Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Keith Kaye

General ID · RU

View

Dr. Rochelle Walensky

General ID · Harvard Medical School

View

Dr. Erin Bonura

General ID · OHSU

View

Dr. Nizar Wehbi

General ID

View

Dr. Robert Schooley

General ID · UCSD

View

Dr. Priya Nori

General ID · MMC

View

Dr. Jan Patterson

General ID · UTHSCSA

View

Dr. Kavita Bhavan

General ID · Dallas County Hospital District

View

Dr. John Goldman

General ID · UPMC Central PA

View

Dr. Gonzalo Bearman

General ID · Medical College of Virginia

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Vaccine Effectiveness Against Pediatric Influenza-A-Associated Urgent Care, Emergency Department, and Hospital Encounters During the 2022-2023 Season: VISION Network.

    Mar 20, 2024DOI: 10.1093/cid/ciad704Journal ArticleCID

    Written by: Katherine AdamsZachary A WeberDuck-Hye YangNicola P KleinMalini B DeSilvaKristin DascombStephanie A IrvingAllison L NalewaySuchitra RaoManjusha GaglaniBrendan FlanneryShikha GargAnupam B KharbandaShaun J GrannisToan C OngPeter J EmbiKarthik NatarajanBruce FiremanOusseny ZerboKristin GoddardJulius TimbolJohn R HansenNancy GriselJulie ArndorferSarah W BallMargaret M DunneLindsey KirshnerJessie R ChungMark W Tenforde

    Find these authors on Roon

    SG
    Dr. Shikha Garg
    General ID · VAMC

    METHODS

    Pediatric electronic health record (EHR) data were included from 55 hospitals and 107 emergency department (ED) or urgent care (UC) sites across 3 participating VISION partner healthcare systems: Kaiser Permanente Northern California, Intermountain Healthcare in Utah, and HealthPartners in Minnesota and Wisconsin. Methods used by the VISION Network to evaluate VE have been described [16]. This study protocol was reviewed and approved by Institutional Review Boards (IRBs) at participating sites or under a reliance agreement with the Westat IRB and by CDC.A test-negative case-control study design was used to estimate seasonal influenza VE against influenza A virus illness among children and adolescents aged 6 months–17 years with acute respiratory illness (ARI)–associated ED/UC encounters or hospitalizations during the 2022–2023 influenza season. Influenza B viruses could not be assessed because of low levels of circulation during the season [5]. International Classification of Diseases, 10th Revision (ICD-10) discharge diagnosis codes were used to develop an ARI case definition, which included clinical diagnoses such as pneumonia or influenza, as well as signs and symptoms of ARI such as cough or shortness of breath (Supplementary Table 1). All eligible pediatric patients had an ARI-associated encounter with molecular testing for SARS-CoV-2 and influenza viruses within 10 days prior to or until 72 hours after the encounter. Influenza case-patients were defined as those with a positive influenza A molecular test, while influenza-negative patients were those with a negative test for influenza A and B.Demographic and clinical data were extracted from EHRs using a standardized data dictionary. If a hospitalized patient was readmitted within 30 days of discharge or a repeat visit to an ED/UC occurred within 24 hours, these encounters were treated as a single event. Start dates varied by study setting due to low numbers of reported inpatient influenza cases. The ED/UC encounters were included at all sites starting on 16 October 2022 (ie, the beginning of the Morbidity and Mortality Weekly Report [MMWR] surveillance week occurring 2 weeks following the start of the US influenza season); hospitalizations were included starting on the date of the first influenza case in the study population at each site (date range = 19 October–6 November 2022) (Supplementary Table 2). Site- and setting-specific end dates used a threshold of when weekly influenza test positivity fell below 2%.Patients aged 6 months–17 years eligible to receive a seasonal influenza vaccine were included. Encounters were excluded if the patient did not receive both influenza and SARS-CoV-2 molecular tests within 10 days before or up to 72 hours after their ED/UC encounter or hospitalization date, if either test result was unknown or indeterminate, or if the patient tested positive for influenza B. Patients who received influenza vaccination 1–13 days prior to the index event date (ie, earliest date of most recent influenza test result or ED/UC encounter/hospital admission) were excluded from analysis. Encounters assigned an influenza-specific ICD-10 diagnostic code but which also had a negative influenza molecular test were excluded because the influenza diagnosis occurred in the absence of laboratory confirmation. Additionally, patients with an ICD-10 code indicating COVID-19 (Supplementary Table 1) without molecular test confirmation or those with positive SARS-CoV-2 test results were excluded to reduce potential confounding due to correlation between influenza and COVID-19 vaccination decisions [17].Vaccination status was ascertained using EHR, state or local immunization information system, and claims data from the 2022–2023 influenza season. Patients were classified as vaccinated if they received 1 or more influenza vaccine doses beginning 1 August 2022, and at least 14 days prior to their index event date, and unvaccinated if no record of current season vaccination was found or if the vaccine administration date occurred after the index date. Since vaccination history prior to the 2022–2023 influenza season was not routinely available, we were unable to ascertain whether children aged 6 months–8 years were fully vaccinated [18].We calculated the number of weekly cases of influenza and percent positivity by site and care setting (ED/UC or hospitalization). Demographic characteristics, stratified by vaccination status and influenza test results, were described using frequencies and proportions for binary or categorical variables and medians and interquartile ranges (IQRs) for continuous variables. Standardized mean differences (SMDs) were reported to allow comparison between groups, with an SMD of 0.20 or greater indicating a nonnegligible difference in distribution [19]. To characterize patterns in vaccination coverage across study months, frequencies and proportions of vaccinated patients were calculated and stratified by influenza test results. Vaccine effectiveness was estimated by comparing the odds of 2022–2023 seasonal influenza vaccination among influenza-A–positive patients versus influenza-negative patients, calculated as VE = (1 – adjusted odds ratio [OR]) × 100%. ORs were generated using multivariable logistic regression, adjusted for patient age, study site, and calendar day. Natural cubic splines were used for age and calendar day. To balance baseline site, demographic, and underlying medical condition characteristics, inverse probability of treatment (ie, vaccination) weighting was performed using generalized boosted regression trees and truncated at the 99th percentile. Unadjusted and adjusted VE estimates are presented for each setting (ED/UC encounters and hospitalizations), along with 95% confidence intervals (CIs). Vaccine effectiveness was not examined by subtype (A/H3N2 vs A/H1N1pdm09) because subtype information was not available for most influenza-A–positive cases.To explore whether different inclusion criteria changed VE estimates among young children, 2 sensitivity analyses were performed. First, inclusion criteria for patients younger than 2 years of age, who may present more often with nonrespiratory signs or symptoms of influenza, were expanded to include ICD-10 codes indicating fever and COVID-19–like illness (CLI) (eg, diarrhea, myalgia, and nausea/vomiting; full expanded ICD-10 CLI definition in Supplementary Table 1) [20]. Second, influenza-positive cases were restricted to those with influenza-specific ICD-10 discharge codes for influenza pneumonia or influenza disease (Supplementary Table 1). This latter analysis was done to remove patients in whom influenza was not identified as a contributing diagnosis during the encounter using a more specific case definition. All analyses were performed using SAS version 9.4 (SAS Institute) or R 4.2.1 (R Foundation for Statistical Computing).

