Abstract

Objectives. A randomized, placebo‐controlled, double‐blind trial of the relative effectiveness of glucosamine sulphate and placebo in managing pain in osteoarthritis (OA) of the knee.

Methods. Eighty patients with OA of the knee were recruited from a rheumatology out‐patient clinic and received either glucosamine sulphate 1500 mg daily for 6 months or dummy placebo. The primary outcome measure was patients' global assessment of pain in the affected knee.

Results. Area under the curve analysis for the primary outcome measure showed no difference between placebo and glucosamine [mean difference 0.15 mm, 95% confidence interval (CI) −8.78 to 9.07]. The placebo response was 33%. There was a statistically significant difference between groups in knee flexion (mean difference 13°, 95% CI −23.13 to −1.97), but this difference was small and could have been due to measurement error.

Conclusions. As a symptom modifier in OA patients with a wide range of pain severities, glucosamine sulphate was no more effective than placebo.

Glucosamine sulphate, a nutrient supplement, is available for purchase from health food shops and pharmacies and is taken in the UK, USA and throughout Europe for the relief of musculoskeletal symptoms [1]. It has been estimated that over £10 million per annum is spent in the UK on glucosamine sulphate (Health Perception UK, personal communication) and more than 15 different companies now market these products for the relief of ‘arthritic symptoms’. Previous studies support the efficacy of glucosamine in relieving pain in osteoarthritis (OA) [2]. However, systematic reviews of the evidence have highlighted variable trial designs and the use of inadequate measures of outcome [35]. A recent study in patients with mild to moderate OA suggested that glucosamine had long‐term structure‐modifying and some symptom‐modifying effects [6]. Well‐designed clinical trials of complementary therapies such as glucosamine sulphate are indicated to determine efficacy and allow practitioners to give patients appropriate advice about glucosamine [7].

In this paper we report the results of a pragmatic, randomized, placebo‐controlled, trial of glucosamine sulphate for the treatment of pain in knee OA in UK patients.

Patients and methods

Participants

Between April and October 1998, ambulatory male and female patients aged over 40 yr with radiologically defined, symptomatic OA of at least one knee [8] were recruited from a hospital rheumatology out‐patient department. The inclusion criteria dictated that pain or discomfort should have been experienced in the affected knee on most days for the previous 3 months. The trial design was pragmatic, with inclusion of patients with a wide range of pain severity and all grades of radiological severity of OA of the knee. Table 1 gives the baseline characteristics of the patients in both treatment groups.

Patients who had an arthroplasty of one knee were not excluded as long as the other knee was affected and was symptomatic. Patients with bilateral symptomatic OA were asked to specify their worst knee at baseline and this knee was evaluated throughout the study period. Patients were excluded if they had prosthetic material in both knees, had previously taken glucosamine, had received an arthroscopic washout of the index knee in the previous 3 months or had any intra‐articular injection to the knee within the previous month. Full ethical approval was obtained from North West Surrey Ethics Committee and written informed consent was obtained from all participants.

Table 1. 

Baseline characteristics of the sample (n=80)


 
Mean (s.d.) or %
 
Percentage women 68% 
Age (yr) 62.28 (9.12) 
Duration of OA (yr)  7.62 (8.06) 
Kellgren and Lawrence grade 1  9.3% 
Kellgren and Lawrence grade 2 30.7% 
Kellgren and Lawrence grade 3 37.3% 
Kellgren and Lawrence grade 4 22.7% 
Global pain 53.41 (22.98) 
Pain on movement 60.69 (20.91) 
Pain at rest 34.95 (27.50) 
WOMAC pain  9.24 (3.50) 
WOMAC function 32.90 (13.81) 
WOMAC stiffness  4.45 (1.90) 
Current treatment—analgesia 22.5% 
Current treatment—NSAIDs 46.3% 
Current treatment—cod liver oil 27.5% 
Current treatment—other alternative remedies 22.5% 

