Oral herbal medicines marketed in Brazil for the treatment of osteoarthritis: A systematic review and meta-analysis: Herbal medicines for treatment of osteoarthritis

Abstract

Herbal medications are commonly used to manage symptoms associated with osteoarthritis (OA). This systematic review evaluated the effectiveness and safety of oral medications used in Brazil for the treatment of OA. Randomized clinical trials involving adults with OA treated by a herbal medicine or a control group were eligible. The primary outcomes measured were pain, physical function, swelling, stiffness and quality of life; and the secondary outcomes were adverse events, activity limitations and treatment satisfaction. Sixteen studies were included (n = 1,741 patients) in the systematic review and nine studies in the meta-analysis, representing 6 of the 13 herbal medicines studied: Boswellia serrata (n = 2), Curcuma longa (n = 3), Harpagophytum procumbens (n = 1), Salix daphnoides (n = 3), Uncaria guianensis (n = 2) and Zingiber officinale (n = 5). B. serrata was more effective than both placebo and valdecoxib for improvement of pain and physical function. No difference was observed for H. procumbens, C. longa and U. guianensis compared with control. Z. officinale showed improvement of pain over placebo. The evidence was insufficient to support the effective and safe use of these herbal medicines, because the quality of evidence of studies was low. This study guides managers of the Brazilian public health system and prescribers in decision-making regarding the use of these herbal medicines for OA. Copyright © 2017 John Wiley & Sons, Ltd.

Full text

REVIEW
Oral herbal medicines marketed in Brazil for the
treatment of osteoarthritis: A systematic review
and meta-analysis
Mariana Del Grossi Moura,
1
Luciane Cruz Lopes,
1
Maique Weber Biavatti,
2
Sean Alexander Kennedy,
3
Maria Carolina de Oliveira e Silva,
1
Marcus Tolentino Silva
1
and
Cristiane de Cássia Bergamaschi
1
*
1
Department of Pharmaceutical Sciences, University of Sorocaba, Sorocaba, State of São Paulo, Brazil
2
Pharmaceutical Department, Federal University of Florianopolis, Florianopolis, Santa Catarina, Brazil
3
Department of Medicine, McMaster University, Hamilton, Ontario, Canada
Herbal medications are commonly used to manage symptoms associated with osteoarthritis (OA). This system-
atic review evaluated the effectiveness and safety of oral medications used in Brazil for the treatment of OA.
Randomized clinical trials involving adults with OA treated by a herbal medicine or a control group were eligible.
The primary outcomes measured were pain, physical function, swelling, stiffness and quality of life; and the
secondary outcomes were adverse events, activity limitations and treatment satisfaction. Sixteen studies were in-
cluded (n= 1,741 patients) in the systematic review and nine studies in the meta-analysis, representing 6 of the 13
herbal medicines studied: Boswellia serrata (n= 2), Curcuma longa (n= 3), Harpagophytum procumbens
(n= 1), Salix daphnoides (n= 3), Uncaria guianensis (n= 2) and Zingiber officinale (n= 5). B. serrata was more
effective than both placebo and valdecoxib for improvement of pain and physical function. No difference was
observed for H. procumbens,C. longa and U. guianensis compared with control. Z. officinale showed
improvement of pain over placebo. The evidence was insufficient to support the effective and safe use of these
herbal medicines, because the quality of evidence of studies was low. This study guides managers of the Brazilian
public health system and prescribers in decision-making regarding the use of these herbal medicines for OA.
Copyright © 2017 John Wiley & Sons, Ltd.
Keywords: osteoarthritis; herbal medicine; effectiveness; safety.
INTRODUCTION
Osteoarthritis (OA) is the most common musculoskele-
tal disease and is associated with significant functional
decline and reduced quality of life (Pereira et al.,
2015). It affects more women than men and represents
the main cause of pain, chronic deficiency in adults
and disability in elderly (Zhang and Wang, 2015).
Non-steroidal anti-inflammatory drugs are the drugs
of choice for the pharmacological treatment of OA, de-
spite gastrointestinal and cardiovascular adverse events,
as well as high cost (Rannou et al., 2016). Therefore, al-
ternatives are of great interest. The use of herbal medi-
cines worldwide is substantial and increasing, varying
significantly among countries according to cultural prac-
tices, availability of plants and health programs in devel-
oped and developing countries (Barnes et al., 2016).
Two systematic reviews have evaluated some herbal
medicinal products for the treatment of OA for oral
(Cameron and Chrubasik, 2014) and topical use
(Cameron and Chrubasik, 2013). However, these stud-
ies did not include some plants that are commercially
available in Brazil. The Brazilian Policy of Integrative
and Complementary Practices and the National Policy
of Medicinal and Phytotherapeutic Plants, adopted in
2006, were created to meet the demands of the Brazilian
population. These policies were decisive steps towards
introducing the use of medicinal and phytotherapeutic
plants in the Brazilian Unified Health System (SUS)
(Antonio et al., 2014).
In Brazil, there are 13 herbal medications marketed
for treatment of OA: Harpagophytum procumbens
DC. ex Meisn. (Devil’s claw), Uncaria tomentosa
(Willd.) DC. (Cat’s claw), Salix alba L. (White willow)
(financed by government), Curcuma longa L. (or
Curcuma domestica Valeton) (Turmeric), Chenopodium
ambrosioides L. (Santa Maria herb),Cordia curassavica
(Jacq.) Roem. & Schult. (or Cordia verbenacea DC.)
(Whaler herb), Zingiber officinale Roscoe (Ginger),
Persea gratissima Gaertn.f. (or Persea americana Mill.)
(Avocado), (included in the National List of Medicinal
Plants of Interest to the Unified Health System –
RENISUS), Boswellia serrata Roxb. ex Colebr. (Salai
guggal), Bowdichia virgilioides Kunth. (Black sucupira),
Salix daphnoides Vill (violet willow), Salix purpurea L.
(Purple shell willow) and Uncaria guianensis (Aubl.) J.
