Assessment: Use of epidural steroid injections to treat radicular lumbosacral pain [RETIRED]
Report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology

We read the Report of the TTA subcommittee with interest. [1] We feel that the report by Armon et al could confuse physicians and patients. The abstract reports that there is insufficient evidence to recommend the use of epidural steroid injections to treat radicular cervical pain (Level U), even though the focus of the review is the use of epidural steroid injections to treat radicular lumbosacral pain and the studies included in the synthesis related solely to this focus.

Principles of evidence-synthesis, and specifically evidence- synthesis and evidence-based interventional pain management have been described. [2] Recently, Abdi et al performed a systematic review that separated lumbar epidural steroid injections into caudal (8 randomized trials), interlaminar (11 randomized trials), and transforaminal (6 randomized trials), as these approaches are distinct techniques with variable effectiveness and separate applications. [3]

In addition, Abdi et al included other observational studies and an examination of the methodologies and criteria of both AHRQ and Cochrane reviews. They concluded that there was moderate evidence to support the long-term (i.e.- > 6 weeks) effectiveness of caudal and transforaminal epidural steroid injections and limited evidence to support the effectiveness of lumbar interlaminar epidural steroid injections.

Armon et al only included four studies considered to have met the predetermined inclusion criteria, although previous studies have included larger numbers of randomized trials in systematic evaluations including the Cochrane review [4] and European guidelines. [5] For these reasons, the report by Armon et al may negatively affect the successful conduct of interventional pain management.

References

1. Armon C, Argoff CE, Samuels J, Backonja MM. Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Assessment: use of epidural steroid injections to treat radicular lumbosacral pain: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2007;68:723-729.

2. Manchikanti L, Boswell MV, Giordano J. Evidence-based interventional pain management: principles, problems, potential and applications. Pain Physician 2007; 10:329-356.

3. Abdi S, Datta S, Trescot AM et al. Epidural steroids in the management of chronic spinal pain: A systematic review. Pain Physician 2007; 10:185- 212.

4. Nelemans PJ, Debie RA, DeVet HC, Sturmans F. Injection therapy for subacute and chronic benign low back pain. Spine 2001; 26:501-515.

5. Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-Moffett J, Kovacs F., et al. Chapter 4: European guidelines for the management of chronic nonspecific low back pain. Eur Spine J 2006; 15:S192-S300.

Disclosure: The authors report no conflicts of interest.

Manchikanti et al refer Neurology readers to four papers to dispel confusion that they attribute to our report. Review of the data in these papers affirms our original conclusions.

Regarding cervical transforaminal injections, Abdi et al [3] considered two randomized trials, each comparing two active treatments without placebo controls and four non-randomized trials without controls. These studies do not meet the inclusion criteria of our original assessment, which required placebo controls [1], leaving unchanged our conclusion that there is insufficient evidence to recommend the use of epidural steroids to treat cervical radicular pain.

Differences in inclusion criteria, including a broader case definition and consideration of uncontrolled studies and studies without placebo controls, led to listing of a larger number of studies within the reviews cited by Manchikanti et al of epidural steroid injections for low back pain [3-5], compared to our listing [1]. However, the conclusions of those reviews are not too dissimilar from ours.

Abdi et al [3] concur with our conclusions as applied to interlaminar epidural injections. While they use the language "moderate evidence" in support of efficacy of transforaminal and caudal routes of injection of injection, this appears to be done at variance with the classification system provided in the methods section of their review. According to this system, the presence of conflicting data that showed in some studies lack of efficacy for injections via these routes should have resulted in the wording "limited evidence," similar to that used for the interlaminar route.

Furthermore, the paper does not reflect awareness of methodological limitations in some studies alleging efficacy. The two other reviews -- the European Guidelines [4] and Nelemans et al [5] -- come to conclusions similar to, or more proscriptive, than ours. In particular, Nelemans et al [4] (not a Cochrane review), considering radicular and non-radicular low back pain, concludes that "convincing evidence is lacking regarding the effects of injection therapy on low back pain."

We are encouraged that even individuals who appear to dispute the content or clarity of our findings provide data that, when examined, are supportive of our conclusions. The precise wording of our assessment reflects the terminology and classification system used by the AAN.

Disclosure: The authors report no conflicts of interest.

Armon et al present a medical legal dilemma for the AAN. [1] In published guidelines for qualifications of physician expert witnesses, it is stated that experts should have evidence of competence in an area of relevance to the opinion including publications. [2] Whereas the subject article presents guidelines, the authors are indirectly providing expert opinions in such cases whether the forum is a court, licensing board, peer review body or other lawful agency.

Randomized controlled trials (RCTs) have been lauded in the subject article. Alternative opinions suggest that this may be taken too far. [3] In the present matter, performing a placebo controlled randomized study may not be especially venerable because of patient expectations that arise out of the consenting process. To be ethical, a recruiter must share beliefs that epidural steroids have not been proven to be better than placebos. The consenting process may ultimately induce pharmacological unblinding due to systemic steroid effects. In this process, the recruiter is likely to induce, in the subject, expectations for no benefit or complication. These negative expectations may be an underestimated factor affecting outcomes in placebo controlled randomized studies. Furthermore, many patients may decline during the recruitment process when confronted with unsavory choices. How recruitment induced biases affect the outcomes of research was not discussed in the subject article.

