key: cord-0833791-vnjt4zr4 authors: Sinclair, J. E.; Mayfield, H. J.; Short, K. R.; Brown, S. J.; Puranik, R.; Mengersen, K.; Litt, J. C.; Lau, C. L. title: Quantifying the risks versus benefits of the Pfizer COVID-19 vaccine in Australia: a Bayesian network analysis date: 2022-02-08 journal: nan DOI: 10.1101/2022.02.07.22270637 sha: 68243f10352d850824ebd4a0a77a7c1373070760 doc_id: 833791 cord_uid: vnjt4zr4 The Pfizer COVID-19 vaccine is associated with increased myocarditis incidence. Constantly evolving evidence regarding incidence and case fatality of COVID-19 and myocarditis related to infection or vaccination, creates challenge for risk-benefit analysis of vaccination programs. Challenges are complicated further by emerging evidence of waning vaccine effectiveness, and variable effectiveness against variants. Here, we build on previous work on the COVID-19 Risk Calculator (CoRiCal) by integrating Australian and international data to inform a Bayesian network that calculates probabilities of outcomes for the Delta variant under different scenarios of Pfizer COVID-19 vaccine coverage, age groups ([≥]12 years), sex, community transmission intensity and vaccine effectiveness. The model estimates that in a population where 5% were unvaccinated, 5% had one dose, 60% had two doses and 30% had three doses, the probabilities of developing and dying from COVID-19-related myocarditis were 239-5847 and 1430-384,684 times higher (depending on age and sex), respectively, than developing vaccine-associated myocarditis. For one million people with this vaccine coverage, where transmission intensity was equivalent to 10% chance of infection over two months, 68,813 symptomatic COVID-19 cases and 981 deaths would be prevented, with 42 and 16 expected cases of vaccine-associated myocarditis in males and females, respectively. The model may be updated to include emerging best evidence, data pertinent to different countries or vaccines, and other outcomes such as long COVID. In December 2020, the Pfizer vaccine (BNT162b2; Cormirnaty) became the first COVID-19 vaccine 58 to be authorized for public use [1], and has since had more than 1.5 billion doses delivered to 131 Table 1 [11,15,20-35] and Supplementary Tables S1-9. Table 2 Table S8 ). Based on estimates from model version 2, Figure 2 shows that, in a 180 population aged ≥12 years, with vaccine coverage of 5% unvaccinated, 5% had one dose, 60% had 181 two doses and 30% had three doses, the probability of developing myocarditis related to symptomatic 182 COVID-19 was 239 to 5847 times higher than developing Pfizer vaccine-associated myocarditis, CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) if 5% were unvaccinated, 5% had one dose, 60% had two doses and 30% had three doses (scenario 230 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) higher risk of developing vaccine-associated myocarditis than older groups, and males were at greater 294 risk than females. We note that myocarditis was more common after COVID-19 compared to the 295 background rates, especially in younger men. In comparison, vaccine-associated myocarditis also has 296 a predilection for younger males but at a much lower prevalence than cases associated with 297 symptomatic COVID-19. Importantly, in the main, vaccination is justified in all age groups because 298 myocarditis is generally mild in the young [37-39], and there is unequivocal evidence for reduced 299 mortality in older individuals across all levels of community transmission. 300 301 While the above risk-benefit analyses were conducted assuming the Australian vaccine coverage at 302 the time of writing, outcomes under other coverage rates can be assessed by the model. We compared 303 the number of COVID-19 cases and deaths expected if the chance of infection was 10% over two 304 months under a scenario where 5% are unvaccinated, 5% had a first dose, 60% had two doses and 305 30% had three doses, to those expected under a second scenario where 0% are unvaccinated, 5% had a 306 first dose, 15% had two doses and 80% had three doses ( Figure 4 ). Younger age groups benefited 307 from the steepest decline in expected case rates, with at least 23,000 fewer cases per million in 20-29 308 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted February 8, 2022. ; year-olds. In contrast, older age groups benefited from the greatest decrease in expected deaths from 309 COVID-19, with 337 fewer deaths per million expected in those aged ≥70 years. 