key: cord-0878247-1o5ow4vn
authors: Keh, Ryan Y. S.; Scanlon, Sophie; Datta-Nemdharry, Preeti; Donegan, Katherine; Cavanagh, Sally; Foster, Mark; Skelland, David; Palmer, James; Machado, Pedro M.; Keddie, Stephen; Carr, Aisling S.; Lunn, Michael P.
title: COVID-19 vaccination and Guillain-Barré syndrome: analyses using the National Immunoglobulin Database
date: 2022-02-18
journal: Brain
DOI: 10.1093/brain/awac067
sha: 76c3fae2e7c884fcabe37c55ad329b6aa4376816
doc_id: 878247
cord_uid: 1o5ow4vn

Vaccination against viruses has rarely been associated with Guillain-Barré syndrome (GBS). An association with the COVID-19 vaccine is unknown. We performed a population-based study of National Health Service data in England and a multicentre surveillance study from UK hospitals, to investigate the relationship between COVID-19 vaccination and GBS. Firstly, case dates of GBS identified retrospectively in the National Immunoglobulin Database from 8 December 2021 to 8 July 2021 were linked to receipt dates of a COVID-19 vaccines using data from the National Immunisation Management System in England. For the linked dataset, GBS cases temporally associated with vaccination within a 6-week risk window of any COVID-19 vaccine were identified. Secondly, we prospectively collected incident UK-wide (four nations) GBS cases from 1 January 2021 to 7 November 2021 in a separate UK multicentre surveillance database. For this multicentre UK-wide surveillance dataset, we explored phenotypes of reported GBS cases to identify features of COVID-19 vaccine-associated GBS. 996 GBS cases were recorded in the National Immunoglobulin Database from January to October 2021. A spike of GBS cases above the 2016-2020 average occurred in March-April 2021. 198 GBS cases occurred within 6 weeks of the first-dose COVID-19 vaccination in England (0.618 cases per 100,000 vaccinations, 176 ChAdOx1 nCoV-19 (AstraZeneca), 21 tozinameran (Pfizer), 1 mRNA-1273 (Moderna)). The 6-week excess of GBS (compared to the baseline rate of GBS cases 6-12 weeks after vaccination) occurs with a peak at 24 days post-vaccination; first-doses of ChAdOx1 nCoV-19 accounted for the excess. No excess was seen for second-dose vaccination. The absolute number of excess GBS cases from January-July 2021 was between 98-140 cases for first-dose ChAdOx1 nCoV-19 vaccination. First-dose tozinameran and second-dose of any vaccination showed no excess GBS risk. Detailed clinical data from 121 GBS patients were reported in the separate multicentre surveillance dataset during this timeframe. No phenotypic or demographic differences identified between vaccine-associated and non-vaccinated GBS cases occurring in the same timeframe. Analysis of the linked NID/NIMS dataset suggests that first-dose ChAdOx1 nCoV-19 vaccination is associated with an excess GBS risk of 0.576 (95%CI 0.481-0.691) cases per 100,000 doses. However, examination of a multicentre surveillance dataset suggests that no specific clinical features, including facial weakness, are associated with vaccination-related GBS compared to non-vaccinated cases. The pathogenic cause of the ChAdOx1 nCoV-19 specific first dose link warrants further study.

A C C E P T E D M A N U S C R I P T 121 GBS patients were reported in the separate multicentre surveillance dataset during this timeframe. 23

No phenotypic or demographic differences identified between vaccine-associated and non-vaccinated 24 GBS cases occurring in the same timeframe. 25

Analysis of the linked NID/NIMS dataset suggests that first-dose ChAdOx1 nCoV-19 vaccination is 26 associated with an excess GBS risk of 0.576 (95%CI 0.481-0.691) cases per 100,000 doses. However, 27 examination of a multicentre surveillance dataset suggests that no specific clinical features, including 28 facial weakness, are associated with vaccination-related GBS compared to non-vaccinated cases. The 29 pathogenic cause of the ChAdOx1 nCoV-19 specific first dose link warrants further study. 30

