Content uploaded by Samantha Coulson
Author content
All content in this area was uploaded by Samantha Coulson
Content may be subject to copyright.
RESEARCH ARTICLE
Green-lipped mussel (Perna canaliculus) extract efficacy in knee
osteoarthritis and improvement in gastrointestinal dysfunction:
a pilot study
Samantha Coulson •Phillip Vecchio •
Helen Gramotnev •Luis Vitetta
Received: 2 January 2012 / Accepted: 8 February 2012 / Published online: 26 February 2012
ÓSpringer Basel AG 2012
Abstract
Objective Clinical data demonstrating efficacy for
nutraceutical compounds marketed for the symptom relief
of osteoarthritis (OA) have been largely contentious. Fur-
thermore, no association has been linked between clinical
trial inconsistencies and gastrointestinal (GI) dysfunction.
The aim of this study was to primarily investigate the
efficacy of a high-dose New Zealand green-lipped mussel
(GLM) extract in patients diagnosed with OA of the knee
and concurrently assess GLM impact on GI function.
Methods An open label, single group allocation study
was conducted, that administered 3,000 mg/day of GLM
extract over 8 weeks to 21 subjects diagnosed with knee
OA. Outcome measures were scored using the WOMAC,
the Lequesne algofunctional index, and the Gastrointestinal
Symptom Rating Scale (GSRS) tools. An intention-to-treat
analysis was employed and subject data collected at T
0
,T
4
and T
8
weeks.
Results Paired ttests showed significant improvement
for the Lequesne, WOMAC (p\0.001) and GSRS (p=
0.005) scores. A repeated measures ANOVA analysis
showed significant improvement in scores for the Lequesne
(F=20.317, p\0.001), WOMAC (F=28.383, p\
0.001) and the GSRS (F=9.221, p=0.002).
Conclusion Green-lipped mussel significantly improved
knee joint pain, stiffness and mobility. We report for the
first time that the administration of GLM extract also sig-
nificantly improved GI symptoms by 49% in OA patients.
Given that GI dysfunction is linked to analgesic medication
use, we further conclude that the therapeutic efficacy of the
GLM extract used was possibly correlated to its effects on
GI function by improving GSRS scores from baseline.
Results from this trial highlight the requisite for further
clinical investigations of gastrointestinal tract function in
OA patients.
Key words Green-lipped muscle extract Osteoarthritis
Gastrointestinal function
Introduction
Osteoarthritis (OA) is the most prevalent form of arthritis
and commonly targets the joints of the knee and less fre-
quently joints of the hip, shoulder, spine (most commonly
zygapophyseal, apophyseal, or facet joints of the spine,
especially in the mid and lower cervical spine, and in the
lower lumbar spine, L3 to L5) hands and toes (Buchanan
and Kean 2002a,b; Wieland et al. 2005). With no known
cure or proven disease-modifying therapy, the first-choice
recommended pharmacological treatment for mild to
moderate pain of the knee or hip is acetaminophen, with a
maximum dose of up to 4 g/day (Buchanan and Kean
2002c). In the absence of a beneficial response, or pain and
inflammatory sequelae, alternate therapy with an NSAID is
indicated. Gastrointestinal (GI) safety however, may be
compromised with long-term intake of these medications
as they are associated with gastric or peptic ulcers (Garcia-
Rodriguez et al. 1998; Roderiguez and Hernandez-Diaz
S. Coulson H. Gramotnev L. Vitetta (&)
Centre for Integrative Clinical and Molecular Medicine,
School of Medicine, The University of Queensland and
The Princess Alexandra Hospital,
Brisbane, QLD, Australia
e-mail: l.vitetta@uq.edu.au
P. Vecchio
The Princess Alexandra Hospital, Rheumatology Department,
199 Ipswich Road, Woolloongabba,
Brisbane, QLD 4102, Australia
Inflammopharmacol (2012) 20:71–76
DOI 10.1007/s10787-012-0128-6 Inflammopharmacology
123
2001), irritable bowel syndrome (IBS) (Locke et al. 2000)
including dyspepsia (indigestion as discomfort experienced
in the upper abdomen) (Rahme et al. 2002), chronic con-
stipation (Chang et al. 2007), diarrhoea (Etienney et al.
2003) and morphological changes such as intestinal
ulceration and increased GI inflammation and permeability
(Scarpignato and Hunt 2010; Roderiguez and Hernandez-
Diaz 2001).
