Efficacy of Curcuma for Treatment of Osteoarthritis

Data Sources, Selection, and Extraction

In October 2015, a literature search using a combination of the terms “curcuma,” “osteoarthritis,” “turmeric,” and “clinical trials,” was performed using PubMed, Academic Search Premier, and Google Scholar. No publication date limits were applied to the search. Human, interventional studies investigating efficacy of curcuma for treatment of osteoarthritis were selected for inclusion. Articles had to be published in peer-reviewed scientific journals. Titles and abstracts were examined by 2 authors (KP and WS) to identify citations related to curcuma. Included studies were reviewed and approved by the third author (RDB).

Excluding duplicates between databases, a total of 35 articles were identified. There were 9 results found on PubMed, 19 results on Academic Search Premier, and 12 on Google Scholar. All authors assessed articles for inclusion in the study; primary reasons for exclusion of articles are provided in Figure 1. The majority of studies were excluded because they did not assess efficacy or were not published in peer-reviewed journals. Ultimately, 8 articles were selected for inclusion. It was decided on initial review of the 8 articles to conduct the project as a narrative review. Study results were not pooled in a meta-analysis due to high anticipated heterogeneity among studies from differences in patients, intervention, control, duration, and blinding. The Consolidated Standards of Reporting Trials (CONSORT) Extension for Reporting Herbal Medicinal Interventions ⁹ was used as the primary basis for evaluation of study quality; however, other author-identified limitations were considered.

OPEN IN VIEWER

Figure 1.

Studies that were identified through title and abstract review during the literature search, reasons for exclusion, and the ultimate number of studies included in the review.

Data Synthesis

See Table 2 for a side-by-side comparison of extracted study information and Table 3 for the study evaluations using the CONSORT Extension.

OPEN IN VIEWER

Table 2.

Summary of Clinical Trial Findings.

Reference	10	12	13	14	15	16	17	18
First Author and Year	Kuptniratsaikul 2009	Kuptniratsaikul 2014	Pinsornsak 2012	Nakagawa 2014	Madhu 2013	Belcaro 2010	Kizhakkedath 2013	Kulkarni 1991
Design	R, DB	R, DB, NI	R, DB	R, DB	R, SB		R	R, DB, CO
Treatment	Curcuma domestica extract 500 mg 4 times daily	Curcuma domestica extract 1500 mg per day	Curcuminoids 500 mg twice daily plus diclofenac 25 mg thrice daily	Theracurmin 180 mg per day	Curcuma longa extract 500 mg twice daily	Meriva 500 mg twice daily plus best available treatment	Curcuma longa extract 350 mg plus Boswellia serrata extract 150 mg	Curcuma longa 100 mg every 8 hours in a combination herbomineral formulation
Control	Ibuprofen 400 mg twice daily	Ibuprofen 1200 mg per day	Placebo plus diclofenac 25 mg thrice daily	Placebo	Placebo, glucosamine 750 mg twice daily, or combination	Best available treatment	Celecoxib 100 mg twice daily	Placebo
Sample size	107	367	88	50	90	100	30	42
Primary outcome	Pain level with walking	WOMAC	Mean reduction in pain on an 11-point VAS	Knee pain VAS	100-point VAS	WOMAC	Incidence of pain	11-point VAS
Duration	6 wk	4 wk	3 mo	8 wk	6 wk	8 mo	12 wk	1 mo
Primary outcome result	2.7 ± 2.6 vs 2.0 ± 2.3 (difference of 0.67, 95% CI −0.35 to 1.68; P = .20)	3.36 ± 2.04 vs 0.23 ± 1.97 (mean difference −0.07, 95% CI −0.43 to 0.29, P = .010 for noninferiority)	1.8 vs 2.3 (0.441)	0.20 vs 0.21 (P = .023)	19.48 ± 17.84 vs 46.03 ± 20.84 vs 29.29 ± 20.58 vs 36.33 ± 28.99 (P < .05 for C longa extract and glucosamine compared with placebo and combination)	33.3 vs 68.8 (not directly compared)	21% vs 50% (P > .05)	0.4 ± 0.1 vs 4.2 ± 0.2 (P < .001)

Abbreviations: CO, crossover; DB, double blind; NI, noninferiority; SB, single blind; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index; VAS, visual analog scale.

OPEN IN VIEWER

Table 3.

