Intended for healthcare professionals

Rapid response to:


Modernising vaccine surveillance systems to improve detection of rare or poorly defined adverse events

BMJ 2019; 365 doi: (Published 31 May 2019) Cite this as: BMJ 2019;365:l2268

Rapid Response:

Re: Modernising vaccine surveillance systems to improve detection of rare or poorly defined adverse events

Following Jacob Puliyel's comment about narcolepsy and (adjuvant) H1N1 vaccines, someone should note that the signal has not been caught in all European countries: the French pharmacovigilance system missed nearly all adverse events related to pandemrix/focetria (narcolepsy, Guillain-Barré syndrome, relapse of multiple sclerosis, sudden death...).

Observational epidemiological studies represent a weak tool prone to several bias, especially cohort studies with limited time of follow-up, while on the other side delayed adverse events are exactly the type of events which are expected with products like vaccines which are biologically active for years or decades.

Individual causality assessment tools are just not designed to capture delayed adverse events, as what we find in the first WHO causality assessment tool [1] (which has been withdrawn now), to be classified at least as possible/probable the AEFI should occur within "a reasonable time relationship to vaccine administration" without defining the "reasonable" timeframe.

In the French arm of the VAESCO case-control study assessing the risk of narcolepsy following H1N1 vaccines, the median time of onset after the vaccine administration for exposed cases was 9.8 months and the median time to diagnosis was 11.4 months. [2] Would those events be classified within a "reasonable" timeframe? What about Hernan's study assessing the link between Hepatitis B vaccines and multiple sclerosis? [3] The authors had to find the onset date within 3 years after vaccination, and "The mean (median) time between first symptoms and diagnosis was 5.0 (2.7) years" in this study, which is far from being "temporally associated" with the vaccination.

Delayed and irreversible adverse events (which are exactly expected with auto-immune diseases) just don't fit the Challenge-Dechallenge-Rechallenge sequence which is usually considered to be the "gold standard" for individual causality assessment of ADR [4], and it is still used in the official French method of causality assessment [5], while it's now known that this sequence generally doesn't work for vaccines (for irreversible and delayed adverse events, the outcome of the method is "doubtful"), except for those who face acute AEs for which the symptoms will be exacerbated in a multiple dose immunization schedule.

For example, in the Gardasil 9 V503-001 clinical trial, several cases of MS have been reported, including one occurring within hours of dose 2 (within the qHPV arm) [6], others occurring more than one year after the last dose, and none of these cases have been assessed as "vaccine related". In the same review [7], we learn that an investigator in Denmark "reported three SAEs assessed as vaccine-related: one subject (AN 19756) experienced hypersomnia at more than four years after the third dose of 9vHPV, and two subjects (AN 71508 and AN 71686) experienced postural tachycardia syndrome (POTS) at more than three years after the third dose of 9vHPV and qHPV, respectively."

Also there is a bibliographic reference [8] given in the same document [9] about POTS in which "Two patients experienced significant symptomatic exacerbation following a subsequent Gardasil injection, suggesting a positive re-challenge result." And, on the other side, we are told that for the cases identified (and assessed as vaccine-related events) in this V503-001 trial: "Given the lack of temporal association with vaccination of the events reported in the safety update, these events were unlikely related to vaccination."

So when adverse events are close to the vaccination (cases of MS occurring within hours / days / weeks after a dose) they're "not related" and when they're delayed the assessment is also "not related". And we are still waiting for an update about the safety signal between Gardasil 9 and leukemia (as the observed incidence is 6 - 12 times higher than the expected one) identified by the EMA. [10]

1. WHO. Adverse Events Following Immunization (AEFI): Causality Assessment (2005) Available at
2. Etude NarcoFlu-VF (NarcoFlu VAESCO-France) – Rapport Final [text in French]
3. Recombinant hepatitis B vaccine and the risk of multiple sclerosis: a prospective study. Neurology. 2004 Sep 14;63(5):838-42.
4. Quelle pharmacovigilance pour les vaccins? Thérapie 2007 Mai-Juin; 62 (3): 241–244 [text in French]
5. Bégaud B, Evreux JC, Jouglard J, Lagier G. [Imputation of the unexpected or toxic effects of drugs. Actualization of the method used in France]. Therapie 1985 Apr;40(2):111–8.
6. FDA - Gardasil 9 – Clinical review – STN 125508/0 – page 67 -
7. Ibid. - page 147
8. Blitshteyn S. Postural tachycardia syndrome following human papillomavirus vaccination, Eur J Neurol, 2014; 21: 135-9
9. FDA - Gardasil 9 – Clinical review – STN 125508/0 – page 148 -
10. EMA - Gardasil 9 – Assessment report – EMA/CHMP/76591/2015 – page 114 -

Competing interests: No competing interests

03 June 2019
Surya ARBY
Observatoire de la sécurité du médicament
Paris, France