Terwee CB, Patrick DL, Alonso J, Stratford PW, Knol DL, et al. The
COSMIN study reached international consensus on taxonomy, terminology, and
definitions of measurement properties for health-related patient-reported outcomes. J Clin
Epidemiol 2010 Jul;63(7):737-45.
(79) Hayden JA, Cote P, Steenstra IA, Bombardier C. Identifying phases of investigation helps
planning, appraising, and applying the results of explanatory prognosis studies. J Clin
Epidemiol 2008 Jun;61(6):552-60.
(80) Dunn KM, Croft PR. The importance of symptom duration in determining prognosis. Pain
2006 Mar;121(1-2):126-32.
(81) Hestbaek L, Leboeuf-Yde C, Manniche C. Low back pain: what is the long-term course?
A review of studies of general patient populations. Eur Spine J 2003 Apr;12(2):149-65.
54
(82) Albert HB, Godskesen M, Korsholm L, Westergaard JG. Risk factors in developing
pregnancy-related pelvic girdle pain. Acta Obstet Gynecol Scand 2006;85(5):539-44.
(83) Sowden G, Hill JC, Konstantinou K, Khanna M, Main CJ, Salmon P, et al. Targeted
treatment in primary care for low back pain: the treatment system and clinical training
programmes used in the IMPaCT Back study (ISRCTN 55174281). Fam Pract 2012
Feb;29(1):50-62.
(84) Flynn T, Fritz J, Whitman J, Wainner R, Magel J, Rendeiro D, et al. A clinical prediction
rule for classifying patients with low back pain who demonstrate short-term improvement
with spinal manipulation. Spine (Phila Pa 1976 ) 2002 Dec 15;27(24):2835-43.
55
Manuscript 1: Translation and discriminative validation of the STarT Back Screening Tool into Danish Eur Spine J (2011) 20:2166–2173
DOI 10.1007/s00586-011-1911-6
ORIGINAL ARTICLE
Translation and discriminative validation of the STarT Back
Screening Tool into Danish
Lars Morsø • Hanne Albert • Peter Kent
Claus Manniche • Jonathan Hill
•
Received: 12 January 2011 / Revised: 29 April 2011 / Accepted: 4 July 2011 / Published online: 19 July 2011
Springer-Verlag 2011
Abstract
Objective The STarT Back Screening Tool (STarT) is a
nine-item patient self-report questionnaire that classifies
low back pain patients into low, medium or high risk of
poor prognosis. When assessed by GPs, these subgroups
can be used to triage patients into different evidence-based
treatment pathways. The objective of this study was to
translate the English version of STarT into Danish (STarTdk) and test its discriminative validity.
Methods Translation was performed using methods recommended by best practice translation guidelines. Psychometric validation of the discriminative ability was
performed using the Area Under the Curve statistic. The
Area Under the Curve was calculated for seven of the nine
items where reference standards were available and compared with the original English version.
Results The linguistic translation required minor semantic and layout alterations. The response options were
changed from ‘‘agree/disagree’’ to ‘‘yes/no’’ for four items.
No patients reported item ambiguity using the final version.
Electronic supplementary material The online version of this
article (doi:10.1007/s00586-011-1911-6) contains supplementary
material, which is available to authorized users.
L. Morsø (&) H. Albert P. Kent C. Manniche
Research Department, Spine Centre of Southern Denmark,
Hospital Lillebaelt, Oestre Hougvej 55, 5500 Middelfart,
Denmark
e-mail: [email protected]
L. Morsø H. Albert P. Kent C. Manniche
Institute of Regional Health Services Research,
University of Southern Denmark, Middelfart, Denmark
J. Hill
Primary Care Musculoskeletal Research Centre,
Keele University, Keele, Staffordshire ST5 5GB, UK
123
The Area Under the Curve ranged from 0.735 to 0.855
(CI95% 0.678–0.897) in a Danish cohort (n = 311) and
0.840 to 0.925 (CI95% 0.772–0.948) in the original English
cohort (n = 500). On four items, the Area Under the Curve
was statistically similar between the two cohorts but lower
on three psychosocial sub-score items.
Conclusions The translation was linguistically accurate
and the discriminative validity broadly similar, with some
differences probably due to differences in severity between
the cohorts and the Danish reference standard questionnaires not having been validated. Despite those differences,
we believe the results show that the STarT-dk has sufficient
patient acceptability and discriminative validity to be used
in Denmark.
Keywords STarT Back Screening Tool Linguistic Cultural Translation Psychometric Validation
Introduction
Predicting outcome from an episode of low back pain
(LBP) is of interest to both the clinical and research
communities [1–3]. Generic and specific questionnaires
have been developed to inform that process [4–6], but there
has been a need for simple tools for clinicians to use in the
triage of patients in routine clinical care. The STarT Back
Screening Tool (STarT) is a nine-item patient self-report
questionnaire, recently validated for triage of non-specific
LBP patients in primary care [7]. STarT identifies modifiable prognostic factors from the health domains of pain,
activity limitation and psychosocial factors, which are risk
factors for persistent non-specific LBP. STarT classifies
patients into three groups: low, medium or high risk of poor
prognosis, which are based on patients’ symptom
Eur Spine J (2011) 20:2166–2173
complexity [7]. When assessed by GPs, these groups can be
used to assist decisions about appropriate evidence-based
treatment pathways. Advice and guidance is recommended
for the low-risk group, referral for further treatment that
focuses on physical aspects for the medium-risk group and
referral for a multidimensional treatment that targets both
physical and psychosocial factors for the high-risk group
[7].
STarT was developed in Britain and has been translated
from English into Norwegian, Dutch, French, Spanish,
Welsh, Arabic and Mandarin Chinese [8], but not Danish.
The objective of this study was to translate the English
version of STarT into Danish and test the discriminative
validity of the translated version.
Methods
2167
to request collaboration in the translation project and
access to the original validation data. A steering committee
was formed consisting of two clinicians/academics whose
native language was Danish and two whose native language was English. That committee included a representative of the research group at Keele.
Phase 2: translation (English to Danish)
The original STarT questionnaire was translated from
English into Danish by a native Danish-speaking, professional translator. The translator was conceptually introduced to the STarT target audience and health condition
and asked to take explanatory notes during the course of
the translation process.
Phase 3: back translation (Danish to English)
This study consisted of two stages: (1) a linguistic and
cultural translation and (2) a psychometric validation of
discriminative ability. They were conducted at the Spine
Centre of Southern Denmark, a multi-disciplinary secondary care facility. The method used was based on best
practice as recommended by translation guidelines [9–11].
Linguistic and cultural translation
An overview of the phases in the linguistic and cross-cultural translation is shown in Table 1.
A back translation of the Danish version was then conducted by an independent native-Danish speaking translator whose qualifications included a university degree in
English. The translator was similarly conceptually introduced to the STarT target audience and health condition
but the translation occurred without direct content knowledge of the original STarT version. During the translation
process, explanatory notes were also taken by the
translator.
Phase 4: synthesis
Phase 1: liaison with STarT developers
Contact was established with the research group at Keele
University that developed STarT. This was to determine
whether revisions to the English version were in progress,
The content of the original and reverse-translated English
versions were compared, and differences were noted. These
differences were discussed with two independent native
English-speaking reviewers, one of whom was a clinician
and one who was not. The reviewers commented on the
differences and a synthesis of these differences was
created.
Table 1 Phases in the linguistic and cultural translation
Phases
Tasks
Liaison with STarT
developers
• Contact with the developers
2
3
Translation
Back translation
• Translation from English to Danish
• Back translation from Danish to
English
4
Synthesis
• Comparison of translations
5
Translation
committee
• Review of translated versions
6
Pilot testing
• Testing in clinical setting
7
Final version
1
• Formation of steering committee
• Reaching of consensus and
development of pilot version
• Revision of pilot version
• Testing of final version
Phase 5: translation committee
The original English, the Danish and the reverse-translated
English versions, plus the synthesis of translation differences were presented to a translation committee, who had
been formed to ensure cultural relevance and conceptual
equivalence. The committee consisted of seven bilingual
people (clinicians/academics and lay people) and included
chiropractors, physiotherapists, a surgeon and secretaries.
The translation committee discussed differences in translation, whether these reflected linguistic imprecision or
cultural differences, and where needed, suggested alternative wording. This process continued until consensus was
reached.
123
2168
Phase 6: pilot testing
A Danish pilot version of STarT was tested with 17 randomly selected LBP patients at the Spine Centre to determine the acceptability and comprehensibility of the
translation. The only inclusion criteria were the presence of
LBP and sufficient Danish language skills to complete the
questionnaire. The pilot testing was conducted by two
independent members of the translation committee, both of
whom were clinicians. For each patient, their response
pattern, hesitation and uncertainty while completing the
pilot questionnaire were noted, along with the specific
questions involved. Patients were asked about item ambiguity and difficulty. These findings were again discussed in
the plenary group and the translation was adjusted, until no
further hesitation or uncertainty was observed (until
‘saturation’).
Phase 7: final version
A second wave of testing was conducted on ten randomly
selected patients using the revised questionnaire, but no
further hesitation or uncertainty was observed or item
ambiguity reported. This became our final Danish version
of STarT, called the STarT-dk.
Psychometric validation
The discriminative validity of the STarT-dk was described
using the Area Under the Curve (AUC) statistic derived
from Receiver Operating Curves. The AUC was calculated
for the seven items in STarT-dk where Danish versions of
reference standard questionnaires were available. Each
STarT-dk question was used as the ‘state’ variable (a yes
on the screening question being the ‘state’) and was compared with its score on the appropriate reference standard
[7]. For comparison, the AUC was also calculated for the
English version using the original data used to validate the
English version [7].
The reference standards were the Roland Morris Disability Questionnaire [12–14] (activity limitation), Coping
Strategies Questionnaire [15, 16] (catastrophising), Tampa
Scale for Kinesiophobia [17, 18], (fear of movement) and
Hospital Anxiety and Depression Scale [19, 20] (anxiety
and depression). All reference standards were administrated in Danish. To our knowledge Roland Morris Disability Questionnaire is the only questionnaire which has
been translated into Danish using methods that meet current recommendations for cross-cultural adaptation.
Translation of the Tampa Scale for Kinesiophobia into
Danish was performed at our clinical facility (forward and
backward translation), but the cross-cultural adaptation is
incomplete and the work is currently unpublished. The
123
Eur Spine J (2011) 20:2166–2173
Coping Strategies Questionnaire was sourced in a book of
Scandinavian psychosocial questionnaires [21], but the
quality of the translation is not reported. Four independent
Danish translations of The Hospital Anxiety and Depression Scale were compared and a final Danish version was
made with the purpose to approximate the original as much
as possible. No further information is given on the translation and validation process [22]. These were the only
available Danish language versions of the reference standards used in the English study and it was beyond the scope
of the current study to validate the translations available for
the reference standards.
The AUC represents the ability of the screening question
to discriminate between patients with and without the
symptom or sign being assessed. In statistical terms, it is
sensitivity (true positive rate) divided by 1-specificity (1true negative rate) [23]. An AUC of 1.0 is perfect discrimination and an AUC of 0.5 is discrimination no better
than chance. AUC was chosen as the statistical approach
because STarT is a multidimensional questionnaire consisting of one or two screening questions for each of eight
underlying constructs. Therefore, multivariable analysis
such as Rasch analysis [24] and other tests of internal
validity are not appropriate, as they are designed for unidimensional instruments or instruments with many items
measuring each construct.
Data were double-entered independently into a database
(Epidata 3.1, http://www.epidata.dk ‘‘The EpiData Association’’ Odense, Denmark) by two research assistants.
Missing values were imputed using the multiple imputation
feature of PASW 18.0 (formerly SPSS) at default settings.
Descriptive statistics (proportions, mean and standard
deviation) were used to illustrate cohort characteristics.
Differences between cohorts or subgroups were tested
using Pearson Chi-square test for binomial or ordinal data
and unpaired t test for continuous data. Chi-square tests
were performed using Prism 5.0 (Graphpad Software Inc,
La Jolla, CA, USA). All other statistical analyses were
conducted using PASW 18.0 (IBM Inc., Somers, NY,
USA).
Results
Linguistic and cultural translation
During phases 2 and 3, some minor linguistic differences
emerged for questions 2 (pain in other body regions), 3
(walking), 4 (dressing) and 7 (catastrophising). These differences were presented to the reviewers in phase 4, who
independently believed these differences to be of no consequence. For question 6, ‘Worrying thoughts have been
going through my mind a lot of the time’ (English version)
Eur Spine J (2011) 20:2166–2173
2169
a
b
STarT-dk
The STarT Back Screening Tool
Patientens navn:____________________________ Dato:___________
Patient name:____________________________ Date:___________
Tænk tilbage på de seneste to uger og markér dit svar på følgende spørgsmål
Thinking about the last 2 weeks tick your response to the following questions:
Disagree
0
Agree
1
1. My back pain has spread down my leg(s) at some time in the last 2
weeks
Nej
0
Ja
1
Uenig
0
Enig
1
1. I løbet af de seneste 2 uger har mine rygsmerter bredt sig ned i mit/mine
ben
2. Jeg har haft smerter i mine skuldre eller nakke i løbet af de seneste 2
uger
2. I have had pain in the shoulder or neck at some time in the last 2 weeks
3. I have only walked short distances because of my back pain
3. Jeg har kun gået korte afstande på grund af mine rygsmerter.
4. I løbet af de seneste 2 uger har jeg klædt mig langsommere på end
normalt på grund af Rygsmerter.
4. In the last 2 weeks, I have dressed more slowly than usual because of
back pain
5. It’s not really safe for a person with a condition like mine to be
physically active
5. Det er egentligt ikke sikkert for en person i min tilstand at være fysisk
aktiv
6. Worrying thoughts have been going through my mind a lot of the time
6. Jeg har været bekymret meget af tiden
7. I feel that my back pain is terrible and it’s never going to get any better
7. Jeg føler mine rygsmerter er forfærdelige og de bliver aldrig bedre
8. In general I have not enjoyed all the things I used to enjoy
8. Generelt har jeg ikke nydt alle de ting, som jeg plejede at nyde
9. Overall, how bothersome has your back pain been in the last 2 weeks?
9. Overordnet set, hvor generende har dine rygsmerter været de seneste 2 uger?
Not at all
Slightly
Slet ikke
Lidt
0
0
0
0
Total score (all 9):__________
Moderately
0
Very much
0
Extremely
0
Sub Score (Q 5-9):________
Total score (alle 9):__________
Middel
Meget
0
0
Ekstremt
0
Sub Score (spg. 5-9):________
Fig. 1 a The STarT*, b STarT-dk
was translated as ‘I have been worried a lot of the time’
(Danish version) as this was the closest that was linguistically possible. Linguistic difficulty in translating time
factors emerged in a few questions resulting in slight
potential ambiguity of these questions in Danish. In question 9 the phrase ‘Overall, how bothersome …’ (English
version) has no equivalence in the Danish language and
this was noted by the reviewers and the plenary group.
Therefore, a slightly different wording closer to ‘Overall,
how much of an irritation…’ (Danish version) was used
and seemed to work in the pilot version. Initially, the
translation of question 5 contained a shift from ‘It’s not
really safe…’ (English version) to ‘It is not safe, really…’
(Danish version) but both phrasings were tested and
eventually a translation very close to the original wording
was used.
The translation committee reached consensus on a pilot
questionnaire that was subsequently tested in phase 6 on 17
patients. During that testing, patient hesitation and perception of item ambiguity were noted primarily for question 5 and also the response options ‘Disagree’ or ‘Agree’.
Several patients argued that some questions could not be
answered in Danish as ‘Disagree’/‘Agree’ but rather should
be answered as ‘No’/’Yes’ and some patients suggested
there was an important distinction in Danish between ‘Not
safe’ and ‘Not wise’. Therefore, the translation committee
produced a revised translation that included ‘No/Yes’
response options for four of the questions. The revised
questionnaire was tested in phase 7 and as saturation had
been achieved, no further linguistic adjustments were
made. The English version of the STarT questionnaire is
shown in Fig. 1a and the final Danish edition of the STarT
is shown in Fig. 1b.
Psychometric validation
During this psychometric validation stage, 513 questionnaires were posted and 311 questionnaires returned—a
response rate of 60.6%. The only information available on
the 202 non-responders was their age and gender. Nonresponders were younger (mean age 46.3 SD 14.5) that
responders (mean age 51.4 SD 15.7) (p = 0.003 unpaired
t test) and less likely to be women (non-responders 49.5%
women, responders 59.5% women, p = 0.026 Chi-square).
This resulted in a Danish cohort that was dissimilar to the
English cohort in age (6.4 year difference in mean age) but
almost identical in gender mix.