Adapted CONSORT (Consolidated Standards of Reporting Trials) Checklist. ⁹

	Reference
	10	12	13	14	15	16	17	18
1. Title and abstract
Description of how patients were allocated to interventions	×	×	×	×		×
Latin binomial name	×	×	×	×	×	×
Part of the plant used			×
2. Introduction
Statement of reasoning behind trial with reference to specific herbal product being tested	×	×	×	×	×	×	×	×
3. Methods
Eligibility criteria for participants	×	×	×	×	×	×	×	×
Setting and location where data were collected	×	×	×		×	×
4. Interventions
Latin binomial name and common name							×
Part of plant used	×	×	×			×
Dosage and duration of administration	×	×	×	×	×	×		×
Explanation of how the dose was determined		×
Content of all herbal products per dosage unit form	×	×	×	×	×	×	×	×
Rationale for type of control or placebo		×
5. Objectives
Specific objectives and hypotheses	×	×			×	×	×	×
6. Outcomes
Clearly defined primary and secondary outcomes	×	×				×
7. Sample size
How sample size was determined	×	×
8. Randomization
Methods used for randomization	×	×				×
9. Blinding
Description of who is blinded	×	×		×		×
10. Statistical methods
Description of statistical methods used for analysis	×	×	×	×	×	×	×	×
11. Results
Flow of participants through each stage		×				×		×
12. Recruitment
Dates of period of recruitment and follow-up
13. Numbers analyzed
Number of subjects in each group included in the analysis	×	×	×	×	×	×		×
14. Outcomes
Summary of results for each group with precision	×	×	×	×	×	×	×	×
15. Ancillary analyses
Report of any other analyses performed indicating prespecified and exploratory								×
16. Adverse events
All important adverse events or side effects in each group	×	×	×		×	×		×
17. Discussion
Interpretation of results taking study hypotheses into account	×	×		×	×	×	×	×
Sources of potential bias or imprecision	×					×		×
18. Generalizability
External validity of trial results		×				×
19. Overall evidence
General interpretation of the results in the context of current evidence	×	×	×			×		×
Discussion of trial results in relation to trials of other available products		×				×		×

A randomized, active-controlled study was performed in Thailand to investigate the efficacy and safety of C domestica extracts in pain reduction and functional improvement in patients with knee osteoarthritis. ¹⁰ Eligible patients had to have knee pain of at least 5 on an 11-point Likert-type scale, radiographic osteophytes, and at least one of the following criteria: age greater than 50 years, morning stiffness less than 30 minutes in duration, and crepitus on motion. The key primary endpoints were pain level with walking and pain level on stairs. Patients were randomized to ibuprofen 400 mg twice daily (n = 55) or C domestica extracts 500 mg 4 times daily for 6 weeks (n = 52). Baseline pain scores with walking were 5.3 ± 2.3 in the C domestica group and 5.0 ± 1.9 in the ibuprofen group. In the C domestica group, baseline pain scores with stairs were 5.7 ± 2.1 versus 6.2 ± 2.2 in the ibuprofen group. Approximately 80% were female in both groups. The average age was 60 years in both groups.

In the C domestica group, there was a change from baseline in the pain score with walking of 2.7 ± 2.6 in comparison with 2.0 ± 2.3 in the ibuprofen group (difference of 0.67, 95% CI −0.35 to 1.68; P = .20). ¹⁰ There was a change from baseline in the pain score on stairs in the C domestica group of 2.5 ± 2.2 versus 2.5 ± 2.6 in the ibuprofen (difference of −0.06, 95% CI −1.07 to 0.96; P = .92). When looking at patient satisfaction, 91% of patients receiving C domestica were highly or moderately satisfied in comparison to 80% in the ibuprofen group (P = .15).

A major limitation was a power calculation was performed stating that 50 patients were needed to detect a significance difference of ±1 with a standard derivation of 2. The study was unable to achieve power due to loss of patients to follow up. The study was designed to be double-blinded, but since double-dummy was not utilized on the intervention medications with different frequencies, bias could have been introduced. The frequency of the interventions were different leading to patients potentially being able to identify which therapy they were receiving. Additionally, the ibuprofen dosage is lower than recommended in other countries. The recommended dose in the United States for osteoarthritis is 400 mg initially then every 4 to 6 hours as needed. ¹¹ Increasing the dose could improve generalizability of results. There was a lack of endpoints assessing impact on osteoarthritis beyond basic assessment of pain scores. Last, the study states that it is a noninferiority trial; however, there was no evidence of a noninferiority analysis.

