Osteoarthritis and Cartilage 25 (2017) 1781e1791

Fourteen days of etoricoxib 60 mg improves pain, hyperalgesia and

physical function in individuals with knee osteoarthritis: a
randomized controlled trial
P. Moss y *, H.A.E. Benson z, R. Will x, A. Wright y
y School of Physiotherapy and Exercise Science, Curtin University, Australia
z School of Pharmacy, CHIRI Bioscience, Curtin University, Australia
x School of Medicine and Pharmacology, University of Western Australia, Australia

a r t i c l e i n f o s u m m a r y

Article history: Objective: Mounting evidence points to the heterogeneity of osteoarthritis (OA) pain, increasing the need
Received 3 January 2017 for more comprehensive assessment of the efficacy of standard interventions. This study investigated
Accepted 6 July 2017 whether 14 days of the selective Cox-2 inhibitor etoricoxib (60 mg/day) would modify self-report of pain
intensity and quality, and physical measures of hyperalgesia and function in individuals with knee OA.
Keywords: Design: This double-blind placebo-controlled trial included 80 community-recruited volunteers with
Etoricoxib
painful knee OA (3/10 VAS), randomly allocated to Active or Placebo groups. Self-report measures of
Neuropathic pain
pain, stiffness, function Western Ontario and McMaster University Osteoarthritis Index (WOMAC) and
Knee osteoarthritis
pain quality (PainDETECT, Pain Quality Assessment Scale [PQAS]) and physical measures of locomotion
and local (knee) and widespread (elbow) hyperalgesia were assessed at Days 0, 4 and 14. Repeated
Measures ANOVA analysed group differences.
Results: Significant group  time interaction effects were found for all measures of pain (all p < 0.001),
with WOMAC pain sub-score improving by 30.7% by Day 14 and index knee mechanical hyperalgesia
improving by 32.6%, whilst Placebo group values worsened. Both self-report and physical tests of
function improved (p < 0.001ep ¼ 0.006): WOMAC-function by 28.4%, sit-to-stand and walk time by 13%,
pain during locomotion tasks by 12.4e32.6%. Pain quality also significantly improved for the Active and
declined for the Placebo group (p < 0.001): PainDETECT score reduced by 23.6% and PQAS paroxsysmal
and surface sub-scores by 36.9% and 29.4%. There were also significant improvements in local cold
hyperalgesia and widespread mechanical hyperalgesia (10e13.8%).
Conclusion: Just 14 days of etoricoxib significantly improves pain intensity and quality, function and local
and widespread hyperalgesia, measured by both self-report and physical tests.
© 2017 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Background quality2e5. A larger group has pain primarily driven by peripheral

The chronic pain associated with osteoarthritis (OA) burdens
individuals and healthcare systems worldwide1. Increasing evi-
dence suggests that OA pain is not as homogenous as previously
assumed. Distinct phenotypic groups can be identified through
psychophysical and clinical testing or by self-report of pain
* Address correspondence and reprint requests to: Penny Moss, School of
Physiotherapy and Exercise Science, Curtin University, GPO Box U1987, Perth, WA
6845, Australia. Fax: 61-8-9266-3699.
E-mail addresses: P.Moss@curtin.edu.au (P. Moss), H.Benson@curtin.edu.au
(H.A.E. Benson), robw@bdaus.com.au (R. Will), T.Wright@curtin.edu.au (A. Wright).
inflammatory nociceptive input whereas a smaller group experi-
ences more severe pain showing features of neuropathic-like
symptoms6,7. Standard therapeutic interventions may therefore
not be efficacious for all individuals with OA.
Non-steroidal anti-inflammatory drugs (NSAIDs) have long been
a mainstay for OA treatment8. Traditional NSAIDs reduce inflam-
matory symptoms by inhibiting the cyclo-oxygenase-1 (COX-1)
enzyme. More recently-developed NSAIDs, such as etoricoxib,
selectively inhibit the COX-2 inducible form of cyclo-oxygense8,
potentially improving anti-hyperalgesic action9,10. COX-2 NSAIDs
have been found to cross the bloodebrain barrier and so may in-
fluence central nervous system nociception11. There are risk profile

