Characteristics of the study population
There were 3807 deaths of 12 to 29 year-olds in England that occurred between 8 December 2020 and 25 May 2022 and were registered by 8 June 2022 (Supplementary Fig. 1). Of these, 444 (11.7%) were due to a cardiac event and 1512 (39.7%) were linked to a vaccination record (1510 from NIMS and 2 from the supplementary NHS point of care extract) (Table 1). 62.8% (950) of first doses, 51.6% (505) of second doses and 98.8% (239) of third doses in the death registrations dataset were mRNA based (either the BNT162b2 Pfizer-BioNTech or mRNA-1273 Moderna vaccines), rather than non-mRNA based (the ChAdOx1 Oxford-AstraZeneca vaccine) or another vaccine or unknown. Of those who received both the first and second vaccination (979), 11.3% (111) received a different type of vaccine for each dose (Supplementary Table 1). Counts of deaths by sex and vaccine vector for people who received at least one dose of that vector are included (Supplementary Table 2).
Between 8 December 2020 and 31 March 2022, 1420 hospital deaths among 12- to 29-year-olds were recorded in HES, 630 (44.4%) of which were linked to a vaccination record (629 from NIMS and 1 from the supplementary NHS point of care extract). 63.3% (399) of first doses, 53.5% (228) of second doses and 98.2% (108) of third doses in the hospital deaths dataset were mRNA based. Of those who received both the first and second vaccination (422), 12.0% (51) received a different type of vaccine for each dose (Supplementary Table 1). Participant characteristics were similar across the three case-series (Table 1).
The corresponding analyses of deaths after positive SARS-CoV-2 test, between 8 December 2020 and 31 December 2021, included 3219 registered deaths, of which 369 (11.5%) were due to a cardiac event, and 353 (11.0%) were linked to a preceding positive SARS-CoV-2 test (297 (9.2%) occurred unvaccinated at the date of test registration and 56 (1.7%) vaccinated at the date of test registration). There were 1123 corresponding hospital deaths recorded in HES between 8 December 2020 and 31 December 2021 of individuals who were not infected on admission to hospital, of which 181 (16.1%) were linked to a preceding positive SARS-CoV-2 test (150 (13.4%) unvaccinated on date of test registration and 31 (2.8%) vaccinated). Participant characteristics were again similar across the three case series (Table 2).
Consistent with an increasing prevalence of registration delay due to coroner referral associated with more recent deaths, the observed incidence of ONS registered deaths decreased over time (Supplementary Fig. 2). Hospital deaths recorded in HES are unaffected by coroner delays and did not exhibit such a trend. Similar patterns were observed for the analyses of deaths after a positive SARS-CoV-2 test for the registered deaths and hospital deaths datasets (Supplementary Fig. 3).
Supplementary Fig. 4A shows the number of deaths each week since vaccination. There was a marked decline at around 12 weeks after first dose, when most people would have a second dose. Supplementary Fig. 4B shows the number of deaths each week since positive SARS-CoV-2 test, with markedly higher numbers of all-cause deaths occurring in the first 5 weeks.
Relative incidence of death after COVID-19 vaccination
In the twelve weeks post-vaccination compared with subsequent periods, there were no significant increases in the incidence of any mortality outcome for all vaccine doses combined across each individual week or all twelve weeks combined (all-cause registered death: incidence rate ratio, IRR, 0.88, 95% confidence interval, CI, 0.80, 0.97; cardiac registered death: IRR 1.11, 95% CI 0.87, 1.42; all-cause hospital death: IRR 0.89, 95% CI 0.77, 1.04) (Fig. 1a). There were also no significant increases in incidence of any mortality outcome for any individual dose for all twelve weeks combined or across each individual week in the first 12 weeks after vaccination, except week 12 for cardiac deaths after dose 1 (IRR 3.22 95% CI 1.39, 7.48; Fig. 1b).
We found a significant decrease in the incidence of all-cause registered death, driven by the first two weeks after vaccination (any dose, week 1: IRR 0.47 [0.34, 0.64]; week 2: 0.77 [0.60, 0.99]; Fig. 1a). Similarly, there was a decreased risk of hospital death in the first two weeks after vaccination (any dose, week 1: IRR 0.32 [0.18, 0.60]; week 2: 0.51 [0.31, 0.83]; Fig. 1a).
Subgroup analyses by vaccine vector were generally consistent with the main results. However, for non-mRNA or unknown vaccine vectors (primarily consisting of the ChAdOx1 Oxford-AstraZeneca) for the first dose and all doses combined, there were significant increases in the incidence of cardiac death (first dose: IRR 1.75 [1.14, 2.71]; all doses: 1.71 [1.20, 2.45]; Fig. 2a). A raised risk of all-cause hospital death was observed after a first dose of a non mRNA vaccine (IRR 1.33 [1.01, 1.74]), but this was not seen in all-cause registered deaths (IRR 0.90 [0.75, 1.07]). Subgroup analyses stratified by age group and sex were consistent with the main results except after the first dose in females, in which cardiac death was increased (IRR 1.79 [1.05, 3.05]). This corresponded to 1 additional cardiac registered death for every 363,419 (95% CI 238,784, 3,272,867) females aged 12–29 vaccinated (Fig. 2b). However, there was no evidence of a raised risk in females after second or third doses (IRR 0.73 [0.37,1.43] and 1.28 [0.32, 5.17] respectively) or when looking at all doses combined (IRR 1.31 [0.86, 2.00]). Additionally, there was no evidence of elevated all-cause mortality in this subgroup after the first dose (all-cause registered death: IRR 1.07 [0.88, 1.31]; all-cause hospital death: IRR 1.01 [0.74, 1.38]).