In a double-blind noninferiority trial the efficacy and safety of C domestica extract was compared to ibuprofen. ¹² Patients included in this study had primary knee osteoarthritis according to American College of Rheumatology criteria, were 50 years or greater, and had a pain score of at least 5 on an 11-point Likert-type scale. Patients were excluded if they had abnormal liver/renal function, history of peptic ulcer, allergy to curcumin or ibuprofen, or if they were unable to walk. Patients were randomized to receive 1500 mg/d of C domestica extract (n = 185) or 1200 mg/d of ibuprofen (n = 182). Both groups were given their designated medication in capsule form and instructed to take 2 capsules by mouth 3 times daily after meals. The primary endpoints in this study were the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and a 6-minute walk distance. Noninferiority was determined in WOMAC scores if the scores between the groups were within 0.5 points. These endpoints were measured after 2 and 4 weeks of therapy.

Baseline characteristic were similar between groups. ¹² The average total WOMAC scores were 5.3 ± 1.8 in the C domestica extract group and 5.2 ± 1.7 in the ibuprofen group. The average 6-minute walk distances were 310.9 ± 84.8 and 304.5 ± 83.8 m for the C domestica extract and ibuprofen groups, respectively. After 4 weeks of therapy, the WOMAC total scores were 3.36 ± 2.04 for C domestica and 3.23 ± 1.97 for ibuprofen (mean difference −0.07, 95% CI −0.43 to 0.29, P = .010 for noninferiority). The 6-minute walk distances were 345.4 ± 91.7 and 348.0 ± 86.6 m with C domestica and ibuprofen, respectively (mean difference 7.2 m, 95% CI −7.0 to 21.4 m, P = .320 for equality). The number of abdominal pain/distention adverse effects were significantly lower in the C domestica extract group, but no other significant differences were seen between treatment groups.

One limitation of this study was that the average baseline pain score was approximately 5 and ranged from 3 to 7. While this meets the inclusion criteria, including patients with lower pain scores may have made the result of noninferiority more favorable. Additionally, patients in this trial were allowed to use tramadol as a rescue medication. Although few patients used rescue therapy in the study and likely had minimal impact on results, this approach may not necessarily reflect current practices. ³

A randomized, double-blind, prospective study was conducted to test the hypothesis that combination therapy of curcumin and diclofenac would reduce inflammation in osteoarthritis. ¹³ Patients had to be 38 to 80 years of age meet American College of Rheumatology criteria for knee osteoarthritis. Patients with inflammatory arthritis or had a contraindication to NSAIDs were excluded from the study. The key endpoints were pain (rated on an 11-point VAS) and knee-related quality of life (measured on a 100-point scale), as a component of the Knee Injury and Osteoarthritis Outcome Score (KOOS). There were 44 patients who were randomized to receive diclofenac 25 mg 3 times daily plus placebo and 44 patients randomized to diclofenac 25 mg 3 times daily plus two 250-mg curcuminoid capsules twice daily. Both groups received treatment for 3 months.

Baseline pain scores were 5.3 and 5.5 for placebo and curcuminoids groups, respectively (variability not provided). ¹³ Baseline knee-related quality of life scores were 46.6 with placebo versus 43.8 with curcuminoids. Approximately 83% were female in both groups. The majority of patients (72%) were age 55 to 74 years. Mean pain scores decreased by 1.8 (P < .001) with placebo compared with 2.3 with curcuminoids (P < .001). There was no significant difference in pain scores between groups (P = .441). Knee-related quality of life scores were increased by 12.7 (P = .005) with placebo and 13.5 with curcuminoids (P < .001) compared with baseline. There was no significant difference in knee-related quality of life scores between groups (P = .662).

There were multiple limitations to the study. Sample size calculations and assumed factors were not described. Furthermore, the only baseline characteristics stated were age and gender. Baseline characteristics such as disease severity and prior therapy would be beneficial to better understand the study population and how it compares with clinical practice. Additionally, analysis of variance was used to analyze pain scores; it would have been more appropriate to account for potential confounding variables, such as varying pain scores, at baseline using analysis of covariance. Last, both groups received diclofenac 25 mg 3 times daily, which is a lower than the recommended dosing in the United States of 150 to 200 mg daily in 3 to 4 divided doses. ¹¹ The lack of effect of curcuma with low-dose diclofenac suggests that minimal effect would also be seen with normal dosing.

