The Flu Vaccine and Narcolepsy

Last year it was reported that there was a possible increase in narcolepsy, a sleep disorder characterized by excessive sleepiness, in children who had received the Pandemrix brand of H1N1 flu vaccine in Sweden, Finland, and Iceland. However a review of the data did not find a convincing connection, although concluded there was insufficient data at present and recommended further surveillance. A narcolepsy task force was formed in Finland, and now we have their preliminary report.

They conclude that the evidence suggests there is a connection:

Based on the preliminary analyses, the risk of falling ill with narcolepsy among those vaccinated in the 4-19 years age group was 9-fold in comparison to those unvaccinated in the same age group. This increase was most pronounced among those 5–15 years of age. No cases were observed among those under 4 years of age. Also, no increase in cases of narcolepsy or signs of vaccination impacting risk of falling ill with narcolepsy was observed among those above 19 years of age.

The World Health Organization (WHO) has reviewed these results and concluded:

WHO’s Global Advisory Committee on Vaccine Safety (GACVS) reviewed this data by telephone conference on 4 February 2011. GACVS agrees that further investigation is warranted concerning narcolepsy and vaccination against influenza (H1N1) 2009 with Pandemrix and other pandemic H1N1 vaccines. An increased risk of narcolepsy has not been observed in association with the use of any vaccines whether against influenza or other diseases in the past. Even at this stage, it does not appear that narcolepsy following vaccination against pandemic influenza is a general worldwide phenomenon and this complicates interpretation of the findings in Finland.

I agree with the WHO, who is basically saying that these results are intriguing, but are problematic and should be considered preliminary. They then follow with – more research is needed. Epidemiology is a complex endeavor, and there are lots of wrinkles to this data. The increased risk of narcolepsy was only seen within a certain age range. In Iceland (but not Sweden or Finland) the increase in narcolepsy was also seen in those who were not vaccinated. And further, other countries (47 in total) that also used the Pandemrix vaccine have seen no increase in narcolepsy, including Norway, the UK, Germany, and Canada.

Overall we have a very inconsistent pattern. The vaccine does not appear to be a consistent or unique risk factor for narcolepsy in these populations. The task force concludes from this that there must be another factor or factors that is combining with the vaccine to increase the risk. This is logically possible, but until this factor X is identified it remains speculation.

Narcolepsy is a neurological disorder characterized by excessive daytime sleepiness (narcoleptics sleep all night and all day), cataplexy (a tendency to lose muscle tone and collapse in response to stress), hypnagogia (hallucinations associated with a fusion of the dream state and the waking state, so-called waking dreams), and sleep paralysis (being paralyzed upon awaking from sleep). There is a strong genetic predisposition for narcolepsy. In fact it is only seen with a certain genetic type known as the (HLA) DQB1*0602 genotype.

All of the individuals who developed narcolepsy following the Pandemrix vaccine have the narcolepsy HLA type. Therefore there is the possibility that the vaccine only increases risk within this genetic populations, perhaps combined with other gene variants. Perhaps something else is also triggering the increase in Iceland, but not Finland and Sweden, to explain the rise in narcolepsy there in the unvaccinated.

Another possibility is that there is one or more confounding factors leading to the increase in narcolepsy, and the vaccines are a correlating but not causative factor.

Such is the nature of epidemiology, or observational studies. Variables are not controlled for and confounding factors are always a possibility. That does not mean that observational data is not useful or cannot be definitive – but it requires careful, thoughtful, and thorough collection and analysis of data from multiple different angles. The data we have so far from Finland is very preliminary, and generates more questions than answers. There is certainly sufficient cause for caution and further analysis. But at this point I would not be surprised by any particular outcome, since the data can be interpreted in many ways.


While there is an intriguing correlation between the Pandemrix vaccine and narcolepsy, this correlation is inconsistent – it is isolated to a few countries and to one age group and there is a rise in narcolepsy in Iceland not correlated to the vaccine. Further the cases identified so far are restricted to those with a known genetic predisposition to narcolepsy. This could mean that this population is susceptible to some factor in the vaccine, but it could also mean that they are susceptible to some other trigger, or perhaps were destined to get narcolepsy and the apparent increase in entirely an artifact of observation and reporting.

I agree with the WHO that this data should be considered preliminary – which means it is worthy of further monitoring and research, but we are not able to make any firm conclusions at this time.  I would not be surprised if it turns out to be a real effect of the Pandemrix vaccine. Vaccines are not without risk, although over the decades the risks have proven to be very small and vastly outweighed by the benefits. Obviously it would be hugely useful to identify which ingredient was the culprit and exactly how it triggered narcolepsy in this population. But I would also not be surprised if this turns out to be entirely a red herring. Such is the nature of observational data.

