JYNNEOS doses administered from May 1, 2022 through March 4, 2023 to individuals in the VSD population were included in the AESI analyses. The day of vaccination was defined as day 0. AESI were selected for evaluation because they are of interest for vaccines in general (e.g., anaphylaxis, Guillain-Barré syndrome), or based on adverse reactions that occurred with first- and second-generation smallpox vaccines (e.g., myocarditis and pericarditis), or based on AESI specified in the EUA (e.g., cardiac, thromboembolic, or neurovascular events). The AESI case definitions used are shown in Supplemental Table 1. Adverse reactions associated with older smallpox vaccines that are caused by replication of live vaccinia virus (e.g., progressive vaccinia) were not assessed for JYNNEOS because JYNNEOS contains a non-replicating virus.

A standardized incidence ratio (SIR) analysis was used to evaluate the risk for each of ten outcomes (all AESI except anaphylaxis) following each dose, including myocarditis or pericarditis, acute myocardial infarction, hemorrhagic stroke, ischemic stroke, venous thromboembolism, pulmonary embolism, acute disseminated encephalomyelitis, encephalitis, Guillain-Barré syndrome, and transverse myelitis. The SIR analysis was restricted to adults aged ≥18 years because the number of people aged <18 years who received JYNNEOS was too small to be informative for these rare AESI (the number of people <18 who received JYNNEOS is described separately). We did separate analyses for males and females because most JYNNEOS recipients were male. We compared the observed incidence rate for each outcome to the expected rate. An incident diagnosis of an AESI within 28 days after either dose of JYNNEOS was included. The expected rates were calculated using the background incidence in the VSD population from July 1, 2021 through June 30, 2022; this time period was chosen to most closely represent the conditions during the period of interest based on a prior evaluation of healthcare utilization in the VSD through 2023 which showed that by the later part of 2021, inpatient and ED visit rates returned to prepandemic levels.[6] The expected rates were adjusted for VSD site, age group (18–49, 50–64, ≥65 years), and history of four categories of atherosclerotic cardiovascular disease risk factors (coronary heart disease equivalents, dyslipidemia, hypertension, HIV infection). These risk factors were assessed based on ICD-10 coded medical encounters (Supplemental Table 2). Due to the large number of strata already involved, no other variables were adjusted for, such as potential seasonal variation in background rates. The stratum-specific background population incidence rates were multiplied by the person-time for each corresponding stratum in the vaccinated population, and the weighted average of these values was used to calculate the expected number of cases for each outcome among the vaccinated. The SIR was calculated by dividing the observed incidence rate among the vaccinated by the expected incidence rate. Analysis was done using SAS version 9.4.

Independent of the SIR analysis, medical record abstraction was prespecified for two AESI, anaphylaxis and myocarditis/pericarditis, to calculate crude postvaccination incidence rates using chart verified cases. For anaphylaxis, cases diagnosed on days 0–1 were classified using the Brighton Collaboration case definitions [7, 8] and the clinical history was reviewed for any potential causes other than vaccination. Myocarditis and pericarditis were classified using CDC case definitions.[9] The incidence of myocarditis (with or without pericarditis) with onset within 30 days following each dose was calculated as an additional analysis because risk intervals up to 30 days post vaccination have been used in previous smallpox vaccine studies. Dose 1 follow-up time was censored upon receipt of dose 2.

This analysis evaluated ICD-10 coded medical encounters during the 70 days following vaccination with dose 1 to identify statistically significant temporal clustering of cases of any incident diagnosis following vaccination. The presence of temporal clustering could suggest a potential association between the vaccination and the medical condition. This method has been described previously.[10–12] Here we will summarize the method and state the specific parameters chosen for this analysis.

ICD-10 codes have a hierarchical tree-like structure with 21 broad categories of diagnoses which successively branch into more specific sub-categories, culminating in highly specific diagnosis codes. ICD-10 codes have up to seven digits, and the ICD-10 “tree” branches out into seven levels, with level 1 corresponding to the 21 broad categories, level 2 to blocks of related diagnoses within each of the 21 categories, level 3 to 3-character ICD-10 codes, level 4 to 4-character codes, and so on out to level 7. We excluded ICD-10 codes for neoplasms, external causes of morbidity, and factors influencing health status and contact with health services, because these conditions seem biologically implausible to be caused by vaccination during the follow-up period assessed. Pregnancy, perinatal conditions, and congenital malformations were excluded because the vaccinated population consisted mostly of adult males. The remaining tree contained more than 60,000 diagnoses and diagnosis groups for evaluation.