A randomized, double-blind, placebo-controlled, prospective clinical trial was performed to examine the efficacy of Theracurmin. ¹⁴ Theracurmin is a specially designed surface-controlled water-dispersible curcumin that has increased bioavailability over standard curcumin powder. Patients included in this study had primary medial osteoarthritis, were older than 40 years of age, and had Kellgren-Lawrence grades of II or III on radiographic classification. Patients were excluded if they had a previous knee surgeries, knee injection treatment during the study, knee steroid injections within 2 months of the study, or other steroid use within 4 weeks of the study. Fifty patients were randomized in a 1:1 ratio to receive either Theracurmin or placebo twice daily for 8 weeks of treatment. Theracurmin therapy consisted of 6 capsules taken daily, which equaled 180 mg of curcumin. Patients were prescribed celecoxib (200 mg per day) if needed, and the use of pain patches was allowed. Patients were evaluated at baseline and at weeks 2, 4, 6, and 8. Improvement in patients’ knee symptoms was measured using the Japanese Knee Osteoarthritis Measure (JKOM), the knee pain VAS included in JKOM, and the knee scoring system of the Japanese Orthopedic Association (JOA). The JKOM had 25 questions that focus on pain and stiffness, condition in daily life, general activities, and health conditions. The VAS and how it was measured was not defined. The JOA knee scoring system is a 100 point scale that assesses ability to walk (30 points), ability to climb up and down stairs (25 points), range of motion (35 points), and joint swelling (10 points).

In both treatment groups, the majority of patients were female (78%) and had a Kellgren-Lawrence grade of II (72% Theracurmin, 87% placebo). ¹⁴ No significant difference was found in baseline characteristics between both groups. Baseline VAS scores were 0.52 ± 0.24 and 0.42 ± 0.25 for the Theracurmin and placebo groups, respectively, with scores of 0.20 and 0.21 (no variability results provided) at the end of study (P = .023 for the difference in change between groups). There was no significant difference found in JOA total and subcategory scores between treatment groups. Specific results were not provided for JKOM or JOA scores. After 8 weeks, there were significantly lower usage of celecoxib in the Theracurmin groups (approximately 30%, only graphical results provided) compared with placebo (approximately 60%, P = .0252). No serious adverse events were reported during this study.

One major limitation to this study was that the authors did not conduct a power calculation, define all VAS assessments, or consistently provide measured effect sizes with variability. With the lower number of patients enrolled and lack of significant findings for secondary endpoints, a type II error could be possible. Another limitation was that a majority of the patients were classified as having a Kellgren-Lawerence grade of II. A more balanced patient population between grades II and III may increase the external validity of this study; this imbalance could also have been corrected with use of analysis of covariance. A third limitation to this study was that the authors stated that pain patches were allowed in both groups, but the authors did not address what pain patches were allowed or the use of pain patches in either group.

A randomized, single-blinded, placebo-controlled clinical trial was performed to investigate the safety and efficacy of C longa extract (NR-INF-02) in comparison with placebo, glucosamine sulfate (GS), and combination of NR-INF-02 with GS. ¹⁵ Patients had to be older than 40 years of age with clinical evidence confirming knee osteoarthritis. Patients were excluded from the study if any of the following were met: concurrent medical or arthritic conditions confounding evaluation for knee osteoarthritis, patella-femoral disease, history of significant trauma/surgery to affected knee, or coexisting diseases that could prohibit successful completion of the trial. The key endpoints were severity of osteoarthritis pain, functional assessment of affected knee, and improvement in patient’s overall condition. Pain was rated on a 100 point VAS whereas patient’s functional assessment was measured using the WOMAC scale. There were 30 patients randomized to each of the following groups: placebo capsule twice daily in the morning and at night; one NF-INF-02 500-mg capsule twice daily; 2 GS 375-mg capsules twice daily; one NF-INF-02 500-mg capsule twice daily concurrently with 2 GS 375-mg capsules twice daily. Patients were assessed at baseline, day 21, and day 42. Participants were allowed to use acetaminophen as rescue therapy throughout the study period.

