New Swedish Registry study confirms increased risk of narcolepsy in children and adolescents after vaccination with Pandemrix

ECDC comment

​(Conducted by the Swedish Medical Product Agency in collaboration with the Karolinska Institute)
This observational cohort study utilizing electronic health care records covering 5.8 million Swedes confirmed an increased risk of narcolepsy after vaccination with Pandemrix in children and adolescents (<20 years).   Relative risks in young adults 21-30 years of age (hazard ratio 2.18; 95% CI: 1.00-4.75) and the age group 31-40 year (hazard ratio 1.53; 95% CI: 0.68-3.44) were not statistically significant. No increased risk was observed for individuals 40 years and older. The authors note however, that it is not possible to specify an exact age where an increased risk is no longer present.

​(Conducted by the Swedish Medical Product Agency in collaboration with the Karolinska Institute)

This observational cohort study utilizing electronic health care records covering 5.8 million Swedes confirmed an increased risk of narcolepsy after vaccination with Pandemrix in children and adolescents (<20 years).   Relative risks in young adults 21-30 years of age (hazard ratio 2.18; 95% CI: 1.00-4.75) and the age group 31-40 year (hazard ratio 1.53; 95% CI: 0.68-3.44) were not statistically significant. No increased risk was observed for individuals 40 years and older. The authors note however, that it is not possible to specify an exact age where an increased risk is no longer present.

All Swedes were offered an AS03-adjuvanted pandemic vaccine (Pandemrix) in a national vaccination campaign that started October 2009. This study was conducted in response to the increased number of narcolepsy cases first reported by neurologists to the Medical Product Agency in the summer 2010 in Sweden and made public in August 2010.

Methods

It was an observational cohort study covering a population of 5.8 million individuals (61% of the Swedish population) and utilizing linkage of a number of different electronic health registries.  Records were linked using the personal identification number allocated to all Swedish citizens at birth or upon immigration to Sweden.  Health registries included immunization registries owned by the seven participating country regions, a number of registries owned by the National Board of Health and Welfare (in- and out-patient data from hospitals, death, birth, cancer and utilized medicinal product registries) and a number of registries owned by Statistics Sweden (population, income, educational level, pregnancy and country of birth) (1).
 
Exposure of interest was vaccination with Pandemrix. Different medical outcomes including narcolepsy that had developed during the selected study period of October 1, 2009 to December 31, 2011 were evaluated and compared in the study population of whom 3.3 million were vaccinated and 2.5 million were unvaccinated.
 
Other outcomes evaluated in the study were more than 50 medical conditions/disease groups including mortality (see Note below). Cases were identified using ICD-10 codes registered either as in-patient or out-patient specialist care at one of the hospitals providing health care in the seven participating regions. There were no similar data bases available for the health care provided by general practitioners. The outcomes evaluated were grouped as follows; a) vaccine reactions including allergic reactions, b) narcolepsy, c) other neurological disorders, d) autoimmune disorders and e) other outcomes that had been reported as possible adverse events.
 
Hazard ratios of narcolepsy in vaccinated compared to unvaccinated in various time windows were used to estimate the relative risks.
The study was conducted by the Swedish Medical Product Agency (Ingemar Persson and Nils Fältelius) in collaboration with the Centre for Pharmacoepidemiology, Karolinska Institute, Stockholm, Sweden (Fredrik Granath).

Results

The overall vaccination coverage in Sweden during the 2009 pandemic was estimated to be around 60% and in the chosen study population it was noted to be 57%.  It was highest in the <20y age group (63%), and lowest in the 20-29y age group (~40%).
 
This study showed an almost three-fold (hazard ratio of 2.92 with 95% CI: 1.78-4.79) increased risk for development of narcolepsy in the study period in subjects 20 years and younger at time of vaccination. The authors comment that the increased risk observed in this study is slightly lower compared to results from an earlier registry study in Sweden using a shorter study period and suggest that this discrepancy may be explained by an increase of diagnosed cases of narcolepsy also in unvaccinated children and adolescents given the significant increased media- and professional attention. The increased risk during the study period corresponds to an absolute risk increase of approximately four additional cases of narcolepsy per 100 000 person-years.
 
Relative risks in young adults 21-30 years of age (hazard ratio 2.18; 95% CI: 1.00-4.75) and the age group 31-40 year (hazard ratio 1.53; 95% CI: 0.68-3.44) were not statistically significant, also for the younger adult age group 21-30 the lower CI interval includes 1.   No increased risk was observed for individuals 40 years and older but the authors caution that it is not possible to specify an exact age where an increased risk is no longer present.
 
Of note, no increased risk for most of the other studied neurological and immune related/autoimmune diseases was found. However, for eg mononeuropathy, smell- and taste abnormalities, agranulocytosis and idiopathic trombocytopen purpura among several other conditions smaller statistically significant increased risks were described but and none of them were close to the elevated risk observed for narcolepsy and vaccine reactions (1). The may also be a result of mass significance. The authors conclude that a possible association with the conditions with a low elevated risk and Pandemrix observed cannot be distinguished from the effect of other possible confounding factors. They also stress that the lack of increased risk for other typical autoimmune diseases such as systemic lups erythematosus, reumatoid arthritis, autoimmune thyroiditis speaks for a selective mechanism behind development of narcolepsy.
 
