Physical and psychological factors predict outcome
following whiplash injury
Michele Sterlinga,*, Gwendolen Julla
, Bill Vicenzinoa
, Justin Kenardyb
, Ross Darnellc
a
The Whiplash Research Unit, Department of Physiotherapy, The University of Queensland, Brisbane, Qld 4072, Australia
b
Centre for National Research on Disability and Rehabilitation Medicine, School of Psychology, The University of Queensland, Brisbane, Qld 4072, Australia
c
School of Health and Rehabilitation Sciences, The University of Queensland, Brisbane, Qld 4072, Australia
Received 4 June 2004; received in revised form 23 November 2004; accepted 7 December 2004
Abstract
Predictors of outcome following whiplash injury are limited to socio-demographic and symptomatic factors, which are not readily
amenable to secondary and tertiary intervention. This prospective study investigated the predictive capacity of early measures of physical and
psychological impairment on pain and disability 6 months following whiplash injury. Motor function (ROM; kinaesthetic sense; activity of
the superficial neck flexors (EMG) during cranio-cervical flexion), quantitative sensory testing (pressure, thermal pain thresholds, brachial
plexus provocation test), sympathetic vasoconstrictor responses and psychological distress (GHQ-28, TSK, IES) were measured in 76 acute
whiplash participants. The outcome measure was Neck Disability Index scores at 6 months. Stepwise regression analysis was used to predict
the final NDI score. Logistic regression analyses predicted membership to one of the three groups based on final NDI scores (!8 recovered,
10–28 mild pain and disability, O30 moderate/severe pain and disability). Higher initial NDI score (1.007–1.12), older age (1.03–1.23), cold
hyperalgesia (1.05–1.58), and acute post-traumatic stress (1.03–1.2) predicted membership to the moderate/severe group. Additional
variables associated with higher NDI scores at 6 months on stepwise regression analysis were: ROM loss and diminished sympathetic
reactivity. Higher initial NDI score (1.03–1.28), greater psychological distress (GHQ-28) (1.04–1.28) and decreased ROM (1.03–1.25)
predicted subjects with persistent milder symptoms from those who fully recovered. These results demonstrate that both physical and
psychological factors play a role in recovery or non-recovery from whiplash injury. This may assist in the development of more relevant
treatment methods for acute whiplash.
q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
Keywords: Whiplash injury; Physical and psychological factors; NDI scores; Prediction
1. Introduction
Most individuals recover within a few weeks of whiplash
injury from motor vehicle crashes but a significant
proportion (14–42%) develop persistent pain with 10%
reporting constant severe pain (Barnsley et al., 1994).
The ability to predict which patients will develop chronic
symptoms following whiplash injury is important so that
appropriate early management can be targeted to these
patients. Since the Quebec Task Force’s recommendation in
1995 (Spitzer et al., 1995) that a greater number of
prognostic studies on whiplash were necessary, many
factors such as socio-demographic status, crash-related,
compensation/litigation, psychosocial and physical factors
have been studied for their predictive capacity (Cassidy et
al., 2000; Kasch et al., 2001; Radanov et al., 1995; Schrader
et al., 1996). Despite these investigations, two recent
systematic reviews of prospective cohort studies on
whiplash could agree on only high initial pain intensity as
showing strong evidence for delayed functional recovery
(Cote et al., 2001; Scholten-Peeters et al., 2003).
It is apparent from these systematic reviews that it is still
very unclear which patients are at risk of delayed recovery
0304-3959/$20.00 q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.pain.2004.12.005
Pain 114 (2005) 141–148
www.elsevier.com/locate/pain
* Corresponding author. Tel.: C61 7 3365 4568; fax: C61 7 3365 2775.
E-mail address: m.sterling@shrs.uq.edu.au (M. Sterling).
following whiplash injury. Furthermore, few factors have
been identified that can assist in directing the secondary and
tertiary management of this condition such that the
transition to persistent pain and disability may be averted.
In light of the current biopsychosocial model of
musculoskeletal pain and recent findings of more complex
changes in both the motor and sensory function of persons
with chronic whiplash associated disorders (WAD) (Koelbaek-Johansen
et al., 1999; Nederhand et al., 2002),
investigation of the predictive capacity of a more complete
set of measures reflecting such changes is long overdue.
Chronic WAD have been shown to be associated with motor
dysfunction, generalised sensory hypersensitivity suggestive
of changes in central nociceptive pathways and
psychological distress (Dall’Alba et al., 2001; KoelbaekJohansen
et al., 1999; Radanov et al., 1996). We and others
have previously demonstrated that some of these changes
occur soon after injury and persist into the period of
chronicity in certain patients (Nederhand et al., 2002;
Sterling et al., 2003a,b.) It is not known whether the early
presence of disturbances in motor and sensory function as
well as psychological distress is predictive of poor outcome
following whiplash injury.
