[75] Gabapentin for pain: New evidence from hidden data

Gabapentin for pain: New evidence from hidden data

[75] Gabapentin for pain: New evidence from hidden data


Gabapentin for pain: New evidence from hidden dataGabapentin (Neurontin) was licensed in Canada in 1993 for adjunctive treatment of epilepsy. In 1998 two double blind randomized controlled trials (DBRCT) suggested mild analgesic effects of gabapentin in painful diabetic peripheral neuropathy (PDPN)1 and post-herpetic neuralgia (PHN)2. Subsequently, unapproved use of gabapentin exploded for pain, migraine, and even as a “mood stabilizer”.3

Therapeutics Letter #33 (Jan-Feb 2000) reviewed gabapentin for pain. It noted that gabapentin is eliminated by kidney filtration (half-life 6 hours with normal renal function) and that it reduced pain by a mean of 1-2 points on a pain score of 0-10, over 2 weeks, NNT=4 for “moderate or marked” benefit. The Letter concluded: “Gabapentin benefits at best a minority of patients with painful diabetic or post-herpetic neuropathy. Toxicity, but not analgesia, is dose-dependent.”4 A 2005 Cochrane Systematic Review similarly reported an NNT of 4.3, suggesting that 23% of patients improve.5

Subsequently, U.S. litigation has revealed that Neurontin’s off-label promotion was assisted by selective publication and citation of studies with favorable outcomes.6 Court-ordered access to unpublished studies now allows us to present a more accurate estimate of gabapentin’s clinical effects.7

How Neurontin became a blockbuster

Gabapentin never achieved major commercial success as an anticonvulsant. In 1995 Parke-Davis marketing staff proposed an experimental program to test anecdotal claims of efficacy for “neuropathic” pain and other syndromes. Research results were to be published, “if positive”.8 Immediately after the 1998 JAMA publications, Parke-Davis launched a program of selective publication and intensive marketing, assisted by “Key Opinion Leaders” (KOL).9 Sworn testimony indicated that Parke-Davis used its “clinical liaison” sales representatives and “Key Opinion Leaders” to market Neurontin “for everything”.10 By 2003 annual U.S. sales of gabapentin had expanded from $98 million to $2.7 billion/year.

A gradually broadened category of “neuropathic pain” became gabapentin’s most durable market, reinforced by guidelines that refer to gabapentin as “first line treatment”.11 In B.C. consumption is still rising, at a cost exceeding $30 million during 2009, 63% from public funds (see Figure).

In February 2010 a U.S. court in Boston is scheduled to hear detailed evidence from published and unpublished DBRCT of gabapentin for pain and other unapproved uses in a civil trial of alleged fraud for the off-label marketing of Neurontin prior to 2004.12

Re-evaluation including unpublished trials

Evidence before the Boston jury will include a 2008 critical appraisal and meta-analysis of all known RCTs of gabapentin for chronic neuropathic and acute pain, including detailed study reports that became public only through the U.S. litigation. Details are available in the Drug Industry Documents Database at UCSF.13 DBRCT were typically from 2-8 weeks duration, in patients screened to eliminate many co-morbidities, such as kidney disease. Studies used either varying fixed doses of gabapentin or forced titration, with typical maximum doses of 1800-2400 mg/day.

Chronic “neuropathic” pain:

Benefits: 9 trials (N=1917) assessed mean pain reduction from baseline. Gabapentin reduced weighted mean pain score by -0.78 (-0.99, -0.58) as compared with placebo on a 0-10 point scale. 7 trials (N=1971) assessed patient-reported “moderate or much improvement”: gabapentin 37.7%, placebo 20.2%; difference 17.5%, NNT=6. 3 trials (N=1028) assessed percentage of patients achieving at least a 50% reduction in pain score: gabapentin 31.4%, placebo 18.4%; difference 13%, NNT=8. Efficacy was greater in PHN than for other pain syndromes.

Harms: In 12 trials (N=2362) gabapentin increased adverse events: gabapentin 67.6%, placebo 55.2%; difference 12.4%, NNH=8. Specific adverse events included dizziness (NNH=6), somnolence (NNH=7), confusion or ataxia (NNH=10) and edema (NNH=11).

