Pandemic A/H1N1 2009 influenza vaccination, preceding infections and clinical findings in UK children with Guillain–Barré syndrome
- 1PIND Research Group, Addenbrooke's Hospital, Cambridge, UK
- 2General and Adolescent Paediatric Unit, Institute of Child Health, University College, London, UK
- 3Statistics, Modelling and Economics Department, Public Health England, Colindale, London, UK
- 4Immunisation, Hepatitis and Blood Safety Department, Public Health England, Colindale, London, UK
- Correspondence to Dr Christopher Verity, PIND Research Group, Box 267, Addenbrooke's Hospital, Hills Road, Cambridge CB2 0QQ, UK;
- Received 16 May 2013
- Revised 27 January 2014
- Accepted 28 January 2014
- Published Online First 28 February 2014
Objective To record clinical findings in all new cases of Guillain–Barré syndrome (GBS) or Fisher syndrome (FS) in UK children in the 2 years following September 2009 and determine the proportion temporally associated with recent infections, pandemic H1N1 (2009) strain influenza vaccination or seasonal influenza vaccination.
Design A prospective UK-wide epidemiological study using the British Paediatric Surveillance Unit system.
Patients Children aged 16 years or less meeting the Brighton Collaboration criteria for GBS or FS.
Results 112 children with GBS (66 boys and 46 girls) and 3 boys with FS were identified in 2 years. All but one recovered sufficiently to go home. The annual UK incidence rate of GBS in patients less than 15 years old was 0.45/100 000, similar to other countries. There was evidence of infection in the 3 months preceding onset in 92/112 GBS and 3/3 FS cases. Of those living in England, 7 cases received pandemic A/H1N1 2009 influenza vaccination before GBS symptom onset (3/7 were within 6 months including 1 within 3 months); 2 children received 2010/2011 seasonal influenza vaccination within 6 months of GBS onset. The numbers vaccinated were not significantly greater than expected by chance.
Conclusions The outcome for childhood GBS and FS after 6 months was better than reported in adults. Most UK GBS and FS cases had infections in the preceding 3 months. When considering the children living in England, there was no significantly increased risk of GBS after pandemic A/H1N1 2009 influenza vaccination or 2010/2011 seasonal influenza vaccination.
What is already known on this topic
Guillain–Barré syndrome (GBS) is an important cause of acute flaccid paralysis worldwide.
Influenza-like illnesses and other infections have been associated with an increased risk of GBS.
In 1976, national swine influenza immunisation in the USA was discontinued partly due to increased risk of GBS after vaccination.
What this study adds
There was no evidence of a significantly increased risk of GBS after pandemic A/H1N1 2009 influenza vaccination or 2010/2011 seasonal influenza vaccination.
There was evidence of a preceding infection in the majority of these children with GBS.
The prognosis for recovery after GBS in children is good.
In 1976, national immunisation against swine influenza in the USA was discontinued partly because of the increased risk of Guillain–Barré syndrome (GBS) after the vaccination.1 Surveillance for GBS was therefore commenced before pandemic H1N1 (swine) influenza vaccine was used in the UK. Data on preceding infections were collected as they have also been associated with an increased risk of GBS.2
In England, monovalent pandemic A/H1N1 2009 influenza vaccination containing the oil-in-water adjuvant AS03 (Pandemrix, a split virion, inactivated, adjuvanted vaccine) was introduced in October 2009, initially for people with high-risk clinical conditions then healthy children under 5 years from mid-December 2009.3–5 Between 1 October 2009 and 31 March 2010, H1N1 (swine) influenza vaccine was given to an estimated 855 378 children in England and it was also given to children in the rest of the UK.6 A second pandemic A/H1N1 2009 influenza vaccination was used (Celvapan, a whole virion, inactivated, unadjuvanted vaccine) but relatively few received this, so the data primarily reflect the uptake of Pandemrix. In 2010–2011, H1N1 was incorporated into seasonal influenza vaccine.