We have previously reported the development of changes
in these systems from soon after injury until the development
of chronicity in this whiplash cohort (Sterling et al.,
2003a,b,c, 2004). The aim of this study was to determine the
predictive capacity of the combined comprehensive set of
measures (motor, sensory and psychological), encompassing
the broad biopsychosocial model of musculoskeletal
pain, on outcome (persistent pain and disability) at 6 months
post-whiplash injury.
2. Methods
2.1. Study design
A prospective longitudinal design was used to assess the
whiplash subjects at less than 1 month post-injury and then at 6
months post-injury. Participants attended the research unit to
complete all questionnaires and assessments.
2.2. Subjects
Eighty individuals (56 females, mean age 36.27G12.69 years)
reporting neck pain as a result of a motor vehicle crash participated
in the study. The whiplash subjects were recruited via hospital
accident and emergency departments, primary care practices and
from advertisement. They were eligible if they met the Quebec
Task Force Classification of WAD II or III (Spitzer et al., 1995).
Subjects were excluded if they were WAD IV, experienced
concussion, loss of consciousness or head injury as a result of the
accident and if they reported a previous history of whiplash, neck
pain or headaches that required treatment.
2.3. Dependent variable
The outcome measure (dependent variable) was persistent pain
and disability at 6 months post-injury as measured with the Neck
Disability Index (NDI) (Vernon and Mior, 1991). The NDI consists
of 10 items addressing functional activities such as personal care,
lifting, reading, work, driving, sleeping and recreational activities
as well as pain intensity, concentration and headache (Vernon and
Mior, 1991). There are six potential responses for each item
ranging from no disability (0) to total disability (10). The overall
score (out of 100) is calculated by totalling the responses of each
individual item and multiplying by two. A higher score indicates
greater pain and disability (Vernon and Mior, 1991). The NDI is a
valid, reliable and responsive measure of neck pain and disability
(Pietrobon et al., 2002) and is in line with calls for the focus on
such instruments as outcome measures of functional recovery
following whiplash injury (Scholten-Peeters et al., 2003). Participants
returned to the research unit to complete this questionnaire at
6 months post-injury.
2.4. Independent variables
The independent variables were measures of both physical
(motor and sensory function) and psychological impairment as
outlined below. Three additional variables were also included since
these have been consistently shown in previous studies to be
predictors of poor outcome (Cote et al., 2001). These were age,
gender and the initial intensity of symptoms based on both NDI
score measured within 1 month of injury and the participants’
initial pain intensity (10 point VAS scale; 0, no pain; 10, worst pain
imaginable).
2.5. Physical measures of motor function
Range of active cervical movement (ROM) was measured in
three dimensions using an electromagnetic, motion-tracking device
(Fastrak, Polhemius, USA) according to previously established
methodology (Dall’Alba et al., 2001).
Joint position error (JPE) was measured according to Revel et
al. (1994) by using the Fastrak system and set-up described for
ROM. The subjects’ ability, whilst blindfolded, to relocate the
head to a natural head posture was measured following active
cervical left and right rotation and extension.
Surface electromyography (EMG) was used to measure the
activity of the superficial neck flexor muscles during the
established 5-staged clinical test of cranio-cervical flexion.
(Jull
et al., 2004; Sterling et al., 2003b).
2.6. Measures of sensory function
PPTs were measured using a pressure algometer with a probe
size of 1 cm2
and application rate of 40 kPa/s (Somedic AB, Farsta,
Sweden). PPTs were measured bilaterally over the articular pillars
of C2/3 and C5/6; over the three main peripheral upper limb nerve
trunks and at a remote site (tibialis anterior). These sites have been
previously used in investigation of WAD (Sterling et al., 2003a).
Triplicate recordings were taken at each site and the mean values
used for analysis.
Thermal (heat and cold) pain thresholds were measured
bilaterally over the cervical spine using the Thermotest system
M. Sterling et al. / Pain 114 (2005) 141–148142
(Somedic AB, Farsta, Sweden). Triplicate recordings were taken at
each site and the mean values used for analysis.
The brachial plexus provocation test (BPPT) was performed
as described previously (Sterling et al., 2003a). The range of
elbow extension was measured at the subjects’ pain threshold
using a standard goniometer (Clarkson and Gilewich, 1989). If
the subject did not experience pain, the test was continued until
end of available range. At the completion of this test, the
subjects were asked to record their pain on a 10 cm visual
analogue scale (VAS).