F. Gmel (Cat’s claw).
Despite the common use of herbal medications to
manage OA, the safety and effectiveness of some of
* Correspondence to: Cristiane de Cássia Bergamaschi, Universidade de
Sorocaba –UNISO, Rodovia Raposo Tavares, Km 92.5, 18023-000, Soro-
caba, SP, Brazil.
E-mail: cristiane.motta@prof.uniso.br
PHYTOTHERAPY RESEARCH
Phytother. Res. (2017)
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/ptr.5910
Copyright © 2017 John Wiley & Sons, Ltd.
Received 10 July 2017
Accepted 09 August 2017

these agents is unclear. Therefore, a systematic review
and meta-analysis of oral herbal medications used in
Brazil for the treatment of OA were conducted.
METHODS
Standards and Protocol Register. The systematic review
was performed according to the recommendations spec-
ified in the Cochrane Handbook for Interventional
Reviews and reported according to the Preferred
Reporting Items for Systematic Reviews and Meta-
Analyses statement (Moher et al., 2009). This protocol
was registered in the International Prospective Register
of Systematic Reviews (PROSPERO-CRD42015019793
- http://www.crd.york.ac.uk/PROSPERO/).
Eligibility criteria
Inclusion criteria. Randomized controlled trials
(RCT) of adults (>18 years old) with a diagnosis of
OA according to the criteria of the American College
of Rheumatology (ACR): Western Ontario and
McMaster Universities (WOMAC) (Altman et al.,
1991) or to the equivalent criterion of the European
League Against Rheumatism (Zhang and Jordan,
2010); using one of the 13 oral herbal medications used
by the Brazilian population derived from any of the
plant preparations (whole, powder, extract, crude drug,
standardized mixture and drug extract ratio and
solvent), namely, B. serrata,B. virgilioides,C. longa (or
C. domestica), C. ambrosioides,C. curassavica (or
C. verbenacea), H. procumbens,P. gratissima (or P. amer-
icana), S. alba,S. daphnoides,S. purpurea,U. tomentosa,
U. guianensis and Z. officinale, compared with a control
group (placebo or a non-herbal medication controls).
Exclusion criteria. Studies in which more than 20% of
patients had other associated disease and those that in-
vestigated the simultaneous use of more than one of
the eligible plants were excluded.
Outcomes measures. Our outcomes were consistent
with those proposed by the Cochrane Musculoskeletal
Group systematic intervention reviews for OA (Pham
et al., 2004) and when necessary, the results were evalu-
ated for unification of the different scales.
The primary outcomes were pain (overall, at rest,
movement or night) (Visual Analogue Scale –VAS,
pain sub-scale of the Western Ontario and McMaster
Universities –WOMAC index; the Lequesne index
and other scales); physical function –global disability
or walking disability (VAS, function sub-scale of
WOMAC index; Lequesne index and other scales);
swelling (VAS and other scales); stiffness (VAS, stiffness
sub-scale of WOMAC index; Lequesne index and other
scales); and quality of life.
The secondary outcomes were as follows: adverse
events (withdrawals and serious adverse events –that
cause death, life-threatening problems, hospitalization,
disability or permanent damage); number of patients
reporting any adverse effects, activity limitations;
satisfaction with treatment; use of rescue medication;
duration of symptom resolved; and change in joint struc-
ture (according to the ACR criteria for OA
classification).
Search methods for primary studies
Electronic searches. The following electronic data-
bases were searched: CENTRAL, MEDLINE (via
OVID), EMBASE (via OVID), CINAHL (via OVID),
Web of Science, Health Star (via OVID), AMED (via
OVID), LILACS, CAB abstracts, clinical trial.gov, the
WHO Trial Register and the Brazilian thesis database
(CAPES) for studies published up to 31st May 2016;
without language or publication status restrictions. The
terms that describe OA and herbal medications were
combined individually.
Searching other resources. The reference lists of each
eligible study and review article were also reviewed to
identify additional eligible trials. We wrote to the au-
thors of all eligible trials and the pharmaceutical compa-
nies involved in the production of herbal medicines and
inquired about additional trials of which they are aware.
Five Brazilian scientific journals were also searched by
hand for additional eligible studies (the Journal of Basic
and Applied Pharmaceutical Sciences, Brazilian Journal
of Pharmacy, Brazilian Journal of Pharmacognosy,
Brazilian Journal of Medicinal Plants and Brazilian
Journal of Pharmaceutical Sciences). Unpublished stud-
ies were identified by searching reference lists reported
in the Brazilian legislation and conference proceedings.
Search strategy. The search was conducted individually
for each plant. The following MeSH terms were used:
(1) intervention (scientific name of plant, synonyms of
each medicinal plant and popular name of each medici-
nal plant) and (2) condition (OA, osteoarthrites,
osteoarthritides, osteoarthrosis, osteoarthroses, arthri-
tis, degenerative, arthritides, degenerative, degenerative
arthritides, degenerative arthritis and osteoarthrosis
deformans). The search strategy was adapted for each
database.
Eligibility determination. Four reviewers (C. B., M. G.,
M. B. and S. K.), working in pairs, independently
screened potentially relevant citations and abstracts
and applied the selection criteria. The full texts of all
articles that either reviewer felt might be eligible were
obtained. Two reviewers independently rated the eligi-
bility of each full-text article and resolved disagree-
ments by consensus. In cases of duplicate publications,
the article with the most complete data was used.
Data extraction. Four reviewers (C. B., M. G., M. B. and
S. K.), working in pairs, independently extracted the
data and recorded information regarding patients,
methods, interventions, outcomes and missing outcome
data using standardized and pretested data extraction
forms with instructions. Before starting data extraction,
calibration exercises were conducted to ensure
M. DEL GROSSI MOURA ET AL.
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

consistency between reviewers. We contacted study au-
thors to resolve any uncertainties that arose. Disagree-
ments were resolved by consensus with any unresolved
issues referred to another reviewer.