Comparisons of gabapentin with tricyclic antidepressants did not favor either treatment, although adverse events differed qualitatively.

Acute nociceptive pain:

Four DBRCT (N=1371) compared gabapentin with placebo, acetaminophen, naproxen and hydrocodone, alone or in combinations for acute pain after dental extraction, orthopedic surgery or exacerbations of osteoarthritis. In contrast with conventional analgesics, gabapentin was not efficacious for acute pain. These studies were never published.

Dose dependence:

Multiple DBRCT provide no evidence that larger doses confer greater analgesia, whereas toxicity is clearly dose-dependent.13, 14

Additional DBRCTs since 1999:

A DBRCT (N=87) in acute shingles found that gabapentin titrated from 300 to 1800 mg/day was not better than placebo over 4 weeks, whereas oxycodone CR, titrated from 20 to 120 mg/day, reduced mean pain score by 1.2 points vs. placebo on a scale of 0-10.15 One publicly funded crossover DBRCT (N=57) compared gabapentin and morphine alone or together for chronic neuropathic pain (PDPN and PHN).16 The authors interpreted this very complex experiment as evidence that gabapentin may enhance the analgesic effect of morphine. An alternative interpretation is that gabapentin is ineffective for neuropathic pain vs. placebo.17 The same authors compared gabapentin and nortriptyline alone or together in another complex 3-period crossover DBRCT (N=56) in a similar population.18 Combined nortriptyline/gabapentin reduced mean daily pain score by 0.6 vs. nortriptyline alone, and by 0.9 vs. gabapentin alone (scale 0-10). Careful inspection of the original graphical data suggests that gabapentin effects do not increase with higher doses nor with time. DBRCT of delayed-release gabapentin for PDPN and PHN were completed by July 2007 and October 2009, but only one is partially reported.19

How does pregabalin compare?

Pregabalin (Lyrica) has not been compared with gabapentin for chronic pain. A September 2009 Common Drug Review20 recommended against provincial formulary listing of pregabalin because new studies raise additional questions about its efficacy for neuropathic pain. One unpublished DBRCT comparing pregabalin to an active comparator found that amitriptyline was better than placebo for PDPN, whereas pregabalin was not. Like gabapentin, drug effects are apparent almost immediately. In 2009 British Columbians spent about $10 million on pregabalin.

Conclusions and recommendations

  • Misleading promotion pushed gabapentin to blockbuster status; scientific evidence suggests gabapentin has a minor role in pain control.
  • Gabapentin reduces neuropathic pain by < 1 point on a 0-10 point scale and benefits about 15% of carefully selected patients (NNT=6-8).
  • A similar proportion of people suffer harm (NNH=8).
  • A test of benefit/harm can be made after 1-2 days at a low dose (100-900 mg/day).
  • Benefit is unlikely to increase with higher doses or longer treatment.
  • Opioids afford greater relief in chronic neuropathic pain, with qualitatively different adverse effects.
  • Use particular caution for people at risk of cognitive impairment, balance disturbance, falls, or when edema is undesirable (e.g. peripheral vascular disease in the elderly).
  • Reassess patients already taking gabapentin at least every 2 months. The short elimination half-life allows reassessment of benefit vs. harm by stopping the drug for 1-2 days (longer if kidney function is impaired).
  • Gabapentin has no role in acute nociceptive pain.
  • Benefits and harms of pregabalin are similar to gabapentin, at higher cost.
NNH = number needed to treat to cause one harmful event
NNT = number needed to treat
RCT = randomized controlled trial
DBRCT = double blind randomized controlled trial
PDPN = painful diabetic peripheral neuropathy
PHN = post-herpetic neuralgia
The draft of this Therapeutics Letter was submitted for review to 55 experts and primary care physicians in order to correct any inaccuracies and to ensure that the information is concise and relevant to clinicians.