The participant characteristics of both the Danish and
English cohorts are shown in Table 2. The prevalence of
each STarT-dk classification group in the Danish cohort
was 39.8% (low), 34.0% (medium) and 26.2% (high). The
proportion of patients with leg pain within the last 14 days
123
2170
Eur Spine J (2011) 20:2166–2173
Table 2 Participant characteristics of the Danish and the English validation samples
Danish validation sample
(n = 311)
English validation sample
(n = 500)
Tests for differences between the
cohorts
Female
185 (59.5%)
293 (59%)
p = 0.8912*
Age, mean (±SD years)
51.4 (15.7)
45 (9.7)
p = 0.0001^
39.8%
47%
p = 0.788#
STarT group, proportions
Low
Medium
34.0%
38%
High
26.2%
15%
17.9 (7.2)
Not comparable
Leg, mean (±SD)
12.9 (9.5)
Not comparable
Numbers with leg pain
239 (76.8%)
303 (61%)
56.6 (24.1)
39.6 (25.6)
p = 0.0343^
98 (31.5%)
276 (55%)
p = 0.0002^
37.5 (8.5)
39.5 (6.9)
p = 0.0003^
12.9 (7.9)
10 (7.9)
p = 0.0001^
6.7 (4.4)
8.2 (4.5)
p = 0.0001^
4.2 (4.0)
6.7 (4.3)
p = 0.0001^
Pain intensity (scale 0–30)
Back, mean (±SD)
p \ 0.0000*
RMDQ (scale 0–100§)
Mean (±SD)
Widespread pain neck/shoulders, n (%)
TSK (scale 17–68§)
Mean (±SD)
CSQ (scale 0–36§)
Mean (±SD)
§
HADS anxiety (scale 0–20 )
Mean (±SD)
HADS depression (scale 0–20§)
Mean (±SD)
RMDQ Roland Morris Disability Questionnaire, TSK Tampa Scale of Kinesophobia, CSQ Coping Strategy Questionnaire (catastrophisation
domain), HADS (A) Hospital Anxiety and Depression Scale (anxiety subscale), HADS (D) Hospital Anxiety and Depression Scale (depression
subscale)
*
Z test of proportions
^
Unpaired t test
#
Kruskal–Wallis test
§
High scores are worse
was 79.4%, while 31.5% also reported pain in either the
neck or shoulders. The mean sum score and standard
deviation of the reference standard questionnaires are also
reported and show that there were differences between the
cohorts on most measures. This is likely to reflect the
Danish cohort’s being from the secondary care sector and
the English cohort from the primary care sector.
Across the STarT variables in the Danish cohort, there
was on average 2.19% missing data (range 5.8–0.6%).
Across the reference standard variables this was 2.8%
(range 10.3–1.0%). The AUC results calculated with
missing values and with imputed values were compared.
The largest difference in a point estimate of AUC was 0.08.
Therefore, as the results calculated with missing data were
almost identical to those using imputed values, only results
from the original unimputed data are reported.
Identical analyses were made for seven out of nine
STarT questions for both the Danish and the English
123
cohorts. The discriminative ability (AUC) for each of those
questions in both cohorts is shown in Table 3. In the
Danish cohort, AUC point estimates ranged from 0.735 to
0.855 (CI95% 0.678–0.897) and in the English cohort these
point estimates ranged from 0.840 to 0.925 (CI95%
0.772–0.948). Overall, the AUC point estimates calculated
for five items were similar between the two cohorts, with
overlapping confidence intervals but there were differences
on three psychosocial sub-score items. The discriminative
ability for pain referral, pain localisation, activity limitation and fear of movement was similar. The divergence
was on the anxiety, catastrophisation and depression items,
as these screening questions had lower discriminative
ability in the Danish cohort than in the English cohort.
Therefore, correlations (Pearson r) between each individual
question on each of the reference standard questionnaires
for anxiety, catastrophisation and depression and the sum
score of its questionnaire were calculated. This was to
Eur Spine J (2011) 20:2166–2173
2171
Table 3 Area under Curve for each STarT question compared with its reference standard
Question on STarT
Danish
Reference standard point
estimate (CI95%)
English^
Reference standard point
estimate (CI95%)
1. My back pain has spread down my leg(s) in
the last 2 weeks
Single question: ‘localisation of
pain’ (leg pain stated yes/no)
0.748 (CI95% 0.692–0.805)
Using a single question on current
co-morbid pain sites, positive for leg
pain (yes/no) 0.856 (0.784–0.927)
2. I have had pain in the shoulder or neck at some
time in the last 2 weeks
Sum score of pain localisation 0.793 (CI95%
0.742–0.845)
Pain sites* 0.898 (0.842–0.955)
3. I have only walked short distances because of
my back pain
RMDQ 0.846 (CI95% 0.804–0.889)
RMDQ 0.880 (0.821–0.938)
4. In the last 2 weeks, i have dressed more
slowly than usual because of back pain
RMDQ 0.855 (CI95% 0.814–0.897)
RMDQ 0.846 (0.772–920)
5. It’s not really safe for a person with a
condition like mine to be physically active
TSK 0.775 (CI95% 0.714–0.837)
TSK 0.840 (0.770–0.908)
6. Worrying thoughts have been going through
my mind a lot of the time
HADS ANX 0.837 (CI95% 0.792–0.882)
HADS ANX 0.918 (0.894–0.942)
7. I feel that my back pain is terrible and it’s
never going to get any better
CSQ 0.779 (CI95% 0.726–0.832)
CSQ 0.925 (0.902–0.948)
8. In general I have not enjoyed all the things I
used to enjoy
HADS DEP 0.735 (CI95% 0.678–0.792)
HADS DEP 0.902 (0.876–0.929)
9. Overall, how bothersome has your back pain
been in the last 2 weeks
No reference standard
No reference standard
RMDQ Roland Morris Disability Questionnaire, TSK Tampa Scale of Kinesophobia, HADS Hospital Anxiety and Depression Scale, CSQ Coping
Strategy Questionnaire
^ Analysis performed on the external patient sample
* Estimates not comparable
examine if there was evidence that the STarT question used
to screen for that construct in the English version might not
be the most appropriate in this cultural setting. The results
are shown in Appendix 1.
They indicate that it was uncommon that items other
than those in the original STarT version displayed a
stronger association with the reference standard and that
when it occurred the difference in association was negligible (r \ 0.086). How much these would vary across
consecutive samples in the same language is not known,
but they may be of a similar magnitude. Therefore, there
must have been other reasons for this divergence in discriminative ability.
Discussion
The aim of this study was to translate the English version
of STarT into Danish and test the discriminative validity of
the translated version. We believe the results show that the
STarT-dk has sufficient patient acceptability and discriminative validity to be used in Denmark.
Although the discriminative ability was similar between
the two language versions for most items, there was a
systematic divergence on three psychosocial items and this
may have occurred for a number of reasons. The
divergence was on the anxiety, catastrophisation and
depression items, as these screening questions had lower
discriminative ability in the Danish cohort than in the
English cohort. As one explanation for this divergence in
discriminative ability might be that the screening questions
chosen for the English version of STarT might not be the
most appropriate for the Danish version but, as shown in
Appendix 1, further analysis did not support that explanation. Another possible reason could be linguistic inaccuracies in the translation. This is not likely, as both the
translators and translation committee believe the translation to be linguistically accurate.
As the three items showing divergent discriminative
validity were all psychosocial constructs and as the discriminative ability was systematically lower on all three, it
raises the question as to whether the divergence was the
product of a cultural difference in the way Danish people
answer psychosocial questions. This seems unlikely as,
compared with the English cohort, the Danish cohort
scored higher on some psychosocial constructs and lower
on others. As the two cohorts scored differently on these
psychosocial constructs, it is also possible that the divergence is due to the association between screening question
and reference standard not being linear across the whole
scoring range. Another reason could be the presence of
inaccuracies in the Danish translation of the reference
123
2172
Eur Spine J (2011) 20:2166–2173
standard questionnaires for these three psychosocial constructs. This remains a possibility, as although these
translations have been performed, there are no validation
studies available for those reference standards. It is also
possible that some variability between samples is inevitable, even within the same language population and we do
not have data to quantify that variability.
Questions of the acceptability and clinical importance of
divergence have been discussed in similar contexts [25]. As
the research community has not produced criterion standards to guide decisions on when divergence is clinical
important, such results can only be descriptively reported.
The strengths of this study are the staged process of
language translation and the direct comparison of discriminative validity with data from the original validation
of the English version. A potential weakness of the study is
that the translation and validation were not specifically
inclusive of non-native Danish speakers as recommended
by some methodologists [26, 27]. During the psychometric
validation there were no restrictions on the linguistic
background of the target population, but it is unknown the
extent to which any indirect cross cultural validation may
have occurred. The translation and psychometric validation
was conducted in the secondary care sector as this was the
population of convenience. Subsequent work will investigate its validity in primary care patients and determine
whether the cut points in the English version are the most
appropriate for the Danish version.
Conclusion
The STarT questionnaire was translated into Danish and its
discriminative validity measured. The translation was
judged to be linguistically accurate and the STarT-dk tool
acceptable for patient use. The discriminative validity largely seemed comparable with the original English version
but three psychosocial questions displayed lower discriminative validity. We suspect this is most likely to be a
product of differences in severity between the cohorts and
variability due to the Danish versions of the reference
standard questionnaires not having been validated. Despite
those differences, we believe the results show that the
STarT-dk has sufficient patient acceptability and discriminative validity to be used in Denmark.
Acknowledgments We are grateful for funding from the Region of
Southern Denmark, the Danish Rheumatism Association and the
Association of Danish Physiotherapists. We are also grateful to Ms
Lene Ververs for data management.
Conflict of interest
conflict of interest.
123
The authors declare no financial or ethical
References
1. Fritz JM, Cleland JA, Childs JD (2007) Subgrouping patients
with low back pain: evolution of a classification approach to
physical therapy. J Orthop Sports Phys Ther 37(6):290–302
2. Grotle M, Brox JI, Glomsrod B, Lonn JH, Vollestad NK (2007)
Prognostic factors in first-time care seekers due to acute low back
pain. Eur J Pain 11(3):290–298
3. Kent P, Keating JL, Leboeuf-Yde C (2010) Research methods for
subgrouping low back pain. BMC Med Res Methodol 10(1):62
4. Bendebba M, Dizerega GS, Long DM (2007) The Lumbar Spine
Outcomes Questionnaire: its development and psychometric
properties. Spine J 7(1):118–132
5. Brazier JE, Harper R, Jones NM, O’Cathain A, Thomas KJ,
Usherwood T et al (1992) Validating the SF-36 health survey
questionnaire: new outcome measure for primary care. BMJ
305(6846):160–164
6. Staerkle R, Mannion AF, Elfering A, Junge A, Semmer NK,
Jacobshagen N et al (2004) Longitudinal validation of the fearavoidance beliefs questionnaire (FABQ) in a Swiss-German
sample of low back pain patients. Eur Spine J 13(4):332–340
7. Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE et al
(2008) A primary care back pain screening tool: identifying patient
subgroups for initial treatment. Arthritis Rheum 59(5):632–641
8. The STarT Back Screening tool website. 2010. Ref Type: Internet
Communication
9. Beaton DE, Bombardier C, Guillemin F, Ferraz MB (2000)
Guidelines for the process of cross-cultural adaptation of selfreport measures. Spine 25(24):3186–3191
10. Bullinger M, Alonso J, Apolone G, Leplege A, Sullivan M,
Wood-Dauphinee S et al (1998) Translating health status questionnaires and evaluating their quality: the IQOLA Project
approach. International Quality of Life Assessment. J Clin Epidemiol 51(11):913–923
11. Ware JE Jr, Keller SD, Gandek B, Brazier JE, Sullivan M (1995)
Evaluating translations of health status questionnaires. Methods
from the IQOLA project. International Quality of Life Assessment. Int J Technol Assess Health Care 11(3):525–551
12. Roland M, Morris R (1983) A study of the natural history of back
pain. Part I: development of a reliable and sensitive measure of
disability in low-back pain. Spine 8(2):141–144
13. Roland M, Fairbank J (2000) The Roland-Morris Disability
Questionnaire and the Oswestry Disability Questionnaire. Spine
25(24):3115–3124
14. Albert HB, Jensen AM, Dahl D, Rasmussen MN (2003) Criteria
validation of the Roland Morris questionnaire. A Danish translation of the international scale for the assessment of functional
level in patients with low back pain and sciatica. Ugeskr Laeger
165:1875–1880
15. Rosenstiel AK, Keefe FJ (1983) The use of coping strategies in
chronic low back pain patients: relationship to patient characteristics and current adjustment. Pain 17(1):33–44
16. Swartzman LC, Gwadry FG, Shapiro AP, Teasell RW (1994) The
factor structure of the Coping Strategies Questionnaire. Pain
57(3):311–316
17. Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW,
Oostendorp RA (2003) Psychometric properties of the Tampa
Scale for kinesiophobia and the fear-avoidance beliefs questionnaire in acute low back pain. Man Ther 8(1):29–36
18. Vlaeyen JW, Kole-Snijders AM, Boeren RG, van EH (1995) Fear
of movement/(re)injury in chronic low back pain and its relation
to behavioral performance. Pain 62(3):363–372
19. Bjelland I, Dahl AA, Haug TT, Neckelmann D (2002) The
validity of the Hospital Anxiety and Depression Scale. An
updated literature review. J Psychosom Res 52(2):69–77
Eur Spine J (2011) 20:2166–2173
20. Zigmond AS, Snaith RP (1983) The hospital anxiety and
depression scale. Acta Psychiatr Scand 67(6):361–370
21. Friis-Hasché E, Elsaas P, Nielsen T (eds) (2004) Appendix 4. In:
Clinical health psychology. 1st edn. Munksgaard, Copenhagen,
p 435
22. Groenvold M, Fayers PM, Sprangers MA, Bjorner JB, Klee MC,
Aaronson NK et al (1999) Anxiety and depression in breast cancer
patients at low risk of recurrence compared with the general
population: a valid comparison? J Clin Epidemiol 52:523–530
23. Kirkwood BRSJAC (1988) Measurement error: assessment and
implications. In: Essential Medical Statistics. 2nd edn, Oxford:
Blackwell Science Ltd., 429–446
24. Pallant JF, Tennant A (2007) An introduction to the Rasch
measurement model: an example using the Hospital Anxiety and
Depression Scale (HADS). Br J Clin Psychol 46(Pt 1):1–18
2173
25. Lauridsen HH, Hartvigsen J, Manniche C, Korsholm L, GrunnetNilsson N (2006) Danish version of the Oswestry Disability Index
for patients with low back pain. Part 1: cross-cultural adaptation,
reliability and validity in two different populations. Eur Spine J
15(11):1705–1716
26. Acquadro C, Conway K, Hareendran A, Aaronson N (2008)
Literature review of methods to translate health-related quality of
life questionnaires for use in multinational clinical trials. Value
Health 11(3):509–521
27. Altman DG, Vergouwe Y, Royston P, Moons KG (2009) Prognosis and prognostic research: validating a prognostic model.
BMJ 338:b605
123
Manuscript 2: The predictive and external validity of the STarT Back Tool in Danish primary care Eur Spine J
DOI 10.1007/s00586-013-2690-z
ORIGINAL ARTICLE
The predictive and external validity of the STarT Back Tool
in Danish primary care
Lars Morsø • Peter Kent • Hanne B. Albert •
Jonathan C. Hill • Alice Kongsted • Claus Manniche
Received: 12 September 2012 / Revised: 17 December 2012 / Accepted: 25 January 2013
Ó Springer-Verlag Berlin Heidelberg 2013
Abstract
Purpose The STarT Back Tool (SBT) was recently
translated into Danish and its concurrent validity described.
This study tested the predictive validity of the Danish SBT.
Methods Danish primary care patients (n = 344) were
compared to a UK cohort. SBT subgroup validity for predicting high activity limitation at 3 months’ follow-up was
assessed using descriptive proportions, relative risks, AUC
and odds ratios.
Results The SBT had a statistically similar predictive
ability in Danish primary care as in UK primary care.
Unadjusted relative risks for poor clinical outcome on
activity limitation in the Danish cohort were 2.4 (1.7–3.4)
for the medium-risk subgroup and 2.8 (1.8–3.8) for the
high-risk subgroup versus 3.1 (2.5–3.9) and 4.5 (3.6–5.6)
for the UK cohort. Adjusting for confounders appeared to
explain the lower predictive ability of the Danish high-risk
group.
Conclusions The Danish SBT distinguished between
low- and medium-risk subgroups with a similar predictive
L. Morsø (&) P. Kent H. B. Albert C. Manniche
Research Department, Spine Centre of Southern Denmark,
Hospital Lillebaelt, Institute of Regional Health Services
Research, University of Southern Denmark, A Member
of the Clinical Locomotion Network, Oestre Hougvej 55,
Middelfart 5500, Denmark
e-mail: [email protected]
J. C. Hill
Arthritis Research UK Primary Care Centre, Keele University,
Keele, Staffordshire, UK
A. Kongsted
Nordic Institute of Chiropractic and Clinical Biomechanics,
A Member of the Clinical Locomotion Network, Odense,
Denmark
ability of the UK SBT. That distinction is useful information for informing patients about their expected prognosis and may help guiding clinicians’ choice of treatment.
However, cross-cultural differences in the SBT psychosocial subscale may reduce the predictive ability of the highrisk subgroup in Danish primary care.
Keywords Classification Predictive value of tests Validation Low back pain STarT Back Tool
Introduction
The STarT Back Screening Tool (SBT) has shown promise
in triaging non-specific low back pain (NSLBP) patients in
primary care [1, 2]. In the (UK), the SBT has shown an
ability in primary care to identify modifiable prognostic
factors, classify people into prognostic subgroups [1] and
improve patient outcomes through subgroup-matched
treatment pathways [3].
The SBT was recently translated into Danish, using bestpractice translation methods [4, 5]. The result was a linguistically accurate and culturally acceptable tool with
adequate discriminative validity to be used in Danish primary care [6]. However, the predictive and external validity
of SBT in Danish primary care has not been established [6]
and therefore there is a need for those aspects of the SBT’s
measurement properties to be described [7].