Posted in: Neuroscience/Mental Health, Vaccines

Leave a Comment (23) ↓

23 thoughts on “The Flu Vaccine and Narcolepsy

  1. Scott says:

    Is it not also plausible that the effect is merely a statistical fluctuation? It would seem to me that monitoring for such effects is inherently a case of multiple comparisons. And furthermore that it’s unlikely those multiple comparisons are corrected for, since the point is to make sure any real association is picked up. Hence statistical false positives seem likely to occur.

    I don’t see any discussion of the statistics in either link which would address this point.

  2. penglish says:

    I was thinking along similar lines to Scott.

    I don’t know what the P value was for the findings in Finland. But, assuming it was close to 0.05, then, if you do similar studies in over 20 countries you’d expect one to have this finding, by chance…

  3. Scott says:

    Or, just look at 20 possible conditions in Finland.

    Yes, they looked at narcolepsy specifically because there were reports of it, but unless the population in which the statistics were done was completely distinct from the population in which the monitoring which noticed the cases was done, those are not independent results and therefore it’s still multiple comparisons.

  4. windriven says:

    From the preliminary report:
    “In Finland during years 2009–10, 60 children and adolescents aged 4-19 years fell ill with narcolepsy. These figures base on data from hospitals and primary care, and the review of individual patient records by a panel of neurologists and sleep researchers. Of those fallen ill, 52 (almost 90 percent) had received Pandemrix® vaccine, while the vaccine coverage in the entire age group was 70 percent.”

    Unreported is the number of people in that cohort who might be expected to develop narcolepsy in any given year. It would also be useful to know what percentage of the cohort have the narcolepsy allele and how prevalence of that allele varies geographically. Also, have any linkages been found between surviving H1N1 and developing narcolepsy in at-risk populations? That is to ask, is the apparent risk more likely related to the virus or the adjuvant?

    It seems to me that the report is interesting but little more.

  5. ejwillingham says:

    That hasn’t stopped the news media from going nuts about it:

    Hope your post gets through all the noise.

    Finland seems to have a founder effect going with the higher frequency of that allele.

    Question: Are “they” still using GSK Pandemrix worldwide? I’ve read conflicting information, some asserting that it was only used during the pandemic, others that it’s still in use.

  6. Jan Willem Nienhuys says:

    If you draw 60 times a random number from {1,2,3,4,5,6,7,8,9,10} the probability
    of drawing 52 or more times the number {1,2,3,4,5,6,7} equals 0.0022 . That answers the question about the size of p. However, one expects about 42 times a number less than or equal 7. One possibility is that the exitement about narcolepsy has caused preferentially reporting narcolepsy after vaccination. If that is the case, then doing meaningful statistics becomes very hard.

    If a task force narcolepsy was formed, it sounds as if there was already quite some publicity about it.

  7. aeauooo says:

    Question: Is the DQB1*0602 genotype sufficient to cause narcolepsy?

    What is the probability that these children would have developed narcolepsy at some point in their lives?

  8. aeauooo says:

    A cursory review of the literature:

    Family studies indicate a 20-40 times increased risk of narcolepsy in first-degree relatives and twin studies suggest that nongenetic factors also play a role.

    The genetics of narcolepsy.

    RESULTS: Of the infectious diseases examined, only flu infections and unexplained fevers carried a significant risk. Several of the stressors carried a significant risk including a major change in sleeping habits. When the timing of all risk factors was considered, exposure prior to puberty increased the risk for developing narcolepsy.

    A case-control study of the environmental risk factors for narcolepsy.

    Results: We observed an increased risk of narcolepsy associated with having lived with two or more household smokers (odds ratio, OR = 5.1; 95% confidence interval, CI: 1.6, 12.1); with a grandparent or a sibling who smoked (OR = 3.0; 95% CI: 1.1, 8.3); with a non-family household member who smoked (OR = 3.7; 95% CI: 1.6, 8.6); and with an unrelated smoker for 1–2 years (OR = 3.1; 95% CI: 1.0, 9.0). The risk of narcolepsy was not associated with exposure to smoke at work or with active smoking before age 21 or before age of narcolepsy onset. Conclusion: Passive smoking may be a risk factor for narcolepsy in subjects with HLA DQB1*0602. Future studies could help clarify whether passive smoking is an important etiologic component of narcolepsy among genetically susceptible individuals.”