The first year of the COVID-19 pandemic produced robust information on the neurological and 2 neuropsychiatric sequelae of SARS-CoV-2 infection. 1 In the peripheral nervous system, brachial neuritis, 3 facial palsy and Guillain-Barré syndrome (GBS) were subjects of particular interest. 4

From January 2020, case reports and case series of patients with GBS occurring around the time of 5 SARS-CoV-2 infection raised the possibility of a link between GBS and COVID-19. 2,3 However, a large 6 national study of the UK population as well as in Singapore showed a decreased incidence of GBS during 7 the pandemic and failed to find a definitive link between GBS and COVID-19 infection. 4,5 8 GBS became an adverse event of special interest (AESI) related to vaccination in the 1970s when an 9 excess of GBS cases was detected in the United States during the 1976/1977 A/New Jersey/76 influenza 10 ('swine flu') vaccination campaign within 6 weeks of vaccination. 6 Serial epidemiological analyses 11 established the rate of GBS attributable to the 'swine flu' vaccine was approximately 4.9-5.9 per million 12 vaccines, mostly from 14-28 days post-vaccination. 6 Although GBS has been identified in subsequent 13 annual surveillance of influenza vaccination programmes at a rate of 1 to 1.6 per million doses, 7,8 a 14 pathogenic explanation has not been found. Expert consensus largely derived from the vaccination 15 surveillance is that GBS risk attributable to vaccination exists for up to 6 weeks (42 days). 7 

The global rollout of COVID-19 vaccines triggered extensive monitoring, with GBS as an AESI. Very rare 17 adverse events, not visible even in large clinical trials, can be identified when mass vaccination 18 monitoring systems are in place, and particularly when a unique disease phenotype emerges. Vaccine -19 induced immune thrombotic thrombocytopenia (VIIT) 9 and more recently myocarditis 10 were identified 20 this way, and are unique diseases. VIIT occurs most commonly in association with ChAdOx1 nCoV -19 21 (AstraZeneca) recombinant adenoviral vector vaccine. VIIT manifests as thromboses, including cerebral 22 venous sinus thrombosis, with an estimated incidence of 1 case per 100,000 exposures. 9 Myocarditis 23 seems specific to the tozinameran (Pfizer) and mRNA-1273 (Moderna) vaccines. 10 I P T  simian adenovirus vector designed to evade anti-human adenovirus neutralising antibodies to stimulate  1   a robust immune response.  2   The UK COVID-19 vaccination programme began on the 8 th December 2020 with tozinameran, then  3 ChAdOx1 nCoV-19 in January and subsequently mRNA-1273 vaccinations. Vaccination was delivered 4 sequentially to cohorts of the most vulnerable and elderly followed by deciles of age. 50% of adults over 5 50 years of age had had their first vaccination by mid-February 2021. 6

We aimed to combine multiple national data sources and systematically investigate any temporal 7 relationship between COVID-19 vaccination and excess cases of GBS during the UK COVID-19 vaccination 8 programme. We retrospectively interrogated a large database of patients hospitalised with GBS in 9

England, Scotland and Northern Ireland treated with immunoglobulin from the National 10 Immunoglobulin Database (NID). Using the common NHS identifier, we combined the English data with 11 data from the National Immunisation Management System (NIMS) on all COVID-19 vaccinations data in 12

England. Separately, we characterised a large surveillance dataset of the incident UK GBS cases, 13 presenting both after COVID-19 vaccination, and also without vaccination during the same period, 14 recording the timing of onset after COVID-19 vaccination. 15 

We extracted NID GBS cases from 1 January to 31 October 2021 and recorded diagnosis, their unique 1 identifier and date of first immunoglobulin prescription. These numbers were compared to the historical 2 GBS cases recorded in the NID from 2016 to 2020. 3 UK Department of Health and Social Care (DHSC) guidance for GBS treatment recommends intravenous 4 immunoglobulin (IVIg) as first line therapy for patients with Hughes Grade 4 or more (significant 5 disability), disease progressing towards intubation or ventilation or with high probability of respiratory 6 insufficiency (mEGRIS score ≥3) or predicted poor prognosis (mEGOS ≥4). 13 Although plasma exchange 7 (PLEX) is also considered a first line option, in practice, this is not readily available and is very seldom 8 used. Utilising IVIg use as a proxy for GBS incidence under-estimates the true incidence of GBS, as milder 9 cases are not treated. However, IVIg is estimated to be given to 86% of European and UK GBS cases, 14 10 and the 2021 data can be compared to previous years with similar clinical behaviours for admitted 11 patients. 12

