Manual Therapy 2016 (Feb); 21: 120–127 ~ FULL TEXT
Alice Kongsted, Cathrine Hedegaard Andersen,
Martin Mørk Hansen, Lise Hestbaek
Nordic Institute of Chiropractic and Clinical Biomechanics,
Department of Sports Science and Clinical Biomechanics,
University of Southern Denmark,
The clinical course of low back pain (LBP) cannot be accurately predicted by existing prediction tools. Therefore clinicians rely largely on their experience and clinical judgement. The objectives of this study were to investigate
1) which patient characteristics were associated with chiropractors' expectations of outcome from a LBP episode,
2) if clinicians' expectations related to outcome,
3) how accurate clinical predictions were as compared to those of the STarT Back Screening Tool (SBT), and
4) if accuracy was improved by combining clinicians' expectations and the SBT.
Outcomes were measured as LBP intensity (0-10) and disability (RMDQ) after 2-weeks, 3-months, and 12-months. The course of LBP in 859 patients was predicted to be short (54%), prolonged (36%), or chronic (7%). Clinicians' expectations were most strongly associated with education, LBP history, radiating pain, and neurological signs at baseline and related to all outcomes.
The accuracies of predictions made by clinicians (AUC .58-.63) and the STarT Back Screening Tool (SBT) (AUC .50-.61) were comparable and low. No substantial increase in the predictive capability was achieved by combining clinicians' expectations and the SBT. In conclusion, chiropractors' predictions were associated with well-established prognostic factors but not simply a product of these. Chiropractors were able to predict differences in outcome on a group level, but prediction of individual patients' outcomes were inaccurate and not substantially improved by the SBT. It is worth investigating if more accurate tools can be developed to assist clinicians in prediction of outcome.
From the FULL TEXT Article:
A large number of prognostic factors have been identified in
non-specific low back pain (LBP), but these generally have rather
weak associations with outcome (Kent and Keating, 2008), and
most investigated predictive models have not demonstrated
adequate predictive value to be useful in relation to individual
patients (Hayden et al., 2010). Still, clinicians need to make decisions
about treatment plans every day and are repeatedly faced
with patients wanting to know their most likely prognosis.
Lacking convincingly helpful predictive models, clinicians rely
on experience and subjective judgement when establishing a
patient's prognosis. The LBP prognosis established by general
practitioners (GPs) has been consistently associated with outcome
in studies investigating this, but although comparable to that of
validated questionnaires (Jellema et al., 2007), the predictive accuracy
of clinicians' prognostic estimation was quite low (Schiottz-
Christensen et al., 1999; Jellema et al., 2007). Also, chiropractors
were not able to accurately predict poor treatment outcomes when
asked to register “whether they thought patients were less likely
than average to report a good outcome following a course of care”
(Newell et al., 2013). In addition, there is evidence suggesting that
the prediction differs substantially between clinicians (Hill et al.,
In elderly patients with musculoskeletal pain, GPs' prediction of
outcome has been shown to improve if combining their subjective
judgements with just three factors obtained from the patient history
(Mallen et al., 2013). In non-specific LBP, the STarT Back
Screening Tool (SBT) is an easily completed scale that combines
potentially modifiable prognostic factors. It has shown promising
for assisting GPs' decision about the treatment plan (Hill et al.,
2008, 2011), and it may provide a simple tool to improve clinicians'
prediction of outcome.
Clinicians' estimation of prognosis potentially differs from that
of standardised screening if clinicians value other factors than those
typically registered. It is not investigated to what extent clinicians'
estimation of prognosis is based on established prognostic factors,
but in one study factors such as pain intensity, level of disability,
and number of previous episodes were associated with clinicians'
prediction (Perrot et al., 2009). In a consensus process it was recognised
that clinicians found issues such as ‘generally difficult life
circumstances’ of importance for the prognosis and this construct is
most likely not easily captured by screening tools (Hill et al., 2010).
More accurate prognoses are essential to inform patients about
their condition and to guide treatment, for instance by early identification
of patients with more extensive treatment needs. Therefore,
this study aimed to increase the understanding of clinicians'
expectations of patient outcome in LBP.
This study comprised a cross-sectional and a longitudinal part.
