Roger Chou, MD; Laurie Hoyt Huffman, MS
Medications are the most frequently prescribed therapy for low back pain. A challenge in choosing pharmacologic therapy is that each class of medication is associated with a unique balance of risks and benefits.
To assess benefits and harms of acetaminophen, nonsteroidal anti-inflammatory drugs (NSAIDs), antidepressants, benzodiazepines, antiepileptic drugs, skeletal muscle relaxants, opioid analgesics, tramadol, and systemic corticosteroids for acute or chronic low back pain (with or without leg pain).
English-language studies were identified through searches of MEDLINE (through November 2006) and the Cochrane Database of Systematic Reviews (2006, Issue 4). These electronic searches were supplemented by hand searching reference lists and additional citations suggested by experts.
Systematic reviews and randomized trials of dual therapy or monotherapy with 1 or more of the preceding medications for acute or chronic low back pain that reported pain outcomes, back-specific function, general health status, work disability, or patient satisfaction.
We abstracted information about study design, population characteristics, interventions, outcomes, and adverse events. To grade methodological quality, we used the Oxman criteria for systematic reviews and the Cochrane Back Review Group criteria for individual trials.
We found good evidence that NSAIDs, skeletal muscle relaxants (for acute low back pain), and tricyclic antidepressants (for chronic low back pain) are effective for pain relief. The magnitude of benefit was moderate (effect size of 0.5 to 0.8, improvement of 10 to 20 points on a 100-point visual analogue pain scale, or relative risk of 1.25 to 2.00 for the proportion of patients experiencing clinically significant pain relief), except in the case of tricyclic antidepressants (for which the benefit was small to moderate). We also found fair evidence that acetaminophen, opioids, tramadol, benzodiazepines, and gabapentin (for radiculopathy) are effective for pain relief. We found good evidence that systemic corticosteroids are ineffective. Adverse events, such as sedation, varied by medication, although reliable data on serious and long-term harms are sparse. Most trials were short term (â‰¤4 weeks). Few data address efficacy of dual-medication therapy compared with monotherapy, or beneficial effects on functional outcomes.
Our primary source of data was systematic reviews. We included nonâ€“English-language trials only if they were included in English-language systematic reviews.
Medications with good evidence of short-term effectiveness for low back pain are NSAIDs, skeletal muscle relaxants (for acute low back pain), and tricyclic antidepressants (for chronic low back pain). Evidence is insufficient to identify one medication as offering a clear overall net advantage because of complex tradeoffs between benefits and harms. Individual patients are likely to differ in how they weigh potential benefits, harms, and costs of various medications.
Appendix Table 1.
Appendix Table 2.
Appendix Table 3.
Appendix Table 4.
Appendix Table 5.
Appendix Table 6.
Appendix Table 7.
Appendix Table 8.
Appendix Table 9.
Appendix Table 10.
Appendix Table 11.
Appendix Table 12.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Jan M Bjordal
Inst. Physical Therapy, Bergen University College,
October 16, 2007
Overviews are methodically inferior to systematic reviews and contribute to myths of effective drugs
In a recent overview encompassing systematic reviews and randomized trials of pharmacotherapy in low back pain (LBP), the authors concluded that there is good evidence that acetaminophen, NSAID and muscle relaxants provide moderate pain relief in acute LBP (1). In another clinical guideline article in the same issue of the Annals (2), the recommended indications for acetaminophen and NSAIDs are extended to include chronic LBP, whereas tramadol, opiods, and benzodiazepines are recommended both in acute and chronic LBP despite the lack of "good" or even "fair" evidence. The methodology used by the authors to review the literature may be flawed in the sense that it merely reiterates what other reviewers have published previously. The current overview does not address crucial differences in focus and protocols of the included reviews, and fails to adequately assess the randomized controlled trials which are the primary sources for information. The criteria list which was used does not imply investigations of the primary sources (RCTs) or the correctness of the trial data cited in the systematic review. The quality assessment method here is inadequate to detect or handle data errors in the review. With multiple data errors, faulty method assessments or misclassification of trials, there is no operationalization of the consequences for method scores. Indeed, some reviewing errors may be grave enough to jeopardize the overreview conclusions completely, but the overview scoring system only opens for a reduction of a single point or two. Misinterpretation of data leads to erroneous conclusions, which in turn may propagate the myth of medication efficacy in LBP. Moreover, the involvement of individuals with industry relations in the making of guidelines is controverisal (3). Indeed, the declaration of conflicts of interest in the form of receiving honoraria from the pharmaceutical industry by principal authors of the clinical guidelines (2) is, in our view, of noteworthy and potentially derogatory significance.