http://dx.doi.org/10.1016/j.joca.2017.07.009
1063-4584/© 2017 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
1782 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
differences between COX-1 and COX-2 NSAIDs9 and OA guidelines
emphasise the importance of prescribing the minimum effective
dose over the shortest time12. More extensive data regarding COX-2
efficacy are therefore needed to assist prescribers in risk-benefit
analysis for individual patients.
Assessment of pharmaceutical efficacy has tended to rely on
standard self-report measures of pain, stiffness and function such
as the Western Ontario and McMaster University Osteoarthritis
Index (WOMAC)13 or the Short Form-36 (SF-36) quality of life
questionnaire14. However, application of a broader range of clinical
and psychophysical measures provides an opportunity to evaluate
efficacy more comprehensively and across different pain pheno-
types. Pain quality questionnaires evaluate the extent of
neuropathic-like symptoms experienced by an individual with
OA2,4,15. Physical assessment of locomotion provides an objective
measure of impact on function16. In cross-sectional studies, quan-
titative sensory testing (QST) has demonstrated the presence of
widespread multi-modality hyperalgesia17e19 which may be an
indicator of centrally-driven pain4,5,20. Measuring change in
hyperalgesia at local and distant test sites following a pharma-
ceutical intervention therefore enables evaluation of its impact on
the nociceptive system.
This study therefore aimed to investigate whether a 14-day
course of etoricoxib (60 mg) was more effective than placebo in
altering pain intensity, pain quality, hyperalgesia and functional
limitations in people with knee OA. In addition to evaluating
patient-reported changes, a battery of QST and functional measures
were utilized to evaluate treatment efficacy.
Methods
Trial design
The study used a double-blind, randomized, placebo-controlled
parallel group design. Participants were randomly allocated to
Placebo or Active groups in a 1:1 ratio at baseline by the Royal Perth
Hospital (RPH) Pharmacy Clinical Trials Coordinator, using a
computer-generated algorithm. All investigators remained blind to
group allocation throughout testing and data management.
Interventions
All medications were administered, stored and dispensed by
RPH Pharmacy. Each participant received an identical pack con-
taining 14 days’ supply of once-daily active or placebo etoricoxib
formulation. All participants were also provided with a 14-day pack
of rescue medication: acetaminophen or tramadol, as prescribed by
the study rheumatologist. Participants were asked to refrain from
taking rescue medication within 12 h (acetaminophen) or 24 h
(Tramadol) of testing. They were also asked to keep a simple daily
checklist of study and rescue medication use.
Subjects
Community-dwelling volunteers with painful knee OA (visual
analogue scale (VAS) 3/10) were recruited via local (Perth, WA)
radio advertisements and assessed for suitability by a rheumatol-
ogist, using the American College of Rheumatology classification21.
If both knees showed OA, the knee with the highest pain was
designated the index knee. Volunteers were assessed for adverse
response to NSAIDs during medical screening. Previous effective-
ness of pharmaceutical interventions was not specifically assessed.
Exclusion criteria included: history of systemic inflammatory con-
dition; neurological disorders affecting sensory, motor or cognitive
function; recent lower limb injury or surgery; history of other
chronic pain disorders. All participants provided written informed
consent. Ethical approval was provided by RPH (EC2009/100) and
Curtin University (HR26/2010). The trial was registered with
ClinicalTrials.gov (NCT00927004).
Study protocol
Participants were tested three times: Day 0 (baseline), Day 4
(drug steady-state plasma level), and Day 14 of their intervention.
All participants underwent a washout period equal to 5 half-lives of
their usual analgesic or anti-inflammatory before baseline testing.
At each test session, subjects completed questionnaires regarding
quality of life, pain intensity, pain quality and function, followed by
function and QST testing.
Outcomes measures
Self-report questionnaires
Subjective pain, stiffness and dysfunction was measured using
the WOMAC OA Knee Index which has been widely used to measure
pain and disability, demonstrating good internal validity and
testeretest reliability13,22. A VAS-scored version was used for
greater responsiveness.
Mean pain in the last 7 days was evaluated using a 10 cm VAS,
with end points marked “no pain” and “worst pain imaginable”. For
Day 4 assessment, participants reported mean pain since Day 0.
Quality of life was evaluated with the SF-36 (v2), which has
demonstrated good validity and reliability across many condi-
tions23. The tool measures the self-perceived impact of health
status on quality of life via eight domains, using Likert-type
response categories. SF36 data was only collected at baseline and
Day 14 as it has not been validated over a 4-day time-frame.
Presence of neuropathic-type symptoms was evaluated using
the PainDETECT questionnaire. Although originally developed for
low back pain24, this self-report tool has more recently been
accepted as a valid assessment tool when identifying neuropathic
pain in a wide range of conditions2,4,15,25. VAS, body diagram and
Likert-type responses assess pain intensity, radiation and frequency
of symptoms such as ‘electric shocks’. A total score is calculated and
subjects classified as ‘negative neuropathic’ (12), ‘unclear neuro-
pathic’ (13e18), or ‘positive neuropathic’ (19)24. The Pain Quality
Assessment Scale (PQAS) was used to provide additional pain
quality data26. Participants respond to each of the 17 questions
using a 0e10 scale. Three pain quality sub-scores (‘paroxysmal’,
‘surface’ and ‘deep’) can be calculated. It has been suggested that
differences between subscales may differentiate nociceptive from
neuropathic pain26.
Physical function tests
Physical function was assessed using the 3 tasks of the
Aggregated Locomotion Function Score (ALF)27. Time taken to