Risk of bias in individual studies. Using a modified
version of the Cochrane collaboration risk of bias tool
(Altman et al., 1990; Higgins and Green, 2011), the same
pairs of reviewers independently rated the risk of bias
for each randomized trial, according to the following
criteria: random sequence; allocation concealment; and
blinding of the patient, healthcare professionals,
outcome assessors, data collectors and data analysts; in-
complete outcome data; selective outcome reporting;
and major baseline imbalance. Reviewers gave response
options of ‘definitely yes’,‘probably yes’,‘probably no’
and ‘definitely no’for each of the domains, with ‘defi-
nitely yes’and ‘probably yes’ultimately being assigned
a low risk of bias, and ‘definitely no’and ‘probably no’
a high risk of bias (Akl et al., 2012). One arbitrator
(L. L.) judged unresolved disagreements.
Possible explanations for heterogeneity included the
following: doses (higher vs. lower) with an expected
larger effect with higher doses, duration of the treat-
ment (longer vs. shorter) with an expected larger effect
with longer duration of the treatment; and the risk of
bias, with an expected larger effect in trials at high or
unclear risk of bias versus trials at low risk of bias. We
ranked heterogeneity associated with pooled effect esti-
mates with the use of a χ
2
-test and the I
2
statistic.
Heterogeneity was defined as follows: 0–25% (low
heterogeneity), 50% (moderate heterogeneity) and
75% (high heterogeneity) (Higgins et al., 2003).
Confidence in pooled estimates of effect. The quality of
evidence from RCT for each of the outcomes was also
independently rated using the GRADE (Grading of
Recommendations Assessment, Development and
Evaluation) approach (Guyatt et al., 2011). The
GRADE evaluates five categories of limitations: risk
of bias, inconsistency, indirectness, imprecision and
reporting bias. Disagreements were resolved by discus-
sion with two of the authors (M. D. G. and C. C. B.).
Data synthesis. Analyses were conducted for each
herbal intervention and outcome of interest. Meta-
analyses were conducted using COMPREHENSIVE
META-ANALYSIS STATA software (version 14.1). Random
effects meta-analyses were employed, which are conser-
vative in that they consider within-studies and between-
studies differences when calculating the error term used
in the analysis.
For trials that reported dichotomous outcomes, the
pooled relative risk was calculated with an associated
95% confidence interval (95% CI). For continuous
outcomes and studies that reported the same construct
using different measurement instruments, the standard-
ized mean difference was calculated, expressing the
intervention effect in standard deviation units, rather
than the original units of measurement, with the value
of standardized mean difference depending on the effect
size (difference between means) and standard deviation
of the outcomes (inherent variability among partici-
pants) (Busse et al., 2015). Funnel plots were not created
to explore possible publication bias owing to the small
number of studies. When a meta-analysis was not appro-
priate, a narrative synthesis of studies was provided.
RESULTS
Literature search results
A total of 1385 studies were retrieved after removing
duplicates. Two studies (Biller et al., 2002; Frerick
et al., 2001) for H. procumbens were excluded because
the full text was not found despite an extensive search.
Of the 22 remaining articles, 16 fulfilled the inclusion
criteria and 9 RCT were included in the meta-analysis
(Fig. 1).
Description of the studies
Sixteen RCT, involving 1741 patients with OA were
included in this review. Most of the trials compared
interventions involving herbal medicines with a
placebo-controlled (n= 6) or active-control (n=5)
study. The included articles are related to 7 of the 13
herbal medicines studied. The characteristics of the
studies are described in Table 1.
Bias risk (Fig. 2)
Only three studies (18.75%) adequately fulfilled all val-
idation criteria and thus had minimum bias risk (Biegert
et al., 2004; Schmid et al., 2001; Wigler et al., 2003).
Nine (56.25%) studies did not report sufficient data
about the randomization and allocation process to allow
an assessment and therefore had selection bias (Bliddal
et al., 2000; Chantre et al., 2000; Haghighi et al., 2005;
Kuptniratsaikul et al., 2009; Lardos et al., 2004;
Paramdeep, 2013; Piscoya et al., 2001; Zakeri et al.,
2011). Other studies reported complete data regarding
randomized sequence generation.
Most of the studies (n= 12) described the blinding of
patient and healthcare professionals (except Sontakke
et al., 2007 and Paramdeep, 2013) as open trials in which
both the researchers and participants knew which treat-
ment was administered, whereas the trials by Madhu
et al. (2013) and by Kuptniratsaikul et al. (2009) were
single-blind studies and thus had high risk of bias.
All authors reported whether there was loss of
follow-up and/or patient exclusion during the study, ex-
plicitly reported separately in the study groups. There
was risk of attrition bias in four groups (Bliddal et al.,
2000; Kuptniratsaikul et al., 2014; Kuptniratsaikul
et al., 2009; Zakeri et al., 2011) because the simple input
of lost data was performed inappropriately and did not
discuss the implications of losses to follow-up. Zakeri
et al. (2011), Haghighi et al. (2005) and Piscoya et al.
(2001) failed to report adverse events, representing
reporting bias.
Two studies on the herbal medicine Z. officinale were
financed by pharmaceutical companies (Bliddal et al.,
2000; Wigler et al., 2003). Three clinical trials did not
HERBAL MEDICINES FOR TREATMENT OF OSTEOARTHRITIS
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

use the classification criteria by the ACR for OA diag-
nosis, leading to doubts regarding the classification for
inclusion of patients in the studies (Haghighi et al.,
2005; Kimmatkar et al., 2003; Lardos et al., 2004).
Effect of the interventions
Boswellia serrata (n= 2 studies). The studies found
were not included in the meta-analysis owing to diver-
gences in interventions. Kimmatkar et al. (2003)
randomized 30 patients with knee OA (ACR criteria)
into two groups and compared the effects of B. serrata
(999 mg/day) with placebo in a crossover study (3-week
washout) for 8 weeks (data extracted for the first phase
of the study only). B. serrata was more effective for the
outcomes of pain and physical function, as well as for
reduction in edema severity edema (p<0.001). Only
two patients from the intervention group reported mild
adverse events (nausea and epigastric pain). Sontakke
et al. (2007) compared plant extract (999 mg/day) with
the anti-inflammatory drug valdecoxib (10 mg/day) in
a group of 66 patients (33 patients in each group) with
knee OA (ACR criteria) during a 6-month treatment
period. The results of this open trial were favorable for
the intervention for the criteria pain, physical function
and stiffness on the WOMAC (albeit having slower
onset) (p<0.001). Only one patient from the interven-
tion group had an adverse event (diarrhea).