GBS is an important cause of acute flaccid paralysis (AFP) worldwide, and it is believed that immune stimulation has a central role in its pathogenesis.7 Fisher syndrome (FS) has been classified as a variant of GBS, so both conditions were included in our surveillance in children.8
We used the British Paediatric Surveillance Unit (BPSU) to contact every UK consultant paediatrician monthly. A separate study of GBS in adults was carried out via the British Neurological Surveillance Unit.9 The BPSU methodology has been described previously.10 ,11 We obtained permission to collect identifying information about reported cases so that the Health Protection Agency (HPA), now Public Health England (PHE), could obtain vaccination details from general practitioners (GPs). We used Hospital Episode Statistics (HES) data as an independent means of identifying cases in order to assess completeness of ascertainment.12 Between September 2009 and September 2011 inclusive, we sent questionnaires to paediatricians who reported cases via the BPSU asking about clinical details including the history of vaccinations and infections in the 3 months before the onset of GBS/FS and the results of investigations. We included those children aged 16 years or less who met the Brighton Collaboration criteria case definitions for GBS or FS—see online file.7 If children had been transferred between hospitals, data were obtained from both sites. The HPA sent questionnaires to GPs, obtaining information about pandemic A/H1N1 2009 influenza vaccines and seasonal influenza vaccines (type and batch number) administered to cases and infections preceding GBS/FS symptoms. Six months after initial notification, follow-up questionnaires went to paediatricians and the Hughes Disability Score was used to record outcome.13
The annual incidence in the UK was calculated assuming complete ascertainment and using Office of National Statistics 2010 population data.14 To calculate the expected number of cases vaccinated with pandemic A/H1N1 2009 influenza vaccine prior to onset, coverage data from approximately 100 GPs in England reporting to the Royal College of General Practitioners (RCGP) Network15 ,16 were used. RCGP data had coverage for age 2–17, so for children aged 6 months to 1 year on 1 September 2009 the 3-years-old coverage was used—coverage was quite similar for 2, 3 and 4 years olds. A case-coverage design17 was used in which each case resident in England was matched with the RCGP population coverage based on the case's age on 1 September 2009 and date of onset of symptoms. For example, a case aged 3 with onset on 1 March 2010 would be matched with all children of that age who had been vaccinated by that date—in this instance, a population coverage of 18%. After matching all cases with coverage data, the average matched coverage was calculated (eg, 11.3%) and this multiplied by the number of cases (the 96 who were resident in England) to obtain an expected number of vaccinated cases (eg, 10.8). To determine the expected number vaccinated within 3 and 6 months of onset, coverage at the dates 3 and 6 months prior to onset was calculated and subtracted from coverage at the date of onset. For example, age-matched coverage was 1% on 1 December 2009 and 0% on 1 September 2009 for the 3-year-old case with onset on 1 March 2010, so 17% of the matched population had been vaccinated in the 3 months before onset and 18% in the 6 months before onset. Observed numbers were compared for significance (5% level, two-sided) with the expected numbers by assuming that they would follow a Poisson distribution with a mean equal to the expected number. A similar calculation was used to compare the number vaccinated with seasonal vaccine within 6 months based on seasonal vaccine uptake data.
A brief summary of our first year results has been published.18 We now report on the full 2 years of surveillance with 6 months follow-up. In total, 203 children were notified via the BPSU, some were duplicates or reporting errors; 115 met the Brighton Collaboration criteria for GBS or FS and were included in the study. Details of these cases are given in table 1.