Evidence for a treatment effect is defined as a positive difference in change between the group receiving the active drug and the group receiving otherwise identical treatment with placebo (4). Weaker, indirect evidence of effect can be obtained by analyzing comparative non- inferiority trials or studies which test interventions in similar pathophysiological states at different anatomical localizations.
Acetaminophen There is no direct evidence that acetaminophen is effective in LBP, as no placebo-controlled trials have been performed. The authors refer to a systematic Cochrane review of NSAID and acetaminophen as their primary source (1). This review is not valid, as it was withdrawn from the Cochrane Library in April 2006 (5). As for indirect evidence of acetaminophen effectiveness compared to other treatments in acute LBP, a single small (n=30) high quality trial found acetaminophen to be inferior to a NSAID in acute LBP (6), and one low quality trial did not report any difference between acetaminophen and no treatment (7). Another low quality study showed ambiguous results (8). In the cited Cochrane review (5), 2 low quality trials found no difference between NSAID and acetaminophen (9, 10). However, the current reviewers seem to have missed the fact that one of these trials (9) was not randomized. Comparative studies of acetaminophen and non-pharmacological treatments in high quality RCTs has shown that acetaminophen was inferior to superficial heat (11). In one old trial of lower quality acetaminophen allegedly was slightly inferior to manipulation, but poor reporting of treatment allocation and handling of withdrawals obscures these results (12). As for evidence of acetaminophen effectiveness in non-LBP scenarios, the authors claim that three systematic reviews demonstrate superiority of NSAIDs over acetaminophen in osteoarthritis (OA) with a standardized mean difference (SMD) of approximately 0.3. We have previously reported NSAID effectiveness over placebo in knee OA corresponding to a SMD of 0.31 (13), whereas acetaminophen had clinically insignificant effects in knee OA with a weighted mean difference for pain equalling 3.0 mm [95% CI, 1.4 "“ 4.7] on a visual analogue scale (VAS) (14). In summary, there is no direct evidence and a paucity of indirect evidence for acetaminophen effectiveness in LBP. The authorsÂ´ claim of "good evidence of a moderate effect" of acetaminophen in this condition is, in our view, misconstrued and unsubstantiated, and constitutes ill advice for the concerned clinician.
NSAIDs For NSAIDs, the problem of using an outdated Cochrane review also applies (5, 15). However, as the authors use these data for their conclusions, we will address the shortcomings of this source of information. The authors fail to mention that the SMD analysis for pain was non-significant for NSAIDs in acute LBP was -0.53 [95% CI, -2.74 "“ 1.69]. The authors state that the Cochrane review included 6 trials where NSAIDs were superior to placebo in acute LBP with a relative risk for benefit of 1.24 [95% CI, 1.10 "“ 1.41]. Four out of these 6 trials found non-significant differences, whereas two trials reported significantly positive results (16, 17), thus contributing 2/3 of the overall statistical weight. In the first of these studies, groups of patients received intramuscular injections of either dipyrone (metamizole), diclofenac or saline (placebo). Dipyrone, the sodium sulphonate of the obsolete antipyretic agent antipyrine, is not licensed in the US (and many other countries) due to serious side-effects, and is hardly a typical species of the NSAID class of drugs. However, only the data for the dipyrone group in this particular study was entered into the Cochrane analysis. In addition, one of the trials (18) was not randomized and thus not eligible for the current review. Upon removal of this trial from analysis and substitution of dipyrone data with diclofenac data in the other trial, global improvement is rendered insignificant with relative risk for benefit of 1.11 [95% CI, 0.98 "“ 1.26]. The overview also state that ibuprofen was superior to placebo in one trial of chronic LBP, but the trial report clearly states that both groups received ibuprofen, while additional medication was compared to placebo (19). Thus, based on the cited material, the conclusion that there is "good evidence for moderate effect" of NSAIDs in LBP appears to be erroneous and misleading.