The National Immunisation Management System (NIMS) database is a national point of care system for 13 capturing vaccination data from England. Patients in the NIMS database are also registered with their 14 NHS number. The COVID-19 vaccination data were linked to the English cases of GBS identified from the 15 NID from 8 December 2020 to 8 July 2021 using the unique NHS identifier by NHS England. included. The exposed population, which was used to calculate GBS rates post-vaccination, was taken 23 from weekly published cumulative counts of vaccine usage in England up to this date. 24

Prospective surveillance study 25 We conducted a prospective surveillance study to compare the demographic and phenotypic 26 characteristics of GBS cases reported from 1 January 2021 to 7 November 2021, comparing GBS cases 27 reported as having received COVID-19 vaccination and cases without vaccination. We recognise the 28 included cases are heavily influenced by reporting bias as there was significant interest in the possibility 29

of vaccination-related GBS at the time. The surveillance study should not be used in direct comparison 1 to the retrospective analyses of linked data described above. 2

Reports of GBS were submitted by members of the British Peripheral Nerve Society (BPNS) and the 3 Association of British Neurologists (ABN), with regular reminders to collect information on hospital 4 presentations of GBS during the study period. Data were entered into the International Neuromuscular 5 COVID-19 database (https://www.ucl.ac.uk/centre-for-neuromuscular-6 diseases/news/2020/may/international-neuromuscular-covid-19-database), at the Centre for 7 Neuromuscular Disease, where database questions had been adapted to allow for reporting of COVID-8 19 vaccine-related neuromuscular cases. Surveillance study data collection ended on 7 November 2021 9 to allow time for retrospective case reporting. Anonymised clinical data of demographics, GBS diagnostic 10 criteria, vaccine details, and prior COVID-19 infection, symptoms and management were collected. 11

Cases were classified according to Brighton Collaboration GBS Working Group criteria 15 by the study 12 team to describe level of diagnostic certainty were recorded as previously. 4 

Two cases were excluded from analysis as the reporting clinicians later informed us of a change in 14 diagnosis. 15

Statistical analysis was performed using STATA 16 16 An ecological analysis presenting the number of GBS cases identified in the NID by calendar month was 18 used to compare the incidence of GBS in the UK across the years 2016-2020 with that seen in January-19

July 2021. 20

The Spearman's rank correlation test was used to test for correlation between the time after the first 21 vaccine dose and the incidence of GBS. 22

One-way ANOVA was used to compare yearly GBS rates within age deciles of the linked NID/NIMS data 23 in 2021 compared to 2019 and 2020, to determine if age distribution of GBS differed compared to 24 previous years, using ONS population estimates as the denominators for calculation of rates. As GBS 25 numbers were only available until July 2021, an estimate of annual numbers was produced on the 26 assumption of a stable GBS incidence through the year to enable comparison to prior years. ONS 27 population estimates for 2021 were assumed to be the same as 2020. 28

For the prospective surveillance study, Chi-square and Kruskal-Wallis tests were used to test for 1 correlations between patient demographics, GBS characteristics or treatment details, and exposure to 2 COVID-19 vaccination within 6 weeks of GBS onset. 3 A significance value of p<0.05 was used throughout. 4

The UK Health Research Authority was consulted and advised that the study did not require review by 6 an NHS Research Ethics Committee, as this was an analysis of previously collected, non-identifiable 7 anonymised data. 8

9 Data are available on request to the corresponding author. 10

Retrospective analyses of NID/NIMS datasets 12 Ecological analysis -GBS cases recorded in NID from 2016 to 2020 13 Between 2016 and 2020, the NID recorded a mean of 1283 immunoglobulin-treated cases of GBS per 14

year (95%CI 1159-1408 , mean 107 cases per month in the three participating UK nations). In England, a 15 mean of 1148 GBS cases occurred annually between 2016 and 2020 (95%CI 1022-1274, or 3.1 GBS cases 16 per day). These annualised case counts represent the vast majority of hospitalised GBS cases in England, 17