Temporal clustering of diagnoses was assessed using the tree-temporal scan statistic with a self-controlled design, which looks for a statistically higher incidence of each outcome during each potential post-vaccination risk interval compared to the person time outside of the interval [11, 12]. We assessed case clustering during post-vaccination days 1 through 70 following dose 1. The day of vaccination (day 0) is excluded because the coded data may not clearly distinguish between conditions with onset before versus after vaccination. We evaluated all potential risk intervals with lengths from 2 through 35 days within the 70 days of follow-up time. Monte Carlo simulation was used to adjust for the multiple scanning of the thousands of different diagnoses and the hundreds of different time intervals. The log likelihood ratio is calculated for each node on the tree and then is compared to the maximum log likelihood ratio distribution developed under the null hypothesis. The P-value is determined based on that distribution. We chose a statistical significance threshold of P<0.01 a priori but show all outcomes with P<0.05 in the results to provide additional context. The statistical analyses were conducted using TreeScan software version 2.0 [13].

VSD members who received dose 1 of JYNNEOS from May 1, 2022 through April 30, 2023 were included in this analysis if they were continuously enrolled during the period from 400 days prior to vaccination through 70 days after receipt of dose 1. This allowed for exclusion of prevalent diagnoses which had been documented prior to vaccination and allowed for six weeks of follow-up time after dose 2 when this was administered at the recommended interval.

The primary analysis evaluated incident diagnoses recorded for inpatient or emergency department (ED) visits; incident diagnoses were defined as those where no instance of the same three-character code (level 3 of the tree) occurred in any setting in the patient’s record in the prior 400 days. Secondary analysis included incident diagnoses recorded for any healthcare visit (including outpatient); incident diagnoses were defined as those where no instance of a diagnosis in the same block of codes (level 2 of the tree) occurred in any setting in the patient’s record in the prior 400 days. Clustering was evaluated in levels 3─6 of the tree in the primary analysis and in levels 2─6 of the tree in the secondary analysis, because the highest and lowest levels of the tree were considered, respectively, too broad or too specific to be clinically meaningful. Analyses were done for all ages combined, and age-stratified analyses were done for the following age groups: <18, 18–39, 40–64, ≥65 years.

These studies were reviewed and approved by the Institutional Review Board at each participating VSD site. These activities were reviewed by CDC and were conducted consistent with applicable federal law and CDC policy [14].

As of March 4, 2023, in the VSD population there were 53,583 adults who received JYNNEOS dose 1, of whom, 38,206 also received dose 2. Males received 92% of the doses (Table 1). The prevalence of HIV infection among males was 16% in the vaccinated cohort compared to 0.5% in the VSD population, and among females was 4% compared to 0.06%. The two-dose vaccine series completion rate for males was 73% and for females was 51%. The anatomic sites of vaccine administration were forearm (19%), upper arm (5%), arm not otherwise specified (56%), unknown (19%), and other (<1%). Most doses were given during the summer of 2022, with 75% of doses given from July through September. JYNNEOS was administered alone (without other vaccines on the same day) for 90% of doses; the most common vaccines simultaneously administered with JYNNEOS were influenza (6.4%) and COVID-19 (2.7%).

The SIR analyses did not find statistically significant elevated SIRs for any AESI among either males or females (Table 2).

Medical records were abstracted for two patients identified by myocarditis ICD-10 codes within 30 days postvaccination. Both met the CDC case definition for “probable” myocarditis (without pericarditis). The two cases occurred in males aged between 30─50 years. One patient was hospitalized and the other was treated in the ED. One case had symptom onset 24 days after dose 1 and the other had onset 13 days after dose 2. The incidence rate of myocarditis among males within 30 days following JYNNEOS dose 1 was 20 cases per million doses administered (95%CI: 0.52, 114) and following dose 2 was 28 cases per million doses administered (95%CI: 0.71, 156).

Two potential cases of anaphylaxis identified by ICD-10 codes were evaluated by medical record abstraction. In one patient, the ICD-10 code was used due to a history of allergic reaction and did not represent a new event following vaccination. The other patient was a male who had acute onset of hives, dyspnea, and throat closing sensation three minutes after receiving JYNNEOS dose 1 (without any other vaccines). This allergic event met the original 2007 Brighton Collaboration case definition for anaphylaxis with diagnostic certainty level 2; however, using the updated 2022 Brighton Collaboration case definition, this was not a case of anaphylaxis. Using the 2007 case definition gives an incidence rate of 19 cases per million doses (95%CI: 0.47, 104).

Among children aged <18 years in the VSD, 88 received JYNNEOS dose 1 and 33 received dose 2. The age range among this group was 1 to 17 years. There were no AESI identified in this age group.

After applying the enrollment criteria, this analysis included 36,912 people who received JYNNEOS dose 1, of which 33,685 (91%) were male and 26,377 (71.5%) went on to receive dose 2 (Table 3). Table 4 shows the temporal clustering results. The primary analysis, which included inpatient and ED visits, did not identify statistically significant clusters for any outcome overall or in any age subgroup. The secondary analysis, which included inpatient, ED, and outpatient visits, identified statistically significant clusters for rash and several codes for unspecified adverse events; age group-specific analyses also identified a non-significant grouping of “pain in ankle and joints of foot” in the ≥65-year age group.