New evidence may also go undetected with the overview methodology used. Data which seems to have evaded the scrutiny of Chou & Huffman include a recent trial of NSAID in acute LBP (20) and four NSAID trials in chronic LBP (21-24). Based on these investigations there seems to be fair evidence in favor of a small/moderate effect of NSAIDs in acute LBP corresponding to 10.9 mm [95% CI, 4.2 "“ 17.6] on the VAS, and good evidence of a small effect in chronic LBP of 8.7 mm [95% CI: 6.9 "“ 11.5] on VAS after adjustment for patient selection bias. However, the results from the latter four trials should be considered in light of increasing concern that NSAIDs belonging to the selective COX-2- inhibitor class have a particularly disfavourable long term safety record.
Muscle relaxants For these drugs we face some of the same problems as with NSAIDs. The primary source of information is again a Cochrane-review (25, 26) with meta-analyses. One problem with the overview method used here is the evaluation of the strength of the evidence. According to the overview a "Good" rating is based on consistent positive results among multiple high quality RCTs. But how does it handle the results of meta-analyses with an overall positive result in spite of several RCTs with negative results? The overview states that there is "good" evidence for a "moderate" effect based on 3 high quality and 1 low quality RCTs in acute low back pain. However, 3 out of 4 RCTs in this relative risk analysis for pain have non- significant results with lower 95% confidence intervals below 1. So from the overviewÂ´s definition of "Good" evidence, the evidence should have been "Poor". And even moreso when det RCTs behind the positive overall result is more closely scrutinized. Methodological quality assessments with the scale used in the review only yields moderate reliability with Kappa values at 0.36 to 0.80 (27). In other words, if this single RCT had been rated at 5 out of 11 and then excluded from the analysis then the result of the meta-analysis would have been negative. The positive overall result rests more or less on a single RCT from 1982 where Merck (MSD) was responsible for the statistical analysis. In the review 79% of the weight in the statistical analysis came from this trial which has a borderline methodological quality score at exactly the cut-off value (6 out of 11) introduced in the review (28). Again the lack of transparency in the overview is causing trouble as the relative risks given there are different from those reported in the reviews (0.80 and 0.58 respectively) and 1.25 and 1.72 in the overview. We assume that t he overview authors have probably been inverting the relative risks to fit into the overview results format, but this has been done without reporting the trial data or the method of the statistical analysis. When we tried to do the same in the Revman software version 4.2, we were unable to replicate the results from the overview. The authors also claim that tizanidine was efficacious in 8 trials, but according to the cited reviews (25, 29), only 6 of these trials were placebo-controlled and three of them did not find any significant results (30-32).
Benzodiazepines The same reviews (25, 26) are the dominating sources for benzodiazepines conclusions. Again t, which again rest on 2 RCTs (33, 34) with borderline methodological quality (6 out of 11 criteria). The overview reports different relative risks for pain relief of 1.41 and global improvement of 1.59 after 10 to 14 days than the reviews. But the review analyses failed to include to 51 patients who dropped out in one of these studies (33). When we included the withdrawals in a new analysis, the relative risk for pain relief became non-significant at 1,38 (95% CI 0,99 to 1,92).
Conclusion We could go on and show that the overview evidence is not robust for other medications as well, but we think that the examples above are sufficient to support our main message. Overviews have a weaker methodology than systematic reviews, they are less transparent and errors in reviews are less likely to be detected. If we use the overviewÂ´s definition of "good evidence" strictly, many of the medications above should have their evidence strength downgraded to "fair" or even "poor". The most dangerous aspect of overviews is perhaps that they will add strength to sensitive and unreliable review conclusions. With the small number of RCTs which conclusions eventually are based on, we see no good reasons for not performing a full systematic review with meta-analyses of randomized placebo-controlled trials when guidelines are being made.