    Active and Passive Smoking and Risk of Narcolepsy in People with HLA DQB1*0602: A Population-Based Case-Control Study

    When using the population data for comparison, the relative risk of narcolepsy in first-degree relatives was 85.3. The odds ratio of narcolepsy spectrum disorder in first-degree relatives was 5.8 (95% CI: 1.2 – 29.3) when compared to healthy controls. There existed 6 multiplex families, in which all 10 relatives with narcolepsy spectrum disorders, including all 3 relatives with multiple SOREMPs, were positive for HLA DQB1*0602.

    The familial risk and HLA susceptibility among narcolepsy patients in Hong Kong Chinese.

  9. daedalus2u says:

    I left a similar comment earlier over at Neurologica.

    This may be one of the side effects of removing thimerosal from flu vaccines. One of the things thimerosal was used for was as an antibacterial during the growth of virus on the fertile chicken eggs and during the “work up” of the virus containing chicken cells. Flu virus is grown in living chicken eggs, which are laid by chickens, which live in chicken coops and who have all manner of natural and environmental bacteria living on them and in them and the eggs of course come into contact with those bacteria as they are laid.

    If an egg becomes contaminated with bacteria when it is injected with flu virus, that bacteria could multiply and contaminate the final product. Probably not with living bacteria, but with bacterial antigen. That may not present a problem unless there is autoimmune sensitization.

    One of the risks associated with this specific HLA genotype is a susceptibility to autoimmune sensitization following a Streptococcal infection. If the eggs did become contaminated with Streptococcal bacteria, they could generate antigen which might contaminate the final product with low levels of Streptococcal antigens.

    There were reports at the time that this virus was difficult to grow in eggs, that it was slow growing, and that there might not be enough to go around, and that this strain was pretty deadly. That was one of the reasons that adjuvants were used, to reduce the amount of virus needed, so that more people could be vaccinated faster. It typically takes 1 to 2 eggs per dose, so these 500,000 dose lots represent half a million or a million eggs. Chickens can lay only 1 egg per day, so this represents a lot of chickens laying a lot of eggs.

    Purification of biological products (and all other things too) is always a trade-off of purity for recovery. You can always increase recovery if you allow more impurities to pass through. That may be what happened here. The combination of low virus yield per egg (meaning more eggs per dose), and a need to get as much virus as possible may have increased the non-viral antigens that were recovered too.

    Allergy to eggs is always an issue for this type of flu vaccine, maybe allergy to other trace constituents is an issue too. This isn’t something that could be known ahead of time because it is so rare. It looks like not every batch of vaccine was affected. It might not even be a problem with the virus source, it might be one of the other ingredients. Some of these antigens can be very difficult to remove from raw materials.

  10. Zetetic says:

    Interesting that reports out of Norway don’t parallel Iceland’s results if you maintain there’s a genetic factor. Norwegians and Icelanders share a lot of DNA!

  11. aeauooo says:


    One of the theories about the increased incidence of Guillain-Barre syndrome associated with the 1976 “swine flu” vaccine is that the egg supply was contaminated with Campylobacter.

    C. jejuni infection is recognized as a “trigger” for GBS.

  12. Ian says:

    @Zetetic: as I understand it, Iceland was founded by a relatively small group of folks who were probably Norwegian, and Icelanders are predominately all descended from the group. But this creates textbook genetic drift, as there would be genotypes in Norway that could be mostly absent in Iceland if they happened to not be in that original population.

  13. simil says:

    In the finnish report they provide the absolute number of diagnosed cases from 2006 to 2010. Here’s a snap:

  14. ConspicuousCarl says:

    Can someone help me with the math here? I can’t figure out the “9-fold” thing.

    60 cases total.

    Then they say,
    “Of those fallen ill, 52 (almost 90 percent) had received Pandemrix® vaccine, while the vaccine coverage in the entire age group was 70 percent. Based on the preliminary analyses, the risk of falling ill with narcolepsy among those vaccinated in the 4-19 years age group was 9-fold in comparison to those unvaccinated in the same age group. ”

    I can only estimate the population size for that age range:

    (they have 3 groups which in total go from 5-19, not 4-19)
    (stats for 2005, but since the population only went from 5.1 million in 2000 to 5.3 in 2008, I assume it doesn’t change much)

    I estimate that to be about 500,000 kids in the age group.

    70% were vaccinated, therefore:
    350,000 vaccinated
    150,000 not

    52 / 350,000 is the narcolepsy rate if you are vaccinated
    8 / 150,000 is the rate if you are not


    0.00014857 vs.

    That’s slightly less than what I would call “3-fold”.

    Obviously one of us is wrong, and it is probably me, but I want to know why.

  15. squirrelelite says:

    @Conspicuous Carl,

    Someone probably just didn’t do the math as well as you did.