  1. Backonja M.. Beydoun A, Edwards KR et al. Gabapentin for the symptomatic treatment of painful neuropathy in patients with diabetes mellitus: a randomized controlled trial. JAMA 1998; 280(21): 1831-6. http://jama.ama-assn.org/cgi/content/full/280/21/1831
  2. Rowbotham M, Harden N, Stacey B et al. Gabapentin for the treatment of postherpetic neuralgia: a randomized controlled trial. JAMA 1998; 280: 1837-42. http://jama.ama-assn.org/cgi/content/full/280/21/1837
  3. Landefeld CS, Steinman MA. The Neurontin legacy – marketing through misinformation and manipulation. N Engl J Med 2009; 360(2): 103-6. http://content.nejm.org/cgi/content/full/360/2/103
  4. Therapeutics Letter #33 (December 1999 – February 2000). Treatment of pain in the older patient. /letter33
  5. Wiffen PJ, McQuay HJ, Rees J, Moore RA. Gabapentin for acute and chronic pain. Cochrane Database of Systematic Reviews 2005, Issue 3. Art. No.: CD005452. DOI: 10.1002/14651858.CD005452. http://mrw.interscience.wiley.com/cochrane/clsysrev/articles/CD005452/frame.html
  6. Vedula SS, Bero L Scherer RW, Dickersin K. Outcome reporting in industry-sponsored trials of gabapentin for off-label use. N Engl J Med 2009; 361(20); 1963-71. http://content.nejm.org/cgi/content/full/361/20/1963
  7. Drug Industry Document Archive at the University of California San Francisco. http://dida.library.ucsf.edu/
  8. Brandicourt O. Neurontin Marketing Assessments. July 31, 1995. http://dida.library.ucsf.edu/tid/qhb00a10
  9. Steinman MA, Bero LA, Chren MM, Landefeld CS. Narrative review: the promotion of gabapentin: an analysis of internal industry documents. Ann Intern Med 2006; 145:284-293. http://www.annals.org/content/145/4/284.full
  10. Disclosure of Information by Relator David P Franklin: Pursuant to 31 USC 3730 b(2), available at http://dida.library.ucsf.edu/tid/abb00a10
  11. O’Connor A, Dworkin R. Treatment of neuropathic pain: an overview of recent guidelines. Amer J Med 2009; 122(10 Suppl): S22-32. doi:10.1016/j.amjmed.2009.04.007. http://www.amjmed.com/article/S0002-9343(09)00396-9
  12. News items, e.g. http://www.cbc.ca/health/story/2010/02/22/gabapentin-neurotonin-lawsuit.html?ref=rss#socialcomments
  13. Perry T. Neurontin: expert opinion on efficacy and effectiveness for pain. http://dida.library.ucsf.edu/tid/oxx18p10
  14. Alldredge B. Expert consultant report. http://dida.library.ucsf.edu/search?query=cs%3Aneurontin*&ct=11
  15. Dworkin R, Barbano R, Tyring SK et al. A randomized, placebo-controlled trial of oxycodone and of gabapentin for acute pain in herpes zoster. Pain 2009; 142(3): 209-17. http://www.painjournalonline.com/article/S0304-3959(08)00761-6
  16. Gilron I, Bailey J et al. Morphine, Gabapentin, or their Combination for Neuropathic Pain. N Engl J Med 2005; 352: 1324-34. http://content.nejm.org/cgi/content/full/352/13/1324
  17. See Perry T in reference 13 (above) at page 219/366.
  18. Gilron I, Bailey J et al. Nortriptyline and gabapentin, alone and in combination for neuropathic pain: a double-blind, randomized controlled crossover trial. Lancet 2009; 374(9697): 1252-1261. http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(09)61081-3/fulltext
  19. Sandercock D, Cramer M, Wu J, Chiang Y-K, Biton V, Heritier M. Gabapentin extended release for the treatment of painful diabetic peripheral neuropathy. Diabetes Care 2009; 32: e20. http://care.diabetesjournals.org/content/32/2/e20.long
    See also press releases from 2007, 2008, 2009: http://investor.depomedinc.com/phoenix.zhtml?c=97276&p=irol-news&nyo=0
  20. CEDAC final recommendation. Pregabalin resubmission (Lyrica – Pfizer Canada Inc). Indication: Neuropatic pain associated with diabetic peripheral neuropathy. http://www.cadth.ca/media/cdr/complete/cdr_complete_Lyrica%20Resubmission_September_25-2009.pdf
  • skwschwarz
    Posted at 16:04h, 31 March Reply

    I observed that Therapeutics Letter #75 concludes “Gabapentin has no role in acute nociceptive pain”.