Approximately 40% of the OA patients will self-medi-
cate with dietary supplements, often using them exclusively
(22%) or in combination (16%) with pharmaceutical med-
ications i.e. analgesics (Armstrong et al. 2011). Clinical
efficacy of nutraceuticals for the treatment of OA however
remains collectively inconsistent, impacting the recom-
mendation of their routine use. Popular supplements
marketed towards OA include glucosamine, chondroitin
and green-lipped mussel (GLM) extract (Perna canalicu-
lus), and although clinical trial results variably support their
use in OA, methodological flaws and inconsistent results
preclude the recommendation of such compounds as an
alternative treatment option (Brien et al. 2008; Wandel et al.
2010). In addition to analgesic use, GI function and integ-
rity may possibly be compounded further by the presence of
inflammation (Peuhkuri et al. 2010) and by the underlying
rheumatic disease (Chong and Wang 2008).
The aim of this study was to investigate the therapeutic
effect of a New Zealand GLM extract (GlycOmega
TM
PLUS) in patients diagnosed with OA of the knee(s) in
relation to knee pain, stiffness and function. Furthermore,
we examined GI symptoms and analgesic medication
intake, and the interaction these may have on the therapeutic
outcome of the GLM. It is therefore plausible to hypothesise
that clinical efficacy of GLM in the treatment of knee OA
may be associated with improved GI integrity and function.
Understanding factors that impact digestion, absorption,
metabolism and the health response process is crucial for the
proper interpretation of biological responses, or the lack
thereof, in clinical research (Possemiers et al. 2011).
Patients and methods
Participants
The study group comprised 23 patients with knee OA (9
males, 14 females) who satisfied the inclusion and exclu-
sion criteria. Patients were recruited if their knee OA was
confirmed by a Rheumatologist and they were not taking
any form of herbal/multivitamin/nutritional supplements
4 weeks prior to recruitment. Patients were excluded if
they had uncontrolled systemic disease, were pregnant or
breast feeding, or had allergies/intolerances to shellfish.
Participants were recruited from the Rheumatology Clinic
at the Princess Alexandra Hospital, Brisbane from July
2010 to March 2011.
All 23 patients recruited satisfied the decision tree for-
mat of the ACR classification criteria for idiopathic clinical
OA of the knee (Altman et al. 1986). The ACR clinical
criteria for knee OA were fulfilled by the uniform presence
of knee pain and at least 3 of the following 6 criteria: C
50 years of age, morning stiffness lasting B30 min, crep-
itus on movement, bony tenderness, bony enlargement and
no palpable warmth of synovium.
Each patient received written and verbal explanations
regarding their involvement in the study before signing
informed consent. The study protocol was in compliance
with the Helsinki Declaration and was approved by the
Human Research Ethics Committees of The University of
Queensland and The Princess Alexandra Hospital.
Procedures
Patient demographic data and medical history were
obtained at baseline (T
0
). Anthropometric measurements,
blood pressure and outcome measures were performed at
T
0
,T
4
and T
8
weeks, and blood samples for safety measures
were collected at T
0
and T
8
.
Anthropometric measurements
Height was measured with the subject standing barefoot
using a body meter measuring tape with wall stop. Body
mass was measured using calibrated scales and body mass
index (BMI) was calculated using the formula—mass
divided by height squared (kg/m
2
); waist hip ratios were
calculated by dividing the waist circumference (cm) by the
hip circumference (cm).
Outcome measures
The Western Ontario McMaster Universities Arthritis Index
(WOMAC) (Bellamy 2002) and the Lequesne algofunc-
tional index (Lequesne et al. 1987; Lequesne and Maheu
2003) were designated the primary outcome measures and
were administered via interview format by the clinical
researcher at each participant’s visit (T
0
,T
4
and T
8
). The
WOMAC is a validated questionnaire designed for the
assessment of lower extremity pain and function in OA of
the knee or hip by assessing severity of knee pain (5
questions), stiffness (2 questions) and limitation of physical
function (17 questions) (Bellamy 2002). Maximum scores
of severity for each subscale are 20, 8 and 68, respectively.
The maximum total score of severity for the WOMAC is 96.