complete each task was measured using a stop-watch: sit-to-
stand (walk 2-metres to a chair, sit, stand and return); walk (8-
metre return); stairs (11-step ascent and descent). Each task
was completed 3 times and the mean time calculated. This score
has shown excellent reliability and good responsiveness following
intervention27. Pain during each task was also measured using a
10 cm VAS.
Quantitative sensory tests
All quantitative sensory tests were applied in triplicate using
standardized instructions at 3 standardized sites in randomized
order28: medial joint line of the index knee and contralateral
 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
knee; extensor carpi radialis brevis (ECRB) muscle of the ipsi-
lateral elbow.
Pressure pain threshold (PPT) was assessed using an electronic
digital pressure algometer (Somedic AB, Sweden), a device that
consistently shows good reliability when applied by a skilled
assessor29. The 1 cm2 probe was applied at 90 to the skin at a rate
of 40 kPa/s. Participants were instructed to depress the hand-held
switch as soon as the pressure sensation became painful. Lower
PPT values indicate increased sensitivity.
Punctate pain threshold (PcPT) was measured using von Frey
filaments (Semmes Weinstein monofilaments, Bioseb). The stair-
case method was used, applying tips in ascending and descending
order of magnitude for three repetitions. PcPT was defined as the
smallest von Frey tip (g) perceived as painful. Lower values indicate
greater sensitivity.
Cold detection (CDT) and cold pain (CPT) thresholds were
measured using a Peltier thermode (Medoc, Israel) and standard
Method of Limits30. The probe was attached to each test site using
a Velcro™ strap and the temperature set to reduce from a baseline
of 32 C at a constant rate of 1  C/s. Participants were instructed to
depress the hand-held switch as soon as they first felt cooling
(CDT) or when the cooling sensation had changed to one of
painful cold (CPT). For participants who failed to indicate pain
before the minimum 0 C was reached, a CPT value of 0 C was
assigned. Higher CPT temperature values indicate greater sensi-
tivity to cold.
Warmth detection (WDT) and heat pain (HPT) thresholds were
measured with the Medoc thermode using similar methodology as
Fig. 1. Participant flow.
1783
for cold (baseline 32 C, 1  C/s ramp), and the maximum set at 50 C.
WDT was defined as the temperature at which a participant first
felt increasing warmth and HPT as the temperature at which par-
ticipants perceived the heating sensation to be painful. For partic-
ipants who failed to register pain before the maximum was
reached, an HPT of 50 C was assigned. Lower HPT temperature
values indicate greater sensitivity to heat.
Sample size and statistical analysis
WOMAC-pain score was designated the primary clinical
outcome and PPT the primary physical outcome for the study.
Sample size was calculated from interventional studies demon-
strating 20e22% change in knee PPT28 or WOMAC score31. With
alpha set at p < 0.05, a calculated sample size of 37e45 per group
would provide 80% power to show 20% change (actual difference
for PPT of 38 kPa (standard deviation 58 kPa)28 and 10.1 mm
(standard deviation 17 mm) for WOMAC pain sub-score31.
ShapiroeWilks analysis showed that parametric statistics could
be applied to all but CPT data. Baseline group differences were
analysed using independent t-tests or equivalent non-parametric
test. Repeated Measures ANOVA, with Active/Placebo group as
the between-subjects’ factor and time (Day 0, Day 4, Day 14) as the
within-subject factor, was used to analyse group differences. For
CPT, Friedman's two-way ANOVA by ranks was applied. Post-hoc t-
tests analysed differences between groups at each time-point. SF-
36 data were analysed with paired t-tests.
1784 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
Effect size for key outcome measures was calculated using
Cohen's-d. Numbers needed to treat (NNT) analysis was calculated
with 50% improvement as the target outcome. The following
formula was used:
1
ðn target outcomes for Active groupÞ  ðn target outcomes for Placebo groupÞ
Results
In total 86 participants with knee OA were recruited (Fig. 1). Of
these, six did not complete the study: 2 withdrew before starting, 1
withdrew after 2 days for personal reasons and 3 withdrew before
Day 4 due to mild adverse effects (headache or facial rash). Aside
from these, no additional side-effects were reported, supporting
the good tolerability of 60 mg etoricoxib over 14 days8. Baseline
data from the 6 participants who withdrew was not included in
analysis. The final randomized group allocation ratio was balanced
1:1. Eighty-five percent of participants returned a completed
medication checklist, with 23 participants reporting rescue medi-
cation use at some point (4 Active and 19 Placebo).
Baseline characteristics
The final cohort of 80 comprised 36 males and 44 females, mean
age 65 years. Establishing date of formal diagnosis proved difficult
in this community cohort. Self-report of duration of knee symp-
toms or clinical diagnosis ranged from 2 to 30 years.
The large majority of participants reported moderate baseline
levels of pain and disability: 8.8% of participants reported mean
pain of >7/10, 46.2% reported mean pain of 5e6/10% and 45% re-
ported 3e4/10 pain. From baseline PainDETECT score, 45% of par-
ticipants were classified as “positive neuropathic” (14%) or “unclear
neuropathic” (31%) (Fig. 5). A previous study has examined baseline
data for this cohort5, finding that the 43.8% of participants who
exhibited widespread cold hyperalgesia (CPT > 12.25 C, based on Z-
scores from a matched healthy group) also showed widespread
mechanical and heat hyperalgesia, higher painDETECT scores,
greater pain and disability, and lower quality of life scores. There
were no significant differences between Placebo and Active groups
for these measures of pain and sensitization, or for age, gender or
BMI (Supplementary Data Table (A)).
The majority of subjects (65%) reported regular use of at least
one analgesic medication: slow release high dose acetaminophen
