Intended for healthcare professionals

Letters Nalmefene in alcohol misuse

Nalmefene in alcohol misuse: junk evaluation by the European Medicines Agency

BMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g2017 (Published 10 March 2014) Cite this as: BMJ 2014;348:g2017
  1. Alain Braillon, public health doctor1
  1. 1Northern Hospital, 80000 Amiens, France
  1. braillon.alain{at}gmail.com

“If you torture the data long enough, it will confess.” Ronald Coase.

Spence claims that nalmefene in alcohol misuse is bad medicine.1 In my opinion it is junk evaluation by the European Medicines Agency.

Firstly, the authorisation was granted on subgroup analysis of two randomised controlled studies of six months’ duration.2 This subgroup comprised only a quarter of the patients and at six months up to half of the data were missing in the nalmefene group (table).3 This is not intention to treat analysis. In this very small subgroup too many data are missing and too many patients were lost to follow-up. Access to the raw data is warranted.

Numbers of participants in the main trial and subgroup analysis2 3

View this table:

Secondly, the authors of the subgroup analysis disclosed strong conflicts of interests.2 I note that one, Henri-Jean Aubin, the chairman of the French Association of Alcohology, will chair a symposium sponsored by Lundeck at the association’s next meeting.4

Thirdly, the European Medicines Agency accepted a reduction in drinking at six months as an endpoint, whereas the US Food and Drug Administration does not. The FDA is right, this surrogate endpoint has not been validated. The European Medicines Agency found this surrogate endpoint only in a ”white paper” jointly released by the first author of the subgroup analysis, Wim van den Brink, who is a beneficiary of honorariums from Lundbeck.5 Moreover, the endpoints in the nalmefene trials were obtained from patient reports, a non-reliable method.6

Lastly, in 1998 Contral Pharma tried to develop nalmefene for alcohol related problems but in 2003 nalmefene did not meet phase III clinical endpoints with statistical significance. These data must be disclosed too.

Notes

Cite this as: BMJ 2014;348:g2017

Footnotes

  • Competing interests: None declared.

References

View Abstract