2.7. Sympathetic nervous system function
The sympathetic vasoconstrictor response (SVR) was used as
an indication of sympathetic nervous system (SNS) activity
(Schurmann et al., 1999). Using laser Doppler flowmetry (floLAB
Monitor, Moor Instruments, Devon, England), the skin blood flow
in the fingertips of both hands was measured. A provocation
maneuver (inspiratory gasp), which is known to cause a short
sympathetic reaction and cutaneous vasoconstriction, was performed
(Schurmann et al., 1999). Two quotients were calculated:
the SRF parameter (sympathetic reflex) that represents the relative
drop in the curve after provocation and the QI (quotient of
integrals) that also takes into account the duration of perfusion
decrease (Schurmann et al., 1999). A high QI and low SRF are
indicative of an impaired vasoconstrictor response.
2.8. Psychological questionnaires
The General Health Questionnaire 28 (GHQ-28) is a 28-item
measure of emotional distress in medical settings (Goldberg,
1978).
The TAMPA Scale of Kinesphobia (TSK) is a 17-item
questionnaire that measures the fear of reinjury due to movement
(Kori et al., 1990).
The Impact of Events Scale (IES) is a 15-item questionnaire
that measures current subjective stress related to a specific life
event (Horowitz et al., 1979).
2.9. Data analysis
Regression analyses were used to evaluate the predictive
function of the variable set. A stepwise regression analysis
predicted NDI score at the endpoint of the study (6 months postinjury).
Binomial logistic regression was used to evaluate group
assignment based on previous studies (Sterling et al., 2003a,b).
These groups were formed based on NDI scores at the endpoint
of the study (6 months post-injury) and included: recovered
(!8 NDI), milder pain and disability (10–28 NDI) and moderate/
severe pain and disability (O30 NDI) (Vernon, 1996). This
grouping was validated by a cluster analysis (K-means algorithm),
which showed no significant difference between the analytical
clustering and the NDI groups as proposed by Vernon (1996). The
logistic regression analyses were then subjected to cross-validation
analysis (leave one out) to assess the error rate of the classification
rates and to assess the reliability and generalisability of the
findings.
For all analyses significance was set at P!0.05.
3. Results
Seventy-six subjects completed the 6-month follow-up.
Five volunteers were excluded from the study. Reasons for
exclusion included a whiplash injury of greater than 4 weeks
duration (3 subjects) and a history of previous whiplash
injury (2 subjects). Four subjects withdrew after the first
assessment point. The reasons for withdrawal included
relocation to another city (two subjects), a head injury
several weeks following the whiplash injury (one subject)
and no reason given (one subject).
Seventy percent of participants were female and the
mean age was 36.27G12.69 years. The mean intensity of
neck pain was 3.5 (1.2) on a 10 cm VAS scale and the mean
score on the NDI was 34.15 (2.37). Three participants
(3.9%) could be classified as WAD III with the remainder
being WAD II. All participants reported neck pain with
headaches (55%), shoulder/arm pain (30%), thoracic (54%),
and lumbar spine (30%) also being reported.
The results of the stepwise regression analysis score
revealed that the independent variables of initial NDI score,
age, left rotation ROM, cold pain threshold, QI, and IES
score contributed significantly to the prediction of NDI
score at 6 months and together accounted for 67% of the
variability in NDI scores at 6 months post-injury (Table 1).
These results indicate that a higher NDI score at 6 months
post-injury is associated with a higher initial NDI score,
older age, female gender, decreased active range of left
rotation, decreased cold pain thresholds, less vasoconstriction
with the SVR test and higher levels of acute emotional
distress.
Details of the whiplash groups used for the logistic
regression analyses are provided in Table 2. This study did
not aim to investigate the effect of treatment and subjects
were free to pursue any form of treatment. The types and
numbers of treatments received (including medication) were
similar between the three whiplash groups (Table 3).
The results of the logistic regression analysis showed that
initial NDI score, age, cold pain threshold and IES score
were significant predictors of membership to the whiplash
Table 1
Results of stepwise regression analysis with NDI score at six months as the
dependent variable
Estimate Standard
error
t-value P
(Intercept) 11.74 10.89 1.08 0.285
Initial NDI 0.387 0.083 4.67 0.00002
Age 0.387 0.108 3.42 0.001
Left rotation K0.178 0.106 K1.902 0.05
CPT 0.505 0.199 2.53 0.01
QI K0.147 0.07 K2.098 0.04
IES (total) 0.338 0.094 3.608 0.006
Residual standard error, 10.07 on 65 degrees of freedom; multiple Rsquared,
0.6742; adjusted R-squared, 0.6291; F-statistic, 14.95 on 9 and 65
DF; P-value, 8.33!10K13
.
M. Sterling et al. / Pain 114 (2005) 141–148 143
group with persistent moderate severe symptoms at 6 months
(P!0.05) (Table 4). This model correctly predicted 68.8%
of those with persistent moderate/severe symptoms and
93.2% of those without moderate/severe symptoms at 6
months post-injury with an overall success rate of 88%.