310 311 Sensitivity analysis showed model estimates to be robust against minor changes in the number of 312 Pfizer vaccine-associated myocarditis cases (Table 3 ), but highly affected by changes in vaccine 313 effectiveness against symptomatic infection and death (Table 4) were available on the incidence of Pfizer vaccine-associated myocarditis after the third dose and 327 international data were deemed inappropriate as a substitute (see Table 1 authority-issued data were employed whenever possible (e.g., national data on Pfizer vaccine-437 associated myocarditis). When this was unavailable, data were retrieved from other reliable and 438 publicly available sources (e.g., background rates of myocarditis). Where Australian data were not 439 readily available and international data were not suitable to use for the Australian context, expert 440 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted February 8, 2022. ; https://doi.org/10.1101/2022.02.07.22270637 doi: medRxiv preprint opinion was sought. For example, there were limited data in Australia about Pfizer vaccine-associated 441 myocarditis incidence and CFR after the third dose. While rates were reported in Israel and Singapore, 442 these were deemed inappropriate to use in the model as reported rates from first and second doses in 443 these countries were much lower than in Australia. However, both reported lower incidence of 444 myocarditis after the third dose than the second dose. Therefore, to avoid underestimating the risk, the 445 decision was made by the subject experts to use a conservative assumption that incidence after the 446 third dose was the same as the second dose. For some variables, data analysis was required to obtain 447 probabilities for the CPTs, e.g., converting COVID-19 case incidence into probability of infection 448 over two months for the community transmission intensity node, or averaging data to fit the BN age 449 categories. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted February 8, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted February 8, 2022. ; • Data from large integrated health system in the USA • Data not specifically for delta variant but for a mix so we assumed there would be negligible difference between variants. • Our model focuses on risk of symptomatic infection, but this study reports estimates for total risk of infection (not necessarily symptomatic). Our model may therefore have underestimated vaccine effectiveness against symptomatic infection. • The study reports vaccine effectiveness at <1 month, 1 to <2 months, 2 to <3 months, 3 to <4 months, 4 to <5 months, and ≥5 months since the second dose. When transforming these data to the time categories used in our model (0 to <2 months, 2 to <4 months and 4 to <6 months), we averaged the reported vaccine effectiveness of the respective months in each group. • In transforming the reported age groups to those used in our model, we assumed that in age group 12-19 years, 50% were aged 12-15 years and 50% were aged 16-19 years. Likewise for age group 40-49 years we assumed that 50% of people were aged 40-44 years and 50% were aged 45-49 years. Similar assumptions were used for 50-59 and 60-69 year-olds. • See Table S1 for summary of final assumptions. • Data from Pfizer third dose efficacy study conducted in the USA, Brazil and South Africa • Age 16-55 years: 96.5% effective. Age ≥ 56 years: 93.1% effective • Study conducted when delta was the dominant variant. • We assumed vaccine effectiveness in ages 12-15 years was the same as in ages 16-55 years. • In transforming reported age groups to those used in our model, we assumed that in age group 50-59 years, 60% were 50-55 years and 40% were 56-59 years. • See Table S1 for summary of final assumptions. Vaccine effectiveness against death if infected 1 dose [22] • Data from Ontario study, reporting vaccine effectiveness against hospitalisation or death from delta variant. These data may therefore underestimate effectiveness against death. • Age <60 years: 89% effective. Age ≥60 years: 74% effective. • Data from Public Health England reporting vaccine effectiveness against death from delta variant. • In transforming reported time since second dose into the categories used in our model, we used weighted averages of the vaccine effectiveness in different time groups reported in the study, with weighting being proportionate to the number of weeks in each category. • In transforming the reported age groups to the categories used in our model, we assumed that for age group 60-69 years, 50% were 60-64 years and 50% were 65-69 years. • Data were reported only for age groups ≥16 years (which includes ≥65 years) and ≥65 years. As data were not provided for ages 16-64 years only, we assumed estimates were the same as for the ≥16 years age group. It is therefore possible that vaccine effectiveness for this age group was underestimated due to influence of the lower effectiveness within the ≥65-year-olds. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint • As no data were reported for age <16 years, we assumed that ages 12-15 years had the same vaccine effectiveness as ages 16-64 years. • See Table S2 for summary of final assumptions. • As no data have yet been published on the effectiveness of a third dose against death, we assumed the same effectiveness as 'Two doses (last dose 0 to <2 months ago)'. Relative risk of symptomatic infection by age and sex Data from Australian National Interoperable Notifiable Diseases Surveillance System (NINDSS) [24] reports age and sex distribution of all COVID-19 cases in Australia up to 8 Dec 2021. We subtracted data from the Australian Government Department of Health Epidemiology Reports 32 and 43 [25] reporting age and sex distribution of COVID-19 cases in Australia in 2020, and Jan to June 2021, respectively, to obtain age and sex distribution of cases from 6 June to 8 Dec 2021 to represent the delta variant. We calculated relative risk of infection by age group and sex by estimating the probability of infection in each age-sex group if overall probability of infection in the community was 1%. See Table S3 for final assumptions. Definitions of low, medium, and high transmission as defined by Australian Technical Advisory Group on Immunisation (ATAGI) [11] . Low -similar to first wave in Australia (equivalent to 0.016% of population infected over 2 months). Medium -similar to second wave in Victoria, Australia in 2020 (equivalent to 0.149% of population infected over 2 months). High -similar to Europe in January 2021 (equivalent to 1.920% of population infected over 2 months). Also included transmission scenarios equivalent to: zero transmission; 1%, 2%, 5% and 10% chance of infection over 2 months. Chance of infection over 2 months calculated for different levels of community transmission. See Table S4 for final assumptions. COVID-19 cases reported in Australia from January 2020 to 18/11/2021 were used to provide estimates of age-sex-specific case fatality rates. Data sourced from Australian NINDSS [24] . To convert reported age groups into those used in our model, calculations were based on age distribution of the Australian population [26] . See Table S5 for final assumptions. Multinational network cohort study from Australia, France, Germany, Japan, Netherlands, Spain, the UK and the USA reports background incidence of myocarditis and pericarditis per 100,000 person-years by age group and sex [27] . We assumed that 65% of reported myopericarditis cases were myocarditis, based on proportions from other studies that differentiate between them post-vaccination [28, 29] . We converted incidence to probability of infection per person over 2 months. To convert reported age groups into those used in the model, calculations were based on age distribution of the Australian population [26] . See Table S6 for final assumptions. Risk of dying from (background) myocarditis Study reports incidence of fatal myocarditis in Finland per 100,000 person-years by age group and sex as total risk [30] , but not as case fatality rate. We converted incidence per 100,000 person-years to probability per person over 2 months (in the general population), then used these values for each age-sex subgroup as the numerator and the respective values for node 'Risk of getting (background) myocarditis' as the denominator to calculate case fatality rate. When converting reported age groups to the age groups used in our model, calculations were based on the age distribution of the Australian population [26] . See Table S6 for final assumptions. Risk of getting Pfizer vaccineassociated myocarditis Therapeutic Goods Administration (TGA) reports rates of myocarditis from the Pfizer vaccine per 100,000 doses in Australia, from all doses and second doses [31] . From this we calculated rates from first doses. At the time of writing, the only data available for the third dose in Australia cited four reports of likely myocarditis from a third dose of Pfizer up to 09/01/2022 with 3,651,855 third doses given nationally up to that date (with no breakdown of proportion of doses by brand). As this information is very limited, we assumed the same rate of vaccine-associated myocarditis as the second dose. This assumption was based on data from Israel reporting that rates of Pfizer vaccine-induced myocarditis from the third dose was higher than after the first dose but lower than after the second dose [32] . To provide a conservative estimate and avoid underestimating the potential risk of myocarditis after the third dose, we assumed the same rates as the second dose, i.e. the 'worst case scenario'. See Table S7 for final assumptions. Risk of dying from Pfizer vaccineassociated Case fatality rate from mRNA vaccine-associated myocarditis has not been reported widely, in part due to very low numbers. Data from USA Centers for Disease Control and Prevention (CDC) Vaccine Adverse Event Reporting System (VAERS) [33] . Reported 1195 myocarditis cases after mRNA vaccination (dose number not specified) in those aged under 30 years, of which two likely died from myocarditis, giving a case fatality rate of 0.17% (2/1195). We assumed the same case fatality rate for Pfizer and other mRNA COVID-19 vaccines, and the same case fatality rate in those aged ≥30 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted February 8, 2022. Study reports a six-month all-cause mortality of 3.9% in COVID-19 patients with myocarditis, assuming that deaths were attributable to myocarditis [34] . Published data were insufficient to stratify by age and sex. Age-sex breakdown of the myocarditis cases and deaths were provided by the authors through personal communication. Data