There is an increasing focus on questionnaires that index
prognostic risk in low back pain (LBP), such as the SBT [1]
and the Orebro Musculoskeletal Pain Questionnaire [8].
Questionnaires that can be used in routine care settings
require a trade-off between brevity, simplicity and precision [9] and adequate validation is important if clinicians
are to be confident in their use. Recognising these
123
Eur Spine J
challenges, Justice et al. [10] proposed a multi-layered
approach to external validation that includes a focus on
reproducibility and transportability, and it has been recommended that validity testing occurs in multiple and
setting-specific samples [11].
Validation of the predictive ability of the SBT in the
cultural context of Danish primary care would determine
whether prognostic stratification based on SBT subgroups
is similar to that in the UK, and inform the implementation
of this screening tool in Danish daily healthcare practice [7,
12]. Therefore, the aim of this study was to compare the
predictive validity of the Danish version of the SBT in
Danish primary care to the English version of the SBT in
UK primary care.
Methods
Patient groups
Two Danish patient samples from general medical practice
(GP) and physiotherapy primary care clinics were pooled
into a cohort representing Danish primary care (Danish
cohort). The analyses were performed on this cohort and
compared to an existing NSLBP cohort from UK primary
care (UK cohort) [13]. In both cohorts, almost all patients
were initially triaged by GPs, some completing their study
questionnaires during or after that consultation and others
at physiotherapy practices following referral.
The Danish GP sample was prospectively recruited as
part of a GP audit conducted by the regional health
authority from February to May 2011, while prospective
data collection from 27 Danish physiotherapy clinics
occurred from May to September 2011 until 200 baseline
questionnaires were collected. There was a 72 % (GP) and
86 % (physiotherapy) follow-up at 3 months. As the
physiotherapy sample was the smallest (n = 172), the total
Danish cohort (n = 344) was formed by adding 172
patients randomly selected from the GP cohort to evenly
weight the cohort across both professional disciplines. This
balanced mix of GP and physiotherapy patients was to
improve the generalisability of the results, as recommended
by de Vet et al. [11]. A detailed flowchart for the Danish
cohort is shown in Appendix 1.
The inclusion criteria were people 18–65 years of age
with NSLBP identified either by: (1) specific diagnostic
coding recorded in GP electronic patient records, or (2) by
physiotherapists using the criteria contained in the European guidelines for NSLBP in primary care [14]. Participants completed questionnaires on basic demographic
details, fear of movement (Tampa Scale of Kiniesophobia)
[15], catastrophisation (Coping strategy questionnaire)
[16], anxiety and depression (Hospital Anxiety and
123
Depression Scale) [17] and the SBT [1]. Data from both
settings were independently double-entered into a database
(Epidata 3.1, The EpiData Association, Odense, Denmark)
by two research secretaries.
The UK data were from the BeBack Study [13] conducted by the Arthritis Research UK Primary Care Centre,
Keele University, from which baseline and 3-month follow-up questionnaire scores were extracted. The study was
a prospective cohort of consecutive patients, from a
socioeconomically heterogeneous population, who consulted with low back pain in eight general practices in
England. Six months’ follow-up data from this cohort have
previously been reported and the current study uses the full
3 months’ follow-up data (n = 856) available from that
cohort [13]. Details of the recruitment and data collected
have also been previously published [13].
Three-month outcomes were chosen for our study as this
has been shown to be the most important time point for the
clinical course of LBP in primary care, marking the end of
rapid improvement and heralding the onset of persistent
pain [18].
Data analysis
Previous studies have tested the predictive validity of
questionnaires using a variety of methods [19, 20]. In the
current study, the predictive validity and external validity
of the SBT met the criteria proposed by Justice et al. [10]
for comparing cohorts across countries and at a different
outcome time points than from that previously studied (in
our case, 3 months in the current study and 6 months in the
original SBT study in the UK).
Descriptive analysis of the baseline characteristics of the
Danish and the UK cohort was performed (means and
standard deviations, medians and inter-quartile ranges).
Baseline differences between the two cohorts were examined using Mann–Whitney U, Chi-square or Kruskal–
Wallis Tests, depending on the data type and distribution.
The three statistical methods that had been used to
describe the predictive of the SBT in the original UK
validation study [1] were mirrored in our study. The outcome measures used were all threshold scores on measures
of activity limitation, pain and pain bothersomeness. These
were the constructs chosen in the original study due to their
being recommended in expert consensus statements [21].
We replicated these threshold scores so that results could
potentially be compared across studies and time points.
Firstly, comparison was made between the proportions
of patients with a poor clinical outcome on activity limitation at 3 months in both cohorts, stratified by SBT subgroup. Poor clinical outcome was defined in this context as
a Roland Morris disability questionnaire (RMDQ) [22] sum
score (0–100 scale) of 30 points or more [23] at 3 months’
Eur Spine J
follow-up. The cut point used in original SBT development
study in the UK [1] was 7 on a 0–24 scale but as we used
the proportional recalculation method to convert all RMDQ
scores to a 0–100 scale, that threshold was recalculated to
be 30 points or more. The proportional recalculation
method has been shown to be more accurate in managing
any missing RMDQ answers [23].
Secondly, for both cohorts, the same outcome was used
to estimate the additional risk (relative risk) [24] for poor
outcome resulting for people in the medium or high SBT
risk subgroup compared to the low-risk subgroup.
Thirdly, the area under the curve (AUC) statistic from
receiver operating characteristic (ROC) curves was used to
describe the ability of the baseline SBT sum scores (0–9
scale) to discriminate (sensitivity/1-specificity) [24]
between people with and people without the following
outcomes at 3 months: (1) poor outcome on activity limitation as defined above, (2) LBP still being ‘severe’ (8–10
on a 0–10 point scale), and (3) LBP rated as ‘very’ or
‘extremely’ bothersome on a 5-point pain bothersomeness
scale. All these criteria were used in the original UK validation study [1].
In addition, as the proportions of patients with a poor
clinical outcome on activity limitation and unadjusted
relative risks suggested that the psychosocial subscale of
the Danish SBT might not have the same predictive
ability as the UK version, logistic regression was performed to explore for potential confounding. Due to
suspected confounding by treatment exposure (approximately 60 % of the Danish cohort was referred for
physiotherapy and approximately 18 % in the UK cohort)
and by differential treatment effectiveness at modifying
psychosocial risk factors (treatment being heterogeneous),
these covariates were further explored. Adjusted odds
ratios controlled for Danish care setting (GP and physiotherapy) and change in SBT psychosocial subscale risk
factors (fear of movement, catastrophisation, anxiety,
depression and pain bothersomeness), measured by their
full reference standard questionnaires. An odds ratio
greater than 1 in these regression models means that
particular clinical characteristic increases the odds of
having a poor outcome and an odds ratio less than 1
means that it is protective against a poor outcome. All
covariates were initially entered into the model and
reported, followed by a manual backwards stepwise
reduction (p \ 0.05 to remove) to the most parsimonious
model.
Relative risk estimates and random number generation
were performed using Microsoft Excel 2003 (Microsoft
Corp, Redmond, WA, USA). Logistic regression was performed using STATA 12 (StataCorp, College Station, TX,
USA). All other statistical analyses were conducted using
PASW 13.0 (IBM Inc., Somers, NY, USA).
Results
Baseline differences between the Danish and the United
Kingdom cohorts
The two cohorts were significantly different on a number of
baseline characteristics (Table 1). On an overall cohort
level, the Danish cohort reported higher pain intensity,
higher activity limitation, slightly more prevalent leg pain
and slightly higher catastrophisation. There were also significant differences between the two cohorts in the distribution of people across the three SBT groups, with a lower
proportion of ‘low risk’ patients and a higher proportion of
‘high risk’ patients in the Danish cohort.
Comparison of the unadjusted risk of poor clinical
outcome on activity limitation at 3 months
Overall 47 % in the Danish cohort and 36 % in the UK
cohort had poor outcome at 3 months. Reassuringly, in
both the Danish and UK cohort, the proportion of patients
with a poor outcome was lowest in the low-risk subgroup
and highest in the high-risk subgroup (Fig. 1). This is also
reflected in the relative risks of the medium-risk and highrisk subgroups. Although the proportions and relative risks
vary between the cohorts, the gradient in the trend line
across risk subgroups was similar, indicating that predictive ability of the SBT broadly followed a comparable
pattern in both countries.
At an SBT subgroup level, these unadjusted data suggest
that the Danish high-risk subgroup does not have the same
incremental step size in predictive validity compared with
the median-risk subgroup, as in the UK cohort. While the
proportion of patients with a poor outcome was quite
similar in the low-risk subgroups (Danish 24 %; UK 17 %)
and medium-risk subgroups (Danish 57 %; UK 54 %), it
was considerably lower in the Danish high-risk subgroup
(64 %) compared to that in the UK cohort (78 %). As a
consequence, while the relative risks (RR) for the mediumrisk subgroup are comparable across cohorts, there was
only a marginal step up to the high-risk subgroup in the
Danish cohort, whereas the step up was greater in the UK
cohort (Fig. 1). As the distinction between the mediumand high-risk subgroups is that higher scores on the SBT
psychosocial subscale are required to be classified as being
high-risk, these unadjusted results could initially be interpreted as suggesting that the psychosocial subscale does
not have the same predictive validity in the Danish cohort.
Raising the threshold score on the subscale from 4 to 5 had
almost no effect of the predictive strength of the Danish
high-risk subgroup (from to RR 2.7 [1.8; 3.8] to RR 2.8
[1.8; 4.4]). Therefore, logistic regression was performed to
control for potential confounding.
123
Eur Spine J
Table 1 Baseline
characteristics of the Danish and
UK primary care cohorts
Danish cohort
(n = 344)
UK cohort
(n = 856)
Tests for differences
between the Danish
and UK cohorts*
Age in years
Median, (IQRa)
46 (39–53)
p \ 0.001
199 (57.8 %)
305 (58.8 %)
p = 0.772
149 (44.2 %)
327 (38.2 %)
p = 0.556
50.0 (41–59)
Female
Duration (%)
\4 weeks
4–12 weeks
[12 weeks
STarT subgroup, proportions
66 (19.6 %)
221 (25.8 %)
122 (36.2 %)
285 (33.3 %)
Low
121 (37.5 %)
460 (53.7 %)
Medium
127 (39.3 %)
295 (34.5 %)
75 (23.2 %)
90 (10.5 %)
High
b
Pain intensity (0–10 scale)
p \ 0.001
c
Overall median (IQRa)
Mild (0–5)
Moderate (6–7)
7 (5–8)
5 (3–7)
130 (38.7 %)
527 (61.6 %)
p \ 0.001
98 (29.2 %)
196 (22.9 %)
108 (32.1 %)
127 (14.8 %)
Presence of referred leg pain
241 (72.2 %)
518 (60.5 %)
p \ 0.001
Presence of comorbid neck or shoulder pain
155 (46.8 %)
463 (54.1 %)
p = 0.014
60.9 (39–77)
33.3 (17–54)
p \ 0.001
2 (0.6 %)
14 (4.2 %)
29 (3.4 %)
108 (12.6 %)
p \ 0.001
Moderately
82 (24.7 %)
235 (27.5 %)
Very much
178 (53.6 %)
307 (35.9 %)
Extremely
56 (16.6 %)
166 (19.4 %)
36 (30–41)
40 (36–43)
p \ 0.001
10 (5–15.8)
9 (4–14)
p = 0.009
5 (3–8)
8 (5–11)
p \ 0.001
2 (1–5)
6 (3–9)
p \ 0.001
Severe (8–10)
Activity limitationd (0–100 scale)c
Median (IQRa)
* Tests for differences were the
Man–Whitney U, Chi-square or
Kruskal–Wallis procedures,
depending on data type and
distribution
a
Inter-quartile range
b
Numeric Rating Scale (0–10)
c
High scores are worse
d
Roland Morris disability
questionnaire
e
Tampa Scale of
Kiniesophobia
f
Coping strategy questionnaire
(catastrophisation subscale)
g
Hospital Anxiety and
Depression Scale
Pain bothersomeness
Not at all
Slightly
Fear of movemente (17–68 scale)c
Median (IQRa)
Catastrophisationg (0–36 scale)c
Median (IQRc)
Anxietyg (0–21 scale)c
Median (IQRc)
g
Depression (0–21 scale)
c
Median (IQRc)
Comparison of the adjusted risk of poor clinical
outcome on activity limitation at 3 months
The unadjusted odds ratios (OR) in Table 2 mirror the
distinction between the Danish and UK cohorts seen in the
relative risk results. The predictive ability of the mediumrisk subgroups was similar (Danish OR 4.2 [2.5; 7.3], UK
OR 5.6 [4.0; 7.8]), whereas in the Danish cohort the highrisk subgroup added only a little predictive information
(OR 5.6 [3.0; 10.5]) and was much more predictive in the
UK cohort (OR 16.9 [9.7; 29.3]).
Adjustment for care setting and change scores in the
psychosocial constructs resulted in the predictive ability of
123
the Danish ‘high risk’ group (adjusted OR 15.9 [5.2; 48.2])
approximating that of the UK cohort, when the regression
model included all covariates. The parsimonious model
only retained four covariates (care setting, change in anxiety, change in pain bothersomeness and the interaction
between change in pain bothersomeness and care setting)
with the predictive ability of the Danish ‘high risk’ group
being almost identical to that observed in the UK cohort in
this model (adjusted OR 15.7 [6.6; 37.5]). There was no
significant (p \ 0.05) non-linearity in the covariates included in the adjusted models (regression model not reported).
As there was no significant interaction between the
SBT subgroups and care setting (regression model not
Eur Spine J
Relative Risk
4.5 [3.6; 5.6]
100%
90%
80%
Proportions
70%
Relative Risk
2.4 [1.7; 3.4]
Relative Risk
2.7 [1.8; 3.8]
Relative Risk
3.1 [2.5; 3.9]
60%
50%
40%
Relative Risk
1*
Relative Risk
1*
30%
20%
10%
0%
DK
UK
Low risk
24%
17%
Medium risk
57%
54%
High risk
64%
78%
Fig. 1 Proportions of patients within each STarT Back Tool
subgroup who had a poor clinical outcome on activity limitation
(Poor clinical outcome defined as a Roland Morris disability
questionnaire score [30) at 3 months and their relative risk. Asterisk
the low-risk STarT Back Tool subgroup is the reference category
reported), and care setting was not changed by study
participation, we interpret the reduced odds (0.31 [0.13;
0.71] parsimonious model) of a poor outcome in the
physiotherapy patients as evidence of confounding [25,
26]. As the psychosocial covariates may have changed
as result of treatment and/or natural history, we believe
they may be on the causal pathway and interpret their
effect on altering risk as evidence of effect mediation
[25, 26]. It was not possible to calculate adjusted ORs
for the UK cohort as the same covariate data were not
available.
Discussion
Comparison of the ability of the total baseline SBT
scores to identify people with outcomes above a clinical
threshold at 3 months
The AUC statistics describing the ability of the baseline
SBT scores (0–9 scale) to discriminate between people
with and people without scores above threshold values
on three different 3-month outcomes are shown in
Table 3. For the outcomes of LBP ‘still being severe’
and LBP rated as ‘very or extremely bothersome’ the
discriminative ability was similar across cohorts. However, for the outcome of a RMDQ score above 30 points
(0–100 scale) the difference was more substantial
(Danish AUC 0.71 [0.66; 0.77], UK AUC 0.81 [0.78;
0.84]).
The aim of this study was to compare the predictive of the
Danish version of the SBT in Danish primary care and the
English version of the SBT in UK primary care. There were
baseline differences between the cohorts, which reflect
their being from different health care systems and cultures.
However, this study did not aim to match the cohorts but to
include samples that were likely to be representative of
their clinical populations and to compare the SBT predictive ability in each cohort. Overall, the results of the current study indicate that the ability of SBT to predict
increased risk of poor prognosis at 3 months in Danish
primary care was similar to that seen in UK primary care
for the low- and medium-risk SBT subgroups, whereas we
initially observed almost no difference between the predictive strength of the medium- and high-risk subgroups in
the Danish cohort. However, there was a very large difference between the cohorts in exposure to physiotherapy
treatment and we found that after adjusting for this confounding [26] and also for significant effect mediation [26]
due to change in two psychosocial characteristics, the
predictive strength of the high-risk Danish subgroup was
almost identical to the UK cohort (unadjusted estimate).