1. Chou, R. and L.H. Huffman, Medications for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Ann Intern Med, 2007. 147(7): 505-14.
2. Chou, R., et al., Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med, 2007. 147(7): 478-91.
3. Taylor, R. and J. Giles, Cash interests taint drug advice. Nature, 2005. 437(7062): 1070-1.
4. Clarke, D. and A. Oxman, Cochrane ReviewersÂ´Handbook, 4.2.0. [update March 2003]. In:The Cochrane Library, Issue 2, 2003. Oxford: Update Software, updated quarterly, 2003.
5. van Tulder, M.W., et al., WITHDRAWN: Non-steroidal anti-inflammatory drugs for low-back pain. Cochrane Database Syst Rev, 2007(3): CD000396.
6. Hickey, R.F., Chronic low back pain: a comparison of diflunisal with paracetamol. N Z Med J, 1982. 95(707): 312-4.
7. Evans, D.P., M.S. Burke, and R.G. Newcombe, Medicines of choice in low back pain. Curr Med Res Opin, 1980. 6(8): 540-7.
8. Milgrom, C., et al., Overexertional lumbar and thoracic back pain among recruits: a prospective study of risk factors and treatment regimens. J Spinal Disord, 1993. 6(3): 187-93.
9. Muckle, D.S., Flurbiprofen for the treatment of soft tissue trauma. Am J Med, 1986. 80(3A): 76-80.
10. Wiesel, S.W., et al., Acute low-back pain. An objective analysis of conservative therapy. Spine, 1980. 5(4): 324-30.
11. Nadler, S.F., et al., Continuous low-level heat wrap therapy provides more efficacy than Ibuprofen and acetaminophen for acute low back pain. Spine, 2002. 27(10): 1012-7.
12. Doran, D.M.L. and D.J. Newell, Manipulation in treatment of low back pain: a multicentre study. Brit Med J, 1975. 2: 161-4.
13. Bjordal, J.M., et al., Non-steroidal anti-inflammatory drugs, including cyclo-oxygenase-2 inhibitors, in osteoarthritic knee pain: meta- analysis of randomised placebo controlled trials. BMJ, 2004. 329(7478): 1317-23.
14. Bjordal, J.M., et al., Short-term efficacy of pharmacotherapeutic interventions in osteoarthritic knee pain: A meta-analysis of randomised placebo-controlled trials. Eur J Pain, 2007. 11(2): 125-38.
15. van Tulder, M.W., et al., Nonsteroidal anti-inflammatory drugs for low back pain: a systematic review within the framework of the Cochrane Collaboration Back Review Group. Spine, 2000. 25(19): 2501-13.
16. Babej-Dolle, R., et al., Parenteral dipyrone versus diclofenac and placebo in patients with acute lumbago or sciatic pain: randomized observer-blind multicenter study. Int J Clin Pharmacol Ther, 1994. 32(4): 204-9.
17. Jacobs, J.H. and M.F. Grayson, Trial of an anti-inflammatory agent (indomethacin) in low back pain with and without radicular involvement. Br Med J, 1968. 3(5611): 158-60.
18. Weber, H. and G. Aasand, The effect of phenylbutazone in patients with in acute lumbago-sciatica. A double blind trial. J Oslo City Hosp, 1980. 30(69-72).
19. Berry, H. and D.R. Hutchinson, Tizanidine and ibuprofen in acute low- back pain: results of a double-blind multicentre study in general practice. J Int Med Res, 1988. 16(2): 83-91.
20. Dreiser, R.L., et al., Relief of acute low back pain with diclofenac-K 12.5 mg tablets: a flexible dose, ibuprofen 200 mg and placebo-controlled clinical trial. Int J Clin Pharmacol Ther, 2003. 41(9): 375-85.