(40%), NSAIDs (36%), Tramadol 10% or opioids 2.5%. The most
frequently reported co-morbidities were medication-controlled
diabetes (30%) and high blood pressure (34%). There was no sig-
nificant group difference: diabetes Active 32.5%, Placebo 27.5%;
high BP Active 30%, Placebo 37.5%). Mild (<3/10) intermittent low
back or neck pain were reported by 16% and depression by 14%.
Self-report measures
WOMAC (Fig. 2)
There was also a significant group  time interaction effect for
WOMAC pain: F(2, 79) ¼ 10.85, p < 0.001 with Active group pain
reducing by 20.5% by Day 4% and 30.7% by Day 14. In contrast, pain
increased steadily in the Placebo group to 9.8% higher by Day 14.
Post-hoc t-tests showed a significant group difference by Day 14
(p ¼ 0.006).
WOMAC stiffness and function sub-scores also showed significant
group  time interactions: stiffness F(2, 79) ¼ 15.75, p < 0.001; func-
tion F(2, 79) ¼ 15.21, p < 0.001. Active participants’ stiffness improved
by 44% and function by 28.4% by Day 14 while Placebo participants
scores declined by 9% and 14% respectively. There were significant
between-group differences for stiffness both at Day 4 and Day 14 and
for function at Day 14 (Fig. 2). Total WOMAC score increased by 29.6%
in the Active group whilst declining by 12% in the Placebo group.
VAS pain
There was a significant group  time interaction effect for prior
7-day pain: F(2, 79) ¼ 8.69, p < 0.001. Post-hoc tests showed a sig-
nificant group difference at both Day 4 (p ¼ 0.030) and Day 14
(p ¼ 0.001). This amounted to a 21.7% reduction in pain for the
Active group and a 6.5% worsening for Placebo participants.
SF-36 quality of life
Both physical (PCS) and mental (MCS) sub-scores showed sig-
nificant group  time interactions: PCS F(1, 79) ¼ 21.24, p < 0.001;
MCS F(1, 79) ¼ 9.95, p ¼ 0.002. Active group PCS improved by 12.9%,
whilst declining by 3.6% for the Placebo group. MCS scores
improved by 6.7% for the Active group and declined by 4.3% for
Placebo participants.
Neuropathic-type symptoms (Fig. 3)
PainDETECT scores showed a significant group  time interac-
tion (F(1, 79) ¼ 20.38, p < 0.001). Active group scores improved by
13.7% by Day 4% and 23.6% by Day 14. Placebo values worsened by
7.6% and 14.0%. There was a significant between-groups difference
at Day 14 (Fig. 3). c2 analysis of PainDETECT categories at Days 0 and
14 showed that there was a significant change in category for both
Placebo and Active groups (Fig. 4): for 22.5% of Active participants,
reduction in PainDETECT score by Day 14 resulted in reclassification
from ‘unclear’ or ‘positive neuropathic’ to ‘negative neuropathic’. In
contrast, Placebo group scores increased by Day 14, resulting in
12.5% of Placebo participants moving from ‘negative’ or ‘unclear’ to
‘positive neuropathic’.
All PQAS sub-scales showed significant group  time in-
teractions (Fig. 3). The Deep subscale, reflecting inflammatory pain
quality, showed the greatest change (F(2, 79) ¼ 24.58, p < 0.001):
Active participants improved 34.2% by Day 4% and 45.5% by Day 14;
Placebo participants' scores declined by 15.7% (Day 4) and 27.4%
(Day 14). There were significant between-groups differences in
Deep score both at Day 4 and Day 14. Paroxsysmal (F(2, 79) ¼ 17.77,
p < 0.001) and Surface (F(2, 79) ¼ 12.19, p < 0.001) subscales,
reflecting neuropathic-type pain quality, showed significant
group  time interactions. For the Paroxsysmal, Active participants
improved by 24.1% by Day 4% and 36.9% by Day 14, whilst Placebo
scores declined. Between-group differences were significant both
at Day 4 and Day 14 (Fig. 3). The Surface sub-score for Active par-
ticipants plateaued at 29.4% by Day 4 while Placebo participants’
scores declined steadily to 49.7% by Day 14.
Physical function tests (Table I)
There were significant group  time interaction effects for time
taken to complete all three locomotion tasks: Sit-to-stand (F(2,
 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
Fig. 2. WOMAC pain, stiffness and function sub-score values for Active and Placebo groups (n ¼ 40/group) at Days 0, 4 and 14, showing overall group  time interaction effect and
between-group differences at Days 4 and 14.
79)¼ 7.77, p ¼ 0.003); walk (F(2, 79) ¼ 5.38, p ¼ 0.006); stairs (F(2,
79)¼ 9.49, p < 0.001). The greatest Active group improvements
were for the chair and stairs tasks, both 10% by Day 4% and 13% by
Day 14. In contrast, time changed minimally for the Placebo group
and post hoc t-tests showed no significant between-groups differ-
ences at either Day 4 or 14 (Table I).
Pain during each task showed significantly greater improvements
in the Active group: pain during sit-to-stand (F(2, 79) ¼ 12.45,
p < 0.001); walk (F(2, 79) ¼ 12.07, p < 0.001); stairs (F(2, 79) ¼ 11.14,
p < 0.001). Active group pain during each task decreased by approx-
imately 50% by Day 4 then plateaued. Between-group differences in
pain during each task were significant at each time-point (Table I).
Quantitative sensory tests
PPT (Fig. 5)
Significant group  time interaction effects were seen for PPT at
all test sites: index knee (F(2, 79) ¼ 9.67, p < 0.001); contra-lateral
knee (F(2, 79) ¼ 7.96, p ¼ 0.001); ECRB (F(2, 79) ¼ 16.69, p < 0.001).
1785
Reductions in pain sensitivity for the Active group were greatest at
the index knee (32.6% by Day 14), although reduced by 13.8% and
12.4% at the contra-lateral knee and ECRB respectively. Between-
group differences were significant at each time point for each test
sites, excepting the ECRB at Day 14.
PcPT (Table II)
PcPT also showed significant group  time interaction effects at
each test site: index knee (F(2, 79) ¼ 11.48, p < 0.001); contra-lateral
knee (F(2, 79) ¼ 6.25, p ¼ .004); ECRB (F(2, 79) ¼ 3.10, p ¼ 0.048).
Although percentage change in PcPT was small, by Day 14 there
were statistically significant between-group differences at the OA
knee and ECRB.Table II.
HPT (Table II)
HPT showed minimal change and no significant interaction ef-
fect at any of the sites: index knee (F(2, 79) ¼ 1.72, p ¼ 0.180); contra-
lateral knee (F(2, 79) ¼ 0.277, p ¼ 0.759); ECRB (F(2, 79) ¼ 0.618,
p ¼ 0.531).
1786 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791