The Lequesne algofunctional index includes the measure-
ment of pain (five questions), walking distance (one
question), and activities of daily living (four questions),
72 S. Coulson et al.
123
with versions available for the hip and knee (Lequesne et al.
1987; Lequesne and Maheu 2003). Scores for each question
are added together to provide a combined disease severity
score. Scores of 1–4 are classified as mild OA, 5–7 mod-
erate, 8–10 severe, 11–13 very severe, and 14 as extremely
severe OA. Secondary outcome measures utilised the Gas-
trointestinal Symptom Rating Score (GSRS) questionnaire
(Svedlund et al. 1988) and the SF-12v2
TM
health survey
(Tucker et al. 2010) via interview format. GI function was
assessed using the validated GSRS questionnaire which
contains 15 items designed to evaluate abdominal pain,
gastro-oesophageal reflux, indigestion, diarrhoea and con-
stipation. Each question is rated 0 to 3 in regards to
intensity, frequency, duration, request for relief and impact
on social performance. General quality of life was assessed
using the SF-12 that includes 12 questions with results
expressed in terms of 2 meta-scores: the physical compo-
nent summary (PCS) and the mental component summary
(MCS) (Ware et al. 1996; Tucker et al. 2010).
Assessment of safety
Blood pathology was performed to assess safety (full blood
count, electrolytes and liver function tests). The inflam-
matory markers, C-reactive protein (CRP) and erythrocyte
sedimentation rate (ESR) were also included. Adverse
events were documented by patients in diaries provided
and events were recorded.
Intervention
Patients were allocated to 3,000 mg/day (3 9500 mg
b.i.d) of a proprietary blend of freeze-dried greenshell
mussel meat (Perna canaliculus), GlycOmega
TM
PLUS
(Aroma NZ Ltd), which was dispensed in opaque white
bottles for the 8-week duration. Compliance was monitored
from participant recordings of daily intake of GLM in
diaries provided.
Daily pain and rescue medication recordings
Participants were asked to complete two diaries that
recorded daily intake of GLM, daily knee pain experienced
on a 5 point Likert scale (T
0
–T
4
and T
4
–T
8
) and daily
rescue medication intake for knee pain.
Statistical analysis
The data in this clinical trial were treated according to an
intention-to-treat analysis. Participants were included in the
analysis if they had returned at least one diary (e.g. visit 2).
Paired samples ttests were conducted for significant or
marginally significant interactions. These were performed
to investigate the difference in pre- and post-supplemen-
tation scores. Repeated measures analysis of variance
(ANOVA) was conducted to investigate the changes
between the time periods. The significance level was set at
p\0.05. Normality tests were performed for all the
scoring variables to confirm the assumption of normality,
justifying the use of ttests and analysis of variance tests.
Results
Patient characteristics
Three patients withdrew from the trial with 1 patient pro-
viding their first diary, resulting in a total of 21 participants
(8 males and 13 females) with an age range from 41 to
87 years (mean ±SD, 61.1 ±12.2 years) that provided
data for an evaluable analysis. Weight and BMI did not
change significantly between time points (mean ±SD,
T
0
=96.9 ±27.1 kg; T
8
=96.7 ±27.5 kg) and (mean ±
SD, T
0
=34.0 ±9.0 kg/m
2
;T
8
=34.5 ±9.2 kg/m
2
), res-
pectively. Waist to hip ratio remained unchanged between
T
0
and T
8
(mean ±SD, 0.9 ±0.1). Current medications
were allowed to continue being taken through the trial
period and included anti-cholesterol agents (n=6), anti-
diabetic agents (n=2), diuretic (n=1), calcium channel
blockers (n=3), beta-blockers (n=1), angiotensin II
receptor agonists (n=3), ACE inhibitor (n=3), anti-
inflammatory medications (n=8), proton pump inhibitors
(n=7), anti-depressant medications (n=9), thyroid
medications (n=3), anti-histamine medication (n=2),
anti-coagulant medications (n=1), gout medication
(n=1) and herpes medication (n=1). Recent intake of
antibiotics was reported by 7 patients, while 12 patients
reported influenza vaccinations.
Dropouts
There were 3 dropouts in total from the original 23 patients
recruited, 2 males and 1 female. These participants were
lost to follow-up due to adverse events (2 males) and lack
of interest in the trial (1 female). One of the male dropouts
did return his first diary and data were included for analysis
as intention-to-treat.
