Data were collected by the Investigating Respiratory Viruses in the Acutely Ill (IVY) Network from September 1, 2023, to May 31, 2024. The IVY Network is a multi-pathogen acute respiratory illness (ARI) surveillance program consisting of 26 hospitals from 20 US states, funded by the Centers for Disease Control and Prevention (CDC) and coordinated from Vanderbilt University Medical Center. This program was determined to be a non-research public health surveillance activity by CDC and each participating site and was conducted in a manner consistent with applicable federal law and CDC policy.^a

To compare likelihood of influenza vaccination (exposure) among patients with influenza–associated hospitalization (outcome) vs control patients, patients who tested positive for influenza virus and negative for SARS-CoV-2 by local clinical testing and/or centralized RT-PCR testing were classified as case patients. Patients who tested negative for both influenza virus and SARS-CoV-2 by local clinical testing and centralized RT-PCR were classified as control patients. Patients were eligible for enrollment if they met the following criteria: (1) at least 18 years old, (2) presentation with one or more symptoms of ARI (fever, cough, shortness of breath, new hypoxemia, new pulmonary findings on chest imaging consistent with pneumonia), (3) hospitalization, and (4) a recent (within 10 days of symptom onset and within 3 days of hospital admission) molecular or antigen test for influenza virus, SARS-CoV-2, or respiratory syncytial virus (RSV). Each site followed standard protocols for eligibility assessment, data collection, and specimen collection, with eligible controls enrolled locally within two weeks of eligible influenza, SARS-CoV-2, and RSV cases to ensure comparability between cases and controls.^3,4

In addition to clinical viral testing completed at hospital laboratories, personnel at each site collected nasal swab specimens at enrollment and sent them to Vanderbilt University Medical Center (VUMC) (Nashville, Tennessee, USA). Overnight shipments of specimens were received on dry ice and maintained at −80°C for a ≤14 days, thawed, and aliquoted into 2 ml cryovials. Aliquoted material was immediately returned to −80°C for long-term storage. At the time of thawing, a volume of specimen sufficient for pathogen testing was combined with lysis buffer and held at 4°C for <1 hour, followed by automated nucleic acid extraction. Specimens were received at the central laboratory in original containment. Research specimens were collected in 3 ml viral transport medium and processed at VUMC. Salvaged clinical residuals were received in containers used by the local site diagnostic laboratory and further processed (i.e., aliquoted) at VUMC, where they underwent testing for influenza virus, SARS-CoV-2, and RSV by reverse transcription polymerase chain reaction (RT-PCR). For specimens that tested positive for influenza virus by RT-PCR, an aliquot of primary specimen stored at −80C was shipped overnight on dry ice to University of Michigan (Ann Arbor, Michigan, USA) for whole genome sequencing (WGS). We excluded SARS-CoV-2-positive patients from our control group to prevent the underestimation of influenza VE associated with the correlation of COVID-19 and influenza vaccination. As patients hospitalized with COVID-19 are less likely to be vaccinated against either disease, including SARS-CoV-2-positive patients as controls increases the number of unvaccinated control patients and thus lower VE estimates.⁴ RSV cases remained included as controls. Although RSV vaccination was also available during 2023–2024, there were neither sufficient numbers of patients vaccinated for RSV (<12%) nor RSV cases among controls (<10%) to introduce meaningful bias based on correlated vaccination behaviors.⁴

Vaccination status was determined by review of electronic medical records, Immunization Information System (IIS) searches, and self-report with date and location of administration. Patients were classified as vaccinated if they received one or more doses of the 2023–2024 seasonal influenza vaccine at least 14 days before the start of symptoms (i.e., illness onset). Patients were classified as unvaccinated if they received no doses of the 2023–2024 seasonal influenza vaccine.

In our analysis, we excluded patients who: (1) received influenza testing before illness onset or more than 10 days after illness onset; (2) displayed first ARI symptoms after hospital admission or >14 days before hospital admission; (3) tested positive for SARS-CoV-2; (4) were controls enrolled outside the period of local influenza circulation (i.e., before the first influenza case or after the last influenza case at each hospital); (5) had unknown influenza vaccination status; (6) received an influenza vaccine <14 days prior to illness onset; or (7) were missing covariate data used in the VE model. For those that remained after exclusion, we compared patient characteristics by case status using Chi-square tests for categorical variables and Mann-Whitney U tests for continuous variables. We tested for differences in median vaccination date among age and immunocompetence groups using Kruskal-Wallis and Mann-Whitney U tests, respectively. We tested for differences in vaccination timing (September and earlier vs October and later) and influenza subtype by age group using Chi-square tests.

We used a test-negative design to estimate influenza VE. Our primary outcome was influenza-associated hospitalization, defined as hospital admission for symptomatic, laboratory-confirmed influenza virus infection. We first fit a multivariable logistic regression model to compare the odds of vaccination among cases and controls while accounting for relevant covariates. Models were adjusted for age group (18–49, 50–64, ≥65 years), sex (male/female), race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Hispanic, Other, Unknown), HHS region (Regions 1–10), and biweekly period of enrollment. We then calculated VE for protection against influenza-associated hospitalization as (1 – adjusted odds ratio for vaccination) × 100%. We calculated point estimates and 95% confidence intervals for overall VE and VE stratified by influenza virus, age group, and immunocompetence status using this model.