Most patients were between 50 and 60 years of age. The majority of patients (71%) reported moderate osteoarthritis severity. ¹⁵ Approximately 27% and 14% of patients were taking a local or oral NSAID, respectively. Average baseline VAS pain scores were between 60 and 67 for each group. Average WOMAC scores among all groups at baseline were between 54 and 61 for each group. Following 6 weeks of treatment, VAS pain scores were 46.03 ± 20.84 for placebo, 19.48 ± 17.84 for NR-INF-02, 29.29 ± 20.58 for GS, and 36.33 ± 28.99 in the combination group (P < .01 for each group compared to baseline). Both NR-INF-02 and GS alone had significant decreases in comparison to both placebo and combination therapy (P < .05). End of study WOMAC scores were 47.90 ± 12.59 for placebo, 27.14 ± 16.13 in the NF-INF-02 group, 34.92 ± 19.48 for GS, and 36.21 ± 24.74 in the combination group at day 42 (P < .01 for each group compared with baseline). The NR-INF-02 group had a significant decrease in WOMAC scores in comparison with placebo and combination therapy (P < .05). There were 2 reports (6.6%) of dyspepsia in the NR-INF-02 group. In the combination group, there were 4 total reports of adverse events including cough, dyspepsia, fever, and pedal edema.

There were several limitations to the study. First, the study was single-blinded, which could increase the risk for bias. The lack of double-dummy could have caused the participants to be aware of treatment groups despite the use of single-blinding. Last, while patients were randomized to treatment groups, there were significant differences between groups regarding gender and oral NSAID intake. There were significantly more females in both the GS and combination therapy. Additionally, significantly more patients in the placebo and combination group reported oral NSAID intake, potentially confounding study results.

A comparative, active-controlled clinical trial was performed to evaluate the long-term safety and efficacy of Meriva, a curcumin-phosphatidylcholine complex, for the treatment of osteoarthritis. ¹⁶ Patients had to meet the ACR criteria for grade 1 or 2 primary knee osteoarthritis diagnosed with radiographic investigation. Additionally, eligible patients had to have mild to moderate pain that was not adequately controlled with anti-inflammatory medications. Key exclusion criteria in this study included: cardiovascular disease requiring medication therapy, diabetes, body mass index greater than 25 kg/m², severe metabolic diseases, or severe bone/joint deformation or condition making the patient unable to walk. The control is not clearly defined but was described as “best available treatment.” The 50 patients in the treatment group received the “best available treatment” and Meriva therapy for 8 months. Meriva therapy consists of one 500-mg tablet twice daily after breakfast and dinner, which corresponds to approximately 200 mg curcumin daily. There were 50 patients allocated to each treatment arm. The key endpoints were improvements in functional impairment and symptoms of osteoarthritis. Functional impairment was measured using the Karnofsky Performance Scale Index (KPSI), where 100 is defined as no evidence of disease and 0 means deceased. Signs and symptoms of osteoarthritis were evaluated using a 96-point WOMAC questionnaire. Patients were assessed at baseline and month eight. Participants were allowed to use NSAIDs in addition to the intervention during the study.

Mean age of the patients was 44 years. Median baseline KPSI scores were 73.3 (range 57-79.4) with treatment and 74.2 (range 58-83) with control. ¹⁶ Mean baseline global WOMAC scores were 81.2 in the treatment group and 79.6 in the control group (variability not provided). After 8 months of treatment, KPSI scores increased to 92.2 (range 88-100) in the treatment group (P < .05) and 81 (range 71-86.3) with control (P > .05). Global WOMAC scores decreased to 33.3 with treatment (P < .05) and 68.8 with control (P > .05). Adverse events were not reported in the study.

There were a number of limitations to the study. Blinding, the number of patients screened, and the number of patients that completed this study were not discussed, increasing the potential for investigator bias. Second, the control was not defined and both groups received “the best available therapy,” which makes it difficult to determine clinical relevance of results. Another limitation is the extensive exclusion criteria which decreases external validity of the trial. Last, the study did not directly compare treatment groups; rather, only within group comparisons to baseline characteristics were provided. This limits the ability to draw conclusions regarding comparative efficacy.