Annual incidence of GBS in the UK
In 2010, we identified 50 UK children less than 15 years old with GBS giving an annual incidence rate of 0.45/100 000 based on a UK population of children aged 0–14 years in 2010 of 10 871 000.14
Clinical features and results of investigations
Most GBS cases were too weak to walk (90/112) and most of those (80) received intravenous immunoglobulin; 13 needed ventilation (see table 2). Lumbar punctures and nerve conduction studies did not always show the features typical of GBS. Although not part of the Brighton Collaboration criteria, MRI scans were performed in 45 GBS cases. Fourteen GBS cases and one FS case were diagnosed solely on the basis of history and clinical examination with no supporting investigations.⇓
Infections in the 3 months preceding the onset of symptoms
In the majority of UK GBS/FS cases (92 GBS+3 FS/115), there was a history and/or laboratory evidence of infection in the 3 months before the onset of weakness. Laboratory results sometimes provided evidence of more than one past infection; the single infection in each patient for which there was most consistent evidence is shown in table 3. As shown in table 4, there was evidence of infection (based on history and/or laboratory confirmation) in the 3 months before GBS onset in all eight cases who received just pandemic A/H1N1 2009 influenza vaccination before GBS, in two of the four cases who received just seasonal vaccine before GBS and in both of the two cases who received both vaccines before GBS.
Pandemic A/H1N1 2009 influenza and seasonal influenza vaccinations in GBS and FS cases in the whole UK
Two children received pandemic A/H1N1 2009 influenza vaccination within 3 months of symptom onset (one child received two doses of Celvapan H1N1 unadjuvanted vaccination 5 weeks before onset, another received Pandemrix adjuvanted vaccine 7 weeks before onset) (see table 4). Two received Pandemrix 5 months before symptoms; six others received Pandemrix more than 6 months before symptoms. One child received 2009/2010 seasonal vaccine (no H1N1 antigen) 4 months before GBS onset, and two children received 2010/2011 seasonal vaccine (containing H1N1 antigen) 4 and 5 months before GBS onset.
None of the three received pandemic A/H1N1 2009 influenza vaccine or seasonal influenza vaccine before FS symptom onset.
Actual versus expected number of pandemic A/H1N1 2009 influenza and seasonal influenza vaccinations in GBS cases living in England
The average RCGP-matched coverage for pandemic A/H1N1 2009 influenza vaccine for the 96 GBS cases in England was 11.3% for ever vaccinated, 4.6% for within 6 months and 2.8% for within 3 months, which gave expected numbers vaccinated in our study of 10.8, 4.4 and 2.7, respectively. The observed numbers (in England, not the whole UK) were 7, 3 and 1, which is slightly lower (but not significantly; Poisson test p=0.35, 0.81 and 0.53, respectively) than expected numbers of pandemic A/H1N1 2009 influenza vaccinated children. Using similar data on age-specific seasonal influenza vaccine uptake, the expected number of cases vaccinated within 6 months of GBS onset in 2009/2010 was 0.8: one case was observed. The expected number in 2010/2011 was 0.6: two cases were observed. These observed numbers after seasonal vaccines were not significantly greater than expected by chance (p=0.55 and 0.12, respectively).
Outcome in the GBS and FS cases according to the Hughes Disability Score
One case who was still chair bound was transferred from hospital to a rehabilitation centre (see table 5).13 All the other 114 cases recovered sufficiently to go home. Also, 96 of the 108 GBS cases for whom data were available were at least able to walk unaided and 86 of these were normal or had minor symptoms. One child recovered from GBS but subsequently died of acute lymphoblastic leukaemia.
We compared the number of children in our study aged less than 15 years old living in England (n=84) with those identified by the HES database for England (see figure 1). Our data closely parallel HES data, suggesting that our case ascertainment was satisfactory.
Incidence of GBS
The BPSU AFP study (1991–1994) found that GBS was the commonest cause of AFP in the UK, but just 20 GBS cases were identified per year.19 National surveillance studies for AFP in other countries found that the commonest cause of AFP was GBS.20–24 In our study, the annual incidence rate of GBS in UK children less than 15 years old was 0.45/100 000. A systematic literature review found worldwide annual incidence of 0.34–1.34/100 000.25 In southern California in children less than 15 years old, the rate was 0.6 (95% CI 0.48 to 0.73)/100 0007;26 in seven Latin American countries the rate was 0.91/100 000 in children less than 15 years old and in Paraguay it was 1.1/100 000 children.27 ,28 In Australian children under 15 years of age, the annual AFP incidence was 0.8/100 000 and GBS cases made up 47% of these,22 giving a GBS rate very similar to ours.