Following cross-validation analysis the prediction rate of
the moderate/severe group remained at 68.8% with the
prediction rate of non-moderate/severe cases decreasing
slightly to 88.5% with an overall success rate of 86.7%.
A logistic regression analysis was also performed using
only the variables of age and initial NDI score to compare
the classification rate of these variables to the more
complete set (initial NDI score, age, cold pain threshold
and IES score). This model correctly predicted 93.2% of
those without moderate/severe symptoms at 6 months and
37% of those with moderate/severe symptoms, with an
overall success rate of 81%.
The results of the logistic regression analysis used to
predict those subjects with residual milder symptoms at 6
months from those who had recovered showed that initial
NDI score, QI quotient of the SVR test, GHQ-28 scores and
decreased range of cervical extension were significant
predictors of membership to this group (P!0.05) (Table 5).
This model correctly predicted 89.3% of recovered subjects
and 90.3% of those with persistent milder symptoms with an
overall success rate of 89.8%. Following cross-validation
analysis the successful prediction rates decreased, with
78.6% of recovered subjects predicted and 83.9% of those
with persistent milder symptoms. The overall success rate
was 81.4%.
The following variables demonstrated no significant
predictive capacity in either the stepwise or logistic
regression analyses: Joint position error, EMG activity of
the superficial neck flexor muscles during the craniocervical
flexion test, PPT (all sites), heat pain threshold,
responses to the brachial plexus test and scores of the TSK.
Table 2
The age, gender and classification of subject groups at 6 months according to the NDI scores
Group Number Age (years) (meanGSD) Gender % female NDI classification NDI (meanGSD)
Recovered group 29 29.3G11.72 50 !8 2.9G2.9
Mild pain and disability
group
30 34.3G12.5 77 10–28 16.5G5.6
Moderate/severe pain
and disability group
17 43.7G14.5 94 O30 42.8G12.2
Table 3
The numbers and types of treatment and medication received by the three whiplash groups
Group N (%) who
received
treatment
No. of
treatments
(average/study
period)
Treatment type N (%) N (%) on
medication
Medication type
PT CH AC SA NS Cod AD St Op
Recovered
(NZ29)
14 (48.3) 10.6 29 (100) 0 0 7 (24) 3 4 1 0 1 0
Mild symptoms
(NZ30)
19 (63) 14.4 14 (46.7) 4 (13.3) 1 (3) 13 (43.3) 2 10 2 1 0 1
Mod/severe
symptoms
(NZ17)
9 (52.9) 18.4 8 (47) 1 (5.8) 0 12 (70.5) 2 7 2 2 0 1
Treatment: PT, physiotherapy; CH, chiropractic; AC, acupuncture. Medication: SA, simple analgesics; NS, non-steroidal anti-inflammatory; Cod, codeine;
AD, anti-depressants; St, steroids; Op, opioids.
Table 4
Results of logistic regression analysis showing factors associated with
membership to the whiplash group with persistent moderate/severe
symptoms at six months post-injury
Variables B SE Wald test
(z-ratio)
P-value OR 95%CI OR
Lower Upper
Initial
NDI
0.06 0.027 4.84 0.028 1.06 1.007 1.12
Age 0.13 0.05 6.41 0.01 1.13 1.03 1.23
Cold pain
threshold
0.26 0.10 6.09 0.01 1.29 1.05 1.58
IES score 0.11 0.04 7.71 0.005 1.11 1.03 1.2
OR, odds ratio; CI OR, confidence interval for odds ratio.
Table 5
Results of logistic regression analysis showing factors associated with
membership to the whiplash group with persistent milder symptoms versus
recovery at six months post-injury
Variables B SE Wald test
(z-ratio)
P-value OR 95%CI OR
Lower Upper
Initial
NDI
0.18 0.077 5.75 0.017 1.15 1.03 1.28
GHQ-28 0.23 0.1 4.84 0.028 1.15 1.04 1.28
Extension K0.15 0.07 4.89 0.027 1.1 1.03 1.25
OR, odds ratio; CI OR, confidence interval for odds ratio.
M. Sterling et al. / Pain 114 (2005) 141–148144
Table 6 depicts the mean (SD) values of the variables that
significantly predicted membership to each of the three
whiplash groups.
4. Discussion
This study is the first to show that a combination of
physical (motor and sensory) and psychological factors, in
addition to previously recognised indicators of age and
initial symptom intensity are important in the determination
of outcome following whiplash injury. Initial NDI score was
the measure of symptom intensity that reached significance
with VAS scores of pain intensity failing to demonstrate a
significant predictive capacity. The high correlation (rZ0.8,
P!0.01) between these measures in this study may explain
these findings, but it also suggests that a measure of pain and
disability as opposed to pain intensity alone be used in the
early assessment of whiplash.