 
Mean (s.d.) or %
 
Percentage women 68% 
Age (yr) 62.28 (9.12) 
Duration of OA (yr)  7.62 (8.06) 
Kellgren and Lawrence grade 1  9.3% 
Kellgren and Lawrence grade 2 30.7% 
Kellgren and Lawrence grade 3 37.3% 
Kellgren and Lawrence grade 4 22.7% 
Global pain 53.41 (22.98) 
Pain on movement 60.69 (20.91) 
Pain at rest 34.95 (27.50) 
WOMAC pain  9.24 (3.50) 
WOMAC function 32.90 (13.81) 
WOMAC stiffness  4.45 (1.90) 
Current treatment—analgesia 22.5% 
Current treatment—NSAIDs 46.3% 
Current treatment—cod liver oil 27.5% 
Current treatment—other alternative remedies 22.5% 

Interventions

Intervention group.

Glucosamine sulphate 1500 mg was taken as 500 mg capsules three times daily for 6 months. Glucosamine sulphate capsules manufactured by Health Perception UK contain 500 mg of potassium chloride‐free glucosamine sulphate, 300 mg of vitamin C, 300 mg of calcium carbonate and 5 mg of manganese.

Placebo control group.

Dummy placebo capsules containing calcium carbonate were taken three times daily for 6 months. The dummy placebo capsules were also manufactured by Health Perception UK specifically for the purposes of this study to ensure the placebo was a true dummy for the active glucosamine sulphate.

All participants were permitted to continue their existing non‐steroidal anti‐inflammatory drug (NSAID) medication and encouraged to avoid changing their dose or medication during the course of the study. Use of NSAIDs was recorded at each follow‐up assessment. The patients were allowed access to paracetamol or other proprietary or prescribed simple analgesia and reported use of analgesia was collected using a patient diary and recorded at each follow‐up assessment. Adherence to trial medication was assessed using a returned capsule count at each follow‐up visit.

Outcome measurement and follow‐up

The patients were assessed four times during the study period, at 0, 6, 12 and 24 weeks. At each visit vital signs were recorded (pulse, blood pressure, weight) and blood and urine were tested according to protocol at 0 and 24 weeks (full blood count, liver and kidney function tests and urinalysis by stick testing). Outcome measures were collected at each visit. Four 100 mm visual analogue scales (VAS) were used to measure pain in the affected knee at rest, on movement and the patient's overall assessment of pain and stiffness. VAS are scored from 0 to 100, where 0 indicates no pain and 100 is the worst pain imaginable. The primary outcome measure was the VAS overall assessment of pain in the affected knee. The WOMAC Osteoarthritis Index [9] was used to evaluate functional activity and the McGill pain questionnaire [10] assessed the affective and sensory components of pain. The functional subscale of the WOMAC produces a range of scores from 0 to 68, where 0 indicates no functional disability and 68 indicates extreme difficulty performing daily living tasks. The pain and stiffness subscales of the WOMAC have scores of 0–20 and 0–8, respectively, with 0 representing no pain or stiffness. Similarly, high scores on the McGill pain questionnaire indicate more pain.

All outcome measures are well validated and have been used extensively in clinical trials in rheumatology. The VAS pain scales, the WOMAC questionnaire and the McGill pain questionnaire were self‐administered by the patients.

The range of movement of the knees in flexion and extension was measured using a long arm goniometer following a standardized method recommended for knee OA [11]. In this method, the patient was examined lying down and all measurements were taken from the lateral aspect of the knee using only active movements. The central pivot of the goniometer was placed over the midpoint of the lateral joint margin with the upper arm of the goniometer aligned with the greater trochanter and the lower arm with the lateral malleolus. The neutral position was taken as zero unless the knee was hyperextended (when a minus value was recorded) or flexed (when a positive value was recorded). The patient was then asked to extend and flex their knee to its limit and readings were taken from the goniometer. This method has intra‐ and inter‐rater reliability coefficients of 0.89–0.91.

Use of rescue analgesia and adherence to trial medication were recorded at each follow‐up assessment. Adverse events, patient withdrawals and concomitant illnesses were recorded in accordance with good clinical practice guidelines. The same research nurse, blind to the patients' treatment allocations, made all study measurements on each patient at every visit.

