A Bayesian network analysis of the Pfizer COVID-19 vaccine in the paediatric population

Abstract

The Pfizer COVID-19 vaccines have been associated with an increased risk of myocarditis. However, COVID-19 infection is also associated with complications. A Bayesian network (BN), informed by Australian and international data, was created to determine individual risks and benefits of the Pfizer COVID-19 vaccine in the paediatric. The risk of myocarditis between vaccine-associated, COVID-19 and background rates was compared, as well as secondary outcomes such as hospitalization, and MIS-C. At a population level, hospitalizations, intensive care admissions and MIS-C cases prevented at differing transmission rates and vaccine coverage were analyzed. The model estimated that teenage males were 4.47 times more likely to develop myocarditis from COVID-19 compared to the vaccine. Furthermore, the risk of hospitalizations and MIS-C were more likely in the unvaccinated cohort for all ages. The population level benefits of COVID-19 Pfizer vaccine at mitigating myocarditis are more nuanced, contingent on age, transmission rates and vaccination coverage.

Fig. 1. Bayesian Network for Risks vs Benefits of Pfizer COVID-19 vaccination in children.

Input nodes in orange (n1-n4), intermediate nodes in yellow (n5-n9 and n17) andoutcome nodes in purple (n10-n16).

Fig. 2

Individual probability of vaccine-induced myocarditis compared to probability of COVID-19 myocarditis and background myocarditis stratified by age and sex.

Fig. 3. Comparison of myocarditis cases.

Cases of myocarditis per million stratified by age and sex attributable to background rates, Pfizer COVID-19 vaccine associated myocarditis, SARS-CoV2 associated at transmission scenarios of 0%, 1%, 2%, 5% and 10% over a 3-month period with vaccine coverage of a 51.4%, b 70% and c 90% of the paediatric population.

Fig. 4. Estimates of hospitalisations, intensive care admissions and MSI cases under various transmission and vaccine coverage scenarios.

Number of hospitalisations and hospitalisations prevented over a 3-month period stratified by different scenarios of transmission and vaccination coverage; a the 5–11 year and b 12–17-year age groups. Number of intensive care admissions, and intensive care admissions prevented over a 3-month period stratified by different scenarios of transmission and vaccination coverage; c the 5–11 year and d 12–17-year age groups. Number of MIS-C cases per million and prevented cases stratified by different vaccination and transmission scenarios; e 5–11 year and f 12–17-year age group.

Fig. 5. Example Bayesian Network (BN) for modelling tisk of developing background myocarditis based on age and sex.

The output node “Background myocarditis over three months”, linked (via arrows) to two parent nodes “Age” and “Sex”. The model assumes 33% 5–11 year-olds, 34% 12–15 year olds and 33% 16–17 years old, with 50% male and 50% female.

Fig. 6. Conditional probability table (CPT) or the probability of background myocarditis over 3 months based on age and sex.

The CPT shows that if you are male and aged 5–11 the background risk of myocarditis over 3 months is 0.000115% and risk of not having myocarditis is 99.99985%.