Both a multiple regression analysis using final NDI score
as the independent variable and logistic regression using a
classification system based on final NDI score as criterion
variables showed similar results with a combination of
variables providing the best prediction of outcome. Higher
initial levels of pain and disability, older age, cold
hyperalgesia and acute post-traumatic stress were associated
with a higher NDI score and membership to the group with
persistent moderate/severe symptoms at 6 months. The odds
ratios for these variables ranged from 1.1 to 1.3. It should be
noted that this applies to increased odds for a one-unit
change in the predictor variable. For example if cold pain
threshold decreases by 1 8C the odds of developing
persistent moderate/severe symptoms increases by a factor
of 1.3. With reference to Table 6, it can be seen that the
difference between mean values of cold pain threshold of
recovered subjects compared to those with persistent
moderate/severe symptoms was 9 8C. In this case the odds
ratio would increase to 10.6.
Diminished vasoconstrictive responses, whilst predictors
of higher NDI scores at 6 months in the multiple regression
analysis were not strong predictors of membership to the
group with moderate/severe symptoms. The reasons for this
are unclear. However, the results of the two analyses
together, suggest that all the above mentioned variables be
included in further evaluation of outcome following
whiplash injury. Gender was not predictive of poor outcome
in either analysis supporting the findings of a recent
systematic review (Scholten-Peeters et al., 2003). The
predictive capacity of this variable requires further
clarification.
The combination of variables identified in this study was
superior in predicting a poor outcome when compared to
previous models. A combination of age, gender, psychological
factors or age, gender, accident features have
accounted for approximately 35% of the variation in pain
and disability at 1 year post-accident (Kyhlback et al., 2002;
Table 6
Group means and SD for best predictors of group (recovered, mild pain and disability, moderate/severe pain and disability) membership based on logistic
regression analysis
Measure Recovered: mean (SD) Mild pain and disability: mean (SD) Moderate/severe: mean (SD)
O1 month 6 months O1 month 6 months O1 month 6 months
Initial NDIa,b,c
19.2 (12.5) 2.9 (2.9) 37.2 (19.8) 16.5 (5.6) 54.7 (13.6)a,b
42.8 (12.2)
Agea
29.3 (11.7) 34.3 (12.5) 43.7 (14.5)a
Left rotationa,c
55.5 (11.0) 59.4 (11.2) 51.4 (11.9) 55.4 (11.2) 44.8 (12.4)a
47.5 (12.6)
Extensionb
45.29 (13.3) 48.8 (13.9) 38.75 (13.6) 48.3 (13.2) 32.14 (13.4)b
31.1 1(3)
Cold pain thresholda,c
11.02 (5.7) 10 (5.1) 10.37 (6.1) 11 (6.1) 20.27 (6.4)a
19.89 (6.4)
QIc
58.4 (17.2) 55.7 (16.9) 52.19 (16.9) 56.1 (15) 69.68 (18.2)b
69.44 (17)
SRFc
0.75 (0.17) 0.75 (0.2) 0.76 (0.18) 0.76 (0.17) 0.61 (0.15) 0.63 (0.14)
IESa,c
9.5 (7.3) 0.38 (7.4) 13.5 (14.2) 4.7 (14.3) 28.1 (15.1)a
21.4 (13.1)
GHQ-28 totalb
19.6 (7.3) 12. (7.3) 32.4 (14.3)b
21.4 (14.5) 41.9 (16.2)b
34 (16.5)
Flexion 39.2 (9.5) 41.7 (9.2) 33.2 (11.5) 39.6 (11.4) 30.63 (12.2) 34.5 (12.9)
Right rotation 50.7 (9.9) 56.5 (10.8) 47.4 (11.8) 51.6 (11.8) 43.5 (12.8) 47.2 (12.6)
JPE 3.7 (2) 3.5 (2) 3.2 (2.4) 3.4 (2.2) 4.3 (2.6) 4.5 (2.8)
% EMG 23.3 (3.22) 19.5 (3.34) 25.43 (3.16) 19.19 (3.11) 34.55 (4.8) 28.65 (4.47)
PPT (neck) 167.3 (75.1) 202 (65.4) 152.5 (75.1) 199.4 (85) 91 (62) 135 (60.7)
PPT (median nerve) 197.6 (70.6) 231.8 (65.1) 210.5 (74.7) 244 (64.6) 140.9 (50.5) 169.9 (54.7)
PPT (Tib ant) 333 (75) 416.3 (76.3) 380.3 (84.3) 415.1 (76.4) 241 (48.7) 256.1 (55.3)
Heat pain threshold 41.9 (3.2) 43.1 (3.3) 42.8 (3.5) 43.4 (3.6) 38.6 (3.2) 39.5 (3.4)
TSK 34.4 (6.3) 28.4 (5.6) 38.4 (6.4) 34.3 (6.2) 42.3 (8.3) 39.7 (8.1)
Variables that failed to reach significance are shown in italics.
a
Significant predictors to distinguish between moderate/severe symptoms at 6 months from the other two groups.
b
Significant predictors to distinguish between those with persistent mild symptoms from those who were recovered by six months post-injury.
c
Significant predictors in stepwise regression analysis.