Study power

Power calculations based on published data for patients with OA indicated that 38 patients were required in each treatment group to detect a 20 mm difference in VAS overall pain score with 90% power at 1% significance (two‐sided). Twenty millimetres has been identified as the minimum clinically important difference in pain between treatment groups in knee OA studies [12, 13]. Recalculating the power of this study [14] using pain data obtained from participating patients, confirms that the study had 90% power to detect this difference at the 1% level of significance and 97% power to detect this difference at the 5% level.

Statistical analysis

The data were analysed using SPSS.PC version 8.0 statistical software. Analyses were performed on the basis of intention to treat. In the primary analysis, area under the curve (AUC) statistics were calculated for overall pain VAS. For non‐completers, AUC was calculated by carrying forward their last recorded pain VAS. This enabled data to be used from all patients who had completed at least 6 weeks of follow‐up. Mean AUC pain scores for each treatment group were then compared using two‐sided t‐tests. Secondary analyses involved AUC analyses for all outcome measures for which there were serial data and calculation of change scores for data with two measurements (range of motion). The patients were then classified as responders or non‐responders to treatment on the basis of the Osteoarthritis Research Society International (OARSI) responder criteria for oral NSAIDs in knee OA [15]. These criteria suggest that responders should experience a relative reduction in pain of 45%, with a minimum absolute reduction of 20 mm on a VAS. Patients who experience a more modest reduction in pain (relative reduction 15%, with a minimum absolute reduction of 10 mm) can also be classified as responders if they additionally report an improvement in function (relative change 30%, minimum absolute change 15 points on the WOMAC) or in patient global assessment (relative change 35%, minimum absolute change 10 points). Patient global assessment (of well‐being) was not recorded in this trial, so the criteria for large and moderate reductions in pain and function were used.

Assignment

Simple randomization based on random numbers tables was used to assign patients to treatment groups. The patients were assigned randomization numbers sequentially on recruitment to the study and the randomization codes were held by the company who had manufactured the glucosamine and the dummy placebo (Health Perception UK).

Masking

The trial medication for each randomization number was packaged and labelled by the manufacturer (Health Perception UK) and stored in the pharmacy department in the hospital who were responsible for dispensing the medication at each follow‐up visit. The packaging and appearance of glucosamine and placebo were identical. Allocation was pre‐assigned on the basis of randomization numbers and was concealed from the patients, the investigator and the research nurse who recruited and assessed participating patients, and the pharmacist who dispensed the trial medication. Allocation was revealed when all patients had completed the full follow‐up period.

Results

Participant flow and follow‐up

Participant flow through the trial is depicted in Fig. 1.

Eighty patients with OA of the knee were randomized. All patients received the trial medication immediately after randomization. Five patients (two in the glucosamine group and three in the placebo group) withdrew or were lost to follow‐up. Only one patient (placebo) withdrew because of patient‐perceived lack of efficacy. Other reasons for withdrawal included: side‐effects (constipation in a placebo patient); moved away (glucosamine); unable/unwilling to attend follow‐up assessments (glucosamine); withdrawal of consent because of family pressure (placebo). No efficacy data are available for two patients (one glucosamine and one placebo) who withdrew before their 6‐week assessment and they were therefore not included in the efficacy analysis. The other three patients withdrew after their 6‐week assessment and are included in the primary analysis by calculating the time‐corrected AUC statistics for reported pain to week 6.

Fig. 1. 

Flow chart showing the movement of patients through the trial.

Fig. 1. 

Flow chart showing the movement of patients through the trial.

Analysis

The sample had a mean age of 62 yr (s.d. 9.11, range 42–94 yr) and a median disease duration of 5 yr (range 1–48 yr). Sixty‐eight per cent (n=54) were female and half the sample identified the right knee as their worst/affected side. The participants reflected a spectrum of symptom and disease severity: median global pain score 52 mm (range 6–96), median stiffness 57 mm (range 0–98), median WOMAC function score 36 (range 2–59), 9% (n=7) Kellgren and Lawrence grade 1, 31% (n=23) grade 2, 37% (n=28) grade 3 and 23% (n=17) grade 4. At baseline, 22.5% were taking analgesics, 46% NSAIDs, 27.5% cod liver oil and 22.5% other complementary therapies.