Clinical features and test results
We have summarised the clinical features, investigation results, treatment and outcome of GBS and FS cases. These are unique data in children—we can find no similar population-based prospective series. Cerebrospinal fluid findings, nerve conduction studies and MRI scans of the spinal cord were normal in some children, but only 15 of the GBS/FS cases were diagnosed on the basis of clinical features alone.
In our study, history and/or laboratory results provided evidence of an infection in the 3 months before the onset of weakness in the majority of GBS/FS cases (92 GBS+3 FS/115). Previous studies have reported an association between infections and GBS.2 ,29 In six paediatric hospitals in Australia, neurological complications were reported in 9.7% of 506 children hospitalised with influenza A (H1N1) 2009 infection with paralysis/GBS in 0.4%.30 In France, a case–control study of a mainly adult population found that influenza infections and influenza-like illness are likely risk factors for GBS.31 In a systematic review, up to 70% of GBS cases were caused by antecedent infections.25 A retrospective survey in southern California found preceding infection in 71 (76%) of 93 children, and national surveillance in the USA found antecedent illnesses within 8 weeks of GBS onset in 67% of patients.26 ,32 Recently, a study of GBS cases identified via the Vaccine Safety Datalink in the USA found that after adjusting for antecedent infections there was no evidence for an elevated GBS risk following 2009–2010 monovalent or 2010–2011 trivalent inactivated influenza vaccines. However, there was a strong association between GBS and antecedent infection.33
Association with vaccination
In 1976, in association with national immunisation in the USA, the estimated attributable risk of vaccine-related GBS in adults was just under 1 case per 100 000 vaccinations. The increased risk was primarily within 5 weeks of vaccination.1 Subsequent studies, mainly in adults, using the United Kingdom General Practice Research Database found no association between influenza vaccination and GBS.2 ,34
This study found that the observed number of pandemic A/H1N1 2009 influenza-vaccinated GBS cases was lower than expected—no child with GBS living in England received Pandemrix in the 42-day risk period adopted in other studies of vaccination risk.35 A complementary study to ours in UK adults showed no evidence of an increased risk of GBS following the use of the H1N1 (2009) pandemic vaccine, and a study in Victoria, Australia, that was mainly in adults reached similar conclusions.9 ,36 A population-based cohort study of all age groups in Sweden found no change in GBS risk after vaccination against pandemic influenza with a monovalent adjuvanted vaccine (Pandemrix).37 Similar findings came from a study in five European countries, where mostly adjuvanted vaccines were used, and from a study in France.31 ,38 In contrast, in Quebec, where an AS03 adjuvanted vaccine was mostly used, the 2009 influenza A (H1N1) pandemic vaccine was associated with a small but significant risk of GBS, but the excess risk was in those over 50 years old.39
In other parts of the world, unadjuvanted vaccines were used. Population-based surveillance in the USA showed 0.74 (95% CI 0.04 to 1.56) excess of GBS cases per million influenza A (H1N1) 2009 vaccine doses; however, this was due to an excess of cases in those older than 25 years. The risk was much smaller than during the 1976 campaign and comparable to some previous seasonal vaccine risks.40 Similar findings came from the meta-analysis of data from six adverse monitoring systems in the USA influenza A (H1N1) 2009 monovalent-inactivated vaccines that were associated with a small increased risk of GBS translating to about 1.6 excess of cases of GBS per million people vaccinated. The excess (attributable) cases per million varied according to age group, being 1.05 <18 years, 2.07 18–64 years and 3.09 in those 65 years or older.35 There was no increase in the GBS risk after 89.6 million doses of monovalent vaccine to Chinese of all ages.41
We found a better outcome in children than reported in previous studies of mainly adult populations. Study of a national hospital discharge database in the USA showed that the case–fatality ratio varied from 0.7% among persons < 15 years old to 8.8% among persons > 65.42 In 1995/1996 in Italy, 7 (5.8%) out of 120 patients died within 30 days of illness onset and a worse 2-year outcome was found in those aged 50 years or more.43 A study in southeast England found that 1 year after recruiting 79 GBS cases, 6 (8%) had died, all over 60 years old.44 In the Netherlands, the mortality rate was 3.9%, age being an important risk factor for death.45
Strengths and limitations
The study has important strengths. It used a well-established prospective national surveillance system (the BPSU), there was an independent means of identifying cases (the HES data) demonstrating good ascertainment of cases, detailed clinical information was available from hospital records via questionnaires and the vaccination histories were obtained directly from GPs who administer the vaccines, so there was no recall bias.