Data were not available to perform adjusted analysis in the
UK cohort. Whether this asymmetric analysis (Danish
adjusted predictive estimates and UK unadjusted estimates)
123
Eur Spine J
Table 2 The odds of having poor clinical outcome on activity limitation (Roland Morris disability questionnaire score [30 (0–100 scale)) at
3 months by STarT Back Tool subgroup the Danish and UK cohorts, estimated using logistic regression
Danish cohort (n = 322)
UK cohort (n = 845)
Odds ratio [CI 95 %]
Odds ratio [CI 95 %]
p value
p value
Unadjusted model
STarT Back Tool low-risk subgroupa
1.00
STarT Back Tool medium-risk subgroup
STarT Back Tool high-risk
4.24 [2.45; 7.32]
5.57 [2.97; 10.47]
\0.001
\0.001
5.56 [3.99; 7.76]
16.88 [9.71; 29.34]
\0.001
\0.001
Constant
0.32 [0.21; 0.48]
\0.001
0.22 [0.16; 0.26]
\0.001
1.00
Full model adjusted for care setting and change on STarT Back Tool psychosocial constructs (n = 213)
STarT Back Tool low-risk subgroupa
1.00
STarT Back Tool medium-risk subgroup
8.16 [3.44; 19.30]
\0.001
STarT Back Tool high-risk
15.85 [5.22; 48.17]
\0.001
Included covariates
Care settingb
Change in fear of movement
c
0.36 [0.13; 1.01]
0.052
0.99 [0.90; 1.09]
0.879
Change in catastrophisationd
0.87 [0.78; 0.97]
0.013
Change in anxietye
0.81 [0.65; 1.02]
0.078
Change in depressione
1.03 [0.81; 1.34]
0.077
Change in pain bothersomenessf
0.36 [0.19; 0.65]
\0.001
Interaction between care setting and change in fear of movement
0.92 [0.81; 1.04]
0.195
Interaction between care setting and change in catastrophisation
Interaction between care setting and change in anxiety
1.14 [1.00; 1.13]
1.17 [0.88; 1.55]
0.055
0.284
Interaction between care setting and change in depression
0.77 [0.55; 1.09]
0.142
Interaction between care setting and change in pain bothersomeness
1.89 [0.90; 3.97]
0.092
Constant
1.04 [0.42; 2.58]
0.933
Parsimonious model adjusted for care setting and change on STarT Back Tool psychosocial constructs (n = 296), using manual backwards
stepwise procedure
STarT Back Tool low-risk subgroupa
1.00
STarT Back Tool medium-risk subgroup
7.89 [3.87; 16.11]
\0.001
STarT Back Tool high-risk
15.73 [6.60; 37.47]
\0.001
Care settingb
0.31 [0.13; 0.71]
0.006
Change in anxietye
0.81 [0.73; 0.89]
\0.001
Change in pain bothersomenessf
Interaction between care setting and change in pain bothersomeness
0.27 [0.17; 0.43]
2.48 [1.41; 4.34]
\0.001
0.002
Constant
1.02 [0.49; 2.15]
0.951
Bold values indicate significant level of p \ 0.05
a
Reference value
b
Patient recruitment in the Danish physiotherapy setting compared with the GP setting (reference value)
c
Tampa Scale of Kiniesophobia
d
Coping strategy questionnaire (catastrophisation subscale)
e
Hospital Anxiety and Depression Scale
f
Bothersome question
appropriately explains the differences between cohorts in
SBT performance requires further discussion.
One potential reason for this difference in risk prediction
could have been that the psychosocial subscale that classifies people as high-risk, rather than medium-risk, was not
as precise in the Danish population. This could have been
123
an influence, as the concurrent validation study of the
Danish translation showed that, compared to the original
UK cohort, the discriminative ability of three of the psychosocial subscale questions was less strong [6]. On
average, the association between a ‘yes’ response on a
psychosocial subscale question and scores on their
Eur Spine J
Table 3 Discriminative ability of total baseline scores on the STarT Back Tool to correctly classify people with high scores on three different
dichotomised outcomes at 3 months follow-up
People with a Roland Morris disability questionnaire score[30 at
3 months
Danish primary care cohort AUC
[95 % CI]
UK primary care cohort AUC
[95 % CI]
0.71 [0.66; 0.77]
0.81 [0.78; 0.84]
People with severe back pain at 3 months (8–10 on a 0–10 scale)
0.79 [0.68; 0.89]
0.81 [0.78; 0.84]
People with ‘very’ or ‘extremely’ bothersome pain at 3 months
0.70 [0.64; 0.76]
0.72 [0.68; 0.76]
AUC area under the curve statistic from receiver operating characteristic (ROC) curves
respective full reference standard questionnaires had an
AUC 0.115 less than in the UK cohort. These differences
were believed to be due to disparity in severity between the
cohorts (the Danish cohort in that study being more severe
and chronic) and the Danish reference standard questionnaires not having been validated [6]. However, imprecision
in the psychosocial subscale, and/or cultural differences in
the influence of psychosocial factors on outcome, cannot be
ruled out as explanatory influences on the predictive ability
(unadjusted analysis) of the Danish high-risk SBT
subgroup.
Another reason for the difference in (unadjusted) risk
prediction observed in the current study for the high-risk
subgroup could have been the difference between the
cohorts in exposure to physiotherapy treatment (approximately 60 % of the Danish cohort and approximately 18 %
of the UK cohort). Exposure to physiotherapy was a confounder, as patients in the physiotherapy group had a
substantially lower risk of poor outcome than in the GP
group (OR 0.31 [0.13; 0.71]) and that effect was the same
across SBT subgroups because there was no significant
interaction between care setting and SBT groups.
A further reason for the difference in (unadjusted) risk
prediction could have been due to differential treatment
effectiveness at modifying psychosocial risk factors, in
either the GP or physiotherapy care settings. For example,
the physiotherapy treatment was not targeted to SBT subgroup and was likely to be heterogeneous. There was evidence to support this effect mediation, as change in two
psychosocial constructs (anxiety and pain bothersomeness)
were significant in the adjusted models, and there was a
significant interaction between change in pain bothersomeness and care setting. These data do not allow the
distinction between treatment effects and change due to
natural history, but the interaction between change in pain
bothersomeness and care setting is suggestive of a differential treatment effect.
There is some evidence that such a differential effect is
explanatory of a weakening of the SBT predictive ability in
unadjusted analysis. Secondary analysis of data from the
UK randomised controlled trial [3] comparing the
predictive ability of the SBT groups in the targeted treatment group to that in the usual care group showed that the
predictive ability was reduced by effective treatment
(unpublished data). Put simply, when treatment is effective,
the predictive ability of the SBT is reduced due to the
natural history of the condition being modified.
Direct comparison of these results with those of other
translations is not possible, as this degree of validation has
not been published for other versions. However, our results
are comparable to those found in a USA validation study
using the English language version [27].
Strengths of this study are the rigour of the validation
method, the ability to compare results across both cultures,
and the Danish cohort consisting of most professions
commonly consulted for back pain. A weakness of this
study was the inability to perform adjusted analyses in the
UK cohort. An additional consideration is that as the
cohorts were different at baseline, some of the differences
in SBT predictive validity might be due to factors other
than change in the psychosocial characteristics. We did not
explore these, as the only substantive differences between
the cohorts were in the predictive ability of the high-risk
(psychosocial subscale) subgroups and those differences
were explained by adjusting for psychosocial change.
However, differences between countries/settings are to be
expected and are an appropriate reason for testing the SBT
in different cohorts to determine whether the predictive
ability of the SBT is robust to these differences.
Conclusion
In conclusion, based on previous results from the concurrent validation study [6] and these current results on predictive ability, the SBT is suitable as a triage tool of LBP
patients in Danish primary care. The Danish SBT distinguished between low- and medium-risk subgroups with a
similar predictive ability to the UK SBT. That distinction is
useful information for informing patients about their
expected prognosis and may help guiding clinicians’
choice of treatment. However, cross-cultural differences in
123
Eur Spine J
the SBT psychosocial subscale may reduce the predictive
ability of the high-risk subgroup in Danish primary care.
Whether SBT subgroup-matched treatment pathways are as
effective in the Danish population as in the UK requires
subsequent research.
Acknowledgments The authors thank the Danish Quality Unit of
General Practice, the GPs and physiotherapists for collecting data,
and the research secretaries at the Spine Centre of Southern Denmark
and NIKKB for assistance in handling that data. The authors are also
grateful for funding received by the Region of Southern Denmark and
the University of Southern Denmark.
Conflict of interest
None.
Appendix 1. Formation of the Danish primary care
cohort
No data exists on how
many were invited to
participate in the study.
271 LBP1 patients were
registered in the GP2 audit
No data exists on how
many physiotherapy
patients were invited to
participate in the study
248 (91%) of the registered
patients recruited into the
study and completed
baseline questionnaires
200 patients recruited into
the study and completed
baseline questionnaires
204 patients completed and
returned their 3-months
follow-up questionnaires
(72% follow-up rate)
172 patients completed and
returned their 3-months
follow-up questionnaires
(86% follow-up rate)
172 patients were
randomly selected to match
to the number of people in
the physiotherapy sample
The overall cohort from Danish primary care:
172 LBP patients from GP practices
172 LBP patients from Physiotherapy practices
Total n=344
1
2
LBP = low back pain.
General practitioner.
References
1. Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE et al
(2008) A primary care back pain screening tool: identifying
patient subgroups for initial treatment. Arthr Rheum 59(5):
632–641
2. Hill JC, Foster NE, Hay EM (2010) Cognitive behavioural therapy shown to be an effective and low cost treatment for subacute
and chronic low-back pain, improving pain and disability scores
in a pragmatic RCT. Evid Based Med 15(4):118–119
123
3. Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE
et al (2011) Comparison of stratified primary care management
for low back pain with current best practice (STarT Back): a
randomised controlled trial. Lancet 378(9802):1560–1571
4. Beaton DE, Bombardier C, Guillemin F, Ferraz MB (2000)
Guidelines for the process of cross-cultural adaptation of selfreport measures. Spine (Phila Pa 1976) 25(24):3186–3191
5. Bullinger M, Alonso J, Apolone G, Leplege A, Sullivan M,
Wood-Dauphinee S et al (1998) Translating health status questionnaires and evaluating their quality: the IQOLA project
approach. International quality of life assessment. J Clin Epidemiol 51(11):913–923
6. Morso L, Albert H, Kent P, Manniche C, Hill J (2011) Translation and discriminative validation of the STarT Back Screening
Tool into Danish. Eur Spine J 20(12):2166–2173
7. Terwee CB, Bot SD, de Boer MR, van der Windt DA, Knol DL,
Dekker J et al (2007) Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol 60(1):34–42
8. Linton SJ, Boersma K (2003) Early identification of patients at
risk of developing a persistent back problem: the predictive
validity of the Orebro Musculoskeletal Pain Questionnaire. Clin J
Pain 19(2):80–86
9. Pepe MS, Janes H, Longton G, Leisenring W, Newcomb P (2004)
Limitations of the odds ratio in gauging the performance of a
diagnostic, prognostic, or screening marker. Am J Epidemiol
159(9):882–890
10. Justice AC, Covinsky KE, Berlin JA (1999) Assessing the generalizability of prognostic information. Ann Intern Med 130(6):
515–524
11. de Vet HCW, Terwee CB, Mokkink LB, Knol DL (2011) Measurement in medicine, 1st edn. Cambridge University Press, New
York, pp 150–201
12. Altman DG, Vergouwe Y, Royston P, Moons KG (2009) Prognosis and prognostic research: validating a prognostic model.
BMJ 28(338):b605
13. Foster NE, Bishop A, Thomas E, Main C, Horne R, Weinman J
et al (2008) Illness perceptions of low back pain patients in primary care: what are they, do they change and are they associated
with outcome? Pain 136(1–2):177–187
14. van Tulder M, Becker A, Bekkering T, Breen A, del Real MT,
Hutchinson A et al (2006) Chapter 3. European guidelines for the
management of acute nonspecific low back pain in primary care.
Eur Spine J 15(Suppl 2):S169–S191
15. Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW,
Oostendorp RA (2003) Psychometric properties of the Tampa
Scale for kinesiophobia and the fear-avoidance beliefs questionnaire in acute low back pain. Man Ther 8(1):29–36
16. Rosenstiel AK, Keefe FJ (1983) The use of coping strategies in
chronic low back pain patients: relationship to patient characteristics and current adjustment. Pain 17(1):33–44
17. Zigmond AS, Snaith RP (1983) The hospital anxiety and
depression scale. Acta Psychiatr Scand 67(6):361–370
18. Pengel LH, Herbert RD, Maher CG, Refshauge KM (2003) Acute
low back pain: systematic review of its prognosis. BMJ
327(7410):323
19. Cleland JA, Fritz JM, Brennan GP (2008) Predictive validity of
initial fear avoidance beliefs in patients with low back pain
receiving physical therapy: is the FABQ a useful screening tool
for identifying patients at risk for a poor recovery? Eur Spine J
17(1):70–79
20. Childs JD, Fritz JM, Flynn TW, Irrgang JJ, Johnson KK, Majkowski GR et al (2004) A clinical prediction rule to identify
patients with low back pain most likely to benefit from spinal
manipulation: a validation study. Ann Intern Med 141(12):
920–928
Eur Spine J
21. Deyo RA, Battie M, Beurskens AJ, Bombardier C, Croft P, Koes
B et al (1998) Outcome measures for low back pain research. A
proposal for standardized use. Spine (Phila Pa 1976) 23(18):
2003–2013
22. Roland M, Morris R (1983) A study of the natural history of back
pain. Part I: development of a reliable and sensitive measure of
disability in low-back pain. Spine (Phila Pa 1976) 8(2):141–144
23. Kent P, Lauridsen HH (2011) Managing missing scores on the
Roland Morris disability questionnaire. Spine (Phila Pa 1976)
36(22):1878–1884
24. Kirkwood BRSJAC (1988) Measurement error: assessment and
implications, Essential Medical Statistics, 2nd edn. Blackwell
Science Ltd., Oxford, pp 429–446
25. Kraemer HC, Stice E, Kazdin A, Offord D, Kupfer D (2001) How
do risk factors work together? Mediators, moderators, and independent, overlapping, and proxy risk factors. Am J Psychiatry
158(6):848–856
26. Kraemer HC, Wilson GT, Fairburn CG, Agras WS (2002)
Mediators and moderators of treatment effects in randomized
clinical trials. Arch Gen Psychiatry 59(10):877–883
27. Fritz JM, Beneciuk JM, George SZ (2011) Relationship between
categorization with the STarT Back Screening Tool and prognosis for people receiving physical therapy for low back pain.
Phys Ther 91(5):722–732
123
Manuscript 3: Is the psychosocial profile of people with low back pain seeking care in Danish primary care different from those in secondary care? Manual Therapy xxx (2012) 1e6
Contents lists available at SciVerse ScienceDirect
Manual Therapy
journal homepage: www.elsevier.com/math
Original article
Is the psychosocial profile of people with low back pain seeking care in Danish
primary care different from those in secondary care?
Lars Morsø*, Peter Kent, Hanne B. Albert, Claus Manniche
Research Department, Spine Centre of Southern Denmark, Hospital Lillebaelt, Middelfart, Institute of Regional Health Services Research, University of Southern Denmark,
Clinical Locomotion Science Network, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 10 May 2012
Received in revised form
3 July 2012
Accepted 6 July 2012
Differences between the psychosocial risk factors of low back pain (LBP) patients in primary and
secondary care are under-investigated. Similarly, differences in the psychosocial profile of people
classified into STarT Back Screening Tool (SBT) subgroups in primary and secondary care settings have
not been investigated. The aim of the study was to determine: (1) if movement-related fear,
catastrophisation, anxiety and/or depression in LBP patients are different between primary and
secondary care settings, and (2) if those differences are retained when stratified by SBT subgroup. This
study was a cross-sectional comparison of LBP patients in Danish primary settings (405 general practitioner or physiotherapy patients) and a secondary care setting (311 outpatient spine centre patients).
Psychosocial factors were measured with the Roland Morris Disability Questionnaire, the Tampa Scale of
Kinesiophobia, the Coping Strategies Questionnaire (catastrophisation subscale), and the Hospital
Anxiety and Depression Scale. There were significantly higher scores in secondary care for movementrelated fear (1.3 points (95%CI .1e2.5) p ¼ .030) and catastrophisation (2.0 (95%CI 1.0e3.0) p < .000),
lower scores on anxiety (1.0 (95%CI 1.0e2.0) p < .000) but no difference for depression. These
differences in psychosocial scores were broadly retained when stratified by SBT subgroup. However,
questionnaire-specific reported thresholds for important difference scores indicate the size of these
differences between the care settings were unlikely to be clinically important from a patient perspective.
Longitudinal studies are required to investigate the predictive ability of SBT in secondary care settings
and whether treatment targeted to SBT subgroups is effective in secondary care.
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
STarT Back Screening Tool
Primary care
Secondary care
Psychosocial profile
1. Background
Identification of best practice management for low back pain
(LBP) is challenged by the heterogeneity of this condition and the
difficulty in reaching a definitive diagnosis of a tissue-specific cause
for pain in most individual patients (Deyo, 2002; Delitto, 2005;
Fourney et al., 2011). International guidelines recommend triaging
LBP patients into three broad groups: specific LBP (e.g. ankylosing
spondylitis, cancer, and infection), radiculopathy and nonspecific
LBP (NSLBP) (Quebec Task Force, 1987; van Tulder et al., 2006;
Airaksinen et al., 2006; Krismer and van, 2007; Liddle et al., 2007;
Laerum et al., 2007; Dagenais et al., 2010). The group of patients
classified with NSLBP represents approximately 75e85% of all
primary care patients with LBP (Deyo, 2002). There is considerable
diversity in the clinical characteristics of patients within this NSLBP
* Corresponding author. Research Department, Spine Centre of Southern
Denmark, Oestre Hougvej 55, DK-5500 Middelfart, Denmark. Tel.: þ45 63484582.
E-mail address: [email protected] (L. Morsø).
group and there is evidence that this diversity influences outcome
(Dunn et al., 2011; Hill and Fritz, 2011).