Fig. 3. PainDETECT and PQAS Paroxsysmal, Surface and Deep sub-scale values for Active and Placebo groups (n ¼ 40/group) at each test site at Days 0, 4 and 14, showing overall
group  time interaction effect and between-group differences at Days 4 and 14.

Fig. 4. Change in PainDETECT category from Day 0 to Day 14 for Placebo and Active group participants (n ¼ 40/group).
 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791 1787

Fig. 5. PPT values for Active and Placebo groups at each test site at Days 0, 4 and 14, showing overall group  time interaction effect and between-group differences at Days 4 and 14
(n ¼ 40/group).

CPT (Table II) NNT and effect size (Table III)

Friedman's two-way ANOVA showed a significant improvement
in CPT at the index knee by Day 14 for the Active group (c2(2) ¼ 6.61,
p ¼ 0.037), an improvement of 9.7% alongside Placebo decline of
12.8% by Day 14. However, no significant between-groups difference
was found for CPT at the index knee at either Day (Table III). There
was no significant change over time for either group at the contra-
lateral knee or ECRB, although there was a significant between-
groups difference at the contra-lateral knee at Day 14 (p ¼ 0.042).
Warmth and CDT
There was minimal change in WDT or CDT temperature and no
significant group  time interaction effect at any of the test sites
(WDT: index knee p ¼ 0.161; contra-lateral knee p ¼ 0.370; elbow
p ¼ 0.278. CDT: index knee p ¼ 0.295; contra-lateral knee p ¼ 0.170;
elbow p ¼ 0.121).
NNT and effect size for each key measure are shown in Table III.
Lowest NNT were seen for PQAS-Deep and Paroxsysmal sub-scores
(n ¼ 1.8 and 1.9) and for WOMAC stiffness (n ¼ 2.5) and function
(n ¼ 2.9). Highest effect sizes were found for pain during ALF walk
and stairs, WOMAC function, and PainDETECT score.
Discussion
A 14-day course of etoricoxib (60 mg) resulted in substantial
improvements across a range of locomotor, QST and self-report
measures in individuals with painful knee OA. This included im-
provements: in self-reported pain, stiffness and function; in time
taken and pain during locomotion tasks; in neuropathic-type pain;
1788 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791

