In 2010 we observed a significant increase in childhood narcolepsy in Finland. In 2002–2009, the average annual incidence of narcolepsy in children aged less than 17 years was 0.31 per 100 000 children. In 2010, the incidence rose to 5.3 per 100 000 children being about 17-fold higher as compared to previous years. In adolescents aged 17 to19, the increase was moderate (3-fold), and no increase was seen in adults ≥20 years of age. Based on registry data, a significant (6.6-fold) increase in childhood narcolepsy after Pandemrix vaccination has been reported also from Sweden 
. In a study from the Stockholm area the small number of cases of narcolepsy (six among vaccinated and two in the unvaccinated cohort) did not allow to make reliable conclusions whether clearly increased incidence of narcolepsy was found in that area 
. In the same study the hazard ratio for Bell's palsy among vaccinated against unvaccinated was 1.25 (1.06 to 1.48). Risk for Guillain-Barré syndrome, multiple sclerosis, type 1 diabetes, and rheumatoid arthritis remained unchanged 
. There has been no evidence of narcolepsy or other sleep-related adverse effects in recipients of MF59-adjuvanted A(H1N1) pandemic or other MF59-adjuvanted influenza vaccines 
In 2002 and 2009, narcolepsy was extremely rare in children aged less than 11 years. Only one 9-year-old child had been diagnosed in Finland in 2003. In 2010, twenty-two children aged less than 11 years were diagnosed giving an annual incidence of 3.39/100 000 children in that age group. In the only previous study of the incidence of narcolepsy, Silber and co-workers presented the average annual incidence of narcolepsy to be 1.37/100 000 and 0.74/100 000 for narcolepsy with cataplexy 
. The annual incidence of narcolepsy in 2002–2009 (for all ages) in our study is 0.79 (95% CI 0.62 to 0.96) corresponding well with the published incidence rates 
All diagnoses were based on international criteria 
. The symptoms were relatively severe () but similar to those described in Caucasian children in other recent publications 
and in the 1998–1999 series of 29 children from North China 
. Psychiatric symptoms were common in our patients (48%) as it was in other studies as well () 
. In the Chinese study psychosocial problems were reported in 27/29 (93%) of children and academic problems were reported in 88% of the cases 
. It is possible, however, that the psychiatric symptoms are more severe in post-vaccination narcolepsy than in “normal narcolepsy” prior to 2010. Four of our children have needed long-term psychiatric hospitalisation and antipsychotic treatment was needed in one. More studies are needed to study in more detail the nature of psychiatric symptoms relative to type of narcolepsy. One 12-year-old boy had Type 1 diabetes. He was DQB1*0602 positive, which is interesting since this HLA haplotype is regarded as protective against type 1 diabetes. Knowing the strong genetic association of narcolepsy with HLA type DQB1*0602 
, it is possible that all our narcoleptic children could be DQB1*0602 positive. Hypocretin-1 CSF levels were clearly below 110 pg/ml in all 13 children who had their CSF specimen analysed.
Fatigue and sleepiness and also more severe neurological complications have been associated with influenza 
. Symptoms of the 1918 Spanish influenza consisted also of excessive sleepiness having features similar to narcolepsy. The present 2009 pandemic H1N1 virus is genetically and immunologically more related with the Spanish influenza virus than recent seasonal influenza viruses 
. The present pandemic virus has genes from avian, human and swine influenza viruses 
and the surface hemagglutinin (HA) and neuraminidase (NA) genes are more related to Spanish influenza genes than to present seasonal influenza A virus genes 
. Unlike in the Spanish flu, the present pandemic 2009 H1N1 virus usually causes a mild infection and so far, except in North China 
, no narcoleptic symptoms or increased incidence of narcolepsy has been reported after the two epidemic seasons (2009/2010 and 2010/2011) of the present pandemic virus. It is anyway tempting to hypothesize that the present cases of narcolepsy could have been caused by an H1N1 influenza infection 
. However, narcolepsy was not diagnosed in any of the nearly 8 000 laboratory confirmed H1N1 cases during the first pandemic wave in 2009 
. Also, none of our children had abnormal MRIs or any signs of focal encephalitis/encephalopathy that has been suggested to be involved in the destruction of the hypocretin cells of the hypothalamic area 
Could the increased incidence be explained by the increased awareness of narcolepsy in 2010 compared to previous years - triggered by the intensive public discussion in Finland in August 2010 of the possible association of narcolepsy with Pandemrix vaccination? This is unlikely since in the majority of our patients the symptoms of sleepiness or cataplexy started abruptly before August 2010, and the parents had consulted health care personnel already during the winter or early spring 2010. At that time there had been no news or articles of the increase in narcolepsy incidence or its possible association with the H1N1 pandemic. Also, the symptoms of narcolepsy and cataplexy were so clear that they impaired the daily life of the children. It is very unlikely that similarly severe abrupt sleepiness with cataplexy and behavioural problems would have remained unnoticed before 2009–2010. In addition, the incidence of adult and childhood narcolepsy in Finland in 2002–2009 has been similar to that seen in other countries 