We additionally calculated VE estimates by time since vaccination by substituting the exposure variable in our regression model with a time variable (unvaccinated, and vaccinated 14–60 days, 61–120 days or >120 days before illness onset for overall VE; unvaccinated, and vaccinated 14–120 days or >120 days before illness onset for subtype-specific VE) and compared intervals by VE 95% confidence interval overlap. Analyses were performed using SAS version 9.4 (Cary, North Carolina, USA)⁵ and R version 4.4.1 (R Core Team, Vienna, Austria).⁶

Overall, 13,041 patients were enrolled between September 1, 2023, and May 31, 2024. Of these, 5,351 were excluded from the analysis. Most excluded patients tested positive for SARS-CoV-2 (n=3,112, 58%), had incomplete data for influenza vaccination status (n=817, 15%), or received influenza testing before the illness-onset date or more than 10 days after the illness-onset date (n=693, 13%) (Figure 1). A total of 7,690 patients comprised the analytic data set, including 1,170 influenza cases and 6,920 controls.

Among 1,170 influenza-case patients, influenza type was determined by RT-PCR for 762 patients, including 465/762 (61%) with influenza A(H1N1) and 228/762 (30%) with influenza A(H3N2), and 69/762 (9%) with influenza B. Influenza A(H1N1) and A(H3N2) were distributed similarly across age groups, but most influenza B cases (49/69, 71%) were among patients aged 18–49 years (Supplementary Figure 1). Among the 465 patients with influenza A(H1N1) confirmed by RT-PCR, viral WGS identified a clade in 441 patients, of which 343 (78%) were genetic group 6B.1A.5a.2a.1 and 98 (22%) were genetic group 6B.1A.5a.2a. Among 228 patients with influenza A(H3N2), WGS identified a clade in 200 patients, of which all were genetic group 3C.2a1b.2a.2a.3a.1. Among 69 patients with influenza B, WGS identified a clade in 51 patients, of which all were genetic group V1A.3a.2.

Influenza vaccination was less likely among influenza case patients (391/1170; 33%) compared with control patients (2779/6520; 43%) (P<.001) (Table 1). Case patients, compared with control patients, were younger (median age: 62 vs 65 years, P<.001), more likely to be non-Hispanic Black (28% vs 22%; P<.001), less likely to be immunocompromised (18% vs 25%, P<.001) and less likely to have had ≥1 hospitalization in the previous year (50% vs 61%, P<.001). Patients who received influenza vaccination, compared with unvaccinated patients, were older (median age: 69 vs 61 years, P<.001), more likely to be non-Hispanic White (64% vs 50%; P<.001), more likely to be immunocompromised (28% vs 21%; P<.001), and more likely to have had ≥1 hospitalization in the previous year (65% vs 56%; P<.001) (Supplementary Table 1). Patients aged ≥65 years (compared with other age groups) were vaccinated a median 5 days earlier in the year and were more likely to be vaccinated before October (Figure 2). Immunocompromised patients (compared with immunocompetent patients) were also vaccinated slightly earlier in the year (Supplementary Figure 2).

The overall VE against influenza-associated hospitalization was 40% (95% confidence interval [CI]: 31%–48%). Stratified by age group, VE declined with age and was 53% (95% CI: 34%–67%) among adults aged 18–64 years, 47% (95% CI: 31%–60%) among adults aged 50–64 years, and 31% (16%–43%) among adults aged ≥65 years (Figure 3). By immunocompromised status, VE was 40% (95% CI: 30%–49%) for immunocompetent patients and 32% (95% CI: 7%–59%) for immunocompromised patients. Stratified by virus subtype, VE against influenza-associated hospitalization was 36% (95% CI: 21%–48%) for influenza A(H1N1), 19% (95% CI: −8%–39%) for influenza A(H3N2) and 68% (95% CI: 35%–84%) for influenza B (Figure 4). VE against influenza A(H1N1)–associated hospitalizations was consistent across patients aged 18–49 years (VE: 40% [95% CI: −4%–65%]), 50–64 years (VE: 38% [95% CI: 8%–58%]) and ≥65 years (VE: 35% [95% CI: 14%–51%]). However, VE against influenza A(H3N2)–associated hospitalizations crossed the null for all three age groups (18–49 years: 40% [95% CI: −20%–70%]); 50–64 years: 30% [95% CI: −23%–60%]; ≥65 years: 2% [95% CI: −47%–35%]).

VE against all influenza–associated hospitalizations differed by days elapsed between vaccination and enrollment, with the highest VE among those vaccinated 14–60 days prior to illness onset (VE: 54% [95% CI: 40%–65%]) and a significant decrease among patients vaccinated >120 days prior to onset (VE: 18% [95% CI: −1%–33%]) (Figure 5). Among patients vaccinated 14–120 days prior to onset (median: 68 days [IQR: 42–93 days]), VE was significant against influenza A(H1N1) (42% [95% CI: 32%–55%]), and B (VE: 72% [95% CI: 29%–89%]) but not influenza A(H3N2) (V28% [95% CI: −4%–49%]). Among patients vaccinated >120 days prior to onset (median: 159 days [IQR:140–188 days]), VE crossed the null against A(H1N1) (13% [95% CI: −24%–39%]), A(H3N2) (−2% [95% CI: −52%–31%]), and B (62% [95% CI: −2%–86%]).