We thank Dr Suss for his letter regarding our osteoarthritis decision aid1 that accompanies the systematic review2 and are pleased he finds the systematic review a valuable summary.
Dr Suss states we did not define meaningfully improved pain but in the third sentence of the decision aid article we state meaningful reductions in pain are “generally defined as a 30% or more reduction in pain, but specific definitions of clinically meaningful vary widely across studies.”1 The decision aid itself does include this estimate (about 30%) in the title to assist clinicians when discussing treatment options with their patients.
Dr Suss raises some other specific concerns and states a number of times that he had to read the whole article to understand the tools. While the PEER (Patients, Experience, Evidence, Research) Group is always seeking to simplify evidence and make it as accessible as possible, we believe that a quick review of the instructions for any tool or resource is not unreasonable. The article is 368 words (about 100 more than Dr Suss’ letter) before the graphics. That is substantially shorter than most available guidelines and evidence synopses.
We address the specific concerns raised:
The exercise benefit is implausible.
Yes, it likely is. How to apply the meta-analyses response rate results is much debated. Pulling numbers directly from the metagraph is easiest, uses the raw absolute numbers, and offers a good approximation in most cases. However, many evidence experts believe we should apply the relative risk (or rate ratio) to standardized numbers (drawn from a population). In decision aids, this allows the relative benefits of interventions to be more easily compared but still presents absolute numbers. For our standardized control (placebo) event rate, we used the average of control rates across all studies. It is not without other limitations, however. The foremost is that interventions with good relative benefit but a comparatively low control rate (like exercise) will appear more effective. On the other hand, studies of topical nonsteroidal anti-inflammatory drugs had a higher placebo response rate, so conversion in those cases leads to a slight reduction of the absolute effect.
While we recognize the positive effect of activity on osteoarthritis is likely inflated by this methodology, we believe that the downside is more people might attempt increased activity. If the overestimation encouraged even a few more people to increase their activity, the net gains would only be positive across multiple health outcomes. Additionally, we wanted to apply the methodology equally across all interventions and selectively applying the results would add further bias.
Opioids are potentially harmful.
From the sensitivity analysis of our systematic review, trials shorter than 4 weeks found benefit in opioid therapy. There was no benefit at 4 to 12 weeks or beyond 12 weeks. When all the data were pooled, the short-term trials drove the results to a (marginally) positive benefit. As the evidence team, we believed it was not appropriate to select only certain results for some therapies and not for others. However, we wanted to provide clinicians and their patients with information to recognize that while opioids might have small benefit in the short term, they likely do not have benefit as a long-term pain medicine but do have potential for harm with long-term use.
Glucosamine, chondroitin, and viscosupplementation appear twice.
As above, we believed it would be important for users of the tool to understand the challenges in interpreting the results of studies of these interventions. Many readers will know the evidence is, at best, conflicting. By showing clinicians the lack of effect in publicly funded trials, we gave them and their patients an opportunity to reflect on what that might mean for them. Some will not care and will choose to value the results of all trials while some will place more value on the results of publicly funded studies. We wanted to give clinicians and patients the option to see both and determine for themselves.
We are currently completing a series of large systematic reviews of common chronic pain conditions in primary care (osteoarthritis, back pain, and neuropathic pain). Once these are done, we will create a simplified guideline on chronic pain management in primary care. After the guideline committee is formed, we hope to provide more clarity. As the evidence team, we try to minimize the influence of our potential biases by avoiding overinterpretation of the systematic review results, preferring instead to simply present the results found with the caveats identified. We will select guideline committee members who (like us) do not have financial conflicts of interest. They will be encouraged to make recommendations considering the complexities of all the evidence—particularly regarding opioids, chondroitin, glucosamine, and viscosupplementation. The PEER Group prefers that a guideline committee of family physicians and other clinicians assist in the final application of the evidence. For now, we provide the best available evidence and try to minimize our potential influence or bias on interpretations. In many ways, we are asking clinicians (and patients) to be their own guideline committee with all the available evidence to make good choices.
Even when we start with pooled randomized controlled trials, there is no perfect solution for taking data and translating them to easy-to-understand numbers, particularly when we try to present all the information and minimize any biases (those in the studies and those we might possess). Many other society and groups prefer instead to provide no actual numbers or comparisons, instead using vague terms, advocating some therapies over others, or just listing options. In these cases, with limited or nonexistent information, we cannot come close to an informed choice. The approach we used is a compromise, derived from the best available research on how to present numbers and data to patients,3 allowing them to make the best possible decisions.
Footnotes
Competing interests
None declared
- Copyright© the College of Family Physicians of Canada