To the Editor:

Dr. Möller raises the issue of limitations of our study findings with regards to baseline characteristics of patients in the glucosamine and placebo groups, study medication dose, disease flare as the primary outcome variable, and carryover effect. All of these limitations were discussed in detail in our article. However, we thank Dr. Möller for the opportunity to repeat a discussion of these.

We agree that no validated definition of disease flare exists for OA. However, the recent development of the Osteoarthritis Research Society International (OARSI) outcome criteria for clinical trials in OA has provided credibility to the a priori chosen definition of disease flare in our randomized discontinuation trial (1). The OARSI response criteria for a nonsteroidal antiinflammatory drug trial was defined as an absolute decrease in pain by 20 mm on a visual analog scale (VAS) in association with a relative decrease of pain by 45%. The OARSI criterion for absolute change in pain is identical to our definition of disease flare, which required a worsening of pain on walking by at least 20 mm (0–100-mm VAS scale). In addition, disease flare in our discontinuation trial was also defined by a worsening in the physician global assessment by 1 point on a 1–5 point scale. Both of these flare criteria were also appropriate from the point of view of a minimal clinically perceptible difference, which was reported as 11 mm (0–100-mm VAS scale) for pain on walking and as 0.43 (1–5 scale) for physician global assessment (2).

Dr. Möller is concerned that the distribution of patients by sex and OA radiographic severity was not homogeneous in the 2 study arms. Although we agree that baseline imbalances were present, this does not invalidate our results or conclusions. There are 2 schools of thought for dealing with such baseline imbalances. The first is that one should correct for any baseline imbalance statistically; the second is that randomization creates a compensating prognostic factor that may not be known, in which case one does not need to adjust for the imbalance. The results were analyzed both adjusting for the differences and not adjusting, and in both cases there was no evidence of any benefit from glucosamine. Furthermore, none of the secondary outcomes suggested any benefit, and an evaluation of cartilage biomarkers also failed to distinguish the 2 treatment groups (3).

The majority of participants in both groups were taking 1,500 mg/day of glucosamine. However, as was noted in the article, there was no suggestion that glucosamine dose or duration of prior glucosamine use had any effect on the risk of disease flare, reinforcing the lack of efficacy demonstrated in the study. Furthermore, it should be pointed out that glucosamine was not used in this study to treat “an acute process such as a disease flare” as Dr. Möller states, but rather glucosamine was used to prevent a disease flare from occurring. As such, this study evaluated the efficacy of glucosamine as a maintenance treatment in knee OA, the very effect for which Dr. Möller claims glucosamine is beneficial.

We agree with Dr. Möller that a carryover effect, including a cure, is at least a theoretical explanation for negative results in a randomized discontinuation trial, which is why we did not conclude that glucosamine was ineffective for improvement of knee OA symptoms, Although the results are also compatible with that conclusion. Rather, we concluded that “for patients with knee OA with at least moderate subjective improvement with prior glucosamine use, this study provides no evidence of symptomatic benefit from continued use of glucosamine sulfate over and above that found with placebo.”

Jolanda Cibere MD, PhD*, Jacek A. Kopec MD, PhD*, John M. Esdaile MD, MPH*, Anona Thorne MSc†, Joel Singer PhD†, * Arthritis Research Centre of Canada, University of British Columbia, Vancouver, BC, Canada, † Centre for Health Evaluation and Outcome Sciences, University of British Columbia, Vancouver, BC, Canada.