21. Birbara, C.A., et al., Treatment of chronic low back pain with etoricoxib, a new cyclo-oxygenase-2 selective inhibitor: improvement in pain and disability--a randomized, placebo-controlled, 3-month trial. J Pain, 2003. 4(6): 307-15.
22. Coats, T.L., et al., Effects of valdecoxib in the treatment of chronic low back pain: results of a randomized, placebo-controlled trial. Clin Ther, 2004. 26(8): 1249-60.
23. Pallay, R.M., et al., Etoricoxib reduced pain and disability and improved quality of life in patients with chronic low back pain: a 3 month, randomized, controlled trial. Scand J Rheumatol, 2004. 33(4): 257- 66.
24. Katz, N., et al., Efficacy and safety of rofecoxib in patients with chronic low back pain: results from two 4-week, randomized, placebo- controlled, parallel-group, double-blind trials. Spine, 2003. 28(9): 851- 8; discussion 859.
25. van Tulder, M.W., et al., Muscle relaxants for non-specific low back pain. Cochrane Database Syst Rev, 2003(2): CD004252.
26. van der Windt, D.A., et al., Shoulder disorders in general practice: prognostic indicators of outcome. Br J Gen Pract, 1996. 46(410): 519-23.
27. Maher, C.G., et al., Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther, 2003. 83(8): 713-21.
28. Baratta, R.R., A double-blind study of cyclobenzaprine and placebo in the treatment of acute musculoskeletal conditions of the low back. Curr Ther Res, 1982. 32(5).
29. van Tulder, M.W., et al., Muscle relaxants for nonspecific low back pain: a systematic review within the framework of the cochrane collaboration. Spine, 2003. 28(17): 1978-92.
30. Baptista, R., et al., Terapeutica da lombalgia com a tizanidina (DS 103-282) un novo agente mioespasmolitico. Folha Medica, 1988. 96(3): 119- 23.
31. Berry, H. and D.R. Hutchinson, A multicentre placebo-controlled study in general practice to evaluate the efficacy and safety of tizanidine in acute low-back pain. J Int Med Res, 1988. 16(2): 75-82.
32. Lepisto, P., A comparative trial of DS 103-282 and placebo in the treatment of acute skeletal muscle spasms due to disorders of the back. Ther Res, 1979. 26: 454-9.
33. Salzmann, E., et al., Treatment of chronic low back syndrome with tetrazepam in a placebo controlled double blind trial. J Drug Develop, 1992. 4 219-28.
34. Arbus, L., et al., Activity of tetrazepam in low back pain. Clin Trials J, 1990. 27: 258-67.
Oregon Health & Science University
December 5, 2007
We write to provide corrections to the recent joint guideline by the American College of Physicians and the American Pain Society on diagnosis and treatment of low back pain (1) and supporting evidence reviews (2, 3). In the original print version of the guideline, the target populations were described incorrectly (1). The word "not" was inadvertently dropped from a sentence that described populations that were excluded from the guideline. Children or adolescents with low back pain; pregnant women; patients with low back pain from sources outside the back (nonspinal low back pain), fibromyalgia or other myofascial pain syndromes, and thoracic or cervical back pain are not covered by the guideline. The online version of guideline has already been corrected.
In response to a letter to the editor (4) we re-reviewed the evidence on acetaminophen and believe we originally graded the evidence too positively in the guideline and evidence review (1, 2). However, our guideline recommendations remain the same. Acetaminophen for acute low back pain should be rated "˜fair' rather than "˜good' quality. Acetaminophen for chronic low back pain should be rated "˜fair' rather than "˜good' quality and magnitude of benefit "˜small' rather than "˜moderate' (Appendix Tables 5 and 6 in the guideline (1) and Appendix Tables 10 and 11 in the evidence review (2)). The Abstract/Data Synthesis section of the evidence review should read: "We found good evidence that NSAIDs, skeletal muscle relaxants (for acute low back pain), and tricyclic antidepressants (for chronic low back pain) are effective for pain relief"¦We also found fair evidence that acetaminophen, opioids, tramadol, benzodiazepines, and gabapentin (for radiculopathy) are effective for pain relief (2)." The Abstract/Conclusions section should read: "Medications with good evidence of short-term effectiveness for low back pain are NSAIDs, skeletal muscle relaxants (for acute low back pain), and tricyclic antidepressants (for chronic low back pain)." Similar changes should be applied to the Discussion section. As noted, these changes do not affect Recommendation 6 suggesting acetaminophen as an option for first-line pharmacologic therapy (1). This recommendation is based in large part on the safety profile of acetaminophen, when taken in appropriate dosages in patients without a contraindication.