The strength of this study is that it is based on nationwide registries and thus includes the whole Finnish population. We are confident that practically all severe narcoleptic children with a fast onset in late 2009–2010 have been identified. Symptomatic narcolepsy due to brain diseases was excluded by careful neurological examination, EEG, MRI and other examinations 
There are, however, certain clinical aspects that have to be considered. In this study, we are limiting our analysis to the years 2002 to 2010. Some cases have remained undetected, since after 2010 we have continued to diagnose new children who have had their first symptoms during the first half of 2010 (data not presented). In most cases their initial symptoms have been less severe than in the present series, explaining why they have not been sent to specialists earlier. Thus our figure of post-Pandemrix narcolepsy cases is an underestimation. This is consistent with the reports showing that the delay from onset of symptoms to narcolepsy diagnosis may be several years 
. The increased awareness of narcolepsy in Finland, starting in the fall 2010, has lead to a clearly faster diagnosis. The incidence figures in previous years are based on nationwide hospital discharge registry data. Although the diagnostic criteria and practice in Finland has not changed, the general awareness of narcolepsy has increased, which may explain the slightly increased incidence figures in recent years. To avoid bias we compared the incidence figures from 2009 and 2010 also from the same registries in a similar fashion for all years. As stated above, the symptoms of most patients started during the spring (median 38 days after vaccination; see ) before any media attention took place. This means that the possible bias is limited to the speed of diagnosis and not to the disease onset itself. The onset, and the nature of symptoms and other diseases is based on information from the patients, parents, school health nurses and general practitioners, and some inadequacies are possible. In two children EDS started on the day of vaccination. They both fulfilled the diagnostic criteria of narcolepsy with cataplexy. Three children had no clear cataplexy at the moment of diagnosis, but in all of them the diagnosis was confirmed by the lack of CSF hypocretin-1. One of them has developed cataplexy during the follow-up about 18 months after onset of sleepiness. Four of the 50 post-vaccination patients were reported to suffer from ILI. However, microbiological verification of influenza or any other microbial infection had not been done suggesting a lack of clinically important microbial infections in our narcoleptic patients. Systematic determination of CSF hypocretin-1 and HLA typing were not done for all patients since they are not obligatory in the diagnosis when cataplexy is present and also MSLT is verifying narcolepsy 
. Also some children/parents did not give their consent for CSF specimen when it was suggested.
Narcolepsy is considered an immune-mediated, autoimmune disease. In addition to a strong association especially with HLA DQB1*0602/DRB1*15/DR15 – DQ6 HLA type, narcolepsy has been associated with the presence of TRIB-2 antibodies 
, specific T-cell receptor alpha 
, and purinergic receptor P2RY11 
genotypes. The onset of symptoms in our children has been very abrupt which contrasts with most other autoimmune diseases 
and with previous concepts of the natural course of narcolepsy 
. The incidence of many autoimmune diseases has increased over the past decades 
, but the rise has never been as abrupt and as strong as in the case of childhood narcolepsy in Finland in 2010.
What could be the mechanism(s) of a sudden increase in childhood narcolepsy in Finland and Sweden? Co-occurrence is not synonymous to causation. We have therefore carefully differentiated the effects of the pandemic and the effects of the vaccination. What is remarkable in Finland is that only the Pandemrix vaccine was used and that the vaccination coverage was very high in children and adolescents (75%). In the age-group 5 to 14 years the vaccination coverage was more than 80% 
. In children the vaccination coverage in Sweden was approximately 67% 
. In the case Pandemrix vaccination contributed to the onset of narcolepsy, the high vaccination coverage in Finland and Sweden may explain the highly increased narcolepsy incidence in these countries as compared to those countries where the H1N1 vaccination coverage was much lower, e.g. in France 10% were vaccinated, and in Italy 0.3% 
. In the Netherlands, where only few Pandemrix-related narcolepsy cases were found, Pandemrix was used in healthy children aged 6 months to 5 years and in siblings and close relatives of children aged less than 6 months, but not in older children 
. In addition to the mathematical explanation, the lack of an increase in narcolepsy in other countries would suggest that there are other genetic or environmental factors, in addition to the AS03 adjuvanted vaccination, contributing to the onset. Recent microbial infections have been suggested as possible environmental triggers that initiate the symptoms of narcolepsy 
. A likely trigger in our patients could be influenza vaccination, which took place in a close time-relation with the onset of narcolepsy. Vaccination may have induced or accelerated already pre-existing autoimmunity leading to a rapid destruction of the hypocretin cells among genetically susceptible children and adolescents. All our HLA typed patients (n
32) were of HLA DQB1*0602 type, which is the presently known major genetic susceptibility factor 
. We cannot formally rule out the contribution of other infectious agents (H1N1, seasonal influenza, enterovirus, rhinovirus, streptococcal infection 
, or some other microbial infections) together with vaccination that could have lead to the development of narcolepsy.