The overall hazard ratio for vaccine reactions (ICD-10 T88.1) was 6.14 (95%CI 3.96 – 9.52). Authors are confident that they have retrieved most narcolepsy cases since the expected vaccine reactions were retrieved using the data linkage methodology as a positive control.
The authors stress that interpretation should consider that several factors may influence the results such as that risk groups were offered vaccination earlier than others and try to address that in the study by presenting data for individuals vaccinated first 45 days and the following period of the national vaccination campaign.
 
It was noted that vaccinated individuals differed from the unvaccinated in that they had been hospitalized to a greater extent before the defined study period; they had higher reported personal income and were more likely to be born in one of the Nordic countries.

ECDC Comment (15 April 2013):

This is, to our knowledge, the largest national study to date of the association between vaccination with the AS03-adjuvanted influenza A(H1N1)pdm09 vaccine Pandemrix produced by GSK in the Dresden facility in Germany and development of narcolepsy.  It used different national and regional medical outcome databases. Data lags, common in electronic databases have been a limitation of such studies until recently but with the passing of time more accurate results may now be expected.
 
This study showed an almost three-fold (hazard ratio of 2.92 with 95%CI: 1.78-4.79) increased risk for development of narcolepsy in the study period in subjects 20 years and younger at time of vaccination. Relative risk in young adults 21-30 years of age (hazard ratio 2.18; 95% CI: 1.00-4.75) was not statistically significant, but somewhat in line with the earlier results of the French national study (2) and the European VAESCO study (3). Increased statistical power may change these results, which may be possible through international collaborations which have been initiated beyond the conducted VAESCO narcolepsy study (4). The increase observed in the age group 31-40 years of age was not statistically significant either. ECDC is aware of further national studies being conducted in adults in Finland, Norway and the UK where initial investigations focused on the age groups where most cases were reported.
 
Overall, this study found a lower relative risk of narcolepsy than previous Swedish registry studies (5). The authors suggest this is likely to be caused by increased diagnosis of unvaccinated individuals with this disease that often has an insidious onset due to the extensive media and professional attention. The role of media and professional awareness has been discussed by all previous investigators as well as the European Medicines Agency.
 
This study has several limitations. In Sweden, no general practice databases were available for data linkage meaning that some cases of various medical outcomes under study may have been missed. In addition, case validation through medical chart review is not mentioned in the report (at least at this stage).  We know from  the VAESCO narcolepsy study and other studies this is of importance since at least in some countries the positive predictive value of certain diagnoses in the electronic medical records was lower than expected.
Adjustment for confounding factors such as underlying chronic diseases, medication,  calendar time, pregnancy, socioeconomic factors including income and education and ethnicity (country of birth) was performed but the effect on the various outcomes under study was not reported at this stage.
 
Finally, the epidemiological studies reported so far (6) all aimed at quantifying the statistical association between vaccination and the occurrence of narcolepsy. To more fully understand the causality, the possible biological mechanisms behind development of narcolepsy following vaccination with Pandemrix must be better understood. As of today more than ten research groups mainly from the Nordic countries, France and the US (in collaboration with the manufacturer GSK) have initiated studies including comparisons between the two AS03-adjuvanted vaccines Pandemrix and Arepanrix (6).
 
The study results will be discussed by the EMA PRAC (Pharmacovogilance Risk Assessment Committee) in April, 2013.
Comments or questions on this public heath development are welcome and should be addressed to: influenza@ecdc.europa.eu.

 

Note: Shoulder neuritis, chronic fatigue syndrome, burn-out syndrome, chronic pain, dizziness, myopathy, rheumatism, tension syndrome, taste- and smell disturbances, paraparesis, mononeuropathy, muscular disease, sclerodermia, expholiative dermatitis, erythema nodosum, SLE, Sjogren syndrome, inflammatory systemic disease, Addison disease, cutaneous vasculitis, hyperthyreodism, hemolytic anemia, thyreoditis, trombocytopenic purpura, rheumatoid arthritis, asthma, hypothyroidism, arthralgia, ulcerative colitis, reactive arthritis, insulin-dependent diabetes, juvenile idiopathic arthritis, agranulocytosis, idiopathic thrombocytopenic purpura, Crohn’s disease, erythema multiforme including Steven Johnson syndrome, myasthenia gravis, dermatopolymyositis, Steven Johnson syndrome, opticus neuritis, an/hypoestesia, Guillain Barre’ syndrome, multiple sclerosis, epilepsia, parestesia, Bell’s paresis, polyneuropathy, narcolepsy, ADEM, anaphylactic shock, allergic reaction, urticaria