Curcuma longa (or Curcuma domestica n = 3 studies).
Kuptniratsaikul et al. (2009) randomized 107 patients
with OA of the knee (ACR criteria) into two groups:
C. domestica (n= 52) receiving 2000 mg/day of the
extract and ibuprofen (n= 55) receiving 800 mg/day
for 42 weeks. C. domestica extract exhibited efficacy
compared with ibuprofen for pain and physical function
caused by OA. No statistically significant difference was
found between the groups regarding adverse effects
(p<0.36). Kuptniratsaikul et al. (2014), in a randomized
double-blind study of 367 patients with OA of the knee
(ACR criteria), compared the plant extract
(1500 mg/day) with ibuprofen (1200 mg/day) for
4 weeks. The comparison between the groups showed
no significant differences in pain, physical function and
stiffness or in number of adverse effects (p= 0.531;
p= 0.278; p= 0.522, p= 0.222, respectively). Madhu et al.
(2013) compared four groups: C. longa extract
(1000 mg/day), glucosamine (1,500 mg/day),
C. longa + glucosamine (1,000 mg + 1,500 mg) and
placebo. C. longa and glucosamine alone were more
effective for pain control compared with placebo
(p<0.05) (Fig. 3). A total of 13 adverse events was
observed, where the smaller number occurred in the
group using herbal medicine. Because the authors
described only the number of events, this study was
not included in the meta-analysis.
Meta-analysis was conducted for the outcomes of pain
and physical function comparing herbal medicine to
ibuprofen. Ibuprofen proved similar to the plant extract
for the pain outcome OR = 0.89 (95% CI = 0.71–1.36)
and for improvement in physical function OR = 1.09
(95% CI = 0.79–1.52) (Figs 3 and 4), but the evidence
was of low quality. Poor quality evidence showed that
the groups did not differ in adverse events RR = 1.28
(95% CI = 0.55–2.98) (Fig. 5).
Harpagophytum procumbens (n= 1 study). There was
only one study, involving a total of 122 patients with
OA of the knees (72.2%) or hips (27.8%) (ACR
Figure 1. Flowchart of literature search and study selection. [Colour figure can be viewed at wileyonlinelibrary.com]
M. DEL GROSSI MOURA ET AL.
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

criteria) randomized into two groups: a herbal group
(n= 62 patients) (435 mg/capsule of the extract pre-
pared from roots of H. procumbens, 6 capsules/day)
and a diacerein group (100 mg/day) (n= 60 patients),
where both groups used the medication for 4 months.
The global assessment of efficacy revealed similar
results for both treatments. Regarding safety profile,
43.3% of the patients that took diacerein and 25.8% of
those on H. procumbens had adverse events, most of
which were gastrointestinal, leading to the exclusion of
a higher number of participants from the diacerein
group (n= 14) compared with the H. procumbens group
(n= 8) (Chantre et al., 2000).
Salix daphnoides and Salix purpurea (n= 3 studies).
Schmid et al. (2001) randomized 78 patients with OA
of the knee or hips to receive the plant extract (daily
dosage of 240 mg/day of salicin) or placebo for 2 weeks
(after 4–6 days of washout). The pain outcome was
favorable for the intervention group (WOMAC index)
after 2 weeks of treatment (p= 0.047). There was no
statistically significant difference between the groups
regarding adverse events. Another double-blind
randomized study compared the plant extract
(equivalent to 240 mg of salicin) with placebo and
diclofenac (100 mg) during a 6-week treatment period
in adult patients (n= 127) with OA of the knee or hips.
The results showed no statistically significant difference
in pain reduction relative to the diclofenac group. The
number of adverse events was higher in the diclofenac
group, predominantly gastrointestinal (p= 0.001), and
a patient from this group had a severe adverse event
and was admitted to hospital for suspected gastritis
and deep vein thrombosis (Biegert et al., 2004). Another
study compared the use of willow bark (90 mg of
salicin/day), willow bark (180 mg of salicin/day) and
diclofenac (150 mg/day). A total of 60 adult patients
with OA of the knee and hips were randomized and
assessed for 6 months. There was no significant
Table 1. Characteristics of the included studies
Characteristics Studies
Number of
patients
Total number of patients 16 1741
Female (%) 16 74.3% (25.3–
89.4%)
Herbal medication (marker)
Harpagophytum procumbens
(Harpagoside)
1 122
Uncaria guianensis (Not known) 2 90
Boswellia serrata (Boswellic acid –
AKBA)
296
Curcuma longa or Curcuma domestica
(Curcuminoids)
3 594
Salix daphnoides,Salix purpurea
(Salicin)
3 265
Zingiber officinale (Gingerols) 5 578
Clinical condition
Osteoarthritis of the knee 13 1116
Osteoarthritis of the knee and hip 3 625
Region
Asia 8 1130
Germany 3 265
Other European countries 2 189
South America 2 90
Not specified 1 71
Follow-up (days)
Mean 16 56
Median (min–max) 16 42 (14–180)
Use of concomitant medication
Yes 1 2
No 3
Not specified 1
Year of Publication
1997–2006 9 720
2007–2014 7 1021
Funded by the industry
Yes 2 138
Not specified 14 1562
Figure 2. Consensus of the authors about bias risk for each study
included. [Colour figure can be viewed at wileyonlinelibrary.com]
HERBAL MEDICINES FOR TREATMENT OF OSTEOARTHRITIS
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

difference regarding the outcome of pain. Only one pa-
tient from the diclofenac group reported an adverse
event owing to the treatment (Lardos et al., 2004).