If RCTs are performed for the subject issue, two separate interventional study arms could be compared. In this case, the study recruiter can ethically discuss two options that are thought to be beneficial. The latter approach would not scare off study participants because it would induce expectations of improvement. Assuming that placebo effects are not dramatically higher than treatment effects, an observed difference between study groups may be determined.

Pharmacologically induced unblinding would not occur in the latter circumstance. A study, not mentioned by the subject review article, exemplified the above proposal. The former study demonstrated the superiority of selective transformaminal injections compared to interspinous epidural injections. [4] Experts should be aware of this and similar studies.

References

1. Armon C, Argoff CE, Samuels J, Backonja MM. Assessment: Use of epidural steroid injections to treat radicular lumbosacral pain. Neurology 2007; 68: 723-729.

2. Williams, MA. Mackin GA, Beresford, HR, et al. American Academy of Neurology qualifications and guidelines for the physician expert witness. Neurology 2006;66:13-14.

3. Smith G, Pell J. Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomized controlled trials. BMJ 2003; 327: 1459-1461.

4. Thomas E, Cyteval C, Abiad L, Picot MC, Taourel P, Blotman F. Efficacy of transformaminal versus interspinous corticosteroid injection in discal radiculalgia. A prospective randomized, double blind study. Clinical Rheumatology 2003; 22: 299-304.

Disclosure: The author reports no conflicts of interest.

The AAN practice parameters are evidence-based reviews of topics relevant to the practice of neurology. The process is described in detail on the website (http://www.aan.com/go/practice/guidelines) and does not include expert opinion or consensus. The Disclaimer, appended to each assessment, states clearly that the AAN views the assessments as an educational service to its members.

The AAN's Guidelines for Qualification of Physician Expert Witnesses are not factored explicitly in selecting the individuals involved in the preparation, review, and approval of guidelines. [2] The AAN Conflict of Interest Policy for Practice Guidelines is more germane to author panel composition. The readers may be reassured, that (a) the authors of the present assessment have not published previously on the topic of epidural steroids; and (b) that three of four do not derive any income from epidural steroid injections. Multiple stages of review, including independent review by Neurology, before an assessment is approved by the AAN Board of Directors, assure the integrity of the process.

The obvious truth, that when death is a certain outcome [3] an intervention that prevents that outcome does not require a placebo control is not germane to the topic. An editorial cited in our original review [5] clarifies the roles of placebo controls in pain therapy clinical trials.

Dr. Jordan proposes that studies comparing two active treatment arms for pain are more ethical than placebo-controlled studies. [4] The prerequisites for a valid comparison of two active treatment modalities are: (a) both need to have been shown effective against placebo; and (b) the comparison of the two needs to have been preceded by studies to determine the optimal dose or method of administration for each. Lacking (a) – treatment A may appear better, yet be no better than placebo, because treatment B makes patients worse. Lacking (b) – treatment A may appear superior to treatment B because treatment A is applied in an optimal manner, whereas treatment B is applied in a suboptimal manner.

In the study cited by Dr. Jordan [4], the investigators utilized, in the interspinous arm an apparent subtherapeutic dose, equivalent to 27-33% of the dose used currently for this route of administration [6-9], to support a conclusion of superiority for the transforaminal arm. They did not make this explicit. In our opinion, the study's methods do not support its conclusions and we do not perceive it to be a model for ethical superiority.

ACKNOWLEDEMENT: The authors acknowledge Yuen T. So, M.D., Ph.D., and Janis Miyasaki, M.D., MEd, FRCPC, TTA co-chairs, and Gary Gronseth, M.D. and Jacqueline French, M.D., QSS co-chairs, for their critique of earlier versions and affirmation of the final version of this response. This response has been reviewed also by Murray G. Sagsveen, JD, CAE, General Counsel, AAN.

References:

5. Fillingim RB, Price DD. What is controlled for in placebo-controlled trials? Mayo Clinic Proc 2005;80:1119–1121.

6. White AH. Injection Techniques for diagnosis and Treatment of Low Back Pain. Ortho Clin North Am 1983;14:553-567.

7. Buchner M, Zeifang F, Brocai DR, Schiltenwolf M. Epidural corticosteroid injection in the conservative management of sciatica. Clin Orthop 2000;375:149–156.

8. Arden NK, Price C, Reading I, et al. A multicentre randomized controlled trial of epidural corticosteroid injections for sciatica: the WEST study. Rheumatology 2005;44:1399-1406.

9. McAuley D. Corticosteroid converter. GlobalRPH.com 2006. http://www.globalrph.com/corticocalc.htm converter.

Disclosure: The authors report no conflicts of interest.