A randomized, active control clinical trial was conducted to examine the efficacy of C longa and Boswellia serrata combination compared with celecoxib. ¹⁷ Patients included in this study were between the ages of 18 and 65 years, were medically stable, and had moderate knee osteoarthritis, defined as evidence of narrowing of the medial joint space with swelling. Patients were excluded if they had the following: long-standing osteoarthritis with gross deformity, severe osteoarthritis, restricted mobility due to swelling, were nursing, history of rheumatoid/reactive arthritis, significant systemic disease, drug/alcohol abuse, malnutrition, or any other condition that could place the patient at risk or interfere with the study results as determined by the principal investigator. Patients were randomized to receive either a combination formulation of 350 mg of C longa extract and 150 mg of B serrata extract twice daily or celecoxib 100 mg twice daily. Both groups received treatment for 12 weeks. The endpoints examined were symptom scoring and clinical examination. Symptoms scored included joint pain and walking distance. The clinical examination included joint tenderness, crepitus, bilateral measurements for welling using a tape measure, range of movement, thigh measurements, warmth of affected joint, and gait.

A total of 54 patients were screened for this study, and 30 patients enrolled. ¹⁷ Baseline characteristics were reported by the author to be similar between groups. The average age for the combination group was 49.7 ± 8.2 years and for the celecoxib group it was 47.2 ± 9.7 years. At baseline, 86% and 79% were reported to have moderate or severe pain in the combination and celecoxib groups, respectively. At week 12, 21% and 50% of patients had moderate/severe pain in the combination and celecoxib groups, respectively (P < .05 for each group compared with baseline; P > .05 between groups at end of study). Approximately 21% of combination and 29% of celecoxib patients could walk more than 1000 meters at baseline, compared with 93% and 86% of patients after 12 weeks of treatment (P < .05 for each group compared with baseline; P > .05 between groups at end of study). There were no adverse events reported in the study. Despite both groups showing significant improvements in osteoarthritis, the authors concluded that the combination formulation was superior to celecoxib for treating active osteoarthritis.

The authors’ conclusion in this study was a major limitation to the study. Although it is claimed that the combination formulation was superior, there no additional analyses reported beyond the comparison of outcomes between groups which showed no difference. Blinding approach was not discussed, increasing risk for bias. Another limitation was how the data were analyzed. Analysis of variance was used for all data; however, differences in baseline pain level should have been corrected for using analysis of covariance for continuous data (eg, pain scores) and Cochran-Mantel-Haenszel for nominal data (eg, pain categories, such as mild, moderate, etc). On a related note, it is generally more common to assess pain levels on a continuous or ordinal scale, rather than as nominal data. Additionally, the authors did report the use of a power calculation, effect size, and pertinent baseline characteristics. Pertinent baseline characteristics that should have been included were use of pain medications and severity and duration of osteoarthritis. Last, the use of this specific combination formulation could make it difficult to apply these results to clinical practice, as this specific formulation is not widely used.

A prospective, double-blind, placebo-controlled, crossover study was performed to evaluate the effectiveness of a curcumin-containing herbomineral formulation in patients with osteoarthritis. ¹⁸ The herbomineral formulation contained Withania somnifera (Ashwagandha, 450 mg), B serrata (100 mg), C longa (50 mg), and zinc complex (50 mg). Patients were included if they had symptoms of osteoarthritis and were attending the Rheumatology Clinic of the Sassoon General Hospital. Additionally, they were included if they had pain, morning stiffness, stiffness/joint swelling, and disability/loss of function due to joint deformity with radiological changes. Patients were excluded if they had diabetes, hypertension, peptic ulcer, renal failure, liver disorders, or were pregnant. Patients were first placed on a 1-month pretreatment period where all of their previous drug therapies were withdrawn and they were evaluated weekly. Then patients were given either the herbomineral formulation or placebo and instructed to take 2 capsules every 8 hours after food for 3 months. After a 2-week washout period, patients were switched to the other treatment group for 3 more months of therapy. Patients were evaluated weekly using pain scores, morning stiffness, Ritchie articular index, joint score (as defined by American College of Rheumatology), disability score, and grip strength. Additionally, erythrocyte sedimentation rate was monitored monthly, and the patients were also asked for their preference of treatment at the conclusion of the study.