Guidelines also recommend assessment of the psychosocial
characteristics of individual LBP patients, as these have been
shown to be risk factors for poor outcome regardless of treatment
type (Burton et al., 1995; Linton et al., 2000; Vlaeyen and Linton,
2000; van Tulder et al., 2006; Chou et al., 2007; Krismer and van,
2007; Main et al., 2010; Hill and Fritz, 2011). However,
measuring psychosocial factors in the daily clinic can present
practical challenges (Kent et al., 2009; Hill et al., 2010b) as many of
the questionnaires validated for detecting psychosocial characteristics are comprehensive and time-consuming (Hill et al.,
2010a).
There is an increasing focus on the ability of questionnaires to
screen LBP patients for psychosocial risk factors in ways that are
practical in daily care settings. Examples of such screening questionnaires are the Örebro Musculoskeletal Pain Questionnaire
(Linton and Hallden, 1998; Linton and Boersma, 2003) and the
STarT Back Screening Tool (SBT) (Hill et al., 2008). The SBT is
particularly interesting, for in addition to its prognostic ability, this
1356-689X/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.math.2012.07.002
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
2
L. Morsø et al. / Manual Therapy xxx (2012) 1e6
classification tool also has treatment implications and is very short.
It was developed in primary care and is a nine-item multidimensional screening questionnaire that classifies NSLBP patients into
subgroups of people with low, medium and high risk of poor
prognosis based on the complexity of their clinical profile (Hill
et al., 2008). That complexity is based on the physical and
psychosocial characteristics of individual patients. In primary care
in the United Kingdom, the SBT classification has been shown to be
predictive of outcome and there is evidence that treatment
matched to these subgroups can be more clinically effective and
more cost-effective than usual care (Hill et al., 2011).
However, as the SBT has only been validated in primary care
settings, it is not known whether it might also be useful in
secondary care settings. This is an important consideration as some
authors have suggested that there might be differences in patients’
clinical profiles across care settings (Fritz et al., 2011; Kongsted
et al., 2011) and there is a perception that secondary care patients
have a more complex psychosocial profile (Pincus et al., 2002). If
this were so, then in a generic sense, for prognostic stratification to
be useful in secondary care it may need to include different
components and similarly, subgroup-stratified treatment could
require more complex or intense interventions. In the specific case
of the SBT, its classification and predictive ability might be different
depending on the care setting.
In Denmark, patients with recent onset of NSLBP are mostly
managed in primary care setting by General Practitioners, Physiotherapists or Chiropractors. Danish clinical practice guidelines
recommend that NSLBP patients are not referred to specialised
secondary care facilities unless an initial course of primary care
treatment has been undertaken (Poulsen, 1999). Consequently,
patients in Danish secondary care tend to have an extended current
pain episode and persistent pain.
Although the influence of psychosocial factors on patient
outcomes has been comprehensively investigated in both acute
and chronic LBP (Burton et al., 1995; Vlaeyen and Linton, 2000;
Pincus et al., 2002; Grotle et al., 2006; Shaw et al., 2009; Hill and
Fritz, 2011), differences in the psychosocial profile of LBP patients
in different care settings remains under-investigated. For
example, it is possible that LBP patients in secondary care might
score higher on psychosocial risk factors due to influences such as
lengthy episode duration, more persistent pain, non-response to
primary care interventions or a different case-mix; however this
has not been quantified. Describing such differences would be
useful if it provided a better understanding of which clinical
characteristics shape the outcomes achievable in each care sector
or it identified characteristics suitable for targeting in each
setting.
Therefore, in this study we aimed to test the hypotheses that: (1)
there would be significant differences between the movementrelated fear, catastrophisation, anxiety and/or depression of LBP
patients in primary and secondary care settings, and (2) there
would be significant differences between care settings on those
psychosocial factors when stratified by the SBT into low, medium,
high risk classification subgroups.
The primary care cohort (n ¼ 405) was collected from February
to October 2011 at general practitioner (GP) clinics and physiotherapy clinics in the region of Southern Denmark. Questionnaires
were given to 271 patients consulting the GP, and consent was
given by 205 patients (a 75.6% participation rate) who then
completed the questionnaires. Twenty-seven physiotherapy clinics
were each sent 10 questionnaire packs to hand out to NSLBP
patients seeking care at the physiotherapy clinic, and consent was
given by the 200 patients included in this study. No further information about non-responders in primary care is available. The
secondary care cohort was collected from May to September 2010
at the Spine Centre of Southern Denmark, which is an outpatient,
multi-disciplinary department in a public hospital. In that period,
513 patients were eligible to participate. As 311 consented to the
study, the participation rate was 60.6%. The only information
available on the 202 non-responders was their age and gender.
Non-responders were younger (mean age 48.3) than responders
(mean age 53.3), and less likely to be women (non-responders
49.5% women, responders 59.5% women). This study was approved
by the Scientific Ethics Committee of the Region of Southern
Denmark (S-20100036).
2.1. Data collection
The patient self-reported questionnaire pack contained demographic information, Numerical Pain Rating scales (Manniche et al.,
1994), the SBT (Hill et al., 2008), the 23-item version of the Roland
Morris Disability Questionnaire (RMDQ) (Roland and Morris, 1983;
Albert et al., 2003), the Tampa Scale of Kinesiophobia (TSK)
(Swinkels-Meewisse et al., 2003), the Coping Strategies Questionnaire (CSQ) (Rosenstiel and Keefe, 1983), and the Hospital Anxiety
and Depression Scale (HADS) (Zigmond and Snaith, 1983). Patients
received the questionnaire pack from secretaries in the clinics, who
also provided study information and obtained informed consent.
All data were double-entered independently into a database (EpiData 3.1, The EpiData Association Odense, Denmark. www.epidata.
dk) by two research assistants.
2.2. Sample size
This study is a secondary analysis of data collected for other
studies in primary and secondary care. Data from the primary
cohort was principally collected for investigating the predictive and
external validity of the SBT in Danish primary care (unpublished
study). The secondary cohort was principally collected for the
translation and discriminative validation of the SBT into Danish
(Morso et al., 2011).
A post-hoc sample size calculation was performed based on the
available numbers of patients and the standard deviation of the
four psychosocial constructs (TSK, CSQ, HADSanx, HADSdep) across
care settings, to calculate the smallest amount of difference
detectable on each construct. Given these samples and their variance, we had sufficient power to detect a significant difference on
the TSK of 1.8; CSQ 1.8; HADSanx .8; HADSdep .8, with a confidence
level of 95% and a power of .80.
2. Methods and material
2.3. Data analysis
This study is a cross-sectional comparison of the psychosocial
profile of patients seeking care for LBP in Danish primary and
secondary care settings. Patients with an episode of NSLBP (leg
pain) who could read and understand Danish were invited to
participate in the study at their first consultation. Participants were
asked for their consent and completed a set of questionnaires
before their clinical examination. In all settings, participation in the
study did not affect usual clinical assessment and care.
Across the collected variables, there was a mean of 3.1% missing
data (range 1.7%e9.8%). As it is recommended that when using
regression models (Schafer, 1997) missing data should be less than
2%, multiple imputation (van Ginkel and van der Ark, 2012) was
conducted. The data distribution was inspected visually and tested
for skewness and kurtosis using the methods recommended by
Garson (Garson, 2009).
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
L. Morsø et al. / Manual Therapy xxx (2012) 1e6
Descriptive statistics (means and SDs, medians and interquartile ranges (IQR) and confidence intervals for differences)
were used to report the cohort summary characteristics. To test for
differences between the primary and secondary care cohorts, we
used unpaired t-tests, ManneWhitney U, Chi square or two-sample
Exact Tests of proportions, depending on the distribution and type
of data.
Linear regression was used to determine the contribution of care
setting (primary care, secondary care) to the variance in scores on
each psychosocial construct (TSK, CSQ, HADSanx, HADSdep) when
controlling for clinically important co-variates (age, gender,
episodic duration, back pain, leg pain, activity limitation). Where
non-normally distributed, data were log- or square root-transformed to approximate normality, prior to inclusion in regression
models. Modelling that entered all variables simultaneously and
forward-stepwise regression modelling (p ¼ .5 to enter and p ¼ .10
to remove) were both performed.
The following software packages were used for the statistical
procedures. SPSS 13.0 (SPSS Statistics/IBM, Chicago IL, USA) was
used for imputation, tests of normality, transformation of data,
unpaired t-tests, ManneWhitney U, Chi square and linear
regressions. STATA 11 (STATA corp., 4905 Lakeway Drive, College
Station, Texas. 77845 USA) was used for post-hoc sample size
calculations, two-sample Exact Tests of proportions (prtesti) and
the calculation of confidence intervals of median difference
(npshift).
3
3. Results
3.1. Differences between primary and secondary care settings
Overall there was no significant difference in the distribution of
the three SBT subgroups in primary and secondary care (p ¼ .246),
though there was a slightly higher proportion of medium risk in
primary care (42% vs. 32.8%). An unexpectedly high proportion of ‘low
risk’ was found among patients referred to secondary care (39.2%).
As seen in Table 1, when comparing the primary and secondary
care cohorts, there were more women than men in both cohorts but
no significant difference between settings. However, the primary
care cohort was significantly younger, more were employed, had
a shorter duration of their current LBP episode and had more
intense back pain, while leg pain intensity was higher in secondary
care. Unexpectedly, both cohorts had near identical levels of
activity limitation.
On the psychosocial constructs, there were significantly higher
scores in secondary care for movement-related fear and catastrophisation. In contrast, there were lower scores on anxiety in
secondary care, but no significant differences for depression. The
difference between care settings for movement-related fear was 1.3
(.1e2.5) and for catastrophisation was 2.0 (1.0e3.0). The scores for
anxiety showed a difference of 1.0 (1.0e2.0) but also showed that
anxiety was low in absolute terms in both settings, with median
values of 4.0 and 5.0 on the 21-point scale.
Table 1
Participant characteristics of the cohorts from primarya and secondarya care.
Female
Missing
Age mean (SD years)
Missing
Employed, proportionb
Missing
Duration of episode, weeksc
0e2
2e4
4e12
>12
Missing
Back pain (scale 0e10)
Median [iqr]
Missing
Leg pain (scale 0e10)
Median [iqr]
Missing
STarT Back Tool group
Low risk
Medium risk
High risk
Missing
Activity limitationd,e (scale 0e100)
Median [iqr]
Fear of movementd,e (scale 17e68)
Mean (SD)
Catastrophisationd,e (scale 0e36)
Median [iqr]
Anxietyd,e (scale 0e21)
Median [iqr]
Depressiond,e (scale 0e21)
Median [iqr]
Primary care (n ¼ 405)
Secondary care (n ¼ 311)
Test for differences between care settingsf
228 (56.3%)
6
49.0 (14.0)
0
265 (65.4%)
4
185 (59.5%)
2
53.3 (15.7)
4
149 (47.9%)
24
p ¼ .466
119 (29.4%)
58 (14.3%)
75 (18.5%)
143 (35.3%)
10
12 (3.9%)
9 (2.9%)
43 (13.8%)
187 (60.1%)
60
7.0 [5e8]
12
6.0 [4e8]
6
3.0 [0e6]
14
4.0 [1e7]
9
140 (34.6%)
170 (42.0%)
95 (23.5%)
0
122 (39.2%)
102 (32.8%)
86 (27.7%)
1
60.8 [39e74]
60.9 [39e78]
p ¼ .678
35.8 (7.6)
37.1 (8.5)
p [ .030
10.0 [5e15]
12.0 [7e18]
p < .000
5.0 [3e8]
4.0 [1e6]
p < .000
2.0 [1e5]
2.0 [0e5]
p ¼ .446
p < .000
p < .000
p < .000
p < .000
p < .000
p ¼ .055
p < .000
p [ .002
p [ .017
p ¼ .246
iqr ¼ interquartile range.
a
Primary care cohort from general practice and physiotherapist clinics, secondary cohort from a hospital outpatient department.
b
Employed or being student.
c
Current episode.
d
RMDQ ¼ Roland Morris Disability Questionnaire, TSK ¼ Tampa Scale of Kinesophobia, CSQ ¼ Coping Strategy Questionnaire (catastrophisation subscale) HADS ¼ Hospital
Anxiety and Depression Scale.
e
Imputed data used for analysis.
f
T-test, ManneWhitney U or Chi square used, depending on the data and distribution.
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
4
L. Morsø et al. / Manual Therapy xxx (2012) 1e6
The results of the standard linear regression models for each
psychosocial construct are shown in Table 2. In each of those
models, one of the psychosocial constructs was the dependent
variable and the following variables were all entered simultaneously as independent predictor variables: care setting, age,
gender, employment status, pain intensity, activity limitation and
duration of complaint. Overall, 13%e31% of the total variance of
each psychosocial construct was explained by the entered
variables. In these models, the beta-coefficients for care-setting
were only statistically significant for catastrophisation (b ¼ .104,
p ¼ .006) and anxiety (b ¼ .101, p ¼ .020). When regression models
were repeated using a forward stepwise procedure, for the models
of depression and movement-related fear, the variable of care
setting was not selected for inclusion at any stage in the model
building. For the stepwise modelling model for anxiety, care setting
was entered in the 4th of 5 stages, and for the modelling of catastrophisation it was entered in the 6th of 7 stages (results not
shown). These stepwise results collectively indicate that care
setting was not a dominant influence on the way people scored
these psychosocial constructs.
3.2. Differences between primary and secondary care settings when
stratified by STarT back screening tool subgroups
The statistically significant differences in psychosocial scores
that were observed between care settings were broadly retained
when stratified by SBT subgroup (Fig. 1). In SBT subgroups 2 and 3,
there was more movement-related fear and catastrophisation in
secondary care. In contrast, in all SBT subgroups there was less
anxiety in secondary care.
In both care settings, there was a trend towards increasing
psychosocial scores on all four psychosocial constructs across SBT
subgroups from low-risk to high-risk, indicating that the stratification in the primary care setting where SBT has previously been
validated, was mirrored in the secondary care setting.
4. Discussion
This study tested the hypotheses that: (1) there would be
significant differences between the movement-related fear, catastrophisation, anxiety and/or depression of LBP patients in primary
and secondary care settings, and (2) there would be significant
differences between care settings on those psychosocial factors
when stratified by SBT classification subgroup (low, medium, high
risk subgroups). The results support both hypotheses, showing
small but statistically significant differences between care settings
on three of the four psychosocial constructs that we tested, and
those differences were retained when stratified by SBT subgroup.
However, although these differences were statistically significant,
we do not consider them to be clinically important in size. That is
because a score of 4 TSK points or more has been reported by Woby
et al. (Woby et al., 2005) as most accurately indicating a clinically
important difference in fear of movement and so the 1.3 TSK points
difference between our care settings was unlikely to be clinically
important in size. Similarly, 1.7 CSQ catastrophising subscale points
and 1.4 HADS anxiety points were reported by Angst et al. (Angst
et al., 2008) as minimal clinically important differences (MCID),
and our differences were 2.0 (catastrophisation) and 1.0 (anxiety).
But they note that their distribution-based method for estimating
MCID is ‘much more conservative’ than anchor-based methods that
ask patients to rate their perceived change (Angst et al., 2008).
Therefore in our view, the threshold values reported in those
studies suggest that the differences between the care settings seen
in our study were unlikely to be clinically important from a patient
perspective.
In the regression models, the variables of care setting, age,
gender, pain intensity, activity limitation and duration of pain
episode collectively explained a limited amount (13%e31%) of the
variance in the included psychosocial measures. In half the
standard regression models, care setting was not a statistically
significant contributor, and the infrequent and late-stage inclusion of care setting in step-wise regression models also indicate
that the contribution of care setting to the psychosocial scores was
minor.
The distribution across the three SBT subgroups in primary care
was very similar to that reported by Hill et al. (Hill et al., 2008) in
a primary care study in the UK (low risk 40%, medium risk 35% and
high risk 25%). This reinforces the findings from the Danish SBT
translation study, where the construct validity of the Danish
version of the SBT was found to be in concordance with the original
SBT (Morso et al., 2011). Stratification by SBT subgroup showed
increasingly higher scores across SBT subgroups on each psychosocial construct. As this pattern was consistent across care settings,
this also reinforces the construct validity of the SBT. This may
Table 2
The influence of ‘care setting’ on each psychosocial construct, controlling for the covariates: age, gender, employment status,a duration,b back pain,c leg pain,c disability,d using
multiple linear regression.
Fear of movementd
Care settinge
Age
Gender
Employment statusa
Durationb
Back painc
Leg painc
Activity limitationd
Total variance explained
by all variables
Catastrophisationd
Anxietyd
Depressiond
Beta-coefficientf (CI 95%)
p-valueg
Beta-coefficientf
(CI 95%)
p-valueg
Beta-coefficientf
p-valueg
Beta-coefficientf
p-valueg
.036
.066
.116
.114
.080
.052
.057
.452
25%
.357
.001
.089
.005
.045
.201
.157
.000
.104
.048
.118
.152
.135
.161
.109
.311
31%
.006
.165
.002
.000
.001
.000
.005
.000
.101
.031
.135
.039
.094
.067
.079
.241
13%
.020
.438
.002
.376
.032
.129
.076
.000
.062
.014
.091
.032
.116
.003
.004
.415
19%
.141
.716
.026
.445
.006
.941
.920
.000
(2.6
(2.0
(2.7
(2.7
(2.6
(5.4
(2.0
(2.6
to
to
to
to
to
to
to
to
2.6)
1.9)
2.4)
2.5)
2.4)
5.5)
2.1)
1.7)
(1.0 to 1.2)
(.2 to .1)
(1.0 to 1.1)
(1.3 to 1.0)
(1.1 to .8)
(6.3 to 6.0)
(.1 to .3)
(.6 to .0)
(.1 to .3)
(.1 to .1)
(.2 to .2)
(.3 to .2)
(.3 to .1)
(1.3 to 1.2)
(.0 to .1)
(.3 to .2)
(.2 to .3)
(.1 to .08)
(.02 to .2)
(.3 to .2)
(.3 to .1)
(1.4 to 1.4)
(.03 to .05)
(.05 to .3)
a
Proportion employed, defined as either being employed or a student.