Subgroup analyses by vaccine vector broken down by sex (Supplementary Fig. 5) showed a raised risk of cardiac death in females following a first dose of a non mRNA vaccine and for all doses combined (first dose: IRR 3.52 [1.71 – 7.26]; all doses: 3.02 [1.65, 5.53]) but not after a first dose or any dose of mRNA vaccine (first dose: IRR 0.87 [0.41 – 1.85]; all doses: 0.76 [0.43, 1.34]). However, no significant increase in the risk of cardiac death was observed in females after the second dose of an mRNA vaccine (IRR 0.59 [0.24, 1.44]) and a nonsignificant increase was observed after a first dose of a non-mRNA vaccine (1.88 (0.73, 4.87]). This increase in risk corresponded to 1 additional cardiac registered death for every 16,486 (95% CI 13,688, 28,426) females aged 12–29 who received a first dose of a non mRNA vaccine. An increased risk of all-cause hospital death was also observed in females after a first dose, and after any dose of a non-mRNA vaccine (first dose: IRR 1.66 [1.10 – 2.51]; any dose: 1.55 [1.10, 2.18]), however this was not observed in all-cause registered deaths (first dose: IRR 1.13 [0.85, 1.49]; any dose: 1.09 [0.86, 1.37]).
There was also a non-significant elevation in IRR for cardiac death following a second dose of an mRNA vaccine in men (IRR 1.70 [0.98, 2.97]), which, if valid, would correspond to one additional cardiac registered death for every 359,294 (223,043, -) men aged 12-29 who received a second dose of an mRNA vaccine. This result was sensitive to variation of the risk period, with no increase observed with a 6 week risk period (IRR 1.15 (0.60, 2.18)). (Supplementary Fig. 6).
Relative incidence of death after positive test for SARS-CoV-2
Following a positive SARS-CoV-2 test among individuals unvaccinated at date of test registration, there was an increase in the incidence of cardiac death (IRR for the whole 12-week period: IRR 2.35 [1.09, 5.06]), driven by the first week (IRR 11.56 [3.93, 33.99]; Fig. 3a). During the whole 12-week period, higher incidences of all-cause mortality were also observed for both registered deaths (IRR 2.50 [1.93, 3.23] and hospital deaths (IRR 4.50 [3.09, 6.54]), similarly most pronounced in the first week (IRR 6.87 [4.53, 10.42] and 9.02 [4.79, 16.97], respectively).
There was also an increase in the incidence of all-cause death following a positive SARS-CoV-2-test among individuals vaccinated at date of test registration (IRR 1.94 [1.03, 3.67] for all-cause registered deaths; 2.76 [1.14, 6.71] for all-cause hospital deaths; Fig. 3b), similarly most pronounced in the first two weeks. There were insufficient data to analyse cardiac deaths among people vaccinated on the date of test registration.
Relative increases in mortality rates following a positive SARS-CoV-2 test were comparable across all subgroups but with greater uncertainty in younger age groups (Fig. 4).
In people aged 12-29 with a positive SARS-CoV-2 test, the increased risk of all-cause registered death in the following twelve weeks corresponded to 1 additional death for every 11,936 (95% CI 10,373, 14,862) individuals aged 12–29 and unvaccinated on date of positive test registration, and 1 additional death for every 55,661 (37,071, 925,962) individuals aged 12–29 and vaccinated on date of positive test registration.
For all three analyses of death after vaccination, varying defined lengths of risk period were consistent with the main findings in remaining below or not significantly different to 1 (Supplementary Fig. 7). For analyses of deaths after a positive SARS-CoV-2 test for individuals unvaccinated at the date of test registration, estimates of relative incidence of all-cause death were greater for shorter risk period definitions, particularly for hospital deaths. While convergence towards one was observed as the defined risk period was increased, complete convergence was not observed, with the relative incidence appearing to stabilise above one for all-cause and cardiac deaths (Supplementary Fig. 7A). For analyses of deaths after a positive SARS-CoV-2 test for individuals vaccinated at date of test registration (Supplementary Fig. 8B), estimates of the relative incidence of all-cause registered the death and all-cause hospital death were greater for shorter risk period definitions and converged to 1 as the risk period was increased.
Results for the analyses of vaccination and positive SARS-CoV-2 tests were consistent when adjusting calendar time in fortnightly intervals compared to the restricted cubic spline adjustment of the calendar week in the main analysis (Supplementary Figs. 9 and 10). Exclusion of the 23 individuals in the HES analysis of vaccinations who tested positive on the day of hospital admission prior to death gave consistent estimates to the main analysis (Supplementary Fig. 11). Inclusion of the 20 individuals in the HES analysis of positive SARS-CoV-2 tests who tested positive on the day of hospital admission prior to death gave similar but slightly higher estimates for weeks 1-12 compared to the main analysis, but with particularly higher estimates in the first week after testing positive (unvaccinated at date of positive test, weeks 1-12: IRR 5.01 [3.48, 7.23], week 1: IRR 15.26 [8.89, 25.95]; vaccinated at date of positive test, weeks 1–12 IRR: 2.97 [1.24, 7.12]; week 1: IRR 3.90 [0.96, 15.86]) (Supplementary Fig. 12).