M. Sterling et al. / Pain 114 (2005) 141–148 145
Mayou and Bryant, 1996), well short of the 67% R2
demonstrated by our variables. Using logistic regression
analysis we correctly classified 68.8% of those with
persistent moderate/severe symptoms post-injury. This
figure is similar or superior to that determined either by
psychological tests, a combination of symptoms and ROM
and a combination of accident features and pain (Hartling et
al., 2002; Kasch et al., 2001; Olsson et al., 2002). These
studies utilised a dichotomous outcome (recovery or nonrecovery)
based on unvalidated questionnaires. In line with
recent calls for prognostic studies of whiplash to use reliable
and valid outcomes of functional recovery (Scholten-Peeters
et al., 2003), we employed the NDI as our outcome measure.
This allowed discrimination to be made between subjects
reporting recovery at 6 months, those with milder symptoms
and those with moderate/severe symptoms at this time
frame. This differentiation is important since those with
more severe symptoms account for a much greater
proportion of the costs (MAIC, 2002) and their early
identification is of interest to all stakeholders. The logistic
regression analysis also showed that higher initial NDI and
GHQ-28 scores, and less cervical extension differentiated
participants reporting milder symptoms at 6 months from
those who recovered. Understanding of differences between
these two groups is important for both the development of
appropriate interventions as well as for stakeholders
interested in curtailing costs through early claim settlement.
The classification rates of our set of predictor variables
(initial NDI, age, cold pain threshold, IES score) were
superior to those based on initial NDI score and age alone
(previously recognised predictors). Whilst both predictor
sets classified 93% of those without moderate/severe
symptoms at 6 months, the latter two variables predicted
only 37% of those with persistent moderate/severe symptoms
compared to the 68.8% rate of the complete predictor
set. This provides strong argument for the inclusion of
additional measures of cold hyperalgesia and post-traumatic
stress reaction in the assessment of acute whiplash.
The classification rates of our study decreased when
subjected to cross-validation analysis suggesting that the
results need to be viewed as preliminary until further
validation in other whiplash populations is undertaken. It is
notable that previous studies (Hartling et al., 2001; Mayou
and Bryant, 1996; Olsson et al., 2002) did not report the use
of cross-validation analysis on their classification rates and
thus the utility of their results is unknown.
Whilst it was a combination of levels of pain and
disability, physical and psychosocial factors that were
predictive of a poor outcome following whiplash injury,
not all measures showed significant predictive capacity.
Post-traumatic stress reaction was the only psychological
predictor of poor outcome with both fear of movement and
general psychological distress failing to reach significance.
This may be at odds with investigation of the development
of chronicity in low back pain where fear avoidance beliefs
play a prognostic role (Vlaeyen and Linton, 2000) and more
recently in whiplash injury where Nederhand et al. (2004)
found TSK scores to be predictive of persistent symptoms at
6 months post-trauma. It should be noted that the latter study
investigated only two variables (NDI, TSK) as possible
prognostic factors. The inclusion of a measure of posttraumatic
stress may have provided different results. Our
findings also highlight potentially important differences
between other musculoskeletal conditions such as low back
pain and whiplash injury. The traumatic event (MVC) that
precipitates the onset of whiplash may have different
psychological consequences that influence recovery than
those involved in the onset of low back pain, indicating
further research investigating such differences is warranted.
Our findings of specific psychological factors related to
delayed recovery following whiplash injury suggest that
specific treatments provided to those with an early posttraumatic
stress reaction may be more efficacious than a
broadly applied cognitive behavioural approach.
Cervical ROM loss was the only measure of motor
function predictive of higher pain and disability at 6 months
and supports findings of previous studies (Kasch et al.,
2001; Radanov et al., 1995). Neither joint position error nor
EMG activity during cranio-cervical flexion showed
significant predictive capacity. These motor deficits are
present in both idiopathic neck pain (Jull et al., 2004;
Kristjansson et al., 2003) and in whiplash injured persons
with lesser symptoms (Sterling et al., 2003b). The
uniformity of these deficits across neck pain conditions is
the likely reason for their poor predictive capacity of those
with persistent moderate to severe symptoms.
Cold hyperalgesia and impaired sympathetic vasoconstriction
were the strongest sensory predictive variables.