Duration of low back pain episode in weeks.
c
Numeric range scale (0e10 scale).
d
Roland Morris Disability Questionnaire (0e100 proportional scale), Tampa Scale of Kinesophobia (17e68 scale), Coping Strategy Questionnaire (0e36 scale of the catastrophisation subscale), Hospital Anxiety and Depression Scale (0e21 scale).
e
Primary care cohort from general practice and physiotherapist clinics. Secondary care cohort from a specialised outpatient spine centre.
f
Standardised Beta coefficient.
g
Significance level p < .05.
b
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
L. Morsø et al. / Manual Therapy xxx (2012) 1e6
a
5
b
Movement-related Fear
Catastrophisation
67
57
30
CSQ score
TSK score
35
47
Primary care
Secondary care
37
27
25
20
Primary care
Secondary care
15
10
5
17
low risk
medium risk
0
high risk
low risk
STarT group
c
d
Anxiety
20
15
Primary Care
Secondary care
10
5
HADS score
HADS score
high risk
Depression
20
0
medium risk
STarT group
15
Primary care
Secondary care
10
5
0
low risk
medium risk
high risk
low risk
medium risk
high risk
STarT group
STarT group
Fig. 1. Differences between care settings1 on psychosocial constructs, stratified by STarT Back Tool classification groups2. 1) Primary care cohort from general practice and physiotherapist clinics, secondary cohort from a specialized outpatient spine centre. 2) STarT Back Tool classification group: low risk, medium risk, high risk. 3) TSK ¼ Tampa Scale of
Kinesophobia, CSQ ¼ Coping Strategy Questionnaire (catastrophisation subscale) HADS ¼ Hospital Anxiety and Depression Scale. *Significant differences between care settings
determined using T-tests or ManneWhitney U tests depending on the data distribution (p ¼ .05).
indicate that the SBT can accurately classify people in the secondary
care setting, though the prognostic and therapeutic implications of
such classification are untested.
Although patients referred to Danish secondary care usually
have received an initial course of primary care treatment, a large
proportion of those referred to our secondary care setting were SBT
‘low risk’ patients. This seems to be a contradiction, as people
classified into the SBT low-risk category are expected to recover
well in primary care. There could be many reasons for this. One
reason could be that some low-risk patients fail to improve in
primary care due to iatrogenic factors, such as inadequate reassurance and information on self-management or over-treating.
Another reason could be that the SBT does not detect important
clinical characteristics in patients that are associated with persistent LBP in secondary care For example, family or work-related
factors or other characteristics from the full biopsychosocial spectrum might also be important in this phase of an episode.
In this study, there were small average differences between care
settings on specific psychosocial factors, albeit that they were
unlikely to be clinically important in size. We could speculate that
these differences might reflect aspects of the transition from subacute to persistent pain. For example, it could be that anxiety is
a more prevalent adaptive response in relatively recent-onset pain
but this transition to catastrophisation or fear of movement as the
pain persists over time despite reassurance and treatment. Investigation of this would require detailed longitudinal tracking of
inception cohorts.
The use of the SBT as a way to target treatment in primary care
has shown promising results (Hill et al., 2011), in addition to its use
as a prognostic classification tool. Targeting treatment to specific
subgroups of primary care patients based on their SBT classification
resulted in better patient outcomes and was more cost effective
than usual physiotherapy care in a recent randomised controlled
trial (Hill et al., 2011). Maybe psychologically-informed physiotherapy treatment targeted to ‘high risk’ patients in primary care
would reduce the number of referrals required to secondary care.
Possibly this treatment approach has a role for ‘high risk’ patients in
secondary care or alternatively, it could be that more intensive
psychological support is required in this setting. However, investigation of the predictive ability and effectiveness of treatment
targeted to SBT subgroups in secondary care was beyond the scope
of our study as only cross-sectional data were available.
A strength of the current study is that data from 711 patients
were available from most types of care settings that are common
for back pain patients in our health region. However, the notable
exception is chiropractic patients. The data collection also covered
GPs and physiotherapists across a broad geographic area within our
health region and included its major spine centre. Therefore, the
results of the study are likely to have good generalisability within
that region and maybe more broadly within Denmark. Their generalisability internationally would require testing in further studies.
Potential weaknesses of the data are that although recruitment was
designed to be sequential consenting patients, no data were
available as to the adherence to this practice in the primary care
recruitment, and that minimal information is available about nonparticipant bias.
5. Conclusion
LBP patients in secondary care had movement-related fear and
catastrophisation scores that were higher and anxiety scores that
were lower than those in primary care, at a statistically significant
level. When stratified by SBT risk subgroup, patients in secondary
care in the medium risk and high risk subgroups had significantly
higher movement-related fear, catastrophisation and depression,
but in the low and medium risk groups had lower anxiety scores
than those in primary care. Though statistically significant, the size
of these differences was unlikely to be clinically important from
a patient perspective.
Stratification by SBT subgroup showed increasingly higher
scores on each psychosocial construct across groups and this
reinforces the construct validity of the SBT in both care settings.
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
6
L. Morsø et al. / Manual Therapy xxx (2012) 1e6
Longitudinal studies are required to investigate the predictive
ability of the SBT in secondary care settings and whether treatment
targeted to SBT sub-groups is also effective in secondary care.
Acknowledgements
The authors thank Lene Ververs, Jytte Johannesen and Orla Lund
Nielsen for assistance in handling the questionnaires. We would
also like to thank Lise Hestbæk and Alice Kongsted for assistance
with collecting data from the GPs and to physiotherapists from
primary care in the Region of Southern Denmark for collecting data
in their clinics.
References
Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-Moffett J, Kovacs F, et al.
Chapter 4. European guidelines for the management of chronic nonspecific low
back pain. European Spine Journal 2006;15(Suppl. 2):S192e300. Available
from: PM:16550448.
Albert HB, Jensen AM, Dahl D, Rasmussen MN. Criteria validation of the Roland
Morris questionnaire. A Danish translation of the international scale for the
assessment of functional level in patients with low back pain and sciatica.
Ugeskrift for Laeger 2003;165(18):1875e80. Available from: PM:12772398.
Angst F, Verra ML, Lehmann S, Aeschlimann A. Responsiveness of five conditionspecific and generic outcome assessment instruments for chronic pain. BMC
Medical Research Methodology 2008;8:26. Available from: PM:18439285.
Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of outcome in
acute and subchronic low back trouble. Spine (Phila Pa 1976) 1995;20(6):
722e8. Available from: PM:7604349.
Chou R, Qaseem A, Snow V, Casey D, Cross Jr JT, Shekelle P, et al. Diagnosis and
treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Annals of Internal
Medicine 2007;147(7):478e91. Available from: PM:17909209.
Dagenais S, Tricco AC, Haldeman S. Synthesis of recommendations for the assessment and management of low back pain from recent clinical practice guidelines. Spine Journal 2010;10(6):514e29. Available from: PM:20494814.
Delitto A. Research in low back pain: time to stop seeking the elusive “magic bullet”.
Physical Therapy 2005;85(3):206e8. Available from: PM:15733045.
Deyo RA. Diagnostic evaluation of LBP: reaching a specific diagnosis is often
impossible. Archives of Internal Medicine 2002;162(13):1444e7. Available
from: PM:12090877.
Dunn KM, Jordan KP, Croft PR. Contributions of prognostic factors for poor outcome
in primary care low back pain patients. European Journal of Pain 2011;15(3):
313e9. Available from: PM:20728385.
Fourney DR, Andersson G, Arnold PM, Dettori J, Cahana A, Fehlings MG, et al.
Chronic low back pain: a heterogeneous condition with challenges for an
evidence-based approach. Spine (Phila Pa 1976) 2011;36(21 Suppl.):S1e9.
Available from: PM:21952181.
Fritz JM, Beneciuk JM, George SZ. Relationship between categorization with the
STarT Back Screening Tool and prognosis for people receiving physical therapy
for low back pain. Physical Therapy 2011;91(5):722e32. Available from: PM:
21451094.
Garson D. Testing for assumptions: Statnotes, from North Carolina State University,
Public Administration Program; 26-2-2009.
van Ginkel JR, van der Ark L. SPSS syntax for missing value imputation in test and
questionnaire data. Department of Methodology; 2012.
Grotle M, Vollestad NK, Brox JI. Clinical course and impact of fear-avoidance beliefs
in low back pain: prospective cohort study of acute and chronic low back pain:
II. Spine (Phila Pa 1976) 2006;31(9):1038e46. Available from: PM:16641782.
Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE, et al. A primary care back
pain screening tool: identifying patient subgroups for initial treatment.
Arthritis and Rheumatism 2008;59(5):632e41.
Hill JC, Dunn KM, Main CJ, Hay EM. Subgrouping low back pain: a comparison of the
STarT Back Tool with the Orebro Musculoskeletal Pain Screening Questionnaire.
European Journal of Pain 2010a;14(1):83e9.
Hill JC, Fritz JM. Psychosocial influences on low back pain, disability, and response
to treatment. Physical Therapy 2011;91(5):712e21. Available from: PM:
21451093.
Hill JC, Vohora K, Dunn KM, Main CJ, Hay EM. Comparing the STarT back screening
tool’s subgroup allocation of individual patients with that of independent
clinical experts. Clinical Journal of Pain 2010b;26(9):783e7. Available from: PM:
20842014.
Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of
stratified primary care management for low back pain with current best
practice (STarT Back): a randomised controlled trial. Lancet 2011;378(9802):
1560e71. Available from: PM:21963002.
Kent PM, Keating JL, Taylor NF. Primary care clinicians use variable methods to
assess acute nonspecific low back pain and usually focus on impairments.
Manual Therapy 2009;14(1):88e100. Available from: PM:18316237.
Kongsted A, Johannesen E, Leboeuf-Yde C. Feasibility of the STarT back screening
tool in chiropractic clinics: a cross-sectional study of patients with low back
pain. Chiropractic & Manual Therapies 2011;19:10. Available from: PM:
21526986.
Krismer M, van TM. Strategies for prevention and management of musculoskeletal
conditions. Low back pain (non-specific). Best Practice & Research Clinical
Rheumatology 2007;21(1):77e91. Available from: PM:17350545.
Laerum E, Storheim K, Brox JI. New clinical guidelines for low back pain. Tidsskrift
for den Norske Laegeforening 2007;127(20):2706. Available from: PM:
17952159.
Liddle SD, Gracey JH, Baxter GD. Advice for the management of low back pain:
a systematic review of randomised controlled trials. Manual Therapy 2007;
12(4):310e27. Available from: PM:17395522.
Linton SJ, Boersma K. Early identification of patients at risk of developing a persistent back problem: the predictive validity of the Orebro Musculoskeletal Pain
Questionnaire. Clinical Journal of Pain 2003;19(2):80e6. Available from: PM:
12616177.
Linton SJ, Buer N, Vlaeyen J, Hellsing AL. Are fear-avoidance beliefs related to the
inception of an episode of back pain? A prospective study. Psychology & Health
2000;14(6):1051e9. Available from: PM:22175261.
Linton SJ, Hallden K. Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. Clinical
Journal of Pain 1998;14(3):209e15. Available from: PM:9758070.
Main CJ, Foster N, Buchbinder R. How important are back pain beliefs and expectations for satisfactory recovery from back pain? Best Practice & Research
Clinical Rheumatology 2010;24(2):205e17. Available from: PM:20227642.
Manniche C, Asmussen K, Lauritsen B, Vinterberg H, Kreiner S, Jordan A. Low back
pain rating scale: validation of a tool for assessment of low back pain. Pain
1994;57(3):317e26. Available from: PM:7936710.
Morso L, Albert H, Kent P, Manniche C, Hill J. Translation and discriminative validation of the STarT Back Screening Tool into Danish. European Spine Journal
2011. Available from: PM:21769444.
Pincus T, Burton AK, Vogel S, Field AP. A systematic review of psychological factors
as predictors of chronicity/disability in prospective cohorts of low back pain.
Spine (Phila Pa 1976) 2002;27(5):E109e20. Available from: PM:11880847.
Poulsen PB. Low-back pain. Frequency, management and prevention from an HTA
perspective, vol. 1(1). Danish Commitee for Health Education; 1999. Danish
Institute for Health Technology Assessment 1999 1(1). Finn Børlum Kristensen,
Mogens Hørder, and Leiv Bakketeig.
Available from: PM:2961086 Quebec Task Force. Scientific approach to the assessment and management of activity-related spinal disorders. A monograph for
clinicians. Report of the Quebec Task Force on Spinal Disorders. Spine (Phila Pa
1976) 1987;12(7 Suppl.):S1e59.
Roland M, Morris R. A study of the natural history of back pain. Part I: development
of a reliable and sensitive measure of disability in low-back pain. Spine (Phila Pa
1976) 1983;8(2):141e4. Available from: PM:6222486.
Rosenstiel AK, Keefe FJ. The use of coping strategies in chronic low back pain
patients: relationship to patient characteristics and current adjustment. Pain
1983;17(1):33e44. Available from: PM:6226916.
Schafer JL. Analysis of incomplete multivariate data. London: Chapman & Hall; 1997.
Shaw WS, van der Windt DA, Main CJ, Loisel P, Linton SJ. Early patient screening and
intervention to address individual-level occupational factors (“blue flags”) in
back disability. Journal of Occupational Rehabilitation 2009;19(1):64e80.
Available from: PM:19082875.
Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW, Oostendorp RA.
Psychometric properties of the Tampa Scale for kinesiophobia and the fearavoidance beliefs questionnaire in acute low back pain. Manual Therapy
2003;8(1):29e36.
van Tulder M, Becker A, Bekkering T, Breen A, del Real MT, Hutchinson A, et al.
Chapter 3. European guidelines for the management of acute nonspecific low
back pain in primary care. European Spine Journal 2006;15(Suppl. 2):S169e91.
Available from: PM:16550447.
Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. Pain 2000;85(3):317e32. Available from: PM:
10781906.
Woby SR, Roach NK, Urmston M, Watson PJ. Psychometric properties of the TSK-11:
a shortened version of the Tampa Scale for Kinesiophobia. Pain 2005;117(1e2):
137e44. Available from: PM:16055269.
Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatrica
Scandinavica 1983;67(6):361e70. Available from: PM:6880820.
Please cite this article in press as: Morsø L, et al., Is the psychosocial profile of people with low back pain seeking care in Danish primary care
different from those in secondary care?, Manual Therapy (2012), http://dx.doi.org/10.1016/j.math.2012.07.002
Manuscript 4: The predictive ability of the STarT Back Screening Tool in a Danish secondary care setting The predictive ability of the STarT Back Screening tool in a Danish secondary care setting.
Lars Morsø1, Peter Kent1, Hanne B Albert1, Claus Manniche1
1
Research Department, Spine Centre of Southern Denmark, Hospital Lillebaelt, Middelfart, Institute
of Regional Health Services Research, University of Southern Denmark, a member of the Clinical
Locomotion Network.
Corresponding author:
Lars Morsø, Research Department, Spine Centre of Southern Denmark, Oestre Hougvej 55 DK5500 Middelfart, phone: +4563484582, e-mail: [email protected]
1
Abstract
Introduction
The predictive ability of the Start Back Tool (SBT) in secondary care settings has not been
investigated. The aim of this study was to compare the predictive ability of the SBT in Danish
secondary and primary care settings.
Methods
Poor clinical outcome at 6 months (>30 points on a 0-100 Roland Morris Disability Scale) was
calculated in secondary care (n=960) and primary care (n=172) cohorts. The cohorts were stratified
into SBT subgroups and estimates of additional risk for poor outcome were calculated (relative risk
[RR], unadjusted and adjusted odds ratios). The discriminative ability was determined using the
area under the curve (AUC) statistic.
Results
In secondary care 69.0% and in primary care 40.2% had poor outcome on activity limitation.
Although significant, the predictive ability of the SBT in secondary care (medium risk RR=1.5,
high risk RR=1.7) was not as strong as in primary care (medium risk RR=2.3, high risk RR=3.5).
Adjusting for episode duration and pain intensity only changed the predictive ability marginally in
secondary care. The discriminative ability of the SBT was similar in both cohorts despite
differences in the predictive ability.
Conclusion
The SBT had less predictive ability in a Danish secondary care setting compared to a Danish
primary care setting for persistent activity limitation at 6 months follow-up. Investigation of SBT
targeted treatment implications in secondary care were not investigated in this study.