Table I
Time taken (mean, SD) and VAS pain during (mean, SD) each Aggregated Locomotion Function task for Active (n ¼ 40) and Placebo (n ¼ 40) groups at Days 0, 4 and 14, showing
within-group percentage change (improvement (þ) or decline ()) and between-groups difference at each time-point

Day 0 Day 4 % Change (0e4) t p Day 14 % Change (0e14) t p

Sit-to-stand task
Time (s) Placebo 6.41 (1.98) 6.31 (1.98) þ1.56% 0.272 0.786 6.30 (2.11) þ1.56% 0.641 0.523
Active 6.94 (2.29) 6.20 (1.66) þ10.66% 6.02 (1.68) þ13.26%
Pain (VAS 0e10) Placebo 2.34 (1.93) 2.74 (2.64) 17.09% 2.49 0.015* 2.59 (2.59) 10.68% 2.40 0.019*
Active 2.91 (2.26) 1.49 (1.77) þ48.80% 1.36 (1.92) þ53.27%
Walk task
Time (s) Placebo 12.58 (3.38) 12.43 (2.99) þ1.19% 0.12 0.908 13.00 (3.95) 3.34% 0.727 0.469
Active 13.24 (3.85) 12.52 (3.57) þ5.44% 12.41 (3.24) þ6.27%
Pain (VAS 0e10) Placebo 2.94 (2.19) 2.55 (2.48) þ13.27% 2.00 0.049* 3.17 (2.81) 7.8% 3.31 0.001*
Active 3.21 (2.44) 1.53 (2.07) þ52.34% 1.32 (2.13) þ58.88%
Stairs task
Time (s) Placebo 17.26 (8.54) 17.65 (8.92) 2.26% 0.331 0.742 17.89 (8.60) 3.65% 0.693 0.490
Active 18.97 (10.07) 16.99 (9.08) þ10.44% 16.51 (9.28) þ12.97%
Pain (VAS 0e10) Placebo 4.39 (1.76) 4.30 (2.47) þ2.05% 3.76 <0.001* 4.51 (2.85) 2.73% 3.66 <0.001*
Active 4.38 (2.70) 2.34 (2.10) þ46.58% 2.30 (2.47) þ47.49

Table II
Punctate and thermal pain thresholds (mean, SD) at the index knee, contra-lateral knee and ECRB test sites at Days 0, 4 and 14, showing within-group percentage change
(improvement (þ) or decline ()) and between-groups difference at each time-point (n ¼ 40/group)

Day 0 Day 4 % Change (0e4) t p Day 14 % Change (0e14) t p

PcPT (kPa)
Index knee Placebo 4.82 (0.37) 4.79 (0.28) 0.62% 0.658 0.513 4.75 (0.31) 1.45% 2.26 0.026*
Active 4.62 (0.45) 4.83 (0.33) þ4.55% 4.91 (0.32) þ6.49%
Contra-lateral knee Placebo 4.90 (0.31) 4.84 (0.33) 1.22% 0.487 0.628 4.85 (0.33) 1.02% 0.258 0.797
Active 4.71 (0.38) 4.85 (0.30) þ2.97% 4.87 (0.31) þ3.40%
ECRB Placebo 4.73 (0.88) 4.69 (0.86) 0.85% 2.351 0.021* 4.72 (0.88) 0.21% 2.270 0.026*
Active 4.93 (0.29) 5.05 (0.39) þ2.43% 5.07 (0.38) þ2.84%
HPT ( C)
Index knee Placebo 44.8 (3.0) 45.7 (3.4) þ2.0% 0.154 0.878 45.9 (2.8) þ2.5% 0.403 0.688
Active 45.8 (3.1) 45.6 (2.5) 0.44% 46.1 (2.7) þ0.7%
Contra-lateral knee Placebo 45.2 (3.2) 45.4 (3.4) þ0.44% 0.386 0.701 46.2 (2.7) þ2.2% 0.844 0.401
Active 45.8 (2.8) 45.7 (2.6) 0.22% 46.9 (2.4) þ2.4%
ECRB Placebo 45.5 (4.1) 46.0 (3.8) þ1.1% 1.51 0.134 46.1 (3.4) þ1.31% 1.72 0.090
Active 47.3 (3.4) 47.2 (3.1) 0.21% 47.3 (2.6) 0.0%
CPT ( C) ManneWhitey U ManneWhitey U
Index knee Placebo 12.4 (8.9) 13.6 (10.1) 9.7% p ¼ 0.321 13.9 (9.6) 12.1% p ¼ 0.165
Active 11.3 (9.6) 10.8 (9.6) þ4.4% 10.3 (9.6) þ8.9%
Contra-lateral knee Placebo 10.8 (9.6) 12.6 (10.1) 16.7% p ¼ 0.174 13.2 (9.8) 22.2% p ¼ 0.042*
Active 8.7 (8.3) 9.3 (8.6) 6.9% 9.0 (9.0) 3.5%
ECRB Placebo 10.1 (8.1) 11.5 (8.9) 12.7% p ¼ 0.461 11.5 (9.1) 12.7% p ¼ 0.342
Active 9.8 (8.7) 9.8 (8.3) 0.0% 9.4 (8.6) þ4.1%