Reference 62 in the evidence review on medications for low back pain is incorrect and should refer to a different trial by the same first author (5).
In the evidence review on medications, we inverted (calculated 1/relative risk) results for "not achieving pain relief" as reported in a Cochrane review (6) in order to report the likelihood of achieving pain relief. This conversion was incorrect because relative risk is not a symmetric statistic. The evidence review (2) will be corrected to state results as originally reported in the Cochrane review: for skeletal muscle relaxants, relative risks for not achieving pain relief 0.80 [CI, 0.71 to 0.89] at 2-4 days and 0.67 [0.13 to 3.44] at 5-7 days and relative risks for not achieving global efficacy 0.49 [CI, 0.25 to 0.95] at 2-4 days and 0.68 [CI, 0.41 to 1.13] at 5-7 days; and for benzodiazepines, relative risks for not achieving pain relief 0.71 [CI, 0.54 to 0.93] and for not achieving global efficacy 0.63 [CI, 0.42 to 0.97] at 8-14 days (6). Similarly, in the evidence review on non-pharmacologic therapies (3), results for a systematic review by Kool et al (7) on exercise therapy should state a relative risk of 0.73 [CI, 0.56 to 0.95] for not returning to work after 1 year . None of these corrections affect conclusions of the evidence reviews or guidelines.
All corrections have been applied to the online version of the articles.
Roger Chou, MD Oregon Health & Science University, Portland, Oregon
Paul Shekelle, MD, PhD Veterans Affairs Health Care System and RAND, Santa Monica, California
Amir Qaseem, MD, PhD, MHA The American College of Physicians, Philadelphia, Pennsylvania
Douglas K. Owens, MD, MS Veterans Affairs Palo Alto Health Care System, Palo Alto, and Stanford University, Stanford, California References
1. Chou R, Qaseem A, Snow V, et al. Diagnosis and treatment of low back pain: A joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med. 2007;147:478 -491.
2. Chou R, Huffman LH. Medications for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Annals of Internal Medicine. 2007;147:505-514.
3. Chou R, Huffman LH. Non-pharmacologic therapies for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians Clinical Practice Guideline. Annals of Internal Medicine. 2007;147:492-504.
4. Bjordal JM, Lopes-Martin RAB, Klovning A, Roland PH, Slordal L. Overviews are methodically inferior to systematic reviews and contribute to myths of effective drugs. Annals Online. 17 Oct 2007.
5. Berry H, Bloom B, Hamilton EBD, Swinson DR. Naproxen sodium, diflunisal, and placebo int he treatment of chronic back pain. Ann Rheum Dis. 1982;41:129-132.
6. van Tulder MW, Touray T, Furlan AD, Solway S, Bouter LM. Muscle relaxants for non-specific low-back pain Cochrane Database of Systematic Reviews. 2003(4).
7. Kool J, de Bie R, Oesch P, Knusel O, van den Brandt P, Bachmann S. Exercise reduces sick leave in patients with non-acute non-specific low back pain: a meta-analysis. Journal of Rehabilitation Medicine. 2004;36(2):49-62.