    Windriven noted that 90% of the 52 cases had received the vaccine. Someone probably just divided the 90% who had received the vaccine by the 10% who had not and pulled out a 9x increase in risk.

    That ignores the different numbers in the two different risk groups, but it does give the 9x number.

  16. desiree says:

    ConspicuousCarl, i left the exact same comment on dr. novella’s other blog. we got the exact same results. the press release does mention that the results were “most prominent” in the 5-15 age group, so maybe they mean to say the increase is in that group? also, i looked at the report even though i don’t read a word of finnish, and it looks like the 5age of vaccination was higher in some of the younger groups. maybe re-crunching just that 5-15 age group, with it’s 75% (or so) rate of vaccination, will give a 9-fold increase. i’ll check later if the little one takes a nap.

    also in the finnish report, they give incidence over the past 5 or so years. simil linked to it. the total incidence in the 0-19 age group was 5 cases, 6, 14, and 11 from 2006 to 2009. if there were 8 cases in 30% of the population, you can extrapolate that there would have been 26.67 cases in 100% of the population even without the apparent effect of the vaccine. so it certainly seems something else might be going on.

  17. desiree says:

    by “5age” above i mean “%age”

  18. Ufo says:

    desiree, the rise of narcolepsy cases occured ONLY in the 5-19 age group.

    Here are the numbers:

    62 kids (age 5-19) got narcolepsy in 2010.

    About 90% of the kids with narcolepsy were vaccinated.

    The number of vaccinated kids in total: 627 120

    The number of non-vaccinated kids in total: 297 725

    Could it really be that they didn’t adjust the numbers appropriately? Hmmm, I contacted one health care professional about this earlier today, maybe he knows what’s going on.


    Ufo from Finland

  19. MansThulin says:

    I just wrote a blog post trying to explain the mathematics behind the conclusion about a 9-fold increase in risk:

    What it all comes down to is that the two groups (vaccinated/unvaccinated) were studied for different time periods, which has to be taken into account.

  20. Ufo says:

    Thanks MansThulin, your analysis mirrors the response I got from the “officials”. They are adjusting their public statement to include more details about this. It should be online in a few days.



  21. antipodean says:

    Hi Steve

    You’ll be interested to know that there are some very serious people taking this very seriously. Having said that it could all be nothing and everybody is acknowledging that.

    You left out of your description of Narcolepsy the most important bit for understanding why this situation is being closely watched.

    Classic cataplectic narcolepsy is caused by the death of the small number of orexin neurons in the hypothalamus (these are also called hypocretin neurons due to simultaneous discoverey). As such patients with classic narcolepsy usually have undetectable levels of orexin in their cerebrospinal fluid.

    The current thinking is that the death of the orexin neurons is caused by an autoimmune process and that’s where the HLA subtypes get very interesting.

    This is how the Stanford people describe it for the lay public:

    What is HLA?
    HLA stands for ” Human Leukocyte Antigens”. HLA antigens are molecules produced by the HLA genes. HLA molecules are expressed on the surface of white blood cells to coordinate the immune response. DR and DQ are two different types of HLA molecules. HLA genes are very important systems to keep the immune system in check. The HLA molecules are very particular in that different individuals generally carry different HLA “subtypes” (for example DR1, DR2, subtypes of HLA-DR; DQ1, DQB1*0602, subtypes of HLA-DQ). The fact HLA molecules are slightly different from one individual to another makes our immune system slightly different from each other.

    What is the best HLA marker in narcolepsy?
    The best HLA marker for narcolepsy is HLA-DQB1*0602. Over 90% of patients with narcolepsy-cataplexy carry HLA-DQB1*0602. This marker is more specific and sensitive than the old marker HLA-DR2, especially in African Americans.


    Because most people with these HLA subtypes don’t have narcolepsy this seems to suggest some sort of trigger that only works in genetically susceptable people. One of the factors might be ‘flu or something somehow related to ‘flu and there are risk factor observations of this.

    Now that might seem a little ‘out there’. But there are very complex back and forth relationships between sleep/wake and immunity. There are also some largely historical diseases such as encephalitis lethargica (von economo or sleeping sickness seen in the movie Awakenings for movie buffs) which may or may not have been caused by influenza.

  22. Ufo says:

    Which leads me to a question I’ve had on my mind for a while, but haven’t remembered to find out more about:

    Are there any reliable ways to distinguish whether these patients got narcolepsy from the swine flu itself or from the vaccine?



  23. antipodean says:

    UFO-“Are there any reliable ways to distinguish whether these patients got narcolepsy from the swine flu itself or from the vaccine?”

    That’s assuming that either had anything to do with it. We don’t know.

Comments are closed.