Baseline comparisons between the treatment groups on demographic variables and all outcome variables confirmed that the randomization had been effective; there were no significant differences between treatment groups on any of the outcome, disease, treatment or demographic variables.

Efficacy data

There were no statistically significant differences in the primary outcome between groups at any time point during the study (Fig. 2).

AUC analyses for the primary outcome and all outcome variables for which repeated measures data were available revealed no significant differences between the treatment groups (Table 2).

Statistically significant differences were found between treatment groups in the degree of knee flexion [mean difference 13°, t=−2.36, P=0.02, 95% confidence interval (CI) −23.13 to −1.97] post‐treatment. The glucosamine group had a small increase in flexion [mean increase (s.d.) 4° (15.62)], whilst flexion in the placebo group decreased by 8.55° (s.d. 28.76). However, these changes are smaller than the minimum detectable change in flexion using a goniometer and could therefore be due to measurement error. There was no significant difference in the use of rescue analgesia between the glucosamine (mean number of paracetamol tablets taken 43, s.d. 63.92, range 0–252) and placebo groups (mean number of paracetamol tablets taken 45, s.d. 75.64, range 0–264). Nor were there any changes in NSAID use in either of the treatment groups.

OARSI response criteria [15] for relative and absolute reductions in pain and disability were used to classify the patients. The patients could be classified as responders to treatment in two ways:

if their percentage reduction in global pain VAS (baseline to 24 weeks) was greater than 45% and was a minimum absolute reduction of 20 mm);

if their percentage reduction in global pain VAS was less than 45% but greater than 15% (and was a minimum absolute reduction of 10 mm) and their percentage reduction in WOMAC function score was greater than 30% (and was a minimum absolute reduction of 10 points).

Treatment response is summarized in Table 3. There were no statistically significant differences in response between placebo and glucosamine (χ2 statistic 0.006, P=0.94).

Fig. 2. 

Plot of mean VAS global pain scores for the affected knee in the glucosamine and placebo groups at each of the assessment points.

Fig. 2. 

Plot of mean VAS global pain scores for the affected knee in the glucosamine and placebo groups at each of the assessment points.

Table 2. 

Comparisons between the glucosamine and placebo treatment groups on AUC analyses for repeated measures outcomes using two‐tailed t‐tests. Mean AUC statistics give the mean total amount of that variable experienced by each treatment group over the study period (n=39 for both groups)

Outcome variable
 
Treatment group
 
Mean
 
s.d.
 
t
 
P
 
95% CI for the mean difference
 
Global pain Placebo 1065.45 398.07    
 Glucosamine 1081.28 577.69 −0.14 0.89 −236.66 to 205.01 
Pain on movement Placebo 1080.45 456.21    
 Glucosamine 1091.10 629.77 −0.09 0.93 −255.82 to 234.52 
Pain at rest Placebo 561.75 361.76    
 Glucosamine 713.02 562.25 −1.43 0.16 −362.30 to 59.75 
McGill affective Placebo 65.25 56.83    
 Glucosamine 63.30 56.87 0.15 0.88 −23.36 to 27.26 
McGill sensory Placebo 380.18 149.20    
 Glucosamine 342.75 191.33 0.98 0.33 −39.02 to 113.87 
WOMAC pain Placebo 179.32 69.96    
 Glucosamine 184.88 98.79 −0.29 0.77 −43.72 to 32.62 
WOMAC function Placebo 625.20 301.92    
 Glucosamine 665.10 394.42 −0.51 0.61 −196.42 to 116.62 
WOMAC stiffness Placebo 82.28 33.88    
 Glucosamine 87.98 47.70 −0.62 0.54 −24.15 to 12.75 
Outcome variable
 
Treatment group
 
Mean
 
s.d.
 