We thank the authors of the letters for their comments on our low back pain guideline (1) and evidence reviews (2, 3). Dr. Bjordal notes some methodologic concerns with our reviews. To clarify, we performed a systematic review of a broad range of low back pain topics, and when prior systematic reviews were available and of sufficient quality, we included them. The idea that new systematic reviews of the primary literature should always be conducted when developing clinical practice guidelines is both unsupported by any empirical evidence and if implemented could be a poor use of scientific resources (4). Guideline panels need relevant, current, high-quality reviews of the evidence; if existing reviews fulfill those criteria then it is wasteful to ignore them and conduct new reviews. We included systematic reviews published in or after the year 2000 and identified higher-quality reviews using a validated quality rating instrument (5, 6). Although systematic reviews should be updated, there is no compelling reason to ignore higher-quality Cochrane reviews (7) that met our criteria for inclusion and were "withdrawn" only because they did not meet an updating deadline, not because of methodological deficiencies or the publication of contradictory trials (8).
Dr. Bjordal suggests that we graded evidence for acetaminophen too positively. In his letter, Dr. Bjordal describes one trial as evaluating acute low back pain when it actually evaluated chronic low back pain (9). Otherwise, our descriptions of the evidence are similar (Appendix Tables 10 and 11 (2)). We agree that our evidence ratings for acetaminophen were generous given some inconsistency among trials of acute low back pain, and lack of direct evidence and small benefits for chronic low back pain. We re-rated evidence for acetaminophen for acute low back pain fair quality with moderate benefits, and for chronic low back pain fair quality with small benefits (see Correction). Because of acetaminophen's favorable safety profile compared to other pharmacologic therapies, these changes do not change our recommendation to consider it as a first-line option for pharmacologic therapy (1).
For NSAIDs, skeletal muscle relaxants, and benzodiazepines, Dr. Bjordal's focus on single outcomes from placebo-controlled trials reported in Cochrane reviews ignores much of the available evidence. Our assessments are based on both placebo and active-controlled trials, non- Cochrane systematic reviews, data on various outcomes related to pain, function, and global efficacy, and indirect evidence from patients with other pain conditions (Appendix Tables 10 and 11 (2)). We also evaluated consistency between trials and across higher-quality systematic reviews (10). In addition, post-hoc analyses, such as those presented by Dr. Bjordal, can be misleading and should be interpreted cautiously. For example, excluding trials based on small differences in quality scores is problematic given unpredictable associations between summary quality rating scores and estimates of effects (11). We did not report data on mean improvement in pain scores from a Cochrane review of NSAIDs because of substantial, unexplained heterogeneity (p<0.0001) (7).
As Dr. Bjordal surmised, we inverted relative risks (1/relative risk) for "˜no pain relief' with skeletal muscle relaxants and benzodiazepines (as reported in a Cochrane review (12)) to present results for a positive outcome (achieving pain relief) (2). However, this transformation was incorrect, as relative risks (unlike odds ratios) are not a symmetric statistic. We will correct the article to show original results as reported in the Cochrane review (see Corrections). This correction does not change any conclusions, but we thank Dr. Bjordal for noting this error.
We disagree with Dr. Bjordal's assertion that there is enough evidence to establish efficacy of low-level laser therapy (LLLT) and transcutaneous electrical nerve stimulation (TENS). In the case of LLLT, there is substantial diversity across trials in doses and types of laser, some inconsistency among higher-quality trials, and the possibility of publication bias. Our conclusion of insufficient evidence is similar to a recently published Cochrane review (13). For TENS, the highest quality placebo-controlled trial found no benefit in chronic low back pain (14). In addition, it is inappropriate to pool studies of disparate populations and therapies (TENS and neuromuscular stimulation), as proposed by Dr. Bjordal, and two of the trials proposed for pooling found no benefits on pain or function with TENS versus placebo (15, 16).
We disagree with Dr. Ernst that our conclusions regarding rare risk of serious adverse events with spinal manipulation are misleading or downplay the risk of cerebrovascular events (3). Our review deals with low back pain and treatment with lumbar spinal manipulation. There are no reports of cerebrovascular events following lumbar spine manipulation or in patients being treated for low back pain (17, 18). Cervical manipulation is not a subject of our review or practice guideline.