The objectives of the cross-sectional part were to determine:
(1) which individual patient characteristics were associated with chiropractors'
expectations of outcome from a LBP episode, and
(2) how closely the chiropractors' expectations could be predicted by a
combination of these patient characteristics.
Since no empirical
data could support the choice of investigated patient characteristics
this part should be considered hypothesis generating.
The objectives of the longitudinal part were to determine:
(1) the association between chiropractors' expectations of clinical course and outcome after 2 weeks, 3 months, and 12 months,
(2) how clinicians' expectations (3-level subjective judgement) performed compared to that of the SBT (3-level standardised tool), and
(3) to what extent combining clinicians' expectations with the SBT increased the amount of variation explained in outcome.
The study is based on a cohort study which has previously been
described (Eirikstoft and Kongsted, 2014), and consisted of patients
visiting one of 40 chiropractors at 17 Danish chiropractic clinics
due to a new episode of LBP. In Denmark, chiropractors belong to
the public primary health care sector, patients can consult without
referral and the expenses are partly covered (approximately 20%)
by national health insurance. The participating clinics were
members of a group of research clinics affiliated with the Nordic
Institute of Chiropractic and Clinical Biomechanics. The chiropractors
attended a one day course introducing study procedures
and a research assistant visited all clinics prior to study start to
repeat general information and ensure that clinical examination
procedures (relevant for the main study) were adequately standardised.
Collection of data on clinicians' expectations was not
introduced from the beginning of patient recruitment and the
present study therefore concerns a slightly smaller cohort than
reported on before. According to the local ethical committee the
study did not require ethical approval. (Danish National
Baseline questionnaires were completed in the reception area
prior to the first consultation. Completed questionnaires were
returned to the receptionist in a closed envelope and were not
available to the chiropractor. There were no specific instructions to
patients whether they could discuss the content of the questionnaire
with the chiropractor. Clinicians obtained a patient history as
they found appropriate and did a standardised clinical examination
(Eirikstoft and Kongsted, 2014). Results from the clinical examination
and the clinician's expectations were registered in a web
based registration form. Treatment was unaffected by study
participation and the chiropractors had no access to questionnaire
data. Follow-up questionnaires were mailed to participants after 2
weeks, 3 months, and 12 months. Participants not responding were
contacted by a research assistant to make sure they had received
the questionnaire. All questionnaires were sent directly to the
research department and patients were informed that their responses
would not be revealed to the chiropractors and only reported
in an anonymous form.
Consulters with non-specific LBP or lumbar nerve root
involvement (based on usual clinical practice for diagnostic triage)
aged 18–65 years who could read Danish and were able to respond
to SMS-questions on a mobile phone (for reasons unrelated to this
part of the study) were potential participants. Patient were not
eligible if pregnant, if acute surgical referral was needed, or if
having had more than one contact to a health care provider due to
LBP within the preceding three months. In addition participants
were excluded from the analyses if the clinician's expectation was
Clinicians' expectations of clinical course
The final question of the clinical examination form was (translated
from Danish) ‘What outcome do you expect for this patient?’
with four response options: 1) Short/uncomplicated course, 2)
Prolonged but without lasting consequences, 3) Long-lasting/
sustained consequences (high risk of chronicity), or 4) Don't
know. The expectation categories are hereafter referred to as short/
uncomplicated, prolonged, and long-lasting/chronic.
Patient-reported baseline variables and their categorization are
listed in Table 1. Body mass index was collected as weight and
height. LBP and leg pain intensity were measured on 0e10 Numeric
Rating Scales (Jensen et al., 1998), and activity limitation on the
Roland Morris disability questionnaire (RMDQ) as a proportional
score 0e100 (Kent and Lauridsen, 2011). Fear avoidance was
measured by the physical activity section of the Fear Avoidance
Beliefs Questionnaire (FABQ) (0e24) (Waddell et al., 1993),
depression by the Major Depression Inventory (MDI) 0e50 (Bech
et al., 2001),, and general health using the health thermometer of
EQ-5D (0e100) (Rabin and de Charro, 2001).
The SBT was scored and categorised as recommended by the
developers of the tool (Hill et al., 2008). Items on aspects that were
not considered covered by other questionnaires (item 2: Shoulder
or neck pain, item 7: catastrophising, and item 9: bothersome pain)
were separately included in the cross-sectional analysis of factors
associated with clinicians' expectations.