A recent Chinese study observed an epidemiological link with pandemic influenza and narcolepsy, without significant relation to vaccination, showing a 3-fold increased incidence of narcolepsy 5 to 7 months after the 2009 epidemics 
. ILI, without microbiological confirmation, was observed in only 4 of our patients (10%). There was, thus, no evidence of simultaneous or closely time-related microbial infections in most of our patients. Unfortunately, serum or CSF specimens for detailed viral and streptococcal antibody analyses were not available since there was no suspicion of microbial etiology of the narcoleptic symptoms. Almost a similar increase in the incidence of narcolepsy has been reported in Sweden, where only Pandemrix was used, as in Finland 
. Although isolated cases of narcolepsy have been diagnosed also after other vaccinations than Pandemrix, there is no evidence of an increased risk of narcolepsy with any other vaccine than the AS03 adjuvanted Pandemrix 
. In order to draw a link between an environmental effect one must show that the incidence has increased significantly as compared to previous years. This was the case in the Chinese study, where the increase was not explained by Pandemrix 
. Combining their and our results, it seems that several different or multiple triggering factors may exist for the narcolepsy to develop. Based on our study and the study by Nohynek and co-workers 
Pandemrix vaccine was likely one of the triggers. The role of possible other closely time-related triggers remains to be studied in future epidemiological and experimental studies.
The adjuvant (AS03) in the Pandemrix vaccine is very potent, since it frequently induces local inflammatory reactions and occasional systemic side effects like fever. We can speculate that the inflammatory response was so strong that it included central nervous system affection. It is, however, difficult to say whether the systemic response produced by the adjuvant could have caused cellular damage. The hypocretin-1 levels were very low or undetectable in our narcoleptic patients, which may indicate a rapid destruction of hypocretin cells within weeks or a few months after vaccination. This does not, however, mean that the adjuvant would be causing hypocretin cell damage directly. Rather, there is a possibility that already an ongoing autoimmune process was accelerated by the nonspecific inflammatory responses induced by the vaccine or its specific components leading to the destruction of hypocretin producing cells. Thus, theoretically any inflammatory process whether it is iatrogenic or infectious (such as influenza and streptococcal infection) could non-specifically enhance the autoimmune process leading to narcolepsy. There is also a possibility that some other genetic factors, in addition to the commonly accepted HLA-DAB1*0602 risk allele 
are enriched in Finland and Sweden making Nordic children susceptible to an acute inflammation/autoimmune-related narcolepsy.
Why were adult onset cases not increased? Theoretically it is possible that the vaccine precipitated onset in people who would have developed it later, anyway. In this case we should see a drop in adult incidence later during the coming years. Another possibility is that some children with multiple genetic predisposition factors are especially vulnerable to develop narcolepsy. In some other autoimmune diseases, such as in type 1 diabetes, early age onset is also often seen. It is also possible that the onset is more insidious in older adolescents and adults and thus there may be a delay in the diagnosis. In this case we would expect an increase in the incidence in adults during the coming years. All this mandates continued clinical and epidemiological surveillance in the future.
We observed a 17-fold increase in the annual incidence of narcolepsy in 2010 as compared to previous years in children aged under 17 years of age. A common feature in the history of our 54 newly diagnosed childhood narcoleptic patients was that 50 children had received an adjuvanted pandemic influenza vaccine (Pandemrix) within 8 months before the onset of symptoms. In most cases, the development of symptoms was fast. We consider it likely that Pandemrix vaccination contributed to the increased incidence of narcolepsy in Finland in 2010 in HLA DQB1*0602 positive children. Our observations warrant further studies on the role of different environmental factors as well as pathogenetic studies to understand how a vaccination/adjuvant and other environmental triggers can cause narcolepsy.