Dr. Bjordal suggests that recommendations for therapy favor pharmacologic over non-pharmacologic options (1). In fact, we recommend either type of therapy as options (Figure 1, Box 9 (1)), and strength of evidence and magnitude of benefits were graded similarly for several pharmacologic and non-pharmacologic therapies (Appendix Tables 5 and 6). However, recommendation 7 on non-pharmacologic therapies in general was graded "weak" because of relatively weak evidence for some suggested options (Appendix Tables 10 and 11 (3)), higher costs compared to first- line pharmacologic therapies, and less convenience (most non-pharmacologic options involving multiple provider visits). It would be appropriate to select a non-pharmacologic over pharmacologic therapy in patients who express such a preference, but the trade-offs should be discussed (19). Superficial heat is already recommended as a self-care option (Figure 2, Interventions box (1)).
Douglas K. Owens, MD, MS Veterans Affairs Palo Alto Health Care System, Palo Alto, and Stanford University, Stanford, California
4. Silagy CA, Stead LF, Lancaster T. Use of systematic reviews in clinical practice guidelines: case study of smoking cessation. BMJ. 2001;323:833-836.
5. Jadad AR, McQuay HJ. Meta-analyses to evaluate analgesic interventions: a systematic qualitative review of their methodology. Journal of Clinical Epidemiology. 1996;49:235-243.
6. Oxman AD, Guyatt GH. Validation of an index of the quality of review articles. Journal of Clinical Epidemiology. 1991;44(11):1271-1278.
7. van Tulder MW, Schotten RJPM, Koes BW, Deyo RA. Non-steroidal anti -inflammatory drugs for low-back pain Cochrane Database of Systematic Reviews. 2000(2).
8. Shojania KG, Sampson M, Ansari MT, Ji J, Doucette S, Moher D. How quickly do systematic reviews go out of date? a survival analysis. Ann Intern Med. 2007;147:224-233.
9. Hickey RF, Hickey RF. Chronic low back pain: a comparison of diflunisal with paracetamol. New Zealand Medical Journal. 1982;95(707):312 -4.
10. GRADE Working Group. Grading quality of evidence and strength of recommendations. BMJ. 2004;328:1490-1494.
11. Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA. 1999;282:1054-1060.
12. van Tulder MW, Touray T, Furlan AD, Solway S, Bouter LM. Muscle relaxants for non-specific low-back pain Cochrane Database of Systematic Reviews. 2003(4).
13. Yousefi-Nooraie R, Schonstein E, Heidari K, et al. Low level laser therapy for nonspecific low-back pain. Cochrane Database of Systematic Reviews. 2007;2.
14. Deyo RA, Walsh NE, Martin DC, Schoenfeld LS, Ramamurthy S. A controlled trial of transcutaneous electrical nerve stimulation (TENS) and exercise for chronic low back pain. New England Journal of Medicine. 1990;322:1627-34.
15. Herman E, Williams R, Stratford P, Fargas-Babkak A, Trott M. A randomized controlled trial of transcutaneous electrical nerve stimulation (CODETRON) to determine its benefits in a rehabilitation program for acute occupational low back pain. Spine. 1994;19(5):561-568. 16. Moore SR, Shurman J. Combined neuromuscular electrical stimulation and transcutaneous electrical nerve stimulation for treatment of chronic back pain: a double-blind, repeated measures comparison. Arch Phys Med Rehabil. 1997;78:55-60.
17. Ernst E. Cerebrovascular complications associated with spinal manipulation. Physical Therapy Reviews. 2004;9(1):5-15.
18. Ernst E. Adverse effects of spinal manipulation: a systematic review. Journal of the Royal Society of Medicine. 2007;100:330-338.
19. Schunemann HJ, Jaeschke R, Cook DJ, et al. An official ATS statement: Grading the quality of evidence and strength of recommendations in ATS guidelines and recommendations. Am J Respir Crit Care Med. 2006;174:605-614.
Chou R, Huffman LH. Medications for Acute and Chronic Low Back Pain: A Review of the Evidence for an American Pain Society/American College of Physicians Clinical Practice Guideline. Ann Intern Med. 2007;147:505-514. doi: 10.7326/0003-4819-147-7-200710020-00008
Download citation file:
Published: Ann Intern Med. 2007;147(7):505-514.
Results provided by:
Copyright © 2017 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only