The meta-analysis showed benefit for the use of the
herbal medicine for reducing pain OR = 0.47 (95%
CI = 0.19–0.85) (very low quality evidence) and for
improvement of physical function OR = 0.54 (95%
CI = 0.31–1.02) (moderate-quality evidence) compared
with placebo (Figs 3 and 4). No difference between the
plant extract and diclofenac was observed for the out-
come of pain OR = 2.43 (95% CI = 1.00–6.02) (Fig. 3).
No difference in adverse events was evident between
the plant and placebo groups RR = 0.94 (0.67–1.34)
(moderate-quality evidence) (Biegert et al., 2004;
Schmid et al., 2001) (Fig. 5).
Uncaria guianensis (n= 2 studies). Piscoya and Herman
(1997) randomized 45 patients in a study comparing the
use of U. guianensis extract to placebo for 4 weeks. Pain
reduction was favorable for intervention group after
4 weeks of use of the plant extract (p<0.05). A higher
number of patients had adverse events in the
phytotherapy group (n= 10) compared with the placebo
group (n= 3), but the authors performed no statistical
tests. Piscoya et al. (2001) assessed patients diagnosed
with knee OA (ACR criteria) randomized to receive
aqueous extract of U. guianensis (single daily oral
100 mg dosage) (n= 30 patients) or placebo (n= 15 pa-
tients) for 4 weeks. In the global evaluation of the
disease, doctors and patients reported improvement in
pain only from the second week (p<0.01). The study
did not report adverse events.
The meta-analysis showed no benefit for the use of
U. guianensis over placebo to reduce pain at rest
OR = 0.31 (95% CI = 0.07–1.39) or at night OR = 0.08
(95% CI = 0.00–1.82) (very low-quality evidence)
(Fig. 3).
Zingiber officinale (n= 5 studies). Haghighi et al. (2005)
randomized 120 patients with OA of the knee and hip
into three groups: Z. officinale (1000 mg capsule,
equivalent to 30 mg of the extract), ibuprofen
(1200 mg) and placebo, during a 30-day treatment
period. There was improvement in pain with the use of
plant compared with placebo only (p<0.001). The
study did not report any adverse events. Wigler et al.
(2003) randomized 29 patients with OA of the knee to
receive Z. officinale extract at 1000 mg (n= 14) and
placebo (n= 15) for around 48 weeks. During the first
12 weeks, the differences between the groups were not
statistically significant for the outcome of pain (as a
crossover study, data were extracted only for the first
phase of the study). Only one adverse event (heartburn)
was reported by a patient in the Z. officinale group.
Bliddal et al. (2000) compared ginger extract
(500 mg/day) with placebo and ibuprofen (400 mg/day)
in patients with OA of the hip and knee in a double-
blind, controlled, crossover study (1-week washout).
Figure 3. Forest plot of meta-analysis for pain reduction (VAS/WOMAC) of herbal medicines to treatment of osteoarthritis. [Colour figure can
be viewed at wileyonlinelibrary.com]
M. DEL GROSSI MOURA ET AL.
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

Figure 4. Forest plot of meta-analysis for improved physical function of herbal medicines to treatment of osteoarthritis. [Colour figure can be
viewed at wileyonlinelibrary.com]
Figure 5. Forest plot of meta-analysis for number of adverse events of herbal medicines to treatment of osteoarthritis. [Colour figure can be
viewed at wileyonlinelibrary.com]
HERBAL MEDICINES FOR TREATMENT OF OSTEOARTHRITIS
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

Data were observed only for the first phase of the study.
Ibuprofen was superior to both the herbal medicine and
placebo for improvement of pain (p<0.05). No differ-
ence was observed among the groups in relation to the
adverse events observed in 34 patients (Z. officinale
n= 10, placebo n= 9, ibuprofen n= 15). Zakeri et al.
(2011) randomized 204 patients with OA of the knee
to receive 500 mg/day of ginger extract (n= 103) or pla-
cebo (n= 101) for 6 weeks. The ginger group showed
improvement for pain (OR = 2.56), stiffness (OR = 2.01)
and physical function (OR = 2.56) (p<0.05). The study
did not report adverse events. Paramdeep (2013) con-
ducted an open trial study of 60 patients with OA of
the knee randomized into three groups: diclofenac
50 mg + placebo (n= 20), Z. officinale 750 mg + placebo
(n= 20) and Z. officinale 750 mg/day + diclofenac 50 mg
(n= 20), for a 12-week period. The pain improvement
was statistically significant for the group diclofenac + pla-
cebo groups compared with the plant + placebo
(p<0.05). There was no difference in adverse events
among the groups (p>0.05). Meta-analysis was per-
formed only for the outcome of pain (Haghighi et al.,
2005; Zakeri et al., 2011) and showed no difference be-
tween the plant extract and placebo (very low quality
evidence) (Fig. 3).
DISCUSSION
This study evaluated the plants commercially available
in Brazil for the treatment of OA and led to the retrieval
of clinical trials for 7 out of the 13 plants studied
(H. procumbens, B. serrata, U. guianensis, S. daphnoides,
C. longa and Z. officinale). The meta-analysis was only
possible for U. guianensis,S. daphnoides,C. longa and
Z. officinale.
Boswellia serrata (999 mg daily dose) was more effec-
tive for reducing pain, stiffness, swelling and improving
physical function compared with both placebo and
valdecoxib. However, the duration of treatment differed
between the studies (2 and 8 months, respectively)
(Kimmatkar et al., 2003; Sontakke et al., 2007). Both
studies reported mild adverse events (nausea and gastric
distress) with use of the plant extract, as also reported
by Basch et al. (2004). The study of Cameron and
Chrubasik (2014) included five studies of this plant.
However, some of these studies involved herbal medi-
cine associations were not included in the present study.
Our findings indicated that the available evidence re-
mains insufficient to confirm the benefit of using this
herbal therapy, because no further clinical trials have
been published about the plant.