Whilst we have shown previously in this cohort that
mechanical, heat hyperalgesia and responses to brachial
plexus provocation are present from soon after injury and
persist in those with poor recovery (Sterling et al., 2003a),
these variables were not predictors of outcome. The reasons
for this are not clear but some possible explanations can be
explored. The sensory hypersensitivity seen in patients with
whiplash injury is proposed as being due to the augmentation
of central pain processing mechanisms (Moog et al.,
2002; Sterling et al., 2003a). Cold hyperalgesia and SNS
disturbances have not been previously investigated in
patients with whiplash injury, with most studies using
some form of mechanical or electrical stimulation (Banic et
al., 2003; Moog et al., 2002). It is possible that the two
former measures are better clinical indicators of central
nervous system hyperexcitability and this requires further
investigation. Additionally cold hyperalgesia and SNS
disturbances could be an indication of peripheral nerve
tissue damage (Djouhri et al., 2004), a potential consequence
of whiplash injury (Ide et al., 2001; Taylor and
Taylor, 1996).
The findings of this study show that in a proportion of
people (20–25% of the cohort), whiplash injury induces
concomitant sensory changes reflective of underlying
M. Sterling et al. / Pain 114 (2005) 141–148146
disturbances in pain processing, a post-traumatic stress
reaction, movement loss and higher levels of pain and
disability. The casual relationship between these factors is
yet to be determined and cannot be inferred from this study.
The reason for such changes in some participants but not
others in this cohort is not known. Whilst the rest of the
cohort showed evidence of muscle impairment and general
psychological distress post-accident, they did not demonstrate
post-traumatic stress or widespread sensory changes
(Sterling et al., 2003a). It appears that those with poor
recovery have experienced a more complex injury that leads
to the development of profound physical and psychological
changes.
These findings have implications for management of
acute whiplash injury. Concomitant high levels of pain and
disability, movement loss, sensory disturbance and a posttraumatic
stress reaction indicate that a multiprofessional
approach is required for these patients in the acute stage of
injury. This approach may need to include psychological
intervention to address the post-traumatic stress reaction,
physical therapy to restore movement loss as well as
adequate pain management (medication) in view of the
sensory disturbances and high pain levels seen in this subgroup
of the whiplash injured.
Acknowledgements
This work was supported by Suncorp Metway General
Insurance and the Centre for National Research on
Disability (CONROD).
References
Banic B, Petersen-Felix S, Andersen O, Radanov B, Villiger P, ArendtNielsen
L, Curatolo M. Evidence for spinal cord hypersensitivity in
chronic pain after whiplash injury and in fibromyalgia. Pain 2003;107:
7–15.
Barnsley L, Lord S, Bogduk N. Clinical review. Whiplash injury. Pain
1994;58:283–307.
Cassidy JD, Carroll LJ, Cote P, Lemstra M, Berglund A, Nygren A. Effect
of eliminating compensation for pain and suffering on the outcome of
insurance claims for whiplash injury. N Engl J Med 2000;20:1179–213.
Clarkson H, Gilewich G. Musculoskeletal assessment: joint range of
motion and manual muscle strength. Baltimore, MD: Williams and
Wilkins; 1989.
Cote P, Cassidy D, Carroll L, Frank J, Bombardier C. A systematic review
of the prognosis of acute whiplash and a new conceptual framework to
synthesize the literature. Spine 2001;26:E445–E58.
Dall’Alba P, Sterling M, Trealeven J, Edwards S, Jull G. Cervical range of
motion discriminates between asymptomatic and whiplash subjects.
Spine 2001;26:2090–4.
Djouhri L, Wrigley D, Thut P, Gold M. Spinal nerve injury increases the
percentage of cold-responsive DRG neurons. NeuroReport 2004;15:
457–60.
Goldberg D. Manual of the General Health Questionnaire. Windsor: NFERNelson;
1978.
Hartling L, Brison R, Ardern C, Pickett W. Prognostic value of the Quebec
classification of whiplash associated disorders. Spine 2001;26:36–41.
Hartling L, Pickett W, Brison R. Derivation of a clinical decision rule for
whiplash associated disorders among individuals involved in rear-end
collisions. Accid Anal Prev 2002;34:531–9.
Horowitz M, Wilner N, Alvarez W. Impact of Event Scale: a measure of
subjective stress. Psychosom Med 1979;41:209–18.
Ide M, Ide J, Yamaga M, Takagi K. Symptoms and signs of irritation of the
brachial plexus in whiplash injuries. J Bone Joint Surg (Br) 2001;83:
226–9.
Jull G, Kristjansson E, Dall’Alba P. Impairment in the cervical flexors: a
comparison of whiplash and insidious onset neck pain patients. Man
Ther 2004;9:89–94.
Kasch H, Flemming W, Jensen T. Handicap after acute whiplash injury.
Neurology 2001;56:1637–43.
Koelbaek-Johansen M, Graven-Nielsen T, Schou-Olesen A, ArendtNielsen
L. Muscular hyperalgesia and referred pain in chronic whiplash
syndrome. Pain 1999;83:229–34.