Outcome measures
Paired ttests were performed on the Lequesne, WOMAC
(total and subscores), GSRS, and SF-12 (PCS and MCS
scores) between T
0
and T
4
,T
4
and T
8
, and T
0
and T
8
.
Lequesne and WOMAC (total) showed significant changes
between all intervals (p\0.001) with WOMAC subscores
of pain (p\0.001), stiffness (p=0.002) and physical
Green-lipped mussel extract for knee osteoarthritis 73
123
function (p\0.001) indices all statistically significant;
GSRS showed significant changes from T
0
to T
4
(p=
0.004) and from T
0
to T
8
(p=0.005). MCS scores signifi-
cantly changed from T
0
to T
4
(p=0.012) only and PCS
scores were not significant. The percentage change in scores
demonstrated significant efficacy on all outcome measures
(Table 1). A repeated measures ANOVA that was per-
formed over the three time points gave significant results for
all scores except PCS (F=0.364, p=0.649). Lequesne
and WOMAC score differences were particularly strong
(p\0.001) as was the GSRS (p=0.002) (Table 1).
Rescue medication
Rescue medication for knee pain associated with OA was
allowed and recorded daily by the participants. Of the 21
patients, 7 did not use any rescue medication over the
8-week trial period, while 14 did and this included panadol
(paracetamol dose 500 mg), panadol osteo (paracetamol
665 mg), panadeine forte (paracetamol 500 mg ?codeine
phosphate 30 mg), prednisone and various NSAIDs (aspi-
rin 100 mg; indomethacin 25 mg; diclofenac sodium
50 mg; ibuprofen 200 mg].
Safety and adverse events
All blood parameters remained normal throughout the trial.
CRP and ESR values did not significantly change over the
8-week period (mean ±SD T
0
=4.08 ±3.49 mg/L; T
8
=
3.31 ±2.69 mg/L) and (mean ±SD T
0
=14.59 ±8.61
mm/h; T
8
=14.21 ±7.66 mm/h), respectively. Blood
pressure remained stable throughout duration of the trial
(mean ±SD T
0
=systolic 130.4 ±18.3 mm Hg and
diastolic 80.8 ±11.8 mm Hg; T
8
=systolic 125.8 ±
17.3 mmHg and diastolic 76.0 ±11.3 mm Hg). Adverse
events included reflux (n=1); abdominal pain, reflux and
diarrhoea (n=1) and gout (n=2).
Discussion
We confirm that a standardized high dose (3,000 mg/day)
GLM preparation used in this study was efficacious in
treating OA symptoms of the knee. The improvement in
symptom management from baseline was robustly signifi-
cant for knee pain, stiffness and mobility.
A credible biological anti-inflammatory mechanism of
action is reported for GLM in arthritic animal models,
which have also demonstrated gastro-protective effects
against NSAIDs-induced GI damage (Rainsford and
Whitehouse 1980). A GLM-induced gastro-protective
mechanism was noted in our sample group. Despite plau-
sible mechanisms for the treatment of OA as an inhibitor of
the lipoxygenase and cyclo-oxygenase pathways, a recent
systematic review calls for further methodical investiga-
tions to provide clear evidence of efficacy, particularly
Table 1 Change in primary and secondary outcome measures from baseline (T
0
) to week 4 (T
4
) and from baseline to week 8 (T
8
)
Outcome Time Mean ±SD ES-Cohen’s d(correlation) tvalue 95% CI pvalue
Lequesne Index T
0
13.16 5.16
T
4
10.30 6.04 1.021 (0.875) 3.94 1.21, 3.94 0.001
T
8
9.13 5.79 1.413 (0.864) 5.72 2.38, 5.12 \0.001
WOMAC (total) T
0
41.63 19.75
T
4
30.0 17.30 1.074 (0.829) 4.10 5.08, 15.72 0.001
T
8
22.80 17.57 1.957 (0.867) 7.78 12.87, 22.33 \0.001
WOMAC (pain) T
0
8.64 3.99
T
4
4.90 3.34 1.745 (0.829) 5.34 2.13, 4.87 \0.001
T
8
3.20 3.21 1.805 (0.649) 7.33 3.72, 6.68 \0.001
WOMAC (stiffness) T
0
3.82 1.74
T
4
2.85 1.69 0.669 (0.643) 3.01 0.30, 1.70 0.007
T
8
2.35 1.63 0.816 (0.428) 3.63 0.63, 2.37 0.002
WOMAC (physical) T
0
29.18 14.96
T
4
22.25 12.91 0.879 (0.840) 3.20 2.04, 9.76 0.005
T
8
17.25 13.70 1.877 (0.901) 7.38 7.81, 13.99 \0.001
GSRS T
0
7.41 6.57
T
4
3.15 3.92 0.795 (0.478) 3.28 1.39, 6.31 0.004
T
8
3.45 3.14 0.843 (0.532) 3.21 1.23, 5.87 0.005
GSRS gastrointestinal symptom rating score, ES effect size (Cohen’s d), Paired ttests between T
