Neurological complications after first dose of COVID-19 vaccines and SARS-CoV-2 infection

The study addresses concerns about rare neurological adverse events following COVID-19 vaccination and SARS-CoV-2 infection. Although randomized trials demonstrated vaccine effectiveness, they were underpowered to detect very rare events. Reports of cerebral venous sinus thrombosis and neuroinflammatory events (including transverse myelitis) after ChAdOx1nCoV-19, as well as case reports of Guillain-Barré syndrome (GBS) and surveillance signals of neurological events after SARS-CoV-2 infection, motivated a comprehensive assessment. The objective was to quantify risks of hospital admission or death from specific neurological outcomes in predefined periods after first-dose vaccination with ChAdOx1nCoV-19 or BNT162b2, and after a SARS-CoV-2-positive test, using linked national datasets in England, and to replicate key findings in an independent Scottish cohort.

Prior studies demonstrated effectiveness of ChAdOx1nCoV-19 and BNT162b2 in reducing COVID-19 infections, hospitalizations and deaths. However, rare neurological adverse events have been reported post-vaccination, including cerebral venous sinus thrombosis and transverse myelitis with ChAdOx1nCoV-19, and case reports of Guillain-Barré syndrome and Bell’s palsy after COVID-19 vaccines. Surveillance studies also suggested associations between SARS-CoV-2 infection and neurological conditions including GBS and myelitis. Existing evidence was limited by small numbers and potential selection/recording biases, underscoring the need for robust population-based analyses.

Design: Self-controlled case series (SCCS) including only individuals (aged ≥16) who experienced one of the neurological outcomes and who received at least one vaccine dose between 1 December 2020 and 31 May 2021 in England. Outcomes were first hospital admission or death due to: acute CNS demyelinating events; encephalitis, meningitis and myelitis (excluding encephalopathy); Guillain-Barré syndrome; Bell’s palsy; myasthenic disorders; hemorrhagic stroke; and subarachnoid hemorrhage, defined by ICD-10 codes. Data sources and linkage: The National Immunisation (NIMS) Database provided vaccination records; SARS-CoV-2 PCR test results came from Public Health England; hospitalizations from Hospital Episode Statistics (HES); and deaths from the Office for National Statistics (ONS). Individual-level deterministic linkage combined these datasets. Exposures: First dose of ChAdOx1nCoV-19 or BNT162b2, and first SARS-CoV-2 infection (PCR positive) during the study period. The Moderna mRNA-1273 vaccine was excluded due to small numbers. Risk windows: Post-exposure intervals were day 0, 1–7, 8–14, 15–21 and 22–28 days. A pre-risk period (−28 to −1 days before exposure) accounted for potential postponement of vaccination/testing due to prodromal illness. Baseline periods were from study start to 29 days before exposure and from 29 days after exposure to study end/censoring. Day 0 was separated to mitigate reverse-causation bias from testing at admission. Statistical analysis: Conditional Poisson regression with offset for risk-period length estimated incidence rate ratios (IRRs) and 95% CIs for each risk period relative to baseline, adjusting for calendar week to account for seasonality and pandemic-related healthcare pressures. Exposure terms for both vaccines and infection were included in the same models. Absolute effects were estimated as excess events per 10 million exposed using established methods compatible with SCCS IRRs. Subgroup interaction analyses by age (≤50 vs >50 years) and sex were conducted using likelihood ratio tests. Sensitivity analyses: Tested robustness to censoring on death, exclusion of second doses or restricting to single-dose follow-up, truncating study period to address recording delays, extending the risk window to 35 days, restricting infection analyses to those infected before vaccination, and restricting vaccine analyses to those without prior infection. Replication: An independent SCCS analysis in Scotland (linked vaccination, hospitalization, mortality, and testing datasets) replicated key associations using the same definitions and modeling approach, recognizing smaller population size and lower infection rates.