Kori S, Miller R, Todd D. Kinesphobia: a new view of chronic pain
behaviour. Pain Manag 1990;Jan/Feb:35–43.
Kristjansson E, Dall’Alba P, Jull G. A study of five cervicocephalic
relocation tests in three different subject groups. Clin Rehabil 2003;17:
768–74.
Kyhlback M, Thierfelder T, Soderlund A. Prognostic factors in whiplash
associated disorders. Int J Rehabil 2002;25:181–7.
MAIC. Whiplash—review of CTP Queensland data to 31 Dec 2001.
Brisbane: MAIC; 2002.
Mayou R, Bryant B. Outcome of whiplash neck injury. Injury 1996;27:
617–23.
Moog M, Quintner J, Hall T, Zusman M. The late whiplash syndrome: a
psychophysical study. Eur J Pain 2002;6:283–94.
Nederhand M, Hermens H, Ijzerman M, Turk D, Zilvold G.
Cervical muscle dysfunction in chronic whiplash associated
disorder grade 2. The relevance of trauma. Spine 2002;27:
1056–61.
Nederhand M, Ijzerman M, Hermens H, Turk D, Zilvold G. Predictive
value of fear avoidance in developing chronic neck pain disability:
consequences for clinical decision making. Arch Phys Med Rehabil
2004;85:496–501.
Olsson I, Bunketorp O, Carlsson S, Styf J. Prediction of outcome in
whiplash associated disorders using West Haven–Yale Multidimensional
Pain Inventory. Clin J Pain 2002;18:238–44.
Pietrobon R, Coevtaux R, Carey T, Richardson W, de Vellis R. Standard
scales for measurement of functional outcome for cervical pain or
dysfunction: a systematic review. Spine 2002;27:515–22.
Radanov B, Sturzenegger M, Di Stefano G. Long-term outcome after
whiplash injury. A 2-year follow-up considering features of injury
mechanism and somatic, radiologic, and psychological findings.
Medicine 1995;74:281–97.
Radanov B, Begre S, Sturzenegger M, Augustiny K. Course of
psychological variables in whiplash injury—a 2-year follow-up
with age, gender and education pair-matched patients. Pain 1996;64:
429–34.
Revel M, Minguet M, Gergory P, Vaillant J, Manuel J. Changes in
cervicocephalic kinesthesia after a proprioceptive rehabilitation
program in patients with neck pain: a randomized controlled study.
Arch Phys Med Rehabil 1994;75:895–9.
Scholten-Peeters G, Verhagen A, Bekkering G, Van der Windt D,
Barnsley L, Oostendorp R, Hendriks EJ. Prognostic factors of whiplash
associated disorders: a systematic review of prospective cohort studies.
Pain 2003;104:303–22.
Schrader H, Obelieniene D, Bovim G, Surkiene D, Mickeviciene D,
Miseviciene I, Sand T. Natural evolution of the late whiplash
syndrome outside the medicolegal context. Lancet 1996;347:
1207–11.
Schurmann M, Gradl G, Andress H, Furst H, Schildberg F. Assessment of
peripheral sympathetic nervous system function for diagnosing early
post-traumatic complex regional pain syndrome type I. Pain 1999;80:
149–59.
M. Sterling et al. / Pain 114 (2005) 141–148 147
Spitzer W, Skovron M, Salmi L, Cassidy J, Duranceau J, Suissa S, Zeiss E.
Scientific Monograph of Quebec Task Force on whiplash associated
disorders: redefining ‘whiplash’ and its management. Spine 1995;20:
1–73.
Sterling M, Jull G, Vicenzino B, Kenardy J. Sensory hypersensitivity
occurs soon after whiplash injury and is associated with poor recovery.
Pain 2003a;104:509–17.
Sterling M, Jull G, Vizenzino B, Kenardy J, Darnell R. Development of
motor system dysfunction following whiplash injury. Pain 2003b;103:
65–73.
Sterling M, Kenardy J, Jull G, Vicenzino B. The development of
psychological changes following whiplash injury. Pain 2003c;106:481–9.
Sterling M, Jull G, Vicenzino B, Kenardy J. Characterisation of acute
whiplash associated disorders. Spine 2004;29:182–8.
Taylor J, Taylor M. Cervical spinal injuries: an autopsy study of 109 blunt
injuries. J Musculoskelet Pain 1996;4:61–79.
Vernon H. The Neck Disability Index: patient assessment and
outcome monitoring in whiplash. J Musculoskelet Pain 1996;4:
95–104.
Vernon H, Mior S. The Neck Disability Index: a study of reliability and
validity. J Manipulative Physiol Ther 1991;14:409–15.
Vlaeyen J, Linton S. Fear-avoidance and its consequences in
chronic musculoskeletal pain: a state of the art. Pain 2000;85:
317–32.
M. Sterling et al. / Pain 114 (2005) 141–148148