0
and T
4
scores and T
0
and T
8
scores and 95% CI
for the differences
74 S. Coulson et al.
123
optimal dosing (Brien et al. 2008; Whitehouse et al. 1997).
In that review, doses in the range of 1,050–1,150 mg/day
were reported. In this study, we report that a high dose of
3,000 mg/day was efficacious and attenuated pain, stiffness
and joint mobility. Efficacy was already established at
4 weeks of supplementation with significant reductions in
pain scores and was maintained when assessed at week 8
(Table 1). The high dose of the GLM employed in this
pilot trial may provide additional information for future
controlled phase IIb dosing studies.
We acknowledge that a weakness of this study is the lack
of a placebo comparator (Zhang et al. 2008). However, as a
preliminary study, it provides additional clinical evidence to
support the efficacy of GLM and to formulate a hypothesis
that has not yet been previously investigated, namely GI
dysfunction in OA. GI dysfunction in clinical trials inves-
tigating nutraceutical efficacy can be a significant limiting
factor. Prescribed analgesic medications can disrupt GI
function (Wolfe et al. 1999; Chang et al. 2007; Etienney
et al. 2003; Scarpignato and Hunt 2010; Roderiguez and
Hernandez-Diaz 2001) and hence this may significantly
affect nutraceutical metabolism, absorption and efficacy.
We have demonstrated that GLM was efficacious in
improving OA symptoms and function, and that it may
have important GI sequelae that assist with GLM utiliza-
tion and hence efficacy. GI dysfunction was significantly
improved with study progress in all subjects.
Ethical approval and clinical trial registration
Approval for this prospective study was obtained from the
Ethics Committee of The University of Queensland and
Princess Alexandra Hospital Human Research Ethics
Committees. The clinical trial was registered with the
Australian and New Zealand Clinical Trail Registry
(ANZCTRN: 12611000517976).
Acknowledgments We thank Shoshannah Beck and Janet Schloss
from The University of Queensland for reading the manuscript and
assisting with data entry and subject recruitment.
Conflicts of interest Funding and study medication for the project
was received from the clinical trial sponsor Aroma New Zealand Ltd.
The sponsor had no involvement in the collection, analysis or inter-
pretation of the data; writing the report; or the decision to submit the
paper for publication.
References
Altman R, Asch E, Bloch D et al (1986) Development of criteria for
the classification and reporting of osteoarthritis. Classification of
osteoarthritis of the knee. Diagnostic and Therapeutic Criteria
Committee of the American Rheumatism Association. Arthritis
Rheum 29(8):1039–1049
Armstrong AR, Thie
´baut SP, Brown LJ (2011) Australian adults use
complementary and alternative medicine in the treatment of
chronic illness: a national study. Aust NZ J Public Health
35:384–390
Bellamy N (2002) WOMAC: a 20-year experiential review of a
patient-centered self-reported health status questionnaire. J Rheu-
matol 29(12):2473–2476
Brien S, Prescott P, Coghlan B et al (2008) Systematic review of the
nutritional supplement Perna canaliculus (green-lipped mussel)
in the treatment of osteoarthritis. QJM 101:167–179
Buchanan WW, Kean WF (2002a) Osteoarthritis: II: pathology and
pathogenesis. Inflammopharmacol 10:19–48
Buchanan WW, Kean WF (2002b) Osteoarthritis: III: radiological and
clinical definition. Inflammopharmacol 10:53–78
Buchanan WW, Kean WF (2002c) Osteoarthritis IV: clinical thera-
peutic trials and treatment. Inflammopharmacol 10:75–149
Chang J, Locke GR, Schleck CD et al (2007) Risk factors for chronic
