You don't need to be signed in to read BMJ Blogs, but you can register here to receive updates about other BMJ products and services via our site.

Primary Care Corner with Geoffrey Modest MD: Low Back Pain Treatment per AHRQ Review

17 Mar, 16 | by EBM

By Dr. Geoffrey Modest

The agency for healthcare research and quality (AHRQ) just published an evaluation of the literature on noninvasive treatments for low back pain (LBP), looking at both pharmacologic and nonpharmacologic approaches (see ).


  • As we know only too well, LBP is really common. 84% of adults will have LBP at some time, and >25% have had it within the past 3 months
  • 1998 US health care expenditures were $90 billion, not including the rather large indirect costs of lost work productivity, etc.
  • Those with acute LBP (up to 4 weeks) typically have rapid improvement. Those with subacute (4-12 weeks) tend to have slower improvement, and those with chronic (>12 weeks) often do not get much more improvement and account for much of the costs of LBP treatment
  • Predictors of chronicity: psych comorbidities, maladaptive coping strategies (e.g. fear avoidance: avoiding activities because they fear they will make things worse), catastrophizing (e.g., anticipating the worst possible LBP outcome), presence of nonorganic symptoms (without physiologic basis), high baseline functional impairment, and low general health status. [By the way, we used to suggest that people with low back pain have strict bedrest, to the point of using urinals at the bedside, for 1-2 weeks. This was subsequently found to be incorrect and counterproductive/harmful, and I would imagine would reinforce the “maladaptive coping strategies” noted. I wonder if we created more chronic LBP by our older incorrect advice???]
  • Imaging: big issues of false attribution — degenerative changes, facet joint arthropathy, bulging or herniated discs are really common in people with and without [And, in my experience, I have had several patients with very severe lumbar spinal stenosis both by symptoms and by really awful looking MRIs, who have spontaneously gotten dramatically better, in a few cases becoming asymptomatic without any treatment]
  • 156 publications met the inclusion criteria of AHRQ, most in patients with nonradicular LBP
  • Results:
    • Pharmacologic therapies:
      • Acute and subacute LBP:
        • NSAIDs, opioids (only studied: buprenorphine patch), and skeletal muscle relaxants were associated with small effects on pain (vs placebo).
        • NSAIDs were associated with small effects on function
        • Acetaminophen and systemic steroids had no benefit vs placebo
      • Chronic LBP:
        • NSAIDs and tramadol had moderate effects on pain, vs placebo
        • Opioids, duloxetine, and benzos had small effects on pain
        • NSAIDs, opioids, tramadol, and duloxetine had small effects on function
        • Tricyclics had no effect (though older studies did show small benefit), and there were not enough data to determine if gabapentin on pregabalin helped; evidence was inconsistent on opioids vs NSAIDs
      • Radicular symptoms
        • Limited data, but no difference in pain or function with systemic steroids vs placebo. One study found that benzos delayed return to work. Insufficient evidence for gabapentin or pregabalin
      • Harms of therapies: overall not very well reported
    • Nonpharmacologic therapies: (limited studies overall)
      • Acute LBP
        • Spinal manipulation, heat, massage, low-level laser therapy are associated with some benefit vs sham treatments or no intervention/usual care
        • Exercise, massage, and heat had moderate effect on pain and function
        • The data on using ice/cold packs is insufficient for a recommendation [though, in my experience, many patients do find it helpful]
      • Chronic LBP:
        • Exercise, yoga, tai chi; psychological therapies (especially restoration or cognitive-behavioral therapies); multidisciplinary rehab; acupuncture; spinal manipulation; and low-level laser therapy have small to moderate effect for improving pain or function
      • ​Evidence for taping, electrical muscle stimulation, passive physical modalities had too little data; one study on ultrasound or TENS showed no benefit vs sham ultrasound/TENS

So, these assessments raise several issues:

  • Systematic reviews and meta-analyses have one really large limitation: they only include those studies which were actually done. I.e., even though some meds may work, or different doses of meds than those studied may work, they get no support in the review if they were never tested in a formal study. And, for better or worse, much of what we do in medicine does not have the support of rigorous studies.
  • In the studies included in the systematic reviews (which include onlymethodologically “good” studies), several may yield the “no benefit” designation if the drug was no better than placebo. But many placebos get reasonable responses, on the order of 20-30% often, so suggesting acetaminophen, for example, has helped several of my patients feel better and be able to function through perhaps the very real “placebo” effect (for more on the placebo effect, see for a discussion of the genetics, or​ for a couple of examples).
  • One currently very hot topic is the use of opioids for chronic low back pain. And it is a really important issue given their attendant huge social costs (addiction, overdoses, etc.) and is particularly relevant given the large numbers of people on chronic opioids for chronic LBP (one of the most common reasons for chronic opioid prescribing). But for acute low back pain, there are only 2 studies (of “low” stength-of-evidence), using only buprenorphine patches and showing a small effect on pain relief. And only 4 studies showing moderate effect on the pain of chronic LBP. I’m not sure how that squares with several of my patients who derive significant benefit even though I really have no doubt that they are taking the meds regularly and have real, functionally-disabling pain that we are treating. And there are no really good trials looking at opioids vs NSAIDs or other drugs, or even what dose of opioids is adequate for pain relief
  • More invasive procedures were not included in this review. And, for example, well-designed studies suggest that epidural injections do not help radicular back pain. But I have certainly had several patients who were in constant pain and unable to function well, who have gotten very dramatic and durable relief from injections. Is that a placebo effect?? Maybe. But it really restored function and basically eliminated their pain, so I will continue to prescribe it when all else fails, and typically before a referral for surgery

So, as with many medical papers, this one sheds some light but largely raises the many lacunes in our knowledge base, ones that we really need answered to practice high quality, appropriate care. (I hate to say the usual caveat after pretty much all editorials in the major medical journals, but “further studies should be done….”)​

By submitting your comment you agree to adhere to these terms and conditions
You can follow any responses to this entry through the RSS 2.0 feed.
EBM blog homepage

Evidence-Based Medicine blog

Analysis and discussion of developments in Evidence-Based Medicine Visit site

Creative Comms logo

Latest from Evidence-Based Medicine

Latest from EBM