t
 
P
 
95% CI for the mean difference
 
Global pain Placebo 1065.45 398.07    
 Glucosamine 1081.28 577.69 −0.14 0.89 −236.66 to 205.01 
Pain on movement Placebo 1080.45 456.21    
 Glucosamine 1091.10 629.77 −0.09 0.93 −255.82 to 234.52 
Pain at rest Placebo 561.75 361.76    
 Glucosamine 713.02 562.25 −1.43 0.16 −362.30 to 59.75 
McGill affective Placebo 65.25 56.83    
 Glucosamine 63.30 56.87 0.15 0.88 −23.36 to 27.26 
McGill sensory Placebo 380.18 149.20    
 Glucosamine 342.75 191.33 0.98 0.33 −39.02 to 113.87 
WOMAC pain Placebo 179.32 69.96    
 Glucosamine 184.88 98.79 −0.29 0.77 −43.72 to 32.62 
WOMAC function Placebo 625.20 301.92    
 Glucosamine 665.10 394.42 −0.51 0.61 −196.42 to 116.62 
WOMAC stiffness Placebo 82.28 33.88    
 Glucosamine 87.98 47.70 −0.62 0.54 −24.15 to 12.75 
Table 3. 

Classification of patients as treatment responders using OARSI response criteria

Response criteria
 
Glucosamine responders
 
Placebo responders
 
Step 1   
Percentage reduction in global 29% 22% 
   pain >45% and minimum absolute   
   reduction of 20 mm on pain VAS   
Step 2   
Percentage reduction in global  3% 11% 
   pain <45% but >15% (and a   
   minimum absolute reduction of 10 mm)   
   and percentage reduction in WOMAC   
   function score >30% (and a minimum   
   absolute reduction of 10 points) 
Total responders from steps 1 and 2 32% 33% 
   combined   
Response criteria
 
Glucosamine responders
 
Placebo responders
 
Step 1   
Percentage reduction in global 29% 22% 
   pain >45% and minimum absolute   
   reduction of 20 mm on pain VAS   
Step 2   
Percentage reduction in global  3% 11% 
   pain <45% but >15% (and a   
   minimum absolute reduction of 10 mm)   
   and percentage reduction in WOMAC   
   function score >30% (and a minimum   
   absolute reduction of 10 points) 
Total responders from steps 1 and 2 32% 33% 
   combined   

Adherence and adverse events

Adherence to prescribed medication is estimated at between 40 and 70% [16, 17]. Adherence to glucosamine and placebo in this trial (assessed by tablet counts) exceeded 80% (glucosamine 84.03%, s.d. 11.76; placebo 83.38%, s.d. 14.51).

No serious adverse effects of treatment were reported during the trial and there were no differences between the treatment groups in the numbers or severity of adverse events reported. There were no significant differences between blood or urine results between groups and no adverse effects of treatment were identified when these results were analysed on an individual patient basis. Table 4 lists all reported side‐effects.

Table 4. 

Frequency of reported side‐effects from the trial medication in the glucosamine and placebo treatment groups

 Glucosamine
 

 
Placebo
 

 
Side‐effect
 
Frequency
 
%
 
Frequency
 
%
 
None 15 37.5 13 32.5 
Increased musculoskeletal 22.5 22.5 
   pain     
Cold/flu 10 15 
Headache 10 7.5 
Constipation 7.5 
Diarrhoea 2.5 
Urinary tract infection 2.5 
Depression 2.5 
Headache and gastrointestinal 7.5 
Skin 2.5 
 Glucosamine
 

 
Placebo
 

 
Side‐effect
 
Frequency
 
%
 
Frequency
 
%
 
None 15 37.5 13 32.5 
Increased musculoskeletal 22.5 22.5 
   pain     
Cold/flu 10 15 
Headache 10 7.5 
Constipation 7.5 
Diarrhoea 2.5 
Urinary tract infection 2.5 
Depression 2.5 
Headache and gastrointestinal 7.5 
Skin 2.5 

Discussion

Although glucosamine has shown some promise as a chondroprotective agent in OA [6], most people take it in the hope of obtaining relief from pain in OA. A meta‐analysis of previous studies of glucosamine suggests that it may have some analgesic efficacy in OA [3]. This was supported by a recent, 3‐yr trial of glucosamine sulphate which reported statistically significant but small symptomatic improvement for glucosamine compared with placebo [6]. In contrast, the results from our study did not show any clinically or statistically significant analgesic effect of glucosamine in OA. There are a number of possible explanations for this apparent discrepancy.