Population: 32,552,534 individuals received a first vaccine dose in England (ChAdOx1nCoV-19 n=20,417,752; BNT162b2 n=12,134,782). Among the vaccinated, 2,005,280 (~6%) had a SARS-CoV-2-positive test (≈91% before vaccination). ChAdOx1nCoV-19: - Guillain-Barré syndrome (GBS): Increased risk at 15–21 days (IRR 2.90; 95% CI 2.15–3.92) and 22–28 days (IRR 2.21; 1.59–3.09); for 1–28 days overall IRR 2.04 (1.60–2.60). Estimated 38 excess GBS cases per 10 million vaccinated. - Bell’s palsy: Increased at 15–21 days (IRR 1.29; 1.08–1.56) but no overall association in 1–28 days (IRR 1.07; 0.94–1.21). - Myasthenic disorders: Increased at 15–21 days (IRR 1.57; 1.07–2.30); no overall 1–28 day association (IRR 1.23; 0.94–1.62). - Hemorrhagic stroke and subarachnoid hemorrhage: No associations across specified periods; 1–28 day IRRs 1.02 (0.90–1.15) and 1.01 (0.86–1.18), respectively. - Acute CNS demyelinating events: No association (1–28 days IRR 0.97; 0.78–1.22). BNT162b2: - Hemorrhagic stroke: Increased at 1–7 days (IRR 1.27; 1.02–1.59) and 15–21 days (IRR 1.38; 1.12–1.71); overall 1–28 days IRR 1.24 (1.07–1.43). Estimated 60 excess hemorrhagic strokes per 10 million vaccinated. - No significant associations for GBS (1–28 days IRR 0.86; 0.54–1.36), Bell’s palsy (1.06; 0.90–1.26), myasthenic disorders (1.18; 0.88–1.59), acute demyelinating events (1.02; 0.75–1.40), or subarachnoid hemorrhage (1.05; 0.84–1.30) in 1–28 days. SARS-CoV-2 infection: - Markedly elevated risks across outcomes with highest IRRs on day 0 (likely reflecting testing at admission). For 1–28 days overall: GBS IRR 5.25 (3.00–9.18), encephalitis/meningitis/myelitis IRR 2.70 (1.78–4.11), myasthenic disorders IRR 3.01 (1.70–5.36). Bell’s palsy 1–28 days IRR 1.34 (0.91–1.97) despite high day-0 and early-period IRRs; hemorrhagic stroke 0.85 (0.57–1.26); subarachnoid hemorrhage 1.51 (0.96–2.36); acute CNS demyelinating events 1.67 (0.93–3.00). - Estimated excess events per 10 million in 1–28 days after positive test: encephalitis/meningitis/myelitis 123; GBS 145; (discussion also cites myasthenic disorders ~163). Composite outcome (GBS + Bell’s palsy): Increased co-occurrence after ChAdOx1nCoV-19 at 15–21 days (IRR 28.86; 5.45–152.74) and 22–28 days (IRR 13.35; 2.48–72.92); 1–28 days IRR 12.66 (2.70–59.21). Estimates for infection not reported due to small numbers. Subgroups: BNT162b2-associated hemorrhagic stroke risk higher in females (IRR 1.44 at 1–7 days; 1.84 at 15–21 days) than males (1.13; 0.98). ChAdOx1nCoV-19-associated myasthenic disorder risk elevated in ≤50 years at 15–21 days (IRR 2.44; 1.07–5.55). Replication (Scotland): Association between ChAdOx1nCoV-19 and GBS replicated (1–28 days IRR 2.31; 1.02–5.24), with similar timing; BNT162b2–hemorrhagic stroke association not observed (IRR 0.65; 0.35–1.20).

The findings indicate that first-dose ChAdOx1nCoV-19 is associated with increased short-term risks of Guillain-Barré syndrome, Bell’s palsy, and myasthenic disorders, while BNT162b2 is associated with a short-term increase in hemorrhagic stroke, particularly among females. However, SARS-CoV-2 infection confers substantially higher risks of the same or broader neurological outcomes than vaccination, emphasizing that the absolute neurological risk is greater from infection than from vaccines. The ChAdOx1nCoV-19–GBS association was robust across sensitivity analyses and replicated in Scotland, strengthening causal inference. Potential mechanisms (e.g., immune cross-reactivity with adenoviral vectors for ChAdOx1nCoV-19 or thrombocytopenia-related bleeding after mRNA vaccination) remain speculative and require further study. Clinically, awareness of these rare complications can guide earlier diagnosis and resource planning (e.g., intensive care needs for GBS). Policy-wise, results support ongoing vaccination programs given the far greater neurological risks following infection.

In this large population-based SCCS study linking national vaccination, testing, hospitalization, and mortality data, several rare neurological adverse events were identified following first-dose vaccination, notably GBS after ChAdOx1nCoV-19 and hemorrhagic stroke after BNT162b2, alongside markedly higher risks following SARS-CoV-2 infection. Replication in Scotland corroborated the ChAdOx1nCoV-19–GBS association. Findings inform clinical vigilance and risk–benefit evaluations for vaccination. Future research should investigate biological mechanisms (particularly for ChAdOx1nCoV-19–associated GBS and sex differences in BNT162b2-associated hemorrhagic stroke), assess risks after second doses and boosters, evaluate variant-specific and age-stratified effects, and replicate analyses in other large international datasets.

- Analyses confined to first vaccine doses; limited data on second doses during the study window. - Hospitalization and mortality data used; milder neurological events not resulting in admission may be under-ascertained. - Inability to distinguish GBS variants (e.g., Miller Fisher) due to ICD-10 coding granularity. - Some myasthenic disorder admissions may represent exacerbations of pre-existing disease; prior history excluded only over two years. - Event-dependent observation/censoring (e.g., deaths after hemorrhagic stroke or subarachnoid hemorrhage) could bias estimates, although sensitivity analyses mitigated concerns. - Potential reverse causation around day 0 for infection (testing triggered by hospital admission), addressed by isolating day 0 but may still influence interpretation. - Scottish replication cohort smaller, yielding imprecise estimates and insufficient power to replicate findings with smaller effect sizes. - Few infections after vaccination limited separate analysis of post-vaccination infection risks.