Keywords: STarT Back Tool, Predictive ability, secondary care, targeted treatment
2
Introduction
The STarT Back Tool (SBT) is a screening tool for nonspecific low back pain (LBP) that has been
validated in primary care [1]. On the basis of potentially modifiable prognostic factors, the SBT
classifies people into prognostic subgroups and identifies targeted treatment pathways for those
subgroups [1]. The construct, concurrent and predictive validity of the SBT have been investigated
[2-4], it has been translated into several languages [5-7] and its cross-cultural validity and crosscultural predictive ability have been described [7, 8]. In addition, a high quality randomized
controlled trial that matched treatment pathways to each SBT subgroup showed improved patient
outcomes and cost effectiveness [9]. Overall this evidence suggests that the SBT can provide
important prognostic information in primary care settings [8], changes in SBT overall scores may
provide important clinical decision-making information for treatment monitoring [2], SBT targeted
treatment can be effective [10], and the SBT is well accepted by primary care clinicians. In addition,
research is being undertaken into appropriate implementation strategies for the SBT in primary care
[11].
The SBT was developed for primary care and consequently the majority of research regarding the
SBT has been performed in primary care settings. However, evidence as to whether the SBT is
appropriate for use in secondary care settings has not been established. In general, it is
recommended that external validation of questionnaires across care settings should be undertaken
due to differences in patient mix [12, 13], and this might be particularly true for LBP patients in
secondary care. Definitions of secondary care vary across countries and in the context of the Danish
health system, secondary care is defined as government-funded specialised care requiring a referral.
LBP patients referred to Danish secondary care have longer duration, greater pain intensity and
higher frequency of referred leg pain compared to those in primary care [14]. Predictably, only
patients with more complex back problems and poorer prognosis are referred to secondary care.
3
The two central considerations about the usefulness of the SBT in secondary care are: (i) its
predictive ability (prognostic accuracy), and (ii) its ability to indicate appropriate subgroup-targeted
treatment pathways. A number of studies have tested the predictive ability in primary care [1, 2, 8,
15], but no previous studies have investigated and reported the predictive ability of SBT in
secondary care. Therefore, the aim of the current study was to extend previous investigations of the
SBT by comparing the predictive ability of the SBT in a Danish secondary care setting and a
Danish primary care setting.
Materials and Methods
This study was conducted at a specialised, multidisciplinary, secondary care setting - the Medical
Department of the Spine Centre of Southern Denmark. Patients are referred to the Spine Centre by
GPs, chiropractors or medical specialists for a comprehensive evaluation due to suboptimal
improvement during assessment in primary care. While most patients are evaluated at the Spine
Centre and referred back to primary care for further treatment, some have a very brief course of
treatment at the Spine Centre and some are referred for surgical evaluation.
The secondary care data were self-reported by patients via electronic questionnaires using touch
screen computers at the time of their first consultation at the Spine Centre. The electronic
questionnaires were part of the SpineData database, which is a comprehensive registry of all
patients attending the Medical Department. Prospective data were available for 960 consecutive low
back pain patients with baseline and 6-months follow-up questionnaires from the period January
2012 to November 2012. The only inclusion criteria for secondary care patients participating in the
study were full electronic completion of the SBT at baseline. No diagnostic data were available in
4
the database but using Magnetic Resonance Imaging to document the presence of lumbar pathoanatomic findings, previous descriptive research on this clinical population showed that
approximately 0.5% or less have serious pathology (tumour, fracture, tuberculosis), 15% have
central stenosis, 29% have nerve root compromise and the remainder have non-specific LBP [16]
The results from this secondary cohort were compared to those from an existing primary care
physiotherapy cohort that had been collected for testing the predictive ability of SBT in Danish
primary care [15]. That study combined data from GP and physiotherapy practices with outcome
measured at 3-months but in the current study, only the physiotherapy data were used, as 6-month
follow-up data were only available for that sample. Details of the recruitment criteria and data
collection used in that study have previously been reported [15]. Briefly, data were collected from
May to September 2011 at twenty-seven Danish physiotherapy clinics. Baseline data were available
for 172 patients and for 83% (n=144) at 6 months follow-up. The variables used in the current study
were extracted to match those available in the secondary care sample.
Data from the physiotherapy setting had been entered into a database (Epidata 3.1, The EpiData
Association, Odense, Denmark) by a research secretary, while data from secondary care was
entered directly into the SpineData database by the patients themselves. This study was approved by
the Scientific Ethics Committee of the Region of Southern Denmark (S-20100036) and all patients
gave written informed consent for research use of their data.
Data measurement
All data were collected in identical ways in both cohorts. Age (years) and gender (female, male)
was extracted from each patient’s unique social security (CPR) number. Duration of the current
5
pain episode was calculated from the date of first consultation and the patient self-reported onset
date of the current episode. Patients also self-reported: numbers of days off work during the last 3
months, number of previous LBP episodes, the SBT (9-item version), activity limitation (Danish
23-item version of the Roland Morris Disability Questionnaire [RMDQ]) [17], low back pain
intensity and leg pain intensity (0 to 10 Numerical Rating Scales). The outcome measures collected
at 6 months follow-up were: low back pain intensity, leg pain intensity and activity limitation.
Data analysis
Descriptive analysis (means and standard deviations, medians and inter-quartile ranges) of the
baseline characteristics of both cohorts were tabulated at the level of the total samples and also
stratified by SBT subgroup. Baseline differences between the three SBT subgroups were examined
using Mann-Whitney U, Chi-square or Kruskal Wallis Tests, depending on the data type and
distribution.
Previous studies have tested the predictive validity of health questionnaires using a variety of
methods [18, 19]. The current study mirrored the three statistical methods used in the original
development study of the English language version of the SBT [1], which were also those used in
the validation study of the SBT in Danish primary care [15].
The proportion of patients with a poor clinical outcome at 6 months was calculated, stratified by
SBT subgroup. Poor clinical outcome was defined as persistent activity limitation measured by the
Roland Morris Disability Questionnaire (RMDQ) [20] score at 6-month follow-up. The cut point
used in original SBT development study in the UK was 7 points on a 0-24 scale [1] but as we used
the proportional recalculation method to convert all RMDQ scores to a 0-100 scale, that threshold
6
was recalculated to be 30 points or more. The proportional recalculation method has been shown to
more accurately manage any missing RMDQ answers [21]
The same threshold was used to estimate the additional risk (relative risk) [22] for poor outcome for
people classified into the medium or high SBT risk subgroup compared to the low risk subgroup.
Differences between the risk groups within each cohort were tested using Chi-square test for 2x2
tables.
The area under the curve (AUC) statistic from Receiver Operating Characteristic (ROC) curves was
used to describe the ability of the baseline SBT sum scores (0-9 scale) to discriminate (sensitivity/1specificity) [22] between people with and people without the following outcomes at 6 months
follow-up: (i) activity limitation as defined above, and (ii) LBP intensity still being ‘severe’ (8-10
on a 0-10 point scale). These criteria were those used in the original UK validation study [1].
In addition to those three statistical approaches, odds ratios for poor outcome on activity limitation
for SBT subgroups were also calculated in unadjusted and adjusted form using logistic regression.
This was performed to explore whether the predictive ability of the SBT in secondary care was
confounded by baseline different between the cohorts. All covariates were initially entered into the
model, followed by a manual backwards stepwise reduction (p<.05 to remove) to the most
parsimonious model. An odds ratio greater that one in these regression models means that particular
clinical characteristic increases the odds of having a poor outcome and an odds ratio less than one
means that it is protective against a poor outcome.
7
As it eventuated that the relative risk estimates were lower in our secondary care data than in
primary care and in order to have a predictive reference standard to compare those results to, posthoc we also calculated relative risks of poor outcome on activity limitation using baseline pain
intensity or baseline activity limitation as the predictor. The predictor of baseline pain intensity was
formed by creating three categories that each contained 33% of the participants based on the
distribution of the cohort’s scores on the 0 to 10 pain intensity scale. The same distribution-based
method was used for categorising the baseline activity limitation scores on the 0 to 100 RMDQ
scale to create a three category predictor variable.
The relative risk estimates were calculated using Microsoft Excel 2003 (Microsoft Corp, Redmond,
WA, USA) and logistic regression was performed using STATA 12 (StataCorp, College Station,
TX, USA). All other statistical analyses were conducted using PASW 13.0 (IBM Inc., Somers, NY,
USA).
Results
Baseline differences between the secondary and primary care cohorts
The two cohorts were significantly different at baseline on duration of episode (p<.001), leg pain
intensity (p<.001) and borderline significant on LBP intensity (p=.059). As seen in Table 1, within
each cohort there were reassuringly significant differences between SBT subgroups, with increased
LBP intensity, leg pain intensity and activity limitation across the low risk to high risk SBT
subgroups. At baseline the level of activity limitation was highest in the high risk subgroup in both
care settings with median RMDQ score of 78.3(IQR65-87) in secondary care and 77.8(IQR70-84)
in primary care (0 to 100 scale).
8
TABLE 1 ABOUT HERE
Six month outcome differences between the secondary and primary care cohorts
At 6 months follow-up there were differences between cohorts in LBP intensity, leg pain intensity
and activity limitation (p<.001) (Table 2). The higher values for LBP intensity, leg pain intensity
and activity limitation in secondary care were also retained when stratified by SBT subgroup.
TABLE 2 ABOUT HERE
Unadjusted risk of poor clinical outcome on activity limitation at 6 months
At a cohort level, 69.0% in secondary care and 40.2% in primary care had a poor outcome on
activity limitation (Table 2) 6 months after their index consultation. When stratified, the proportion
of those increased from low risk to high risk SBT subgroup within each cohort but with some
distinct differences between the cohorts. Most notable was the large difference in patients with poor
outcome on activity limitation in the low risk subgroup (47.8% in the secondary care cohort, and
20.0% in the primary care cohort) with almost half of the patients in the secondary cohort still
having an RMDQ score above 30 points. That pattern of a larger proportion in secondary care
having a poor outcome was also retained across the other subgroups.
Another important observation was that the gradient of relative risk across the three SBT subgroups
was not nearly as steep in secondary care as in the primary care (Figure 1). Though still
significantly predictive of additional risk of poor outcome in the medium risk (RR=1.5[95%CI 1.3;
1.7]) and the high risk group (RR=1.7[1.5; 2.0]) these unadjusted results indicate that the predictive
9
ability for the SBT subgroups for 6 month outcome is not as strong in secondary care as it was in
primary care (RR medium risk 2.3 [CI95% 1.2; 4.5], high risk 3.5 [CI95% 1.8; 6.6].
It is likely that these two findings of: (i) nearly half the low risk subgroup in secondary care having
a poor outcome and (ii) the reduced predictive ability of the SBT subgroups in secondary care; are
inter-related, as the low risk subgroup is the reference category for the predictive ability. As it was
possible that this relationship was also confounded by the difference in baseline episode duration
and pain intensity between the cohorts, an adjusted analysis was also performed.
FIGURE 1 ABOUT HERE
Unadjusted and adjusted odds of poor clinical outcome on activity limitation at 6 months
The unadjusted odds ratios (OR) shown Table 3 reflect the difference between the cohorts already
reported in the relative risk results. The predictive ability of the medium risk subgroup across
cohorts was not markedly different (secondary care OR 2.7[1.9; 3.9], primary care 3.5 [1.4; 8.9]),
whereas the difference was more distinct in the high risk groups (secondary care OR 4.8 [3.3; 6.8],
primary cohort 9.0 [3.0; 27.6]).
TABLE 3 ABOUT HERE
However, adjustment for baseline duration of episode and pain intensity resulted in only marginally
reduced ORs in the medium risk and high risk subgroups. Episode duration made statistically
significant contributions to the models in both cohorts and baseline LBP intensity to the model in
secondary care. In some cases those changes increased the ORs by 10% but as they occurred in both
10
cohorts, they did not account for the reduced predictive ability of the SBT subgroups in secondary
care. There were no statistically significant interactions between SBT subgroups and episode
duration and this was also reflected in the correlation between SBT total scores and episode
duration being very weak (-.005 in secondary care and .037 in primary care). There was also no
significant interaction between pain intensity and SBT subgroups. Therefore, for predicting
persistent activity limitation at 6 months, baseline episode duration was predictive in both cohorts
and had an influence that was independent of the predictive ability of the SBT subgroups. The same
was true for baseline LBP intensity in secondary care.
To gain a sense of what the predictive ability of other reference standard predictors would be in
secondary care, post-hoc analyses were performed using the three-category distribution-based
predictors of baseline pain intensity and activity limitation. The RR for baseline pain intensity were
medium risk 1.5[1.3; 1.7], high risk 1.6[1.4; 1.8] and for activity limitation were medium risk
1.6[1.4; 1.8], high risk 1.8[1.6; 2.1]. These were nearly identical to those obtained when using the
SBT subgroups as predictors.
The ability of the baseline SBT total scores to identify people with outcomes above a clinical
threshold at 6 months follow-up
The AUC statistics describing the ability of baseline SBT total scores (0-9 scale) to discriminate
between people with and people without scores above threshold values on two different 6-month
outcomes are shown in Table 4. For both outcomes; activity limitation ‘still being present’ and,
LBP ‘still being severe’, the discriminative ability was similar across cohorts.
TABLE 4 ABOUT HERE
11
Discussion
The aim of this study was to compare the predictive ability of the SBT in a Danish secondary care
setting and a Danish primary care setting for the outcome of persistent activity limitation at 6
months follow-up. The results indicate that the SBT subgroups were not as strongly predictive of
poor outcome in the Danish secondary care setting but were as predictive as similarly categorised
baseline pain intensity or activity limitation scores.
The results also show very similar proportions of patients across the cohorts having poor activity
limitation at baseline, both at an overall cohort level and also when stratified into SBT subgroups.
However, at 6 months follow-up these proportions were quite different, reflecting that the recovery
trajectories were less favourable in secondary care, a finding which is in concordance with earlier
findings [23]. While the large proportion of secondary care patients with poor outcome in the high
risk group was similar to the primary care cohort and to that found in other primary care studies [1],
the 47.8% in the secondary care low risk subgroup and 71.3% in the medium risk subgroup who
had a poor outcome were clearly different from that in primary care [15]
The proportion of patients classified into the SBT low risk subgroup who nonetheless experienced
persistent activity limitation was much larger in secondary care (47.8% of ‘low risk’ patients in
secondary care, 20.0% in primary care). While this higher proportion in secondary care might be
expected, it has the consequence of attenuating the relative risks estimates that were possible,
because this subgroup is the reference category (the denominator in the relative risk formula). This
is seen in the results showing that similarly categorized baseline pain and activity limitation were no
stronger at prediction in this cohort, despite it being well recognised that these are strong predictors
12
[24] and that baseline values are the best predictors for the same outcome [2]. It therefore seems
that prediction in this setting is challenging, perhaps due to a combination of more frequent poor
outcome and a wider variability of outcome relative to baseline presentation.
The results also indicated that the predictive ability of the psychosocial subscale component, which
is the distinction between medium and high risk subgroups, was lower in secondary care. This was
previously noted in an earlier primary care study that compared the predictive ability of the SBT in
the UK and Denmark [15]. In that instance, those differences were explained by changes in the
psychosocial factors during the treatment period, probably due to differences in treatment exposure.
In the current study, confounding may also have occurred due a difference in the management of
psychosocial factors but the available data did not allow statistical adjustment for change in these
factors.
Another explanation for the different predictive ability of the SBT in these primary and secondary
care cohorts could be differences in case mix. Although diagnostic codes were not available in the
data from either setting, SBT was originally validated in people diagnosed by GPs as having nonspecific LBP. In our secondary care setting, approximately 45% have MRI evidence of central
stenosis or nerve root compromise [16] and this may have affected their recovery trajectories and
thereby, the predictive ability of the SBT. Another potential factor affecting the predictive ability
could be a social class bias that we believe results in an over-representation of lower
sociodemographics in the secondary care cohort. In pregnancy-related pelvic pain it has been shown
that sociodemographics are influential on outcome [25]. The SBT does not measure these
characteristics and it may be that for it to have better predictive ability in secondary care, these
factors would need to be included.
13
In the regression models that adjusted for baseline differences, only episode duration and baseline
pain intensity were retained as an independent predictive factor alongside the SBT subgroups in
secondary care. Previous studies have shown that both influence outcome and return to work [26,
27], but our findings indicate that in this context, neither exerted an influence that could explain the
differences between care settings in the SBT predictive ability.
Paradoxically, the results in both cohorts of our AUC analysis show similar discriminative ability of
the SBT 9-item sum scores to correctly classify patients on two dichotomised outcome measures
(persistent activity limitation and severe LBP) at 6 months follow-up, despite differences in the
predictive ability of the SBT subgroup classification. This might be interpreted to indicate that the
predictive ability potentially would improve by changing the SBT cut-points, but such post hoc
analysis revealed that neither changing these cut-points nor using median baseline activity
limitation in secondary care as the outcome criterion or both, more than marginally altered the
predictive ability (results unreported).
Previous studies of primary care in the UK and Denmark indicate that 17% to 24% of people
classified into the low-risk group nonetheless had a poor outcome [15]. Therefore, it is to be
expected that some ‘failed’ low risk patients who do not improve are referred to secondary care.
However, given that almost half of the ‘low risk’ patients in secondary care had a poor outcome,
perhaps we need to reframe the language in this setting so that this subgroup are referred to as ‘low
complexity’ and compared to ‘medium and high complexity’ subgroups.