provided through personal communication were based on electronic medical records in TriNetX, reported with patient counts ≥10 rounded up to 10 to safeguard protected healthcare data. The case fatality rate for age-sex subgroups with 10 deaths was thus assumed to be <1.00%, with a value of 1.00% used in the model to assume the worst-case scenario. For males aged 12-19 and 20-29 years, there were zero deaths out of 152 and 661 cases of myocarditis, respectively. To avoid using a 0% case fatality rate in the model, we assumed that 12-19 and 20-29 year old males had the same case fatality rate as 30-39 year old males (1.00%). We believe this is a reasonable assumption because in females there was no significant difference in case fatality rate between ages 12-19 and 20-29 years and 30-39 years. See Table S8 for final assumptions. Age distribution of population Distribution based on Australian Bureau of Statistics national population estimates from September 2021 [26] . See Table S9 for final assumptions. Note age group 0-11 years was excluded from this version of the model because they were not yet eligible for vaccination in Australia at time of writing. This age group can be added into the model when vaccine coverage increases and data on vaccine effectiveness become available. Sex distribution of population Assumed 50% male, 50% female. Pfizer vaccine coverage in population* Assumed 5% had no doses, 5% had one dose only, 60% had two doses only, 30% had three doses for ages ≥12 years. These approximations were based on vaccine coverage data from Australian Government Department of Health COVID-19 vaccination data on 3 Jan 2022 [35], and our estimates of how coverage will increase over the coming months. Community transmission at x% over 2 months* Chance of infection (x%) over 2 months, based on different levels of community transmission. Priors set to even distribution between categories, assuming that community transmission level will be selected when using the CoRiCal tool or running public health-level scenario analyses. See explanation above under 'Risk of symptomatic infection under current transmission and vaccination status'. *Note that prior distributions do not affect results of scenario analysis but enables the model to provide population-level estimates. Assumptions can be changed as the situation evolves. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 8, 2022. ; Australian Government Department of Health. ATAGI statement on revised 607 recommendations on the use of COVID-10 vaccine AstraZeneca Effectiveness 611 of mRNA BNT162b2 COVID-19 vaccine up to 6 months in a large integrated health system 612 in the USA: a retrospective cohort study Thrombosis with thrombocytopenia 614 syndrome (TTS) following AstraZeneca ChAdOx1 vaccination -A risk-benefit analysis for people <60 years in Australia Risk-benefit 618 analysis of the AstraZeneca COVID-19 vaccine in Australia using a Bayesian network 619 modelling framework Designing an evidence-621 Bayesian network for estimating the risk versus benefits of AstraZeneca COVID-19 622 vaccine Assessment of severe outcomes with variants of concern in Ontario Australian Government Department of Health. Coronavirus (COVID-19) case numbers and 639 statistics -cases and deaths by age and sex alerts/novel-coronavirus-2019-ncov-health-alert/coronavirus-covid-19-case-numbers-and-641 statistics#covid19-summary-statistics Australian Government Department of Health. Coronavirus disease 2019 (COVID-19) 643 epidemiology reports Bureau of Statistics. National, state and territory population Characterising the 650 background incidence rates of adverse events of special interest COVID-19 vaccines in eight 651 countries: multinational network cohort study BNT162b2 mRNA COVID-19 vaccine in a nationwide setting Incidence of fatal myocarditis: a population-661 based study in Finland Therapeutic Goods Administration. COVID-19 vaccine weekly safety Vaccines and Related Biological Products Advisory Committee Myocarditis cases reported 670 after mRNA-based COVID-19 vaccination in the US from Prevalence and clinical outcomes of myocarditis and pericarditis in 718,365 COVID-19 674 patients Australian Government Department of Health. COVID-19 vaccination -vaccination data -3 Therapeutic Goods Administration. COVID-19 vaccine weekly safety report reactions/coronavirus-vaccine-summary-of-yellow-card-reporting BNT162b2 vaccine against COVID-19 in Israel Risk assessment and decision analysis with Bayesian networks Common quandaries and their practical solutions in Bayesian network modeling How COVID-19 impacts Chinese travelers' mobility decision-697 making processes: a Bayesian network model. Information and Communication Technologies 698 in Tourism Bayesian network model for personalized COVID-19 risk assessment and contact tracing Extending the range of COVID-19 risk factors in a Bayesian network 703 model for personalized risk assessment Machine reasoning to assess pandemics risks: case of USS 706 Acute myocarditis: aetiology, diagnosis and 708 management