The AAN committee's rigorously ascertained assessment of epidural steroid injection efficacy provides an overdue authoritative guideline. [1] Our own earlier review also affirmed that reports of serious complications are inadequate to establish statistical prevalence. [2]

Evidence-based clinical studies have significant data of collateral value. In the 2007 Physicians Desk Reference, the manufacturer of methylprednisolone acetate (Pharmacia and Upjohn), states that the drug is "an anti-inflammatory glucocorticoid." However, we were unable to discover any valid pathological or clinical evidence that inflammation is a pathological component of clinical radiculopathy/back pain. [2] A likely hypothesis for short-term symptomatic improvement is the toxic effect of the usual hypertonic injectate mixture at the bedside of steroid, anesthetic, and often radiological contrast agent. Indeed, Pharmacia and Upjohn warn that "This product contains benzyl alcohol [also other preservatives], which is potentially toxic when administered locally to neural tissue…Because of possible physical incompatibilities, DEPO-MEDROL Sterile Aqueous Suspension should not be diluted or mixed with other solutions."

Concerning complications they warn of "Adverse reactions reported with intrathecal/epidural routes of administration: arachnoiditis, meningitis, paraparesis/paraplegia, sensory disturbances, bowel/bladder dysfunction, headache, seizures…DEPO-MEDROL Sterile Aqueous Suspension is contraindicated for intrathecal administration. Reports of severe medical events have been associated with this route of administration." Our references indicated that inadvertent intrathecal injection during lumbar epidural infusion occurs in at least 5 to 6% of cases. [2]

Concerning the professional popularity of this practice, a recent editorial in Pain concluded, "Those who have a firm conviction that they must use injections come in three groups. One is the misguided who cannot separate placebo effect, natural history of disease and regression to the mean from therapeutic effect. The second group is those who feel that they must do something…They stick and inject in the hope rather than the belief that what they do is helpful. The third group is in it for the money." [3]

References

1. Armon C, Argoff CE, Samuels J, Backonja MM. Assessment: Use of epidural steroid injections to treat radicular lumbosacral pain. Report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2007;68:723-729.

2. Nelson DA, Landau WM. Intraspinal steroids: history, efficacy, accidentality, and controversy with review of United States Food and Drug Administration reports. J Neurol Neurosurg Psychiatry 2001;70:433-443.

3. Butler SH. Editorial: Primum non nocere—first do no harm. Pain 2005;116:175-176. Disclosure: The authors report no conflicts of interest.

We thank Drs. Nelson and Landau for their gracious appraisal of our work [1], and for referring the readers of Neurology to their own assessment of epidural steroid injections [2]. They raise the intriguing hypothesis that the possible short term benefit of epidural steroid injections may be due to toxicity of the injectate rather than to a therapeutic effect of the steroids. Their hypothesis might be testable by utilizing epidural steroid injections and two control arms. One control arm would receive a control solution that is matched to the active solution in all aspects (volume, osmolality, preservatives) but without any steroids, and the other would receive normal saline. Until that is done, this remains a hypothesis.

Drs. Nelson and Landau proceed to cite the manufacturer's cautions regarding uses of SoluMedrol as listed in the 2007 Physician Desk Reference (PDR). They state in their letter: "Our references indicated that inadvertent intrathecal injection during lumbar epidural infusion occurs in at least 5% to 6% of cases [2]." We asked Dr. Landau to indicate which references they were relying on, and he directed us to two. [4,5] It may be seen, that the citations do not support the wording used by Drs. Nelson and Landau. Moreover, White reported specifically that in 300 patients treated using fluoroscopy, "complications occurred in 0.4%. These were local sacral pain, transient hypotension, and two cases of dural penetration through the lumbar route."

We note further, that, while 2 out of 300 cases are 0.7%, dural penetration it is not the same as intrathecal injection. [4, 5] We recognize that published series may underestimate the frequency of complications, and agree with Drs. Nelson and Landau that case reports of serious complications are inadequate to establish the frequency of their occurrence, due to the absence of a denominator. We stand by our original presentation of the safety profile of epidural steroids [1], which is not intended to supplant the PDR, and appreciate Drs. Nelson and Landau's comments referring the readers also to the PDR.

Finally, our work is an evidence-based assessment of epidural steroid injections. Since the available reviewed studies do not address the motivation of the treatment provider, it was beyond the scope of our work to explore such issues. We hope that additional studies to establish verifiable indications, expected outcomes and likely complications of epidural steroid injections will optimize the use of this treatment.

References

4. Dilke TFW, Burrry HC, Grahame R. Extradural Corticosteroid Injection in management of Lumbar Nerve Root Compression BMJ 1973;2:635-637.

CITATION: "There were no complications … A minority of patients experience transient pain of sciatic distribution during extradural injection. Cerebrospinal fluid was inadvertently tapped on six occasions in the course of this trial [100 patients], and when this happened the needle was withdrawn and an extradural injection was carried out immediately through an adjacent interspinous space." (page 637, column 1)

5. White AH. Injection Techniques for diagnosis and Treatment of Low Back Pain. Ortho Clin North Am 1983;14:553-567.

CITATION: "We did find, however, in our double-blind study, that even by these techniques we are not accurately in the epidural space as much as 25 per cent of the time and do occasionally enter the thecal space, which can complicate the clinical picture." (page 556 column 1) Disclosure: The authors report no conflicts of interest.