The Quebec classification (Spitzer et al., 1987), was registered as
part of the examination form completed by the clinician, and nerve
root involvement and spinal stenosis were combined due to few
LBP intensity (NRS 0e10) and activity limitation (RMDQ 0e100)
at 2-weeks, 3-months, and 12-months follow-up were outcome
measures. The scales were used in their original forms and
dichotomised to define poor outcome as LBP > 0 and RMDQ > 8
(RMDQ > 8 corresponded to >2 on the original 0e24 RMDQ
(Kamper et al., 2010)).
Data analysis [can be omitted without loss of continuity]
Data were double entered into Epidata (Lauritsen, 2008), and
analyses conducted in STATA SE/12.1.
To explore which patient characteristics were related to the
clinicians' expectations of outcome, univariate associations were
first tested by KruskaleWallis rank test (continuous variables) or
Pearson's chi-squared test (dichotomous and categorical). We then
did a multinomial multivariable regression with clinicians' expectations
(4-level categorical) as the dependent variable to investigate
which of the patient characteristics were associated with the
clinicians' expectations of outcome independently of other
measured factors. In this model all measured baseline factors were
introduced simultaneously, and independent variables with p > .2
at all outcome levels were removed manually in a stepwise fashion
without eliminating factors with a risk ratio above 1.5 or below .66.
Next, we calculated what the clinicians' expectation of outcome
would be if this had been simply a product of the patient baseline
characteristics. This was to understand how closely clinicians'
judgements relate to measurable patient characteristics. To do that,
we calculated the probability of belonging to each of the categories
short/uncomplicated, prolonged and long-lasting when based on
the full multinomial model described above. These predictions
were compared to the actual expectations of the clinicians. If the
expectations predicted from patient characteristics perfectly
matched the clinicians' expectations of outcome, the predicted
probability would be 100%, indicating that the clinicians' prediction
was entirely based on the included characteristics. If, on the other
hand, there was no match between predicted outcomes for a given
category and the clinicians' expectations of outcome, the predicted
probability would be 0%.
In the longitudinal part of the study patients in the ‘don't know’
category were excluded since we were interested in the predictive
accuracy of clinicians' expectations only when they were able to
come to a conclusion. Linear and logistic univariate regressions
were performed to test for associations between prediction (clinicians'
as well as prediction by the SBT) and outcome. Potential
dependency between observations from the same clinics was taken
into account using STATA's cluster option for robust variance estimation
with clinics as cluster level.
The discriminative ability calculated as area under the curve
(AUC) was compared between the clinicians' expectations and the
SBT. Positive likelihood ratios (LHþ) for a poor outcome were
calculated for each category of the clinicians' expectations and the
SBT. The positive predictive value (PPV) for the short/uncomplicated
category predicting a good 2-weeks outcome and for longlasting/
chronic predicting a poor 12-months outcome were calculated
to represent the outcome that seemed to correspond best to
the expectation categories provided to the clinicians. Similar PPVs
were calculated for the SBT model. Finally, the amount of explained
variance in continuous outcome measures (adjusted R-squared)
was compared between the models and also used to quantify the
effect of combining clinicians' expectations and the SBT.
For the multivariable analysis missing values on baseline factors
were imputed by multiple imputations based on fully conditional
specifications with five chained iterations without replacing
missing values on outcome measures (Moons et al., 2006). No item
had more than 7% missing values.
Baseline information was available from 890 participants of
which 31 (3%) were excluded from the analyses because there was
no registration of the clinician's expectations. Characteristics of the
study cohort (n = 859) and those excluded appear from Table 1.
Patients with missing information had statistically significant less
leg pain than the study cohort but otherwise no differences were
detected. Follow-up questionnaires after 2 weeks, 3 months and 12
months were available from respectively 83%, 79%, and 74% of
participants (Figure 1). Non-responders at the 12-months follow-up
did not differ from responders regarding baseline LBP intensity,
leg pain intensity, activity limitation, duration of LBP, sick leave, SBT
risk group, or the clinician's expectations. However, nonresponders
were on average 5.5 years younger and a larger proportion
was male (63% vs. 54%), smoker (27% vs. 17%), and reported
heavy physical workload (31% vs. 19%).