    This conclusion apparently is made without consideration of a fair number of RCTs and several metaanalyses/systematic reviews on the analgesic effects of gabapentin in the perioperative setting. Since postoperative pain represents an archetypical example of acute nociceptive pain, I would suggest that this fact should be explicitly acknowledged, and ideally addressed in a future letter.

    Stephan K. W. Schwarz, M.D., Ph.D., FRCPC;
    Assistant Professor,
    Department of Anesthesiology, Pharmacology & Therapeutics,
    The University of British Columbia;
    Staff Anesthesiologist & Director of Anesthesia Research, St. Paul’s Hospital;
    Scientific Program & Academic Anesthesia Chair, British Columbia Anesthesiologists’ Society

  • r.hewko
    Posted at 22:59h, 04 May Reply

    The Therapeutics Letter 75 – Gabapentin for pain offers a unique perspective regarding the role of gabapentin and pregabalin for neuropathic pain -“minor role for pain control”. This perspective is not shared by any pain management organization or group in the world. Specifically the Canadian Pain Society, the American Pain Society, the America Anesthesia Society and the International Association for the Study of Pain all recommend gabapentin and pregabalin along with tricyclic antidepressants as first line agents for the treatment of neuropathic pain. The TI perspective of narcotics for neuropathic pain is also unique. At best narcotics at least pure Mu agonists are considered second or third line agents options. Although there are some fifteen studies supporting the effectiveness of narcotics for neuropathic pain, the studies cover a range of two to eight weeks. The problem with these agents is that there is no evidence that the benefit is sustained over time without progressive escalation in dose which ultimately may become problematic for either the patient or the prescribing physician. Neither tricyclic antidepressants or gabapentin/pregabalin are associated with tolerance. Over the last fifteen years of attending pain conferences I have heard the issue discussed repeatedly and the guidelines of these organizations representing thousands of pain specialists around the world are reflected in their guidelines. While there may not be double blind placebo control studies regarding the issue of narcotic tolerance in the management of neuropathic pain, the recommendations of the various pain societies does represent the other arm of evidence based medicine – clinical expertise which the TI letter seems to ignore to the potential detriment of the neuropathic pain patient in BC.

    R. Hewko MD FRCPC
    Clinical Professor Dept. of Psychiatry UBC
    Clinical Director CL Psychiatry Service VGH

  • Thomas L. Perry MD, FRCPC
    Posted at 21:49h, 07 December Reply

    Better late than never?
    We regret the 9 year delay in replying to Dr. Schwartz and Dr. Hewko.
    Use of gabapentin (and pregabalin) in the perioperative period is a different, although related issue. Most of the early studies assessed reduction of short term opioid use in the post-anesthetic recovery room with gabapentin vs. placebo. We consider that a surrogate outcome, of much less importance than mortality, recovery from surgery, serious adverse events, length of recovery room of hospital stay, or other more clinically important outcomes. The most common expected adverse effects were nausea, sedation, and balance disturbance, but edema from gabapentin or pregabalin can also complicate or confound clinical interpretation of volume status or right heart function.
    Dr. Hewko’s comments about opioids were well taken, but it remains in 2019 that we have virtually no evidence about long-term pain treatment with any modality (drugs, interventional techniques, or other treatments). Short-term experiments directly comparing opioids with gabapentin found opioids slightly more effective, but in many such experiments the placebo effect or regression to the mean (spontaneous improvement) are powerful. This was shown, for example, with oxycodone vs. gabapentin vs. placebo in acute shingles. During the first 2 weeks, oxycodone was slightly more effective than placebo, whereas gabapentin was not:
    However, relying on pain specialty society “guidelines” was a trap which led not only to the irresponsible prescription of opioids for chronic pain, but to overuse of all other drugs prescribed for pain, especially in combination. When we have looked closely at evidence, pain guidelines emerge along with guidelines for psychiatric drugs as amongst the most highly conflicted and least evidence-based advice one can find. Unfortunately there has been little progress in this field toward improvement of guidelines in the direction of Guidelines We Can Trust.
    See: https://www.bmj.com/content/367/bmj.l6576 (BMJ open access December 3, 2019)
    So what is the state of evidence about gabapentin (and pregabalin)?
    Evidence about gabapentin has been greatly clarified since 2010. The Boston jury convicted Pfizer of racketeering fraud over promotion of Neurontin, and the conviction was upheld on serial appeals. Multiple subsequent systematic reviews are consistent with our conclusions about neuropathic pain, and extend the evidence of gabapentin’s ineffectiveness to migraine headache prevention and fibromyalgia, and low back or radicular pain.
    A few of useful references available online with open access include:
    1. Cochrane systematic review for neuropathic pain, revised 2017: https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007938.pub4/epdf/full