There were 42 patients in this study that had average age of 48.4 ± 2.6 years, disease duration of 4.6 ± 0.8 years, and were predominately female (n = 32). ¹⁸ A significant decrease in pain scores from 4.4 ± 1.2 to 0.4 ± 0.1 was seen in the treatment formulation (P < .001) versus placebo (4.2 ± 0.2). There was also a significant decrease in disability score from 3.1 ± 0.9 to 0.9 ± 0.2 (P < .05) in the formulation versus placebo (2.6 ± 0.8). No significant differences were found in morning stiffness, Ritchie articular index, joint score, or grip strength. A majority of patients (n = 39) indicated that they preferred the formulation therapy over placebo. No major adverse events that would cause discontinuation were reported.

A major limitation to this study is that since a formulation of multiple products is used, it is difficult to determine which ingredients most contribute to results. Also, there was no mention of effect size or power calculation performed, and there was no distinction between primary or secondary endpoints. Because of these limitations, it is unclear if a type II error existed for endpoints that did not find statistically significant differences. The authors also did not address how some assessment criteria were performed. One example is the pain score. The authors did discuss their pain scale or how patients would use the scale. Additionally, in this crossover trial, baseline characteristics were only provided at the beginning of the overall study, rather than each treatment period. It is possible that results from these time frames could have varied. One final limitation to note is that the exclusion criteria included some very common disease states, such as diabetes and hypertension. This is a major cause of decreased external validity.

Discussion

The clinical trials included in this review represent the highest level of available evidence regarding the use of curcuma for knee osteoarthritis. Curcuma-containing products generally demonstrated statistically significant improvement in osteoarthritis-related endpoints compared to placebo, ^{14 –16,18} with one exception. ¹¹ When compared with active control, curcuma-containing products were similar compared with NSAIDs, ^10,12,17 and potentially to glucosamine. ¹⁵ However, due to consistent presence of major study limitations, the clinical significance of these results remains uncertain. Key study limitations identified were small sample size, poor overall study design, and issues with baseline characteristics. Another issue is the use of subjective scales to assess pain and functionality. Validated scales, such as WOMAC, were used in half of the studies while other studies relied on subjective measures to assess pain. Additionally, there was a lack of consistency with study formulations, which can decrease the generalizability of the results. Last, the safety profile in patients with multiple comorbidities was not fully evaluated in the clinical trials.

In its Extension for Reporting Herbal Medicinal Interventions, CONSORT provides recommendation for the reporting of dietary supplement clinical trials. ⁹ An adaptive CONSORT checklist was performed to objectively assess the strengths and weaknesses of the included clinical trials (Table 2). All trials provided a statement of reasoning, eligibility criteria, the content of all herbal products per dosage unit form, described the statistical methods used, and provided a summary of results per group. The majority of studies did not account for dates of recruitment and follow-up, did not state the Latin binomial and common name, or provide a rationale for the type of control. Overall the greatest weaknesses identified by the CONSORT were lack of clearly defined outcomes, no sample size determined, and failure to disclose randomization methods.

Future curcuma studies should use commonly available formulations, conduct appropriate sample size calculations, and include patients with various disease severities and pain score. While historically oral NSAIDs have been the main comparator group for curcuma, it would be beneficial to compare it to first-line therapies for osteoarthritis, such as acetaminophen. Furthermore, to better evaluate the safety profile of curcuma with other medical therapies, patients with multiple comorbidities should be included in future studies. Non-pharmacological therapy is an important component in the treatment of osteoarthritis. It would be beneficial for the clinical trials supporting curcuma to address nonpharmacological recommendations for patients alongside the intervention medication.

In summary, further studies need to be performed to define curcuma’s place in therapy for osteoarthritis. Curcuma has shown a well-tolerated adverse event profile, and demonstrated potential benefit in reduction of pain due to osteoarthritis. However, because of major limitations of previous trials, patients should use currently recommended nonpharmacological and pharmacological therapies prior to consider curcuma as alternative therapy.

Conclusion

Published clinical trials evaluating curcuma formulations for treatment of osteoporosis have found similar efficacy compared to NSAIDs, and potentially to glucosamine; however, they also contain significant limitations that call into question validity of the results While statistical significant differences in outcomes were reported in a majority of studies, the small magnitude of effect and presence of major study limitations hinder application of these results. Further rigorous studies are needed prior to recommending curcuma as an effective alternative therapy for knee osteoarthritis.