Curcuma longa extract had similar effectiveness to
ibuprofen for the treatment of osteoarthritic pain and
pain-related functional impairments (Kuptniratsaikul
et al., 2014; Kuptniratsaikul et al., 2009). While in vitro
studies suggest that curcumin is an effective therapy
for OA (Jackson et al., 2006; Lev-Ari et al., 2006;
Mathy-Hartert et al., 2009), there are few RCT for this
plant (only two new studies have been published since
the last review), and most of these were of poor method-
ological quality owing to high risk of bias, precluding
confirmation of their findings.
Despite the very low-quality evidence affirming that
the plant extract and placebo had similar safety profiles,
Peddada et al. (2015) reported that non-steroidal anti-
inflammatory drugs can cause adverse gastrointestinal
events such as peptic ulcers and subsequent risk of
bleeding and perforation, which can be significant for
patients with OA owing to prolonged use of these drugs.
Moreover, these effects were observed less frequently
with the use of curcumin owing to their selective inhibi-
tion of COX-2 (Goel et al., 2001).
Chantre et al. (2000) found that H. procumbens was
similar to diacerein in reducing pain but allowed less
usage of diclofenac and acetaminophen. Although
Cameron and Chrubasik (2014) included two other
studies (Biller et al., 2002; Frerick et al., 2001) (for which
we were unable to locate the full text, as previously
described); these authors did not perform a meta-
analysis because of the disparities in the interventions
found for the clinical trials. Therefore, the evidence
available evidence remains insufficient to confirm the
efficacy and safety of this herbal medicine for OA.
Chantre et al. (2000) used a daily dose of 57 mg of the
marker harpagoside. This dose is in accordance with
Gagnier et al. (2004), who suggested a daily dose of at
least 50 mg harpagoside, and is consistent with the
European Pharmacopoeia, which considers the dose of
50–100 mg harpagoside suitable for the treatment of
OA. The safety profile was similar to placebo, and no
serious adverse events were evident. This finding was
also observed by Vlachojannis et al. (2008) in an evalu-
ation of 28 RCT showing no increase in the incidence
of adverse events with this herbal medicine.
The quality of evidence was insufficient to suggest
effectiveness of S. purpurea (at daily dose of 240 mg of
salicin –active marker) for improving pain compared
with placebo, owing to the small number of participants
and high heterogeneity across studies. Similarly, the
effectiveness of the intervention for the controlling pain
compared with that of the diclofenac could not be
confirmed in the present analysis because the studies
retrieved had inherent randomization problems, moder-
ate heterogeneity and small sample size. However, there
was moderate-quality evidence to support improvement
in physical function relative to placebo (Biegert et al.,
2004; Schmid et al., 2001). The findings of the present
study differ from SR published previously owing to the
inclusion of the study by Lardos et al. (2004).
The number of adverse events did not differ with the
use of S. purpurea compared with placebo (Biegert et al.,
2004; Schmid et al., 2001) (moderate-quality evidence)
but was considered safer when compared to diclofenac
(Biegert et al., 2004). Although the studies observed
no serious adverse events with the use of this plant,
corroborating results of Boullata et al. (2003), anaphy-
lactic reactions may occur in patients with a history of
allergy to salicylates, contraindicating the use of willow
bark preparations in these patients.
The meta-analysis showed that U. guianensis was
more effective for pain reduction compared with pla-
cebo (Piscoya et al., 2001; Piscoya and Herman, 1997).
However, the evidence is of very poor quality owing to
randomization problems, high heterogeneity and small
sample size. The study of Piscoya and Herman (1997)
was not included in previously published SR. Other
efficacy outcomes were not assessed because the studies
were not adequately reported.
Although the results showed benefits from the use of
Z. officinale compared with placebo for pain reduction
M. DEL GROSSI MOURA ET AL.
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

(Haghighi et al., 2005; Zakeri et al., 2011), there are few
RCT and most of these had selection, attrition and
reporting bias, as well as inconsistency and inaccuracy.
Therefore, well-designed trials are needed to elucidate
the effects of ginger extract on OA.
Many of the results of this review were inconclusive
owing to the small number of RCT evaluating the use
of herbal medicines for OA and to the low to very low
quality of evidence. In addition, most of the RCT
differed in their design, interventions, dosage and
duration of treatment, hampering meta-analysis for
most outcomes assessed. The safety evaluation of the
use of plants was also affected because some studies
failed to report adverse effects, leading to reporting bias
(Haghighi et al., 2005; Piscoya et al., 2001; Zakeri et al.,
2011).
The findings of this review showed that the most of
the RCT had significant methodological flaws, and the
majority was not conducted with products of standard-
ized quality, because the preparations had been
produced exclusively for the study in question. It is im-
portant to note that the few adverse events reported
were based on studies with limited treatment duration
and hence the prolonged intake of herbal medicines
was not evaluated. Another limitation of this study was
the fact that some RCT did not report other relevant
outcomes for patients, such as quality of life and satis-
faction with the treatment.
It is also noteworthy that the exact components of the
tested plants were unclear where, in most of the clinical
trials, the preparations were produced exclusively for
the studies. The preparation of the plant was not
discussed in some studies (Bliddal et al., 2000;
Kuptniratsaikul et al., 2014; Lardos et al., 2004;
Paramdeep, 2013), although the content of active
compounds was given. The studies of Chantre et al.
(2000), Madhu et al. (2013), Sontakke et al. (2007) used
industrial pharmaceutical products, while Kimmatkar
et al. (2003), Zakeri et al., 2011 and Wigler et al., 2003
used products produced by pharmacies or by the
institution where the research was conducted.
This study evaluated the oral use of herbal medicines
sold in Brazil for OA and, although some of these plants
had been evaluated in a previously published SR, the
present study provides updated information about
them. No RCT were found for the plants B. virgilioides,
C. ambrosioides,C. curassavica,P. gratissima,S. alba
and U. tomentosa. Few clinical trials on B. serrata,
C. longa, H. procumbens, S. daphnoides, S. purpurea,
U. guianensis and Z. officinale were found, pointing to
the need for further RCT to prove the effectiveness
and safety of these plants.