First, the trial by Reginster et al. [6] was performed on patients with milder symptomatic and less severe structural OA than the patients in our trial. Seventy per cent of the patients in the trial by Reginster et al. had Kellgren and Lawrence grade 2 OA compared with 29% of the patients in our trial and whilst 21% of the patients in our study were Kellgren and Lawrence grade 4, none of the patients in the trial by Reginster et al. was greater than grade 3. Similarly, whilst the patients in our trial had mean WOMAC pain, function and stiffness scores that were around or over the midpoint on the scale, the mean scores for the patients in the trial by Reginster et al. were around or below the 40% mark. In addition, the patients in our trial were more likely to have received previous pharmacological treatment for their OA than the patients in the trial by Reginster et al. and a greater proportion of these were taking NSAIDs (60%) or a combination of NSAIDs and analgesics (22%) than the patients in the trial by Reginster et al. (24% NSAIDs, 8% NSAIDs and analgesics). Our trial was pragmatic and designed to reflect the range of severity of OA symptoms among the sort of UK patients who might be buying and taking glucosamine. The discrepancy between our results and those of the trial by Reginster et al. might suggest that glucosamine has an analgesic effect in mild to moderate OA but not at the more severe end of the spectrum. This hypothesis should be tested in studies adequately powered to perform subgroup analyses for pain and OA severity.

Second, the improvement in pain reported in the glucosamine group in the trial by Reginster et al. was of borderline statistical significance (P=0.047) and small (around 30 mm on a scale of 0–500 mm). There was no indication whether such a change was detectable by the patients, or whether they considered it important or adequate.

Although there were differences in inter‐participant variance in pain scores in our population of OA patients compared with the published data on which the sample size calculation was based, our trial still had 90% power to detect a difference in treatment effect between the groups at the 1% significance level and 97% power to detect a difference at the 5% significance level, suggesting the trial would have had sufficient power to detect a statistically significant treatment effect between the groups.

Glucosamine appears to be a safe compound when used in patients with OA and there were no adverse effects in the treatment group as judged by both symptoms and blood and urine testing. This compares favourably with the adverse effect profiles of other analgesic treatments for OA such as NSAIDs. Adherence to glucosamine treatment was high and probably reflects the lack of side‐effects and a favourable perception of glucosamine as a safe complementary supplement.

In this study we observed a strong analgesic placebo response with a total of 33% patients classified as responders according to OARSI responder criteria. This may be an indication of selection bias in this trial (patients with an affinity for complementary therapies may have been more likely to agree to participate) and has implications for the design of all trials of complementary therapies. The degree of response in some of these patients was impressive and we undertook a qualitative retrospective study on a sample of study patients prior to unblinding, the results of which are reported elsewhere [18].

Interest in complementary therapies in many areas of medicine is growing and patients with chronic diseases, concerned about the possible adverse effects of many orthodox treatments, frequently seek alternatives that are seen as ‘natural’ and therefore harmless [19]. Clinicians have a responsibility to offer rational, evidence‐based advice about complementary therapies, but in many cases this is restricted by the lack of scientifically rigorous evidence. This study has provided some evidence of the safety and relative analgesic effectiveness of glucosamine and, in the process, has highlighted the need for further research to give a definitive answer as to the analgesic role of glucosamine in OA.

Correspondence to: R. A. Hughes, Department of Rheumatology, Ashford and St Peter's Hospital Trust, Guildford Road, Chertsey, Surrey KT16 0PZ, UK.

We thank Health Perception UK for providing all the trial medication and packaging, for shipping the medication to the pharmacy, and for generating and managing the randomization list. Our thanks to the pharmacists at St Peter's Hospital for storing and dispensing all study medication and to all the patients who gave up their time to participate in the trial.

