BMJ meta-analysis: Reboxetine “an ineffective and potentially harmful antidepressant”

Reboxetine, racemic mixture.

Thanks to Ben Goldacre and Vaughan Bell, I learned this morning of an interesting paper in British Medical Journal that analyzed substantial unpublished data from Pfizer on their norephinephrine reuptake inhibitor antidepressant, reboxetine. Sold in Europe as a pair of enantiomers under the trade name Edronax, reboxetine's US application was rejected by the FDA in 2001 for "lack of compelling evidence of efficacy." The authors, all from the Institute for Quality and Efficiency in Healthcare in Köln, Germany, requested from Pfizer all unpublished clinical data on the antidepressant from trials comparing its efficacy and safety to either placebo or active SSRI drugs. (The institute is better known by the acronym, IQWiG, Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen (IQWiG).) An accompanying analysis of the article discusses the role of the IQWiG. ResearchBlogging.orgThe paper stemmed from a report the Institute had conducted in 2009 for the German Federal Joint Committee (G-BA) for efficacy and safety of reboxetine, mirtazepine, and bupropion given acutely, generally eight to ten weeks. Both Essex Pharma and GSK provided IQWiG investigators with all unpublished data for the analysis of the latter two drugs but Pfizer had not (PDF, English translation).
The retrieval of previously unpublished trials was hampered by the fact that during preparation of the preliminary health technology assessment report, the manufacturer of reboxetine did not provide a complete list of unpublished trials as requested by IQWiG. Secondary publications clearly indicated that further potentially relevant unpublished trials existed. As the preliminary report showed that reboxetine had been tested in at least 16 trials including about 4600 patients, but data on almost two thirds of these patients were not accessible, the institute initially concluded that no meaningful assessment of reboxetine was possible. After the publication of the preliminary report, the manufacturer decided to cooperate and provided most of the missing data (one venlafaxine controlled trial was not available as a full publication).
Indeed, roughly three-quarters of the data had not been published (from 3,033 of 4,098 subjects). When now pooled with published data, the analysis revealed that reboxetine was no better in remission and response rates than placebo and was inferior to SSRIs such as fluoxetine, paroxetine, and citalopram. More concerning was the finding that reboxetine was 1.8-fold more likely than fluoxetine to cause trial withdrawals due to adverse events. Note, however, that the adverse events were pooled and not analyzed separately - suicidal ideation, attempts, and deaths were too low in incidence to make any comparison and this was not an intent of the meta-analysis. (For the record, the investigators reported, "A total of 18 serious adverse events related to suicide (suicidal tendencies, suicide attempts, or completed suicides) were noted (six for reboxetine; four for placebo; eight for SSRIs).") As one might expect, the authors hammer on the concept of publication bias, where trial sponsors publish only the most favorable data and call for international regulatory requirements that all clinical data be considered in regulatory approvals and be made available to health assessment organizations. Look for more discussion today on the release of clinical outcomes and safety data by drug manufacturers.
Eyding, D., Lelgemann, M., Grouven, U., Harter, M., Kromp, M., Kaiser, T., Kerekes, M., Gerken, M., & Wieseler, B. (2010). Reboxetine for acute treatment of major depression: systematic review and meta-analysis of published and unpublished placebo and selective serotonin reuptake inhibitor controlled trials BMJ, 341 (Oct 12) DOI: 10.1136/bmj.c4737 Wieseler, B., McGauran, N., & Kaiser, T. (2010). Finding studies on reboxetine: a tale of hide and seek BMJ, 341 (Oct 12) DOI: 10.1136/bmj.c4942

Author: David Kroll

Share This Post On


  1. Sneaky buggers…. thanks for that post Dave.

    I have personal experience of trying to be an honest player in publishing clinical trials. It is very very hard to get a negative trial published. Editors are quite willing to bend to inane reviewer comments when directed at a negative study- the same study if positive will be accepted forthwith.

    Any move toward the acceptance of trials for publication before the results are known (or by keeping the results out of the hands of reviewers until they decide whether it is a basically acceptable trial) would be something I would support wholeheartedly.

  2. antipodean, thanks for the insight. Indeed, we all have experience where even the most relevant negative data is difficult to publish. Glad to hear that you have done your best to contribute to transparency.

    I don’t know the European drug registration process as well as in the US but it seems to me that FDA may have had access to the complete data when making their decision not to approve reboxetine in the States.

  3. I am surprised that companies turned-over unpublished data.

    Antipodean, my first thought was your idea is good. However, if I understand it correctly what you are suggesting is to accept a manuscript based on approval of the protocol. It seems that paves the way to publication of inconclusive results (all CAM research concludes “Further research is needed) as well as accepting clinically insignificant results.

  4. JJM

    You could still be rejected for having your conclusions not match your data. And clinically insignificant results are not clinically insignificant because they highlight the lack of effect of that treatment. This is just as important to know as a large treatment effect.

  5. Antipodean,

    Of course it is important to know when a treatment is ineffective, I was thinking, though, about publication in a top-line journal which eschews mundane, if useful, results. Also, some work is clinically insignificant simply because it is redundant (I know a chemist who revels in saying “We knew that 30 years ago”). Other times, work may be ‘justified’ by some convoluted logic that would not normally occur to someone; viz. using oral Mg Citrate solution to treat glaucoma.

    Overall, publication of negative results is good. Really important results (lack of effect of echinacea) is even acceptable to JAMA. If a journal has a policy of deciding the value of results (good or bad) there should be no need to blind reviewers.

  6. antipodean and JJM: Funny that we should be having this conversation about the time that my German oncologist/scientist colleague, Martin Fenner, posted at his PLoS blog calling for a pre-print archive of clinical trials.

  7. I like Fenner’s idea, I like antipodean’s idea, too; I just don’t think the complication of pre-approval of the protocol is necessary. Good journals should be open to publishing solid, negative results. And some are, they are just not usually the top journals (with exceptions, e.g., the echinacea study in JAMA).

  8. David, thanks for mentioning my post suggesting a preprint server for clinical trials. The FDA since September 2008 requires reporting of clinical trial results within 12 months after the last patient has ben treated. I haven’t yet seen any report that analyzed the compliance with this policy.

    • Hi Martin

      Even in good faith it is very very difficult to report clinical trials in under a year. I realise an expert reviewer and guideline writer, such as yourself, will find this extremely frustrating. It is hard to get this done quickly to the requisite standard.

      JJM “Good journals should be open to publishing solid, negative results.” They should be. But they aren’t. The top journals will publish negative results only when they overturn a long held clinical or folklore tradition (i.e. the echinacea study). They have to be surprising or newsworthy. The bigger better study of a novel technique that overturns a previous smaller less rigourous study will often be rejected. You can see this clearly in citation practices as well- even when published in the same journal the smaller positive trial will be cited faster than the larger negative study.


  1. Quick Links | A Blog Around The Clock - [...] BMJ meta-analysis: Reboxetine “an ineffective and potentially harmful antidepressant” [...]