Electronic letters to:

Review:
Andrea D. Furlan, Juan A. Sandoval, Angela Mailis-Gagnon, and Eldon Tunks
Opioids for chronic noncancer pain: a meta-analysis of effectiveness and side effects
CMAJ 2006; 174: 1589-1594 [Abstract] [Full text] [PDF]
*eLetters: Submit a response to this article

Electronic letters published:

[Read eLetter] The "at best available upper limit of effect"?
Hardo W. Sorgatz   (20 September 2006)
[Read eLetter] The Fundamental Problem With Opioid Trials for Chronic Pain
Tushar mehta   (12 June 2006)

The "at best available upper limit of effect"? 20 September 2006
Previous eLetter Top
Hardo W. Sorgatz
Technical University of Darmstadt

Send letter to journal:
Re: The "at best available upper limit of effect"?

sorgatz{at}psychologie.tu-darmstadt.de Hardo W. Sorgatz

Dear Editor: Despite the fact that our comments may methodically emphasize a "case for failed meta-analyses" (Naylor, 1995), we do not doubt the significance of the review for clinical decisions and guideline development. Although it is frequently doubted whether studies concerning the application of opioids in chronic non-cancer pain can be processed in a "state of the art" manner with meta-anlaytical procedures, Furlan et al. have performed this task and have achieved a clinically significant result. The following remarks concern the method and the result:

The meta-analytical Review is surprising with its positive evaluations of the original studies and by the aggregation of data from different abstraction levels. The data accumulation is questionable, because statistically filtered indices (Gimbel et al., 2003) are calculated together with the mean values and standard deviations from the raw data of other rct-studies. Values from studies with a change of substance within one investigational phase (Raja et al., 2002; Watson 2003) or with active placebo (Watson et al., 2003) were summarized with values from classical placebo comparisons. The aggregated studies partially exhibit high dropout rates; the cumulated individual findings are not always statistically significant.

The efforts of the review authors in placing the overall result on an adequate study basis are undeniable. The analgesic effect of opioids on chronic noncancer pain is reported as an effect size of 0.60. This could be classified as a so moderate effect size that its clinical relevance might be questionable. Transformed into a ‘Number Needed to Treat’ measure, an effect size of .60 (under the assumption: Mean = 5 on a 10- point VAS, SD = 1) results in a NNT over 15.

Nevertheless, this meta-analysis is valuable for clinical decisions, because it defines an "at best available upper limit of effect", whereby the effect of individual studies on clinical decisions is limited. Whether a pain reduction by opioids of ES = 0.60 in a period of a few weeks is clinically relevant, can only be decided by comparisons with the activity spectra (analgesic and adverse effects) of other analgesics. Chronic pain is a multifactorial, only subjectively describable event associated with a measurement error of nearly 60% which limits the magnitude of the effects that can be expected. Therefore, systematic estimates of the "at best available upper limits of effect" for all analgesics that are frequently used in chronic pain over a period of several weeks have to be compared with the reported ones of this review. An average pain reduction of ES = 0.60 could then be more significant for the subjective success of therapy than the numerical expression would lead one to believe. co authors: Henriette Reinecke, Daniel Ullrich; Affilation for both: LONTS Guideline Office, c/o Inst. of Psychology, Technical University of Darmstadt

Conflict of Interest:

None declared
The Fundamental Problem With Opioid Trials for Chronic Pain 12 June 2006
Next eLetter Top
Tushar mehta
Substance Use Medical Service, St. Joseph's Health Centre, Toronto ON

Send letter to journal:
Re: The Fundamental Problem With Opioid Trials for Chronic Pain

tushar_toronto{at}hotmail.com Tushar mehta

June 8, 2006 Respected authors (with a special hello to Angela Mailis!)

Thanks for your work on a very difficult topic. I appreciate and agree with your comments that the studies you cite are problematic; they are too short to detect tolerance, addiction is not evaluated, etc. Furthermore, the studies are funded by the pharmaceutical companies, which may cause large biases as studies can be designed to show a favorable result – not difficult to do in the case of opioid research.

I believe there is another fundamental problem with opioid research for chronic non-cancer pain (CNCP). Opioids have a definite cognitive effect that is difficult to compare with either placebo or active placebo such as NSAIDs. They improve mood due to a euphoric effect, and even a nominal dose, such as of 5mg of oxycodone, is capable of producing this. After some time tolerance develops, and even a person using Q8h continuous release opioids will experience intermittent withdrawal, as well as smaller euphoric effects, during the peak and trough drug levels of their daily dosing. This has been my impression.

Under such conditions, patients will generally endorse benefit of opioids for the complex phenomena of CNCP, which may include psychogenic factors. Euphoria is equated with pain relief and greater wellbeing, and the dysphoria and pain of withdrawal is equated worse CNCP. Hence, patients are likely to say “I feel better when I take the opioids and worse when they wear off; hence they are helping my pain.”

Furthermore, it is often difficult to recall how bad their pain was while opioid free, and people on opioids for CNCP feel are likely to endorse and overall improvement even if they are worse off with the peaks and trough effects of their opioids. This may be because they cannot imagine a life without the opioids, since they now think that being without opioids would feel like their trough level of opioids – a state of withdrawal.

How else could the research be done? Firstly, I think that studies done with fentanyl patches, or a continuous infusion of opioids via pump would overcome some of these problems. The dose would have to be titrated so gradually that the patient would barely notice the nociceptive or psychogenic effects of the opioid. A suitable placebo would be necessary.

Secondly, in a trial comparing opioid in pill form to active placebo, perhaps the active placebo should contain one ingredient that gives nociceptive relief (i.e. an NSAID) and a second ingredient that has a calming but euphoric effect (i.e. a benzodiazepine such as a lorazepam, because it has a faster and more noticeable effect).

In traditional trials comparing oral opioids to placebo or NSAIDs, an objective and detailed assessment of “level of functioning” and overall “quality of life” are better primary outcomes than pain itself. Pain as a primary outcome measure is too skewed by the mechanisms I proposed above, and may have very little role in determining opioid effectiveness. For all of these measures of effectiveness, care must be taken to measure a study participant’s functioning/life-quality/pain at a single point in time. There should be nothing in the questioning method to that leads participants to compare how they were to their baseline levels. Psychologically, this could also lead patients to biased reporting.

Many thanks and all the best!

Dr. Tushar Mehta MD, CCFP

Substance Use Medical Service

St. Joseph’s Health Center

Toronto, Ontario

Conflict of Interest:

None declared