There are some limitations to our study. We may have missed children with GBS or FS with mild illness. Results of some investigations were not available, and we only saw scan reports rather than seeing the actual scans (however, this does not affect our conclusions as the scan findings are not used as part of the Brighton Classification). We obtained a history and/or laboratory evidence of an infection via the questionnaire completed by the paediatricians who reported cases. This included specific questions about infections in the 3 months before the onset of GBS/FS symptoms and asked for the results of laboratory investigations. In some cases, there was evidence of more than one infection and we had to decide which infection was the likely one, based on all the evidence. In the 3 months preceding the onset of GBS, there was evidence of infection in all but 19 of the 115 GBS/FS cases; however, we cannot demonstrate that this was a statistically increased number of infections because we have no control group without GBS. There was evidence of infection in the 3 months preceding symptoms in all but 2 of the 14 children receiving pandemic A/H1N1 2009 influenza and/or seasonal influenza vaccine before GBS onset. It is not clear that this would be a confounding factor, but if it was then it would reduce any vaccine-associated risk. Although it is also a potential confounder, we were not able to determine how many GBS cases had a preceding H1N1 influenza infection because there is evidence that many cases in the pandemic were subclinical—a cross-sectional serological study in England showed that around one child in three was infected with 2009 pandemic H1N1 in the first wave of infection in regions of high incidence, 10 times more than were suspected from clinical surveillance.46
Outcome after GBS in children was better than has been reported in adults. There was no evidence for an increased risk of GBS/FS after pandemic A/H1N1 2009 influenza vaccination or after 2010/2011 seasonal influenza vaccine containing H1N1 antigen. In contrast, there was evidence of a preceding infection in the 3 months preceding the onset of symptoms in 96 of the 115 GBS/FS cases.
Many thanks to reporting paediatricians, Mr Richard Lynn, BPSU Scientific Co-ordinator, and Professor Douglas Fleming, Royal College of General Practitioners, for supplying RCGP data for vaccine coverage.
Contributors All the authors were involved in planning the study and writing the paper. LS, AMW and CV obtained ethics and other consents. AMW and LS sent questionnaires to paediatricians and recorded their data. JS sent questionnaires to general practitioners. NA performed the statistical analyses.
Funding This is independent research commissioned and funded by the Department of Health Policy Research Programme (Ref. no.: 0190047).
Competing interests None.
Ethics approval Trent Research Ethics Committee (Ref. no.: 09/H0405/45) and the National Information Governance Board (Ref. no.: ECC/BPSU 5-02 (FT1). Public Health England has approval for England from the National Information Governance Board for Health and Social Care (NIGB) (PIAG ref: PIAG 03-(c)/2001), which allows staff access to patient-identifiable information for vaccine safety monitoring purposes.
Provenance and peer review Not commissioned; externally peer reviewed.