Global Vaccine Safety

World Health Organization Global Advisory Committee on Vaccine Safety: Response to the paper by MA Hernán and others in Neurology 14th September 2004 issue entitled “Recombinant Hepatitis B Vaccine and the Risk of Multiple Sclerosis”

September 2004

1. The Global Advisory Committee on Vaccine Safety (GACVS) of the World Health Organization has given careful consideration to the article published by MA Hernán and others in the 14th September 2004 issue of Neurology (63(5):838-842) on the risk of multiple sclerosis associated with recombinant hepatitis B vaccine. The findings are based on a case control study conducted within the General Practice Research Database (GPRD) in the United Kingdom. Dr Hernán has previously presented his results and interpretation of them to the GACVS at the December 2003 meeting of the committee. On the basis of the data and argument given by Hernán et al in their article, the GACVS does not believe that the findings provide convincing support for the hypothesis that immunisation with recombinant hepatitis B vaccine is associated with an increased risk of multiple sclerosis.

2. The main misgivings of the committee with the Hernán paper are as follows:

  • The authors’ conclusions depend on the results of altogether 11 cases of MS adult patients vaccinated against hepatitis B. Such a sample size is too small for a definite interpretation to be made, one way or the other. On the basis of these 11 cases, Hernán and his colleagues calculate an odds ratio of 3.1 (95% CI 1.5, 6.3 – a wide confidence interval) of MS in patients vaccinated with hepatitis B within 3 years before the index date, compared with an odds ratio of 1 and 0.6, respectively, for influenza and tetanus vaccines. The committee noted that there is inherent risk, however small, of misclassification of vaccination status and that minimal differences in the recorded data would render the results and conclusions insignificant.
  • The practice of vaccination against hepatitis B in the United Kingdom at the time of the study was targeted towards high-risk individuals. They included health care and laboratory workers, travellers to endemic regions, those with liver damage, dialysis patients, prostitutes and drug addicts. This cannot be regarded as a representative sample of the general population, and the skewed selection might have biased the results of the study. For example, health care workers might bring neurological symptoms to the attention of their physicians earlier than others, or the recall of symptoms might be different. Other potential sources of bias with such a select group cannot be excluded. (This argument has been made in the accompanying editorial in Neurology that covers the Hernán article; refer RT Naismith and AH Cross. Neurology, 2004(63): 772-773.) In general, since vaccination of health care workers in the UK usually takes place in occupational health departments, there is likely to be incomplete information on the vaccination status in this group in the GPRD.
  • Of the original 713 cases of MS 163 were selected, and eventually only 11 vaccinated patients were used for deriving actual hazard data. This selection process, however careful it might have been, is fraught with methodological problems and with the risk of inadvertent bias. An inadequate account is given in the article of the cases that were excluded from analysis. Moreover, no population attributable risk is provided.
  • Since no association of the onset of MS was found with other vaccinations (influenza and tetanus) in the study, the authors’ suggestion that the association of MS with HBV might be explained by the aluminium or thiomersal content of the latter must be considered unsupported by other studies and even by their own evidence. (It should be pointed out that the authors’ conclusions regarding the other vaccines suffer from the same criticism of small study sample that applies to the HBV analysis.)
  • The date of first symptoms was related in the study to the last dose given. There is no information provided of the total number of doses and the time that previous doses were administered, precluding any judgment of a dose-response effect. The biological plausibility of the findings might be questioned by the observation that the risk only appeared greater more than one year after the last immunisation. This delay does not support the hypothesis of a triggering by HBV and is not in accordance with the French pharmacovigilance data which prompted this study.

The GACVS has noted that the findings and conclusions of the Hernán paper are at variance with those of a number of others; namely Ascherio (2001)1, De Stefano (2003)2, Touze (2002)3, Sturkenboom (1999)4, Confraveux (2001)5, Zipp (1999)6, Sadovnick (2000)7. Whereas each of these papers has its own methodological problems, it is notable that the overall conclusion of each is consistently different to those of Hernán. Data accumulated globally during the last 20 years has also provided evidence supporting the safety of hepatitis B immunization in infants and adolescents. Nevertheless, it is important that the questions raised by Hernán and colleagues are put to the test and the GACVS will be keeping a watching brief on the issues. For the time being, the GACVS has advised the WHO that the evidence and argument submitted by Hernán et al are insufficient to support the hypothesis of a link between hepatitis B vaccination and MS, and do not justify discontinuation or modification of immunisation programmes with HBV. The latter have had a demonstrated profound beneficial public health benefit worldwide.


Page last updated: 01 October 2008