Rodney Hughes conceived the original idea for the study and wrote the protocol. Angela Huggett and Gillian Glover were the research nurses who recruited the patients, performed the assessments and entered all the outcome data on to the computer. Alison Carr carried out the data analysis. The paper was written by Rodney Hughes and Alison Carr. Rodney Hughes will act as guarantor for the paper.

The research nurses for this trial were funded by an educational grant from Health Perception UK. All trial medication, including dummy placebo, was provided by Health Perception UK. Health Perception UK did not have access to the raw data from the trial and were not involved in any of the data analysis.

Health Perception UK is a manufacturer of glucosamine sulphate.

References

1
Briffa J. A critical review of the use of glucosamine sulphate in the treatment of osteoarthritis.
Int J Alt Comp Med
 
1997
;
15
:
15
–6.
2
McQuay HJ, Moore RA (eds). Glucosamine and arthritis.
Bandolier
 
1997
;
4
:
1
–3.
3
McAlindon TE, LaValley MP, Gulin JP, Felson DT. Glucosamine and chondroitin for treatment of osteoarthritis: a systematic quality assessment and meta‐analysis.
J Am Med Assoc
 
2000
;
283
:
1469
–75.
4
Adams ME. Hype about glucosamine.
Lancet
 
1999
;
354
:
353
–4.
5
Hughes RA. Oral glucosamine sulphate in the treatment of osteoarthritis—A review of the literature.
Ann Rheum Dis
 
1999
;
58(EULAR Suppl.)
:
124 (abstract)
.
6
Reginster JY, Deroisy R, Rovati LC et al. Long‐term effects of glucosamine sulphate on osteoarthritis progression: a randomised, placebo‐controlled clinical trial.
Lancet
 
2001
;
357
:
251
–6.
7
Health Which? August
1999
:
4
.
8
Kellgren JH, Jeffrey M, Ball J.
Atlas of standard radiographs
 , Vol. 2, The epidemiology of chronic rheumatism. Oxford: Blackwell Scientific,
1963
.
9
Bellamy N, Buchanan WW, Goldsmith CH, Campbell J. Validation study of WOMAC: a health status instrument for measuring clinically important patient‐relevant outcomes following total hip or knee arthroplasty in osteoarthritis.
J Orthop Rheum
 
1988
;
1
:
95
–108.
10
Melzack R. The short‐form McGill Pain Questionnaire.
Pain
 
1987
;
30
:
191
–7.
11
Bellamy N.
Musculoskeletal clinical metrology
 . London: Kluwer,
1993
.
12
Bellamy N, Carette S, Ford PM et al. Osteoarthritis antirheumatic drug trials. II. Tables for calculating sample size for clinical trials.
J Rheumatol
 
1992
;
19
:
444
–50.
13
Bellamy N, Carette S, Ford PM et al. Osteoarthritis antirheumatic drug trials. III. Setting the delta for clinical trials—results of a consensus development (Delphi) exercise.
J Rheumatol
 
1992
;
19
:
451
–7.
14
Altman DG. Calculating power. In: Altman DG, ed.
Practical statistics for medical research
 . London: Chapman & Hall,
1997
:460.
15
Dougados M, LeClaire P, van der Heijde D, Bloch DA, Bellamy N, Altman D. A report of the Osteoarthritis Research Society International standing committee for clinical trials response criteria initiative.
Osteoarthritis Cartilage
 
2000
;
8
:
395
–403.
16
Donovan J, Blake DR. Patient compliance: deviance or reasoned decision‐making?
Soc Sci Med
 
1992
;
34
:
507
–13.
17
Conrad P. The meaning of medications: another look at compliance.
Soc Sci Med
 
1985
;
20
:
29
–37.
18
Carr AJ, Hughes RA, Stowers K. What influences response to a complementary therapy, glucosamine sulphate?
Rheumatology
 
2000
;
39(Suppl.)
:
127
.
19
Hitchon CA, Cheang M, Wardell J, El‐Gabalawy HS, Canvin JMG. Alternative therapy use in rheumatoid arthritis: prevalence and patient characteristics.
Arthritis Rheum
 
1998
;
41(Suppl.)
:
S79
.

Comments

0 Comments