Scotland and Northern Ireland, resulting in an estimated GBS incidence rate of 1.99 per 100,000 18

individuals per year (95%CI 1.79-2.18). This is comparable to previous European and North American 19 studies with incidence rates of between 0.84-1.91 per 100,000 individuals per year. 18 As previously 20 described, the UK experienced an overall reduction in cases in 2020 during the height of the COVID-19 21 pandemic, resulting in the lowest case number (1053 cases in 2020) and estimated incidence (1.57 cases 22

per 100,000 individuals per year) in the five-year period. 4 23

Ireland from the years of 2016 to 2020, and compares these to the monthly incident case numbers from 2 January to October 2021 during the UK vaccination programme. A total of 996 GBS cases were recorded 3 from January to October 2021, fewer than observed from January-October 2016 to 2019 (pre-pandemic 4 range: 1054-1182 cases) but higher than for January-October 2020 (pandemic range: 856 cases). 5

The number of GBS cases in January 2021 was significantly lower than the mean number seen in the 6 same month 2016-2020, continuing the trend of lower GBS rates from 2020. However, England, 7

Scotland and Northern Ireland experienced a spike of GBS cases in March and April 2021, before rates 8 fell again below the lower range of the 95% CI for the 2016-2020 mean incident GBS case number in July 9

to October 2021. 10

Annual GBS cases in England incident from 2019, 2020 and January to July 2021 were stratified by age 12 decile, enabling age-specific incidence rate to be compared to recent years (Table 2) . To enable a 13 comparison across years, an estimate of annual case numbers for 2021 was produced based on 14 assumption of minimal seasonal variation. An excess of cases in the 50-59 and 60-69 age groups was 15 identified compared to 2019 and 2020, with statistically significant differences between age groups 16 (p=0.00033). 17

Using the linked NID/NIMS data, the first record of GBS occurring within 6 weeks after a COVID-19 18 vaccination was in January 2021. Of note, not all GBS patients in NID had a vaccination record. 198 cases 19 of GBS identified in the linked NID/NIMS data study period occurred within 6 weeks of the first dose of 20 any COVID-19 vaccine (0.618 cases per 100,000 vaccinations in 6 weeks, all ages). A total of 32.1 million 21 first dose vaccinations were recorded during the reporting period (20.3 million ChAdOx1 nCoV-19, 11.5 22 million tozinameran and 0.3 million mRNA-1273). Of the 198 linked GBS cases, 176 followed a first dose 23 ChAdOx1 nCoV-19 vaccine (rate 0.868 per 100,000) and 21 followed a first dose tozinameran vaccine 24 (rate 0.183 per 100,000). Only one case was reported within 6 weeks of mRNA-1273vaccination. Only 23 25 GBS cases were reported within 6 weeks of any second vaccine dose. 26 Table 1 and Fig. 2 summarise patient characteristics of GBS cases in England occurring within 6 weeks of 27 first COVID-19 vaccination. The GBS incidence after first vaccination was highest in males receiving the 28 ChAdOx1 nCoV-19vaccine (1.069 per 100,000 doses). 29

The daily number of incident GBS cases, with an 84-day post-vaccine follow-up from dose 1 and 2 of 1 COVID-19 vaccination was plotted (Fig. 3) . A peak of GBS cases is observed around 24 days following a 2 first dose, with higher numbers of cases seen in the period of 2 to 4 weeks after vaccination than in 3 other periods. First doses of ChAdOx1 nCoV-19vaccine account for the majority of this increase. A 4 similar pattern is not seen following a second dose of any vaccine. Using the Spearman's rank 5 correlation test of randomness, the occurrence of GBS was random for all times after the first-dose 6 tozinameran vaccine (p=0.84) (no peak associated with tozinameran vaccine) and after the second-dose 7 vaccination of all vaccines (p=0.85). However, it was non-random for 'all first-dose' vaccination 8 (p=0.009) and the first dose of the ChAdOx1 nCoV-19 vaccination (p=0.004). 9