constipation and a possible role of analgesics. Neurogastroen-
terol Motil 19:905–911
Chong VH, Wang CL (2008) Higher prevalence of gastrointestinal
symptoms among patients with rheumatic disorders. Singapore
Med J 49:419–424
Etienney I, Beaugerie L, Viboud C et al (2003) Non-steroidal anti-
inflammatory drugs as a risk factor for acute diarrhoea: a case
crossover study. Gut 52:260–263
Garcia-Rodriguez LA, Cattaruzzi C et al (1998) Risk of hospitaliza-
tion for upper gastrointestinal tract bleeding associated with
ketorolac, other nonsteroidal anti-inflammatory drugs, calcium
antagonists, and other antihypertensive drugs. Arch Intern Med
158:33–39
Lequesne MG, Maheu E (2003) Clinical and radiological evaluation
of hip, knee and hand osteoarthritis. Aging Clin Exp Res
15(5):380–390
Lequesne MG, Mery C, Samson M et al (1987) Indexes of severity for
osteoarthritis of the hip and knee. Validation value in comparison
with other assessment tests. Scand J Rheumatol Suppl 65:85–89
Locke GR, Zinsmeister AR, Talley NJ et al (2000) Risk factors for
irritable bowel syndrome: role of analgesics and food sensitiv-
ities. Am J Gastroenterol 95:905–911
Peuhkuri K, Vapaatalo H, Korpela R (2010) Even low-grade
inflammation impacts on small intestinal function. World J
Gastroenterol 16:1057–1062
Possemiers S, Bolca S, Verstraete W et al (2011) The intestinal
microbiome: a separate organ inside the body with the metabolic
potential to influence the bioactivity of botanicals. Fitoterapia
82:53–66
Rahme E, Pettitt D, LeLorier J (2002) Determinants and sequelae
associated with utilization of acetaminophen versus traditional
nonsteroidal antiinflammatory drugs in an elderly population.
Arthritis Rheum 46:3046–3054
Rainsford KD, Whitehouse MW (1980) Gastroprotective and anti-
inflammatory properties of green lipped mussel (Perna canalic-
ulus) preparation. Arzneimittelforschung 30:2128–2132
Roderiguez GL, Hernandez-Diaz S (2001) Risk of upper gastrointes-
tinal complications among users of acetaminophen and non-
steroidal anti-inflammatory drugs. Epidemiology 12:570–576
Scarpignato C, Hunt RH (2010) Nonsteroidal antiinflammatory drug-
related injury to the gastrointestinal tract: clinical picture,
pathogenesis and prevention. Gastroenterol Clin N Am
39:433–464
Svedlund J, Sjodin I, Dotevall G (1988) GSRS: a clinical rating scale
for gastrointestinal symptoms in patients with irritable bowel
syndrome and peptic ulcer disease. Dig Dis Sci 33:129–134
Tucker G, Adams R, Wilson D (2010) New Australian population
scoring coefficients for the old version of the SF-36 and SF-12
health status questionnaires. Qual Life Res 19:1069–1076
Green-lipped mussel extract for knee osteoarthritis 75
123
Wandel S, Juni P, Tendal B et al (2010) Effects of glucosamine,
chondroitin, or placebo in patients with osteoarthritis of hip or
knee: network meta-analysis. BMJ 341:c4675
Ware JE, Kosinski M, Keller SD (1996) A 12-Item Short-Form
Health Survey: construction of scales and preliminary tests of
reliability and validity. Med Care 34(3):220–233
Whitehouse MW, Macrides TA, Kalafatis N et al (1997) Anti-
inflammatory activity of a lipid fraction (lyprinol) from the NZ
green-lipped mussel. Inflammopharmacology 5:237–246
Wieland HA, Michaelis M, Kirschbaum BJ et al (2005) Osteoarthri-
tis—an untreatable disease? Nat Rev Drug Discov 4:331–344
Wolfe MM, Lichtenstein DR, Singh G (1999) Gastrointestinal
toxicity of nonsteroidal antiinflammatory drugs. N Engl J Med
340:1888–1899
Zhang W, Moskowitz RW, Nuki G et al (2008) The placebo effect
and its determinants in osteoarthritis: meta-analysis of random-
ised controlled trials. Ann Rheum Dis 67:1716–1723
76 S. Coulson et al.
123