H. procumbens,U. tomentosa and S. alba are herbal
medicines funded by the Brazilian Ministry of Health
and may be prescribed to SUS users but should not be
recommended for the treatment of OA. A further five
plants included in this review (C. longa,C. ambrosioides,
C. curassavica,Z. officinale and P. gratissima) are part of
the RENISUS and considered to be plants with
potential for SUS products. B. serrata,B. virgilioides,
S. daphnoides, S. purpurea and U. guianensis are not
funded by the Brazilian government nor belong to
RENISUS but are commercially available on the
market and used by the Brazilian population.
None of the studies found were conducted in Brazil.
The use of these plants is based only on traditional
use. Therefore, clinical research assessing the benefits
and safety of these herbal medicines should be encour-
aged. Of the 16 plants studied, only three (C. longa or
C. domestica, U. guianensis and Z. officinale) showed
benefits for some of the outcomes evaluated; however,
the quality of this evidence was low or moderate,
hampering the analysis of the effectiveness and safety
of herbal medicines.
The evidence was not sufficient to support the safe
and effective use of these herbal medicines for OA.
Thus, this study can guide managers of the Brazilian
public health system, prescribers and users of these
drugs in decision-making regarding the use of these
herbal medicines, whose value for the treatment of
OA remains unclear.
Acknowledgement
This work was financially supported by a grant of Governmental
Graduate Education Institutions Program in Brazil known as
PROSUP-CAPES.
Conflict of Interest
The authors declare no conflict of interest.
REFERENCES
Akl EA, Sun X, Busse JW, et al. 2012. Specific instructions for
estimating unclearly reported blinding status in randomized
trials were reliable and valid. J Clin Epidemiol 65:
262–267.
Altman R, Alarcon G, Appelrouth D, et al. 1990. The American
College of Rheumatology criteria for the classification and
reporting of osteoarthritis of the hand. Arthritis Rheum 33:
1601–1610.
Altman R, Alarcon G, Appelrouth D, et al. 1991. The American
College of Rheumatology criteria for the classification and
reporting of osteoarthritis of the hip. Arthritis Rheum 34:
505–514.
Antonio GD, Tesser CD, Moretti-Pires RO. 2014. Phytotherapy
in primary health care. Revista de saude publica 48:
541–553.
Barnes J, McLachlan AJ, Sherwin CM, Enioutina EY. 2016.
Herbal medicines: challenges in the modern world. Part 1.
Australia and New Zealand. Expert Rev Clin Pharmacol 9:
905–915.
Basch E, Boon H, Davies-Heerema T, et al. 2004. Boswellia: An
evidence based systematic review by the natural standard re-
search collaboration. J Herb Pharmacother 6:63–83.
Biegert C, Wagner I, Ludtke R, et al. 2004. Efficacy and safety of
willow bark extract in the treatment of osteoarthritis and rheu-
matoid arthritis: results of 2 randomized double-blind con-
trolled trials. J Rheumatol 31: 2121–2130.
Biller AI, Schulz VRN, Roots I, Loew D. 2002. Results of two ran-
domized controlled studies and of a post-marketing surveil-
lance study investigating a Devil’s claw extract. In
Phylopharmaka VII, Schulz V, Rietbrock N, Roots I, Loew D
(eds). Steinkopf-Verlag: Darmstadt; 81–92.
Bliddal H, Rosetzsky A, Schlichting P, et al. 2000. A random-
ized, placebo-controlled, cross-over study of ginger extracts
and ibuprofen in osteoarthritis. Osteoarthritis Cartilage 8:
9–12.
Boullata JI, McDonnell PJ, Oliva CD. 2003. Anaphylactic reaction
to a dietary supplement containing willow bark. Ann
Pharmacother 37: 882–885.
HERBAL MEDICINES FOR TREATMENT OF OSTEOARTHRITIS
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)

Busse JW, Bartlett SJ, Dougados M, et al. 2015. Optimal strate-
gies for reporting pain in clinical trials and systematic reviews:
recommendations from an OMERACT 12 Workshop. J
Rheumatol 42: 1962–1970.
Cameron M, Chrubasik S. 2013. Topical herbal therapies for
treating osteoarthritis. Cochrane Database Syst Rev
Cd010538.
Cameron M, Chrubasik S. 2014. Oral herbal therapies for treating
osteoarthritis. Cochrane Database Syst Rev Cd002947.
Chantre P, Cappelaere A, Leblan D, Guedon D, Vandermander J,
Fournie B. 2000. Efficacy and tolerance of Harpagophytum
procumbens versus diacerhein in treatment of osteoarthritis.
Phytomedicine : Intl J phytotherapy and phytopharmacology
7: 177–183.
Frerick H, Biller A, Schmidt U. 2001. A treatment schedule for
coxarthrosis: A double-blind study with Devil’s claw
[Stufenschema bei coxarthrose: Doppelblindstudie mit
Teufelskralle] 5:34–41.
Gagnier JJ, Chrubasik S, Manheimer E. 2004. Harpgophytum
procumbens for osteoarthritis and low back pain: a systematic
review. BMC Complement Altern Med 15: 13.
Goel A, Boland CR, Chauhan DP. 2001. Specific inhibition of
cyclooxygenase-2 (COX-2) expression by dietary curcumin in
HT-29 human colon cancer cells. Cancer Lett 172: 111–118.
Guyatt GH, Oxman AD, Kunz R, et al. 2011. GRADE guidelines: 7.
Rating the quality of evidence–inconsistency. J Clin Epidemiol
64: 1294–1302.
Haghighi M, Khalvat A, Toliat T, Jallaei S. 2005. Comparing the
effects of ginger (Zingiber officinale) extract and ibuprofen
on patients with osteoarthritis. Arch Iran Med 8: 267–271.
Higgins JP, Thompson SG, Deeks JJ, Altman DG. 2003. Measur-
ing inconsistency in meta-analyses. BMJ 327: 557–560.