A strength of this study was the use of a pre-exiting validation model to test the predictive ability of
the SBT classification categories as this allowed the comparison of results across care settings and
two previous studies [1, 15]. Two other primary care studies used different methodological
14
approaches to asses the predictive ability of SBT [2, 8]. In one study, SBT sumscores were used as
a continuous scale in longitudinal modelling of a non-uniform outcome period [8]. In the other
study the SBT sumscores and the outcome measures were used as continuous scales in multiple
linear regression modelling to monitor of change during treatment and avoid the borderline
misclassification of cases [2]. Our study was not designed to monitor change but to investigate the
predictive ability of the baseline SBT classification categories (low, medium and high risk
subgroup) and therefore we mirrored the method used in the original validation studies [1, 15].
A limitation of this study is that it was not designed to investigate the treatment implications of the
SBT. Although the SBT predictive ability was not as strong as in primary care, this was
investigated by us in secondary care where care pathways were uninfluenced by SBT subgroup.
Therefore, it is possible that an ‘SBT-type’ of classification might have clinically useful treatment
implications in secondary care, although such risk-based classification may require including
different constructs. In addition, as secondary care settings and the characteristics of their patients
vary greatly within and between countries, caution should be exercised in generalising these results.
Conclusion
In our multidisciplinary Danish secondary care setting, the SBT classification subgroups were less
able to predict persistent activity limitation at 6 months follow-up than in a Danish physiotherapy
primary care setting. This finding remained even after adjusting for baseline differences in episode
duration and LBP intensity. In both settings, episode duration was a predictive factor that was
independent of SBT subgroup classification, and baseline pain intensity also was in secondary care.
The usefulness of SBT subgroup targeted treatment in secondary care was not investigated in this
study.
15
Reference List
(1) Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE, et al. A primary care back pain
screening tool: identifying patient subgroups for initial treatment. Arthritis Rheum 2008
May 15;59(5):632-41.
(2) Beneciuk JM, Bishop MD, Fritz JM, Robinson ME, Asal NR, Nisenzon AN, et al. The
STarT Back Screening Tool and Individual Psychological Measures: Evaluation of
Prognostic Capabilities for Low Back Pain Clinical Outcomes in Outpatient Physical
Therapy Settings. Phys Ther 2012 Nov 2.
(3) Hill JC, Dunn KM, Main CJ, Hay EM. Subgrouping low back pain: a comparison of the
STarT Back Tool with the Orebro Musculoskeletal Pain Screening Questionnaire. Eur J Pain
2010 Jan;14(1):83-9.
(4) Wideman TH, Hill JC, Main CJ, Lewis M, Sullivan MJ, Hay EM. Comparing the
responsiveness of a brief, multidimensional risk screening tool for back pain to its
unidimensional reference standards: The whole is greater than the sum of its parts. Pain
2012 Nov;153(11):2182-91.
(5) Bruyere O, Demoulin M, Brereton C, Humblet F, Flynn D, Hill JC, et al. Translation
validation of a new back pain screening questionnaire (the STarT Back Screening Tool) in
French. Arch Public Health 2012;70(1):12.
(6) Gusi N, Del Pozo-Cruz B, Olivares PR, Hernandez-Mocholi M, Hill JC. The Spanish
version of the "STarT Back Screening Tool" (SBST) in different subgroups. Aten Primaria
2011 Feb 4.
(7) Morso L, Albert H, Kent P, Manniche C, Hill J. Translation and discriminative validation of
the STarT Back Screening Tool into Danish. Eur Spine J 2011 Jul 19.
(8) Fritz JM, Beneciuk JM, George SZ. Relationship between categorization with the STarT
Back Screening Tool and prognosis for people receiving physical therapy for low back pain.
Phys Ther 2011 May;91(5):722-32.
(9) Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of
stratified primary care management for low back pain with current best practice (STarT
Back): a randomised controlled trial. Lancet 2011 Oct 29;378(9802):1560-71.
(10) Sowden G, Hill JC, Konstantinou K, Khanna M, Main CJ, Salmon P, et al. Targeted
treatment in primary care for low back pain: the treatment system and clinical training
programmes used in the IMPaCT Back study (ISRCTN 55174281). Fam Pract 2012
Feb;29(1):50-62.
(11) Foster NE, Mullis R, Young J, Doyle C, Lewis M, Whitehurst D, et al. IMPaCT Back study
protocol. Implementation of subgrouping for targeted treatment systems for low back pain
patients in primary care: a prospective population-based sequential comparison. BMC
Musculoskelet Disord 2010;11:186.
16
(12) Moons KG, Altman DG, Vergouwe Y, Royston P. Prognosis and prognostic research:
application and impact of prognostic models in clinical practice. BMJ 2009;338:b606.
(13) Henrica C.W.de Vet, Caroline B.Terwee, Lidwine B.Mokkink, Dirk L.Knol. Validity.
Measrement in medicine. First ed. New York: Cambridge Universiry Press; 2011. p. 150201.
(14) Morso L, Kent P, Albert HB, Manniche C. Is the psychosocial profile of people with low
back pain seeking care in Danish primary care different from those in secondary care? Man
Ther 2012 Jul 31.
(15) Morso L, Kent P, Albert HB, Hill JC, Kongsted A, Manniche C. The predictive and external
validity of the STarT Back Tool in Danish primary care. 2013.
(16) Albert HB, Briggs AM, Kent P, Byrhagen A, Hansen C, Kjaergaard K. The prevalence of
MRI-defined spinal pathoanatomies and their association with modic changes in individuals
seeking care for low back pain. Eur Spine J 2011 Aug;20(8):1355-62.
(17) Albert HB, Jensen AM, Dahl D, Rasmussen MN. [Criteria validation of the Roland Morris
questionnaire. A Danish translation of the international scale for the assessment of
functional level in patients with low back pain and sciatica]. Ugeskr Laeger 2003 Apr
28;165(18):1875-80.
(18) Childs JD, Fritz JM, Flynn TW, Irrgang JJ, Johnson KK, Majkowski GR, et al. A clinical
prediction rule to identify patients with low back pain most likely to benefit from spinal
manipulation: a validation study. Ann Intern Med 2004 Dec 21;141(12):920-8.
(19) Cleland JA, Fritz JM, Brennan GP. Predictive validity of initial fear avoidance beliefs in
patients with low back pain receiving physical therapy: is the FABQ a useful screening tool
for identifying patients at risk for a poor recovery? Eur Spine J 2008 Jan;17(1):70-9.
(20) Roland M, Morris R. A study of the natural history of back pain. Part I: development of a
reliable and sensitive measure of disability in low-back pain. Spine (Phila Pa 1976 ) 1983
Mar;8(2):141-4.
(21) Kent P, Lauridsen HH. Managing missing scores on the Roland Morris Disability
Questionnaire. Spine (Phila Pa 1976 ) 2011 Oct 15;36(22):1878-84.
(22) Kirkwood BRSJAC. Measurement error: assessment and implications.
Essential Medical
Statistics. 2nd edition 2003 ed. Oxford: Blackwell Science Ltd.; 1988. p. 429-46.
(23) Grotle M, Vollestad NK, Brox JI. Clinical course and impact of fear-avoidance beliefs in
low back pain: prospective cohort study of acute and chronic low back pain: II. Spine (Phila
Pa 1976 ) 2006 Apr 20;31(9):1038-46.
(24) Chou R, Shekelle P. Will this patient develop persistent disabling low back pain? JAMA
2010 Apr 7;303(13):1295-302.
(25) Albert HB, Godskesen M, Korsholm L, Westergaard JG. Risk factors in developing
pregnancy-related pelvic girdle pain. Acta Obstet Gynecol Scand 2006;85(5):539-44.
17
(26) Grotle M, Foster NE, Dunn KM, Croft P. Are prognostic indicators for poor outcome
different for acute and chronic low back pain consulters in primary care? Pain 2010
Dec;151(3):790-7.
(27) Pengel LH, Herbert RD, Maher CG, Refshauge KM. Acute low back pain: systematic
review of its prognosis. BMJ 2003 Aug 9;327(7410):323.
18
Table 1. Baseline characteristics for the Danish secondary and primary care cohorts.
Secondary care cohort
Total cohort
(n=960)
SBT
Low risk
n= 252
(27.7%)
SBT
Medium risk
n=296
(32.5%)
SBT
High risk
n=363
(39.9%)
Test for
differences
between
SBT
subgroups*
Age in years
• Mean, (SD1)
52.0 (14.1)
52.3 (14.6)
52.0 (13.8)
51.3 (14.0)
p= .673
Female, proportion
521 (54.3%)
116 (46%)
179 (60.5%)
194 (53.4%)
p= .003
Duration in months
• <1 month
• 1 to 3 months
• > 3 months
47 (5.0%)
139 (14.9%)
746 (80.0%)
5 (2.0%)
36 (14.6%)
205 (83.3%)
18 (6.3%)
56 (19.6%)
212 (74.1%)
22 (6.3%)
41 (11.6%)
289 (82.1%)
p=.012
6 (4-7)
4.3 (2-6)
6.0 (5-7)
7.0 (6-8)
p< .001
5 (2-7)
3.0 (1-5)
5.0 (3-6)
6.3 (4-8)
p< .001
60.9 (44-78)
34.8 (22-52)
60.9 (48-74)
78.3 (65-87)
p< .001
838 (88.6%)
158 (64.2%)
281 (95.9%)
354 (98.9%)
SBT
Low risk
n=54
(34.6%)
SBT
Medium risk
n=64
(41.0%)
SBT
High risk
n=38
(24.4%)
Test for
differences
between
SBT
subgroups*
Pain intensity3 (0-10 scale)4
• Low back pain, median, (IQR2)
•
Leg pain, median, (IQR2)
Activity limitation5 (0-100 scale)4
• Median, (IQR1)
•
Proportion of patients >30
Primary care cohort
Total cohort
(n=172)
Age in years
• Mean, (SD1)
52.0 (15.2)
53.3 (14.6)
50.4 (14.6)
49.5 (16.7)
p=.377
Female, proportion
98 (57.0%)
30 (55.6%)
41 (64.1%)
17 (44.7%)
p=.163
65 (38.9%)
39 (23.4%)
63 (37.7%)
18 (34.0%)
17 (32.1%)
18 (34.0%)
29 (47.5%)
13 (21.3%)
19 (31.1%)
13 (34.2%)
7 (18.4%)
18 (47.7%)
p=.246
5.0 (4-7)
4.0 (3-6)
6.0 (4-8)
6.0 (5-7)
p< .001
3.0 (0-6)
1.5 (0-3)
3.0 (1-6)
4 (1-7)
p< .001
60.9 (39-77)
34.8 (26-49)
67.4 (52-78)
77.8 (70-84)
p< .001
146 (86.3%)
36 (66.7%)
60 (93.8%)
38 (100%)
Duration in months, proportion
• <1 month
• 1 to 3 months
• > 3 months
Pain intensity3 (0-10 scale)4
• Low back pain, median, (IQR2)
•
Leg pain, median, (IQR2)
Activity limitation5 (0-100 scale)4
• Median, (IQR1)
•
Proportion of patients >30
1
Standard deviation
Inter-quartile range
3
Numeric Rating Scale (0-10)
2
19
4
High scores are worse
Roland Morris Disability Questionnaire
* Tests for differences between subgroups were ANOVA or Kruskal Wallis procedures, depending on data type and
distribution
5
20
Table 2. Outcome at 6 months follow-up for the Danish secondary and primary care cohorts.
Total cohort
n=960
SBT
Low risk
n=252
(27.7%)
SBT
Medium risk
n=296
(32.5%)
SBT
High risk
n=363
(39.9%)
4.7 (2-6)
3.0 (2-5)
4.3 (2-6)
5.7 (3-7)
2.7 (0-5)
1.3 (0-4)
2.7 (0-5)
3.7 (1-6)
47.8 (28-70)
26.1 (13-48)
47.8 (22-70)
65.2 (35-83)
656 (69.0%)
120 (47.8%)
209 (71.3%)
292 (81.3%)
Total cohort
n=144
SBT
Low risk
n=48
(36.9%)
SBT
Medium risk
n=52
(40.0%)
SBT
High risk
n=30
(23.1%)
3.0 (2-5)
3.0 (2-4)
2.0 (0-4)
4.5 (2-5)
Leg pain, median, (IQR3)
1.3 (0-4)
0.5 (0-3)
2.0 (1-5)
1.0 (0-5)
Activity limitation4 (0-100 scale)2
• Median, (IQR3)
26.0 (9-57)
17.0 (4-26)
30.0 (9-55)
65.0 (25-75)
51 (40.2%)
9 (20.0%)
21 (46.7%)
18 (69.2%)
Secondary care cohort
Pain intensity1 (0-10 scale)2
• Low back pain, median, (IQR3)
•
Leg pain, median, (IQR3)
Activity limitation4 (0-100 scale)2
• Median, (IQR3)
•
Proportion of patients > 305
Primary care cohort
Pain intensity1 (0-10 scale)2
• Low back pain, median, (IQR3)
•
•
Proportion of patients > 305
1
Numeric Rating Scale
High scores are worse
3
Inter-quartile range
4
Roland Morris Disability Questionnaire
5
Patients with a Roland Morris Disability Questionnaire score >30 were classified as having persistent activity
limitation
2
21
Table 3. The odds of having a poor clinical outcome on activity limitation1 at 6 months follow-up in
the Danish secondary care and primary care cohorts, estimated by STarT Back Tool subgroup using
logistic regression.
Secondary care cohort (n=903)
Odds ratio
p-value
[CI 95%]
Unadjusted model
STarT Back Tool low-risk subgroup2
STarT Back Tool medium-risk subgroup
STarT Back Tool high-risk
Constant
Model adjusted episode duration and baseline
low back pain
STarT Back Tool low-risk subgroup2
STarT Back Tool medium-risk subgroup
STarT Back Tool high-risk
Episode duration
0-1 month
1-3 months
> 3 months
Primary care cohort (n=116)
Odds ratio
p-value
[CI 95%]
1.00
2.72 [1.91; 3.87]
4.76 [3.31; 6.84]
.92 [.72; 1.17]
p<.001
p<.001
p<.488
1.00
3.50 [1.37; 8.93]
9.00 [2.97; 27.25]
.25 [.12; .52]
p=.009
p<.001
p<.001
1.00
2.36 [1.61; 3.47]
3.31 [2.18; 5.03]
p<.001
p<.001
1.00
4.68 [1.66; 13.18]
9.13 [2.60; 32.01]
p=.003
p=.001
1.00
1.08 [.52; 2.23]
2.10 [1.10; 4.02]
p=.835
p=.025
1.00
4.38 [1.19; 16.05]
6.45 [2.20; 18.84]
p=.026
p=.001
Low back pain intensity baseline
1.17 [1.09; 1.26]
Constant
.252 [.12; .53]
1
Roland Morris Disability Questionnaire score > 30 (0 to 100 scale)
2
Reference category
22
p<.001
p<.001
Table 4. Discriminative ability of the STarT Back Tool to correctly classify people with high scores
on two different dichotomised outcomes at 6 months follow-up.
Danish secondary care cohort
AUC1 [95%CI]
Danish primary care cohort
AUC1 [95%CI]
.69 [.66; .73]
.73 [.64; .82]
.72 [.68; .77]
.66 [.46; .85]
People with a Roland Morris Disability
Questionnaire score > 30 (0-100 scale) at 6
months
People with severe back pain at 6 months (8-10 on
a 0-10 scale)
1
AUC = the Area Under the Curve statistic from Receiver Operating Characteristic (ROC) curves
23
Figure 1. Relative risk of poor clinical outcome1 on activity limitation at 6 months
by SBT subgroup in the Danish secondary and primary care cohorts.
4
3.5*
3,5
3
2.3*
2,5
2
1,5
1
1.5*
1.7*
1
Low risk
Medium risk
High risk
1
0,5
0
Secondary care
Primary care
1
More than 30 Roland Morris Disability Questionnaire points on a recalculated 0-100 scale
*P<.05
24
PhD thesis -­‐ Appendices Appendix 1: Scoring of the SBT Appendix 2: The validated version of the Danish translated SBT Appendix 1.
Score flow of the SBT
Total-score
score
3 or less
4 or more
Sub-score
Low
complexity
3 or less
4 or more
Medium
complexity
High
complexity
i
Appendix 2.
The validated version of the translated SBT
STarT Spørgeskemaet
Patientens navn: _______________________________
Dato: _____________
Tænk tilbage på de seneste 2 uger og marker dit svar på følgende spørgsmål:
1
I løbet af de seneste 2 uger har mine rygsmerter bredt sig ned i mit/mine ben
2
Jeg har haft smerter i mine skuldre eller nakke i løbet af de seneste 2 uger
3
Jeg har kun gået korte afstande på grund af mine rygsmerter
4
I løbet af de seneste 2 uger har jeg klædt mig langsommere på end normalt på
grund af rygsmerter
5
Det er egentligt ikke sikkert for en person i min tilstand at være fysisk aktiv
6
Jeg har været bekymret meget af tiden
7
Jeg føler mine rygsmerter er forfærdelige og de bliver aldrig bedre
8
Generelt har jeg ikke nydt alle de ting, som jeg plejede at nyde
Nej
Ja
0
1
□
□
□
□
□
□
□
□
Uenig
Enig
0
1
□
□
□
□
□
□
□
□
9. Overordnet set, hvor generende har dine rygsmerter været de seneste 2 uger?
Slet ikke
Lidt
Middel
Meget
Extremt
□
□
□
□
□
0
0
Total score (alle 9): __________________
0
1
1
Sub Score (spr. 5-9):______________
ii