Table III
NNT and effect size (Cohen's d) for key outcome measures

NNT Effect size

(50% improvement 0e14) Percentage change (0e14) Cohen's d

Responders (n) NNT (n) Active mean Placebo mean Pooled SD

Active Placebo

VAS pain (mean 7 days) 16 1 4.4 29.0 47.2 103.2 0.94

WOMAC Pain 16 2 3.6 32.5 20.6 55.1 0.96
Stiffness 17 6 2.5 34.6 26.8 96.2 0.64
Function 11 2 2.9 28.0 14.8 42.6 1.01
PainDETECT 12 1 3.6 22.3 15.3 38.4 0.98
PQAS Paroxysmal 24 3 1.9 26.4 63.5 139.0 0.65
Surface 21 8 3.1 28.5 65.5 108.8 0.86
Deep 24 2 1.8 45.9 80.6 211.5 0.60
ALF (time) Sit-to-stand 2 0 20 11.2 1.3 13.5 0.73
Walk 1 0 40 5.2 3.0 10.3 0.81
Stairs 2 1 40 12.2 4.7 18.2 0.93
ALF (pain) Sit-to-stand 17 9 5.0 45.0 10.58 70.1 0.49
Walk 20 7 3.1 60.9 18.0 75.0 1.05
Stairs 19 8 3.6 50.0 2.7 54.0 0.98
PPT e OA knee 13 8 8.0 38.1 11.6 43.1 0.61
CPT e OA knee 5 0 8.0 0.99 1.5 4.9 0.51
 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
and in local and widespread hyperalgesia. This demonstrates the
breadth of etoricoxib's efficacy.
There was a strong equivalence in improvements in self-report
and physical measures. For example, both mean pain (VAS) and
PPT at the index knee improved by around 22% for those taking
etoricoxib, whilst Placebo values changed minimally. Physical
function also improved for the Active group and declined for the
Placebo group, whether measured as self-report or physical test.
A recent study by Arendt-Nielsen et al.20 applied a randomized
cross-over design with 39 participants with OA knee to evaluate a
28-day course of etoricoxib 60 mg/day. This study found very
similar improvements in WOMAC-pain (29.4%) and function sub-
scores (27.9%) by Day 28, to that found in the current 14-day study.
They similarly tested locomotion tasks and found minimal change
in time taken but significant improvements in pain.
Although a previous cross-sectional study has reported a strong
correlation between WOMAC and ALF values27, in both the current
and Arendt-Nielsen20 studies the change in WOMAC was consid-
erably greater than for actual function. It may be that the rapidly
reduced pain and stiffness following NSAID use causes an over-
estimation of improved functional ease. That aside, the 10%
reduction in time taken to complete ALF tasks after just 2-weeks
NSAID intervention is on par with improvements in ALF time of
between 8% and 23% reported after 8-week exercise programmes16.
The current study has demonstrated that physical testing of loco-
motion, including report of pain during each task, is a useful
addition when evaluating pharmaceutical interventions.
As anticipated, alongside improvements in pain and function,
Active group participants reported improvements in both physical
and mental quality of life, whereas Placebo participants reported
declines in both. A brief intervention therefore had a clear impact
on quality of life, improving SF-36 physical sub-scores by nearly
13% in just 14 days. The mental sub-scale showed less of an increase
but this perhaps reflects the longer time-frame needed to re-
establish positive patterns of thinking and return to community
participation. These results suggest that changes to pain and
function are clearly associated with quality of life for people living
with chronic pain.
Previous studies have reported that 20e30% of individuals with
knee OA report neuropathic-type symptoms which are associated
with multi-modality hyperalgesia and higher levels of dis-
ability2,20,32e36. In the current study, 14% of all participants scored
as ‘positive neuropathic’ at baseline, with a further 31% scoring as
‘unclear’ according to PainDETECT. By Day 14 only 7.5% of Active
group participants scored as ‘positive neuropathic’ (Fig. 5) with
scores reducing by almost 24%. Changes in the PQAS neuropathic
(paroxsysmal and surface) subscores were even more dramatic,
improving by 37% and 29% respectively for the Active group. The
current study therefore indicates that a 14-day Cox-2 inhibitor
intervention has a significant influence on self-reported neuro-
pathic-like symptoms. It should be acknowledged that, although
PainDETECT is widely used to identify neuropathic-type symp-
toms2,4,15,25, specific criteria for diagnosing neuropathic pain in OA
have yet to be established.
Widespread hyperalgesia may be an important indicator of
centrally-driven pain20,35,37,38. In the current study 14 days NSAID
intervention resulted in small but significant improvements in PPT
beyond the index knee (upper limb ECRB site 12.4%) whilst wors-
ening in the Placebo group by 7.5%. Additionally, the small but
significant improvements in PcPT indicate that etoricoxib modu-
lates mechanical and tactile hyperalgesia beyond the site of pa-
thology. These findings support those of Arendt-Nielsen et al.20
who investigated etoricoxib over a longer time frame and found
improvements in PainDETECT (22%) and widespread PPT (8%)
comparable to the current study.
1789
Arendt-Nielsen also used additional QST measures of centrally-
mediated pain, including temporal summation, and found signifi-
cant improvements20. In the current study, cold hyperalgesia was
used, as it has been reported to be a key indicator of persistent or
severe pain that may be centrally-mediated18,35,37,39. The findings
that CPT improved by nearly 10% at the index knee for the Active
group whilst declining by more than 12% in the Placebo group
suggests an influence on multi-modal hyperalgesia. Although there
was minimal change at either distant site, a longer intervention
may have shown greater improvements.
It has been suggested that, in addition to its peripheral pathway,
etoricoxib has the potential to act centrally as a result of its capacity
to cross the blood brain barrier11,20. It may appear surprising that a
short course of 60 mg etoricoxib reduced both local and wide-
spread hyperalgesia and neuropathic-like symptoms in an estab-
lished pain presentation. However, development of chronic pain
will involve both local and central factors. In OA, widespread pain
sensitisation is likely to be largely driven by persistent peripheral
input. In the current study, participants’ baseline local hyperalgesia
was greater than at distant sites and the improvement following
etoricoxib was greatest locally. This may indicate that inhibition of
the peripheral COX pathway rapidly reduced local nociceptive drive
thereby influencing more generalised sensitisation.
The study had a number of factors that limit generalisability.
Although appropriately powered, the sample size was relatively
small and homogenous, recruited from a single Australian city. As
might be anticipated with a community-recruited cohort who
needed to attend several locations for testing, few participants
exhibited severe levels of pain or disability.
However, the combined evidence from this and other studies19
strongly supports the efficacy in knee OA of a selective Cox-2 in-
hibitor. The current study has shown that a short 14-day course of
etoricoxib is sufficient to have a significant impact on pain,
neuropathic-like symptoms and both local and widespread multi-
modality hyperalgesia.
Conclusions
A 14-day course of etoricoxib (60 mg/day) reduced knee pain by
approximately 30%, whether assessed by self-report or QST. Both
subjective report and locomotion testing showed that function was
also significantly improved. The finding that neuropathic-type pain
quality and widespread mechanical and local cold hyperalgesia
were also improved warrants further assessment of the impact of
NSAIDs on centrally-mediated pain.
Author contributions
All authors were involved in all aspects of study conception, design,
funding, recruitment and data collection, data analysis and inter-
pretation and final article drafting and editing.
Competing interests
The authors have no conflicts of interest to declare.
Role of the funding source
The study was funded by an Investigator Initiated Study Programme
grant from Merck Inc. The application was reviewed in line with the
normal funding process. Merck Inc provided the drugs and an un-
restricted grant to run the study but was not involved in data
analysis, data interpretation, or in the preparation of the manuscript.
Acknowledgements
The authors would like to acknowledge: Merck Inc for providing
the drugs used for the study and providing an unrestricted grant to
1790 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
conduct the study; Ms Lisa Webster for study administrative and