    2. Cochrane systematic review for fibromyalgia, revised 2017:

    3. Cochrane systematic review for prevention of migraine, revised 2013:

    4. Cochrane systematic review for acute post-operative pain, 2010 (from unpublished data revealed by the Neurontin trial):

    5. PLoS Medicine systematic review of gabapentin and pregabalin for chronic low back pain:

    6. CMAJ systematic review of anticonvulsants for back pain and radicular pain 2018:

    Recently more real world evidence is becoming available about the harms of gabapentin and pregabalin, e.g.:
    7. BMJ Swedish population cohort study finding increase of serious harms from suicidal behaviour, unintentional overdoses, injuries, road traffic incidents, and violent crime 2019:

    Scientists at the Johns Hopkins Medical School continue to explore how “cherry picking” of outcomes coloured the interpretation of experimental results even in published trials of gabapentin for pain:
    The lead author Evan Mayo-Wilson will explore this through a RIAT project to re-evaluate and publish their analysis of harm from 6 gabapentin trials that were not properly reported:

    In April 2019 the UK government made gabapentin and pregabalin controlled substances:
    This decision reflects increasing concern about harms from gabapentin (and pregabalin), including drug dependency. Access to the following BMJ articles requires a subscription or library but the abstracts tell part of the story, including the potential that restricting these drugs could further abuse of other drugs affecting the brain:
    https://www.bmj.com/content/347/bmj.f6747 (Bad Medicine commentary)
    https://www.bmj.com/content/365/bmj.l2147 (risks of suicidal behaviour)
    https://blogs.bmj.com/bmj/2019/03/29/ian-hamilton-reclassifying-pregabalin-gabapentin-only-moves-problem-other-drugs/ (BMJ blog commentary)
    https://www.bmj.com/content/365/bmj.l4021 (BMJ editorial on new Swedish population study finding multiple serious harms from gabapentin and pregabalin, see open access reference #7 above)

    Dose dependence and timing of drug effects:
    One of the most useful things we have learned from studying gabapentin and pregabalin clinical trial data is that their effects occur very early in treatment, and at low doses. This is observable both in parallel group RCTs and in crossover trials, e.g.:
    https://www.ncbi.nlm.nih.gov/pubmed/19796802 (Lancet, full article requires library)
    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2906437/ (open access)

    Most important, the same principles apply to other drugs used for symptoms such as pain or depression. When effects are measured early, they are typically present early (both benefits and harms). In chronic pain treatment, most treatments do not work well, but the good clinician will keep trying to help the patient while striving to minimize treatment harms to that patient and to the broader society:
    https://www.bmj.com/content/346/bmj.f2690 (BMJ Analysis 2013: Expect analgesic failure; pursue analgesic success)

    Thomas L. Perry MD, FRCPC
    Chair, Education Working Group
    Editor-in-Chief, Therapeutics Letter
    December 8, 2019

    For conflict of interest declaration, see main TI website. I was a paid consultant from 2008-2010 for plaintiffs in the US Neurontin litigation that led to Pfizer’s conviction for racketeeting fraud. I have had no relationships with pharmaceutical companies since 1998.

Post A Comment