Higgins JPT, Green S. 2011. Cochrane Handbook for Systematic
Reviews of Interventions Version 5.1.0 [updated March
2011]. The Cochrane Collaboration Disponível em: htpp:
wwwcochrane-handbookorg
Jackson JK, Higo T, Hunter WL, Burt HM. 2006. The antioxidants
curcumin and quercetin inhibit inflammatory processes associ-
ated with arthritis. Inf Res official J Eur Histamine Res Soc 55:
168–175.
Kimmatkar N, Thawani V, Hingorani L, Khiyani R. 2003. Efficacy
and tolerability of Boswellia serrata extract in treatment of
osteoarthritis of knee –a randomized double blind placebo
controlled trial. Phytomedicine : Intl J phytotherapy and
phytopharmacology 10:3
–7.
Kuptniratsaikul V, Dajpratham P, Taechaarpornkul W,
Buntragulpoontawee M, et al. 2014. Efficacy and safety of
Curcuma domestica extracts compared with ibuprofen in pa-
tients with knee osteoarthritis: a multicenter study. Clin Interv
Aging 9: 451–458.
Kuptniratsaikul V, Thanakhumtorn S, Chinswangwatanakul P,
Wattanamongkonsil L, Thamlikitkul V. 2009. Efficacy and
safety of Curcuma domestica extracts in patients with knee
osteoarthritis. J Altern Complement Med 15: 891–897.
Lardos A, Schmidlin CB, Fischer M, Ferlas-Chlodny E, et al. 2004.
Efficacy and tolerance of an aqueous willow bark dry extract in
patients with knee or hip arthrosis. ZeitschriftfurPhytotherapie
25: 275–281.
Lev-Ari S, Strier L, Kazanov D, Elkayam O, et al. 2006. Curcumin
synergistically potentiates the growth-inhibitory and pro-
apoptotic effects of celecoxib in osteoarthritis synovial adher-
ent cells. Rheumatology 45: 171–177.
Madhu K, Chanda K, Saji MJ. 2013. Safety and efficacy of
Curcuma longa extract in the treatment of painful knee osteo-
arthritis: a randomized placebo-controlled trial.
Inflammopharmacology 21: 129–136.
Mathy-Hartert M, Jacquemond-Collet I, Priem F, Sanchez C,
Lambert C, Henrotin Y. 2009. Curcumin inhibits pro-
inflammatory mediators and metalloproteinase-3 production
by chondrocytes. Inf Res official J Eur Histamine Res Soc
58: 899–908.
Moher D, Liberati A, Tetzlaff J, Altman DG. 2009. Preferred
reporting items for systematic reviews and meta-analyses:
the PRISMA statement. Ann Intern Med 151: 399–409.
Paramdeep G. 2013. Efficacy and tolerability of ginger (Zingiber
officinale) in patients of osteoarthritis of knee. Indian J Physiol
Pharmacol 57: 177–183.
Peddada KV, Peddada KV, Shukla SK, et al. 2015. Curcumin in
common musculoskeletal disorders: a review of current labo-
ratory, translational, and clinical data. Orthopaedic Surg 7:
222–231.
Pereira D, Ramos E, Branco J. 2015. Osteoarthritis. Acta Med Port
28:99–106.
Pham T, van der Heijde D, Altman RD, Anderson JJ, et al. 2004.
OMERACT-OARSI initiative: Osteoarthritis Research Society
International set of responder criteria for osteoarthritis clinical
trials revisited. Osteoarthritis Cartilage 12: 389–399.
Piscoya J, Rodriguez Z, Bustamante SA, Okuhama NN, Miller MJ,
Sandoval M. 2001. Efficacy and safety of freeze-dried cat’s
claw in osteoarthritis of the knee: mechanisms of action of
the species Uncaria guianensis. Inf Res official J Eur Histamine
Res Soc 50: 442–448.
Piscoya JS, Herman DR. 1997. Estudio multicéntrico comparando
Uncaria guianensis con placedbo en osteoartritis de
rodilla/multicentric study comparing uncaria guianensis vs.
placebo for knee osteoarthritis. Rev méd Inst Peru Segur Soc
6:60–64.
Rannou F, Pelletier JP, Martel-Pelletier J. 2016. Efficacy and safety
of topical NSAIDs in the management of osteoarthritis: evi-
dence from real-life setting trials and surveys. Semin Arthritis
Rheum 45: S18–S21.
Schmid B, Ludtke R, Selbmann HK, Kotter I, et al. 2001. Effi-
cacy and tolerability of a standardized willow bark extract
in patients with osteoarthritis: randomized placebo-
controlled, double blind clinical trial. Phytotherapy Res PTR
15: 344–350.
Sontakke S, Thawani V, Pimpalkhute S, Kabra P, Bachulkar S,
Hingorani L. 2007. Open, randomized, controlled clinical trial
of Boswellia serrata extract as compared to valdecoxib in oste-
oarthritis of knee. Indian J Pharm 39:27–29.
Vlachojannis J, Roufogalis BD, Chrubasik S. 2008. Systematic re-
view on the safety of Harpagophytum preparations for osteo-
arthritic and low back pain. Phytother Res 22: 149–152.
Wigler I, Grotto I, Caspi D, Yaron M. 2003. The effects of Zintona
EC (a ginger extract) on symptomatic gonarthritis. Osteoarthri-
tis Cartilage 11: 783–789.
Zakeri Z, Izadi S, Bari Z, Farhang S, Behzad N, Mohammad G.
2011. Evaluating the effects of ginger extract on knee pain,
stiffness and difficulty in patients with knee osteoarthritis. J
Med Plant Res 5: 3375–3379.
Zhang M, Wang J. 2015. Epigenetics and osteoarthritis. Genes &
Dis 2:69–75.
Zhang Y, Jordan JM. 2010. Epidemiology of osteoarthritis. Clin
Geriatr Med 26: 355–369.
M. DEL GROSSI MOURA ET AL.
Copyright © 2017 John Wiley & Sons, Ltd. Phytother. Res. (2017)