Friday, January 19, 2007
The most recent Annals of Internal Medicine has a review article and metaanalysis about chronic opioid therapy for chronic back pain. This is a very interesting article which is being widely reported as refuting the idea that opioids are effective for chronic back pain, and that they cause a lot of addiction. Here's what they actually found...
a) Surprise surprise the quality of the studies overall was poor, and there was sufficient heterogeneity in them that limited metaanalysis (although they applied it to some things).
b) All the head to head studies of opioids vs. placebo showed they were more efficacious (albeit modestly so); all the studies comparing opioids to active non-opioid meds found opioids equal or better than nonopioid agents. Nevertheless when they re-tooled this data in the metaanalysis they found no benefit to opioids. I'll be frank I'm not sure how all of this works from a statistical standpoint. I'm used to seeing metaanalyses doing one of two things: looking at a many studies with differing conclusions, compiling all the data, and coming out with (presumably) a more definitive conclusion (more definitive because the amount of data/number of patients looked at/etc. is higher); or looking at a bunch of smallish studies which all suggest a finding but lack power to show it, compiling those data, and answering the question of whether or not that finding was accurate now that there are adequate data. In this study, however, particularly for the studies involving placebos, all the studies individually found the same thing (opioids modestly effective), but when they pooled that data they concluded that well, all those studies were wrong. I'm not sure how this works from a statistical standpoint--I'm not particularly doubting it more than expressing curiosity as to how it can happen and would love a pallimed.org reader to comment about it and enlighten us all.
c) Anyway, at best, opioids are modestly effective for chronic back pain, but the studies aren't that great--that's consistent with the prevailing dogma. In my own experience I have had a few patients for whom opioids have been wonderfully helpful, a few for whom they've been modestly helpful, and a few for whom the side effects etc. just weren't worth it...in sum: modestly effective.
d) Now for the 'addiction' stuff. Reported in the abstract, and widely in the press, is that they found the prevalence of lifetime substance abuse to range from 36-56% in these patients, and that current substance abuse "as high as 43%" of these patients. The context that's missing in these figures is the prevalence of substance abuse in non-opioid users with chronic back pain. Only a couple of their reviewed studies actually looked at this as far as I could tell. One found that lifetime prevalence was 54% in opioid users vs 52% in non-opioid users and that current substance abuse was 23% for both groups. Ooops. The other one found that lifetime prevalence of substance abuse was slightly higher (46%) in non-opioid users than opioid users (43%) . Double-oops. As far as I could tell the other studies didn't do any comparisons with non-opioid users (to be fair there was a single study which showed a much higher rate of 'aberrant' medication behavior in opioid users vs. non). Oh yeah, they also point out that only two of the studies actually used validated substance abuse tools to identify which subjects truly had a history of substance abuse.
So the real conclusion, I guess, is that current and lifetime prevalence of substance abuse disorders is relatively common in chronic back pain patients...probably.
Now, onto how these findings are predictably being misrepresented in news coverage of the article.
This is from a relatively respectable source, Reuters (via Medscape):
"Findings from a systematic review of published research suggest that opioids often provide no advantage over non-opioids for relieving chronic back pain, but carry a high risk of addiction."
It's that last bit about the 'high risk of addiction' which is particularly wrong.
This is one of the authors quoted in a press-release:
"Our findings suggest that clinicians should consider other treatments with similar benefits but fewer long-term adverse effects."
What I'm not sure about is what are these safe and efficacious treatments? Insofar as opioids (in the studies they looked at for this review at least) were as good as or better than non-opioids (metaanalysis aside), and long-term NSAID or COX-2-inhibitor therapy probably isn't very safe either particular for the elderly (they actually only looked at one study comparing an opioids with an NSAID and the opioid did better), one is left wondering what it is you're supposed to do? Where are these effective therapies?
No one therapy is clearly superior for chronic back pain, and patients frequently will need to combine a variety of types of drug and non-drug treatments to get decent relief. Opioids are one of these therapies--they have side effects, complications, and risks to be sure--but to suggest that these findings indicate we are making a bunch of addicts out of chronic back pain patients does no one a service.
And adieu to Art Buchwald. Thanks Christian for posting that, I didn't know. I can only hope that I'll be writing columns sticking it to the President from my hospice room.