Cross-sectional part investigating clinicians' expectations
The chiropractors expected a short/uncomplicated course in 54%
of the cohort, a prolonged course in 36%, a long-lasting/chronic
course in 7%, and did not know what to expect in 3%. Patient
characteristics are compared between these groups in Table 1.
Gender, education, number of previous episodes, LBP last year,
duration of the present episode, leg pain, activity limitation, SBT
items, depression, general health, the Quebec classification, and
whether the patient had previously visited a chiropractor were all
associated with the expected prognosis in the univariate analyses
(p < .05) (Table 1).
A multivariable model demonstrated that independently of
other measured factors, clinicians more often expected a prolonged
or a long-lasting course than a short/uncomplicated in patients
with higher BMI, more previous LBP, long duration of the present
episode, more disability, radiating pain, and neurological signs and
of female sex (Table 2). Having more than five years of education
after finishing public school was associated with the clinicians
predicting a short/uncomplicated course. Clinicians felt more often
unable to predict the clinical course in patients with long duration,
shoulder/neck pain or nerve root involvement whereas a long education,
having light physical workloads, and age between 35 and
45 years reduced the likelihood of the clinician choosing the ‘don't
know’ option (Table 2).
We used information from the baseline questionnaire to calculate
which category (short/uncomplicated, prolonged or longlasting)
the clinicians would be expected to assign each patient to
if that decision could be explained as a product of just the factors
registered in the questionnaire. When averaging across all categories,
the probability that the outcome predicted by the combination
of patient-reported baseline factors matched the clinicians'
expectation of outcome was 62% (95% CI: 60e65%). The outcome
category with the lowest probability for matching between the
prediction from patient-reported baseline factors and the clinicians'
expectation of outcome was for the long-lasting/chronic
group at 18% (95% CI: 8e34%). These results imply that clinicians'
expectations were likely based on other, and possibly more complex,
factors than those registered in the baseline questionnaire.
Associations between chiropractors' expectations and outcome
Clinicians' expectations were significantly associated with all
outcome measures at all follow-up points. Table 3 shows the differences
in mean LBP intensity and RMDQ scores between the
short/uncomplicated group and the other groups (b-values), and
the odds ratios for poor outcome in the prolonged and long-lasting/
chronic groups as compared to the short/uncomplicated group. LBP
intensity and activity limitation in the expectation groups during
the follow-up period are illustrated in Fig. 2.
Clinicians' expectations compared to the STarT Back Tool
The abilities of the clinicians' expectations and the SBT to
discriminate patients with a poor outcome from others are presented
in Tables 4 and 5. AUC values quantify the ability to
discriminate between patients with good and poor outcome and
range from .5 (no better than tossing a coin) to 1 (perfect
discrimination). The discriminative ability of the clinicians' expectations
(AUCs .58 to .63) and the SBT (AUCs .50 to .61) were of
similar magnitudes and generally low. Because the number of patients
in some categories was low some of the estimated likelihood
ratios in Tables 4 and 5 are subject to uncertainty as apparent from
the wide confidence intervals.
In the group expected by the clinicians to have a short/uncomplicated
course, 11% were pain free (LBP = 0) after two weeks
(PPV = .11; 95% CI: 8e15%), and similarly in the group predicted to
be in low risk of poor prognosis by the SBT, 10% (95% CI: 6e14%)
were pain free. The group expected to have long-lasting LBP by
clinicians, 83% (95% CI: 67–93%) actually had a poor 12-months
outcome, whereas this was only true for 60% (95% CI: 44–74%) in
the high risk SBT group.
The clinicians' expectations as well as the prediction of outcome
by SBT explained only little of the variation in the continuous
outcomes, meaning that patient outcomes differed substantially
within each prediction category. Clinicians' expectations and the
SBT combined were slightly better than each of the two by themselves
when predicting activity limitation after 3 and 12 months,
but the proportions explained remained low (Table 6).
Chiropractors expected a short uncomplicated course for about
half of their LBP patients and a severe long-lasting course for less
than 10% when asked to choose one of three prognostic categories.