Using case numbers from day 43-84 after first-dose vaccination as a comparison group (assuming this Table 3 . 29 We decided to use GBS case numbers from day 43-84 after vaccination as a comparator group, rather 1 than an externally-derived control based on historical GBS numbers, as GBS case numbers during the 2 COVID-19 pandemic may be different from historical baselines. 4 GBS incidence in this group was 3 estimated at 1.9 cases per day, lower than the historical 2016-2020 average of 3.1 cases per day based 4 on NID data in England, supporting the hypothesis that the baseline GBS case numbers in 2021 continue 5 to be lower than pre-pandemic levels. 6

The total number of cases of GBS in the NID from January to October 2021 is lower than the 10-month 7 total of January to October 2016 to 2019; but is higher than total case numbers during 2020. A monthly 8 increase in GBS cases in March and April 2021 was notable, but total numbers fell back into the normal 9 range thereafter and thus this 'spike' is the only hint of a causative link in simple occurrence data. If all COVID-19 vaccines were associated with an increased risk of GBS across all age groups, a more 21 sustained rise in GBS incidence would be expected. 9,646,715 people, mostly the older and more 22 vulnerable, were vaccinated from early December 2020 to 1 st February, but the increase in GBS cases 23 was not seen until mid-February. 24 Interestingly, there is no recognisable increase in GBS after the second dose which may be a 29 pathophysiological phenomenon, individual susceptibility, or because patients experiencing adverse 30

effects from the first dose did not take the second vaccine dose. We know of a single patient who was 1 reported to have GBS-like illness following both first-and second-dose of the ChAdOx1 nCov-19 vaccine. 2

The patient initially developed a facial diplegia and paraesthesia phenotype with subsequent gait 3 disturbance and elevated CSF protein, and improved with IVIg. Two months later, two weeks after their 4 second dose of ChAdOx1 nCoV-19 vaccination, they developed increasing weakness with neuropathic 5 pain, with elevated CSF protein, demyelinating changes on nerve conduction studies and MRI 6 enhancement of the cauda equina, with only partial response to IVIg treatment. 7

The single patient in our surveillance study who had recurring neuropathic symptoms after second-dose 8 vaccination appears to be a unique report in the UK at this time, and may potentially represent acute-9 onset chronic inflammatory demyelinating polyradiculoneuropathy (A-CIDP) rather than true GBS. A 10 recent study from Israel of COVID-19 vaccinations in patients with previous GBS described a single 11 patient with recurrent GBS-like illness time-linked to both doses of their tozinameran vaccination, but 12 limited information is available regarding the strength of GBS diagnosis. 21 

A small excess of GBS in males is seen in this vaccine-associated GBS cohort; as is documented in the 14 GBS literature, 15 The reason for the association between only ChAdOx1 nCoV-19 vaccination and GBS is unclear. COVID-21 19 infection does not have a strong, or possibly any, increased risk of GBS, 4,5 and the lack of increased 22 risk associated with tozinameran vaccination implies that it is unlikely that the COVID-19 spike protein is 23 the causative factor for the increased risk. The excess incidence is estimated to be 5.8 cases per million 24 doses, similar to the estimates for the 1976 'swine flu' vaccine and higher (but within the same order of 25 magnitude) as the reported excess cases for the modern influenza and yellow fever vaccines. It is far 26 below the 1:1000 cases of GBS in C. jejuni gastroenteritis or Zika-virus. A non-specific immune activation 27 in susceptible individuals might therefore be implicated, but if that were the case similar risks might 28 apply to all vaccine types. It is therefore logical to suggest the simian adenovirus vector may account for 29 the increased risk. Adenoviruses have not been strongly associated with GBS in previous studies, 26 and 30 any association between adenoviral vaccination and GBS 11 has only once been reported. Nevertheless, 31

adenovirus testing is not routinely performed in cases of GBS in the UK, and whether they may account 1 for a proportion of 'idiopathic' or 'serology negative' GBS may be the subject for further study. 2

Although the first retrospective analysis presented here employs two cross-referenced national and 3 mandated datasets, there are still confounders, bias and criticisms as applicable to many 4 epidemiological studies. We were unable to individually validate cases of immunoglobulin-treated GBS. 5