organizational assistance.
Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.joca.2017.07.009.
References
1. Salmon JH, Rat AC, Sellam J, Michel M, Echard JP, Guillemin F,
et al. Economic impact of lower-limb osteoarthritis world-
wide: a systematic review of cost-of-illness studies. Osteoar-
thritis Cartilage 2016;24(9):1500e8.
2. Moreton BJ, Tew V, das Nair R, Wheeler M, Walsh DA,
Lincoln NB. Pain phenotype in patients with knee osteoar-
thritis: classification and measurement properties of painDE-
TECT and self-report leeds assessment of neuropathic
symptoms and signs scale in a cross-sectional study. Arthritis
Care Res 2015;67(4):519e28.
3. Kittelson AJ, Stephens-Lapsley JE, Schmiege SJ. Determination
of pain phenotypes in knee osteoarthritis: a latent class anal-
ysis using data from the osteoarthritis initiative. Arthritis Care
Res 2016;68(5):612e20.
4. Thakur M, Dickenson AH, Baron R. Osteoarthritis pain: noci-
ceptive or neuropathic? Nat Rev Rheumatol 2014;10:374e80.
5. Wright A, Benson HA, Will R, Moss P. Cold pain threshold
identifies a subgroup of patients with knee osteoarthritis
that present with multimodality hyperalgesia and elevated
pain levels. Clin J Pain 2016, http://dx.doi.org/10.1097/AJP.
0000000000000458. e-pub ahead of print.
6. Lluch E, Torres R, Nijs J, Van Oosterwijck J. Evidence for central
sensitization in patients with osteoarthritis pain: a systematic
literature review. Eur J Pain 2014;18(10):1367e75.
7. Goldenberg DL, Clauw DJ, Fitzcharles M-A. New concepts in
pain research and pain management of the rheumatic diseases.
Seminars Arthritis Rheumatism 2011;41(3):319e34.
8. Song GG, Seo YH, Kim JH, Choi SJ, Ji JD, Lee YH. Relative efficacy
and tolerability of etoricoxib, celecoxib, and naproxen in the
treatment of osteoarthritis : a Bayesian network meta-analysis
of randomized controlled trials based on patient withdrawal.
Z Rheumatol 2016;75(5):508e16.
9. Hunter TS, Robison C, Gerbino PP. Emerging evidence in NSAID
pharmacology: important considerations for product selection.
Am J Manag Care 2015;21(7 Suppl):s139e47.
10. Samad TA, Moore KA, Sapirstein A, Billet S, Allchorne A,
Poole S, et al. Interleukin-1beta-mediated induction of Cox-2
in the CNS contributes to inflammatory pain hypersensitiv-
ity. Nature 2001;410:471e5.
11. Renner B, Zacher J, Buvanendran A, Walter G, Strauss J,
Brune K. Absorption and distribution of etoricoxib in plasma,
CSF, and wound tissue in patients following hip surgery-a pilot
study. Schmiedeb Arch Pharmacol 2010;381:127e36.
12. Hochberg MC, Altman RD, April KT, Benkhalti M, Guyatt G,
McGowan J, et al. American College of Rheumatology 2012
Recommendations for the use of non-pharmacologic and
pharmacologic therapies in osteoarthritis of the hand, hip, and
knee. Arthritis Care Res 2012;64(4):465e74.
13. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW.
Validation study of WOMAC: a health status instrument for
measuring clinically important patient relevant outcomes to
anti-rheumatic drug therapy in patients with osteoarthritis of
the hip or knee. J Rheumatol 1988;15(2):1833e40.
14. Laine L, White WB, Rostom A, Hochberg M. COX-2 selective
inhibitors in the treatment of osteoarthritis. Seminars Arthritis
Rheumatism 2008;38(3):165e87.
15. Haanpaa M, Attal N, Backonja M, Baron R, Bennett M,
Bouhassira D. NeuPSIG guidelines on neuropathic pain. Pain
2011;152(1):14e27.
16. McCarthy CJ, Mills PM, Pullen R, Roberts C, Silman A,
Oldham JA. Supplementing a home exercise programme with a
class-based exercise programme is more effective than home
exercise alone in the treatment of knee osteoarthritis. Rheu-
matol (Oxford) 2006;43(7):880e6, http://dx.doi.org/10.1093/
rheumatology/keh188.
17. Bajaj P, Bajaj P, Graven-Nielsen T, Arendt-Nielsen L. Osteoar-
thritis and its association with muscle hyperalgesia: an
experimental controlled study. Pain 2001;93(2):107e14.
18. Arendt-Nielsen L, Nie H, Laursen MB, Laursen BS, Madeleine P,
Simonsen OH, et al. Sensitization in patients with painful knee
osteoarthritis. Pain 2010;149(3):573e81, http://dx.doi.org/
10.1016/j.pain.2010.04.003.
19. Moss P, Knight E, Wright A. Subjects with knee osteoarthritis
exhibit widespread hyperalgesia to pressure and cold. PLoS
One 2016;11(1)e0147526.
20. Arendt-Nielsen L, Egsgaard LL, Petersen KK. Evidence for a cen-
tral mode of action for etoricoxib (COX-2 inhibitor) in patients
with painful knee osteoarthritis. Pain 2016;157(8):1634e44.
21. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al.
Development of criteria for the classification and reporting of
osteoarthritis. Classification of osteoarthritis of the knee. Diag-
nostic and therapeutic criteria Committee of the American Rheu-
matism Association. Arthritis Rheumatism 1986;29:1039e49.
22. Jinks C, Jordan K, Croft P. Measuring the population impact of
knee pain and disability with the Western Ontario and
McMaster Universities Osteoarthritis Index (WOMAC). Pain
2002;100(1e2):55e64.
23. Gandhi SK, Salmon JW, Zhao SZ, Lambert BL, Gore PR,
Conrad K. Psychometric evaluation of the 12-item short-form
health survey (SF-12) in osteoarthritis and rheumatoid
arthritis clinical trials. Clin Ther 2001;23(7):1080e98.
24. Freynhagen R, Baron R, Gockel U, Tolle T. PainDETECT: a new
screening questionnaire to identify neuropathic components
in patients with back pain. Curr Med Res Opin 2006;22(10):
1911e20.
25. Freynhagen R, Tolle T, Gockel U, Baron R. The painDETECT
project: far more than a screening tool on neuropathic pain.
Curr Med Res Opin 2016;32(6):1033e57.
26. Victor TW, Jensen MP, Gammaitoni AR, Gould EM, White RE,
Galer BS. The dimensions of pain quality: factor analysis of the
Pain Quality Assessment Scale. Clin J Pain 2008;24:550e5,
http://dx.doi.org/10.1097/AJP.0b013e31816b1058.
27. McCarthy CJ, Oldham JA. The reliability, validity and respon-
siveness of an aggregated locomotor function (ALF) score in
patients with osteoarthritis of the knee. Rheumatology 2004;43:
514e7, http://dx.doi.org/10.1093/rheumatology/keh081.
28. Moss P, Sluka K, Wright A. The initial effects of knee joint
mobilization on osteoarthritic hyperalgesia. Man Ther 2007;12:
109e18, http://dx.doi.org/10.1016/j.math.2006.02.009.
29. Wylde V, Palmer S, Learmonth ID, Dieppe P. Test-retest reli-
ability of Quantitative Sensory Testing in knee osteoarthritis
and healthy participants. Osteoarthritis Cartilage 2011;19(6):
655e8, http://dx.doi.org/10.1016/j.joca.2011.02.009.
30. Fruhstorfer H, Lindblom U, Schmidt WC. Method for quanti-
tative estimation of thermal thresholds in patients. J Neurol
Neurosurg Psychiatry 1976;39:1071e5.
31. Bingham CO, Sebba AL, Rubin BR, Ruoff GE, Kremer J, Bird S,
et al. Efficacy and safety of etoricoxib 30 mg and celecoxib 200
 P. Moss et al. / Osteoarthritis and Cartilage 25 (2017) 1781e1791
mg in the treatment of osteoarthritis in two identically
designed, randomized, placebo-controlled, non-inferiority
studies. Rheumatology 2007;46(3):496e507.
32. Moss P, Benson HA, Will R, Wright A. Patients with knee oste-
oarthritis who score highly on the PainDETECT questionnaire
present with multi-modality hyperalgesia, increased pain and
impaired physical function. Clin J Pain 2017, http://dx.doi.org/
10.1097/AJP.0000000000000504. e-pub ahead of print.
33. Hochman JR, Gagliese L, Davis AM, Hawker GA. Neuropathic
pain symptoms in a community knee OA cohort. Osteoarthritis
Cartilage 2011;19(6):647e54.
34. Ohtori S, Orita S, Yamashita M, Ishikawa T, Ito T, Shigemura T,
et al. Existence of a neuropathic pain component in patients
with osteoarthritis of the knee. Yonsei Med J 2012;53(4):
801e5.
1791
35. Hochman JR, Davis AM, Elkayam J, Gagliese L, Hawker GA.
Neuropathic pain symptoms on the modified painDETECT
correlate with signs of central sensitization in knee osteoar-
thritis. Osteoarthritis Cartilage 2013;21(9):1236e42.
36. Soni A, Batra RN, Gwilym SE, Spector TD, Hart DJ, Arden NK,
et al. Neuropathic features of joint pain: a community-based
study. Arthritis Rheumatism 2013;65(7):1942e9.
37. Sterling M, Jull G, Kennardy J. Physical and psychological fac-
tors maintain long-term predictive capacity post whiplash
injury. Pain 2006;122(1e2):102e8.
38. Woolf CJ. Central sensitization: implications for the diagnosis
and treatment of pain. Pain 2011;152(3):S2eS15.
39. Goldsmith R, Wright C, Bell SF, Rushton A. Cold hyperalgesia as
a prognostic factor in whiplash associated disorders: a sys-
tematic review. Man Ther 2012;17(5):402e10.