The clinicians' expected outcomes were associated with a number
of previously identified prognostic factors. Long duration, radiating
pain, and nerve root involvement were factors strongly
associated with clinicians' expectations of a severe long-lasting
course, whereas the registered psychological factors had no independent
association with the expected prognosis. This is in
contrast to the SBT in which identification of the 'high risk'
category is based on psychological factors. Long duration and
nerve root involvement also increased the probability of the
clinician stating that he/she did not know which prognosis to
expect. Despite individual associations with previously identified
prognostic factors, combining the patient-reported baseline characteristics
mathematically in a regression model did not accurately
predict clinicians' expectations, especially when considering the
long-lasting/chronic outcome category. These findings suggest
that the clinicians' may have used other pieces of information to
make their prognostic estimates. The clinical process of establishing
a prognosis ought to be further elucidated in qualitative or
mixed methods studies which may inform future development of
The prognosis expected by the clinicians was significantly
associated with the observed clinical course. The mean outcome in
the expectation groups differed by a sufficient magnitude that we
believe the groups were truly clinically different. However, prediction
on an individual level was not accurate (as discrimination
between patients with good and poor outcome was low), at least
not when the clinical course was defined by the applied outcome
measures. Compared to these outcome measures the expectations
about outcome were generally overoptimistic as observed also in a
recent study investigating GPs' prediction of outcome in musculoskeletal
pain (Mallen et al., 2013). One possible explanation for
the lack of accurate prediction is that the clinicians may have
thought of expected ‘course’ as for instance expected number of
treatments, whereas we tested if their predictions were in line with
recovery from pain and activity limitation. The SBT was not helpful
as a tool for increasing the accuracy of the prediction. That finding
was in line with a previous study that did not find the SBT very
useful as a prognostic tool in chiropractic practice (Field and
Newell, 2012). It seems that the relatively low prognostic ability
of the SBT in this cohort is related to many patients presenting with
short duration of LBP (Morso et al., 2015), and it can be speculated
that psychological factors may not be very influential in chiropractic
patients as indicated by the only prospective study that we
know of investigating this (Leboeuf-Yde et al., 2009), conceivably
because psychological distress appears to be infrequent in
chiropractic patients (Bolton, 1994; Leboeuf-Yde et al., 2009;
Kongsted et al., 2011).
The main limitation of this study was that the measure of clinicians'
expectations was not validated and it is uncertain to what
extent the expectation question could be anticipated to match the
outcome measures. It is possible that clinicians define ‘short and
uncomplicated’ differently than being pain free and likewise it is
unknown what clinicians would define as ‘prolonged’ or ‘longlasting’
since this was not specifically defined in the question. It
may not be problematic that the positive predictive valuewas low if
clinicians' definition of the prognostic categories is different from
what was captured by our outcome measures; as long as they make
sure that their patients have the same understanding when
informed about prognosis. Future studies should use an expectation
question that aligns closely with outcome as demonstrated by
Mallen et al. (2013).
Still, the study provided new insights based on a large study
sample with quite complete data and an acceptable drop-out rate
that we did not suspect to influence conclusions. A further strength
was the relatively large number of participating chiropractors
which increased the generalizability of the results. The number of
patients included by each clinician was not sufficient to investigate
potential individual differences between clinicians.
In summary, chiropractors' expectations of the clinical course
was associated with well-established prognostic factors but was
not simply a product of these. Chiropractors were able to predict
differences in outcome up to one year after the initial visit on a
group level but did not predict individual patients' outcome
precisely although as well as the SBT. Therefore it is worth
investigating if more accurate tools can be developed to assist
clinicians in prediction of outcome. Although subject to uncertainty,
chiropractors identified a group with markedly increased
risk of a poor outcome more precisely than the SBT and the
development of prediction rules may benefit from understanding
better how clinicians predict poor long-term outcome. In short,
chiropractors cannot rely solely on their gut feeling when telling
LBP patients what to expect, but they have an insight that may
help researchers in the development of improved prediction
The authors would like to thank the clinics that took part in the
data collection. The Danish Foundation for Chiropractic Research
and Post-graduate Education funded the study through an unrestricted
grant (Grant No. 01/1624).
Bech P, Rasmussen NA, Olsen LR, Noerholm V, Abildgaard W.
The sensitivity and specificity of the Major Depression Inventory, using the Present State Examination as the index of diagnostic validity.
J Affect Disord 2001;66(2e3): 159e64.
Psychological distress and disability in back pain patients: evidence of sex differences.
J Psychosom Res 1994;38(8):849e58.