However, these patients would need to be unwell enough to be admitted to hospital, and the diagnosis 6 was assessed not only by the admitting physicians but also by an Immunoglobulin Assessment Panel 7 who authorise the treatment. Furthermore, the data are controlled against data from earlier years 8

where GBS rates are consistent with other international, methodologically robust epidemiological 9 studies with high levels of ascertainment and clinical validity. We recognise that patients with 'mild GBS' 10 may not attend hospital or be treated, but this has been the same in past years. We also recognise that 11 patients may have been more reluctant to attend hospital, although this was not obviously observed in 12 2021, and paralysed patients are likely to have attended more than those with lesser disability. 13 We explored detailed phenotypes by collecting data on GBS presentations to UK neurologists with a 14 continuation of the BPNS/ABN surveillance study. 121 cases were reported, representing only 13% of 15 GBS cases in that period of 2021. Clinicians were much more likely to report cases with recent 16 vaccination compared to those who were unvaccinated. Nonetheless, this dataset allowed for deeper 17 confirmation of GBS diagnoses, with 79% of the cohort meeting Brighton Collaboration diagnostic 18 criteria level 1, 2, or 3 (compared to 15% of the cases reported to the MHRA 'Yellow Card' system 19 [personal communication]). Within our dataset, there were no differences identified between patients 20 who had recent vaccination and those who had not, in terms of baseline demographics, disease course 21 or treatment. Although BPNS and ABN members reported cases of facial weakness associated with GBS, 22 there was no increase linked to vaccination status. The description of some of these cases best fit the 23 'atypical' GBS presentation of 'bilateral facial weakness with paraesthesia', which is felt to represent 24 <5% of total GBS cases. 27 25

Another recently published UK-based study has analysed neurological complications in the context of 26 recent COVID-19 vaccination and infection, and has reported comparable excess in GBS cases of 3.8 per 27 million doses ChAdOx1 nCoV-19 vaccine. The incidence rate ratio (IRR) of 2.90 at 15-21 days post-28 vaccination suggests a similarly plausible time-locked association to that which we observed, in keeping 29 with the pathological mechanism of GBS. 28 However, they also report a possible increase in GBS cases in 30 relation to COVID-19 infection of 14.5 per million COVID-19 infections, with an IRR of 5.25. 28 We note 31

that neurological manifestations were identified using hospital coding data, which may be less accurate 1 than NID identification of Immunoglobulin Panel-scrutinised GBS cases. Moreover, mortality rate was 2 1.8% in that cohort, lower than the 3-10% mortality generally associated with GBS 27 . In addition, 16% of 3 COVID-19-associated GBS in their cohort (7/43 cases) were diagnosed on the same day as COVID-19 4 diagnosis, which makes it less likely that GBS was caused by a post-infectious immune process triggered 5 by COVID-19 infection. For these and other reasons, we believe that our study's findings on excess risk 6 may provide a more accurate estimate of risk. 7

Our study reports an association between first-dose ChAdOx1 nCoV-19COVID-19 vaccination and GBS, 8 accounting for an estimated excess incidence of 5.8 GBS cases per million first doses. The cause for this 9 association remains unclear, and excess risk remains comparable to previous vaccine-associated GBS. 10 The risk in proportion to the benefits of vaccination is very small. Further studies are required to confirm 11 these observations, to determine causality, to explore the pathogenic mechanisms and to investigate 12 effects of other COVID-19 vaccine preparations in use elsewhere in the world. 13 Crosses represent upper and lower limits of 95% confidence intervals. 10 

Pericarditis After Vaccination for COVID-19

Adverse events following adenovirus 4

type 4 and type 7 vaccine, live, oral in the Vaccine Adverse Event Reporting System

United States

content/uploads/ImmunoglobulinDatabaseAnnualReport201920.pdf. Accessed

Updated Commissioning Guidance for the use of therapeutic immunoglobulin

immunology, haematology, neurology and infectious diseases in England

Commissioning-Guidance-CQUIN-1920

Regional variation of Guillain-Barre´ syndrome

Guillain-Barré syndrome and Fisher syndrome: case definitions 17 and guidelines for collection, analysis, and presentation of immunization safety data

StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC. 20 17. R Core Team R: A language and environment for statistical computing. R Foundation for

Statistical Computing