Danish National Commitee on biomedical research ethics.
Guidelines about notification.
Eirikstoft H, Kongsted A.
Patient characteristics in low back pain subgroups based on an existing classification system. A descriptive cohort study in chiropractic practice.
Man Ther 2014;19(1):65e71.
Field J, Newell D.
Relationship between STarT Back Screening Tool and prognosis for low back pain patients receiving spinal manipulative therapy.
Chiropr Man Ther 2012;20(1):17.
Hayden JA, Dunn KM, van der Windt DA, Shaw WS.
What is the prognosis of back pain?
Best Pract Res Clin Rheumatol 2010;24(2):167e79.
Hill JC, Dunn KM, Lewis M, et al.
A Primary Care Back Pain Screening Tool: Identifying Patient Subgroups For Initial Treatment
(The STarT Back Screening Tool)
Arthritis Rheum. 2008 (May 15); 59 (5): 632–641
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
Clin J Pain 2010;26(9):783e7.
Hill JC, Whitehurst DG, Lewis M 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): 1560–1571
Jellema P, van der Windt DA, van der Horst HE, Stalman WA, Bouter LM.
Prediction of an unfavourable course of low back pain in general practice: comparison of four instruments.
Br J General Pract: J R Coll General Pract 2007;57(534):15e22.
Jensen MP, Miller L, Fisher LD.
Assessment of pain during medical procedures: a comparison of three scales.
Clin J Pain 1998;14(4):343e9.
Kamper SJ, Maher CG, Herbert RD, Hancock MJ, Hush JM, Smeets RJ.
How little pain and disability do patients with low back pain have to experience to feel that they have recovered?
Eur Spine J 2010;19(9):1495e501.
Kent PM, Keating JL.
Can we predict poor recovery from recent-onset non-specific low back pain? A systematic review.
Man Ther 2008;13(1):12e28.
Kent P, Lauridsen HH.
Managing missing scores on the Roland Morris disability questionnaire.
Spine (Phila Pa 1976) 2011;36(22):1878e84.
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 (Apr 28); 19: 10
EpiData data entry. Data management and basic statistical analysis system.
Odense, Denmark. 2008.
Leboeuf-Yde C, Rosenbaum A, Axen I, Lovgren PW, Jorgensen K, Halasz L, et al.
The Nordic Subpopulation Research Programme
Prediction of Treatment Outcome
in Patients With Low Back Pain Treated By Chiropractors --
Does the Psychological Profile Matter?
Chiropr Osteopat 2009;17:14.
Mallen CD, Thomas E, Belcher J, Rathod T, Croft P, Peat G.
Point-of-care prognosis for common musculoskeletal pain in older adults.
JAMA Intern Med 2013;173(12): 1119e25.
Moons KG, Donders RA, Stijnen T, Harrell Jr FE.
Using the outcome for imputation of missing predictor values was preferred.
J Clin Epidemiol 2006;59(10): 1092e101.
Morso L, Kongsted A, Hestbaek L, Kent P.
The prognostic ability of the STarT Back Tool was affected by episode duration.
Eur Spine J 2015 [Epub ahead of print].
Newell D, Field J, Visnes N.
Prognostic accuracy of clinicians for back, neck and shoulder patients in routine practice.
Chiropr Man Ther 2013;21(1):42.
Perrot S, Allaert FA, Concas V, Laroche F.
“When will I recover?” A national survey on patients' and physicians' expectations concerning the recovery time for acute back pain.
Eur Spine J 2009;18(3):419e29.
Rabin R, de Charro F.
EQ-5D: a measure of health status from the EuroQol Group.
Ann Med 2001;33(5):337e43.
Schiottz-Christensen B, Nielsen GL, Hansen VK, Schodt T, Sorensen HT, Olesen F.
Long-term prognosis of acute low back pain in patients seen in general practice: a 1-year prospective follow-up study.
Fam Pract 1999;16(3):223e32.
Spitzer WO, LeBlanc FE, Dupuis M.
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 1987;12(7 Suppl): S1e59.
Waddell G, Newton M, Henderson I, Somerville D, Main CJ.
A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear- avoidance beliefs in chronic low back
pain and disability.
Return to LOW BACK PAIN
Return to OUTCOME ASSESSMENT
Return to CLINICAL PREDICTION RULE