    RESULTS

    Pediatric influenza A cases among both ED/UC encounters and hospitalizations at participating sites peaked during the week ending 10 December 2022, and remained high until late December (Supplementary Figure 1). Site-level influenza test percent positivity for ED/UC encounters dropped below 2% after 21 January 2023 and through 4 March 2023. Among hospitalizations at sites with sufficient data to ascertain trends, influenza test percent positivity dropped below 2% after 4 February 2023. Of 507 influenza-A–positive cases with information on subtype (all from a single site), 419 (82.6%) were A/H3N2 and 88 (17.4%) were A/H1N1pdm09 viruses.Among influenza-positive cases across care settings (ED/UC encounters and hospitalizations), the most common ARI discharge diagnosis codes included those for influenza disease (63.0%) and upper respiratory tract infection (32.2%) (Supplementary Table 3). Among influenza-negative patients, the most common discharge codes were for upper respiratory tract infection (52.7%) and acute respiratory signs and symptoms (26.6%).Of 62 207 pediatric ARI-associated ED/UC encounters, 56 898 (91.5%) occurred during periods of local influenza circulation (ie, where percent positivity among those tested for influenza was >2%). Of these encounters, 12 111 (21.3%) were excluded, most commonly due to lack of influenza molecular testing (n = 7311), having a negative influenza test result and a positive SARS-CoV-2 test result (n = 1833), and influenza vaccination 1–13 days before the index date (n = 1583) (Supplementary Figure 2).A total of 44 787 eligible ED/UC encounters were included in the analysis; 13 547 (30.2%) were influenza A case-patients and 31 240 (69.8%) were influenza-negative patients (Table 1). Overall, 10 517 (23.5%) ED/UC encounters occurred in vaccinated patients; 15.2% of influenza-positive patients versus 27.1% of influenza-negative patients had been vaccinated; and median time since vaccination was 52 days (IQR, 33–75 days). The proportion of vaccinated influenza-negative ED/UC encounter patients ranged across sites from 12.8% to 16.0% within the first week of data capture and increased to 33.3–44.1% during the final week, and was higher in 0–4-year-olds than in 5–17-year-olds (Supplementary Table 4). Among 6848 (65.1%) vaccinated patients with a documented influenza vaccine product, 95.6% received egg-based standard-dose quadrivalent inactivated influenza vaccine (IIV4), 2.0% received a live attenuated influenza vaccine (LAIV), and 2.4% received cell culture–based IIV4. The median patient age was 4 years (IQR, 2–8 years), with vaccination coverage highest in patients aged 6 months–4 years (27.9%) and lowest in those aged 9–17 years (17.9%). Of ED/UC encounters, 53.2% occurred in male patients, 56.5% of patients were White, and 69.9% were non-Hispanic. Of ED/UC patients, 9.8% had an underlying respiratory or non-respiratory medical condition, and 0.4% had a likely immunocompromising condition documented at the index encounter.The adjusted VE against influenza-A–associated pediatric ED/UC encounters was 48% (95% CI, 44–52%) overall (Figure 1). Vaccine effectiveness was higher among younger patients: 53% (95% CI, 47–58%) for those aged 6 months–4 years, 46% (95% CI, 38–53%) among 5–8-year-olds, and 38% (95% CI, 30–45%) among 9–17-year-olds. Expanding the inclusion criteria to nonrespiratory signs and symptoms for patients aged 6 months–2 years or restricting influenza-positive patients to those with influenza pneumonia or influenza disease discharge codes produced similar VE estimates as the primary analysis (Supplementary Figures 4 and 5).Of 2632 pediatric ARI-associated hospitalizations identified, 2048 (77.8%) occurred during periods of local influenza circulation at participating VISION sites (Supplementary Figure 3). Of these, 186 (9.1%) were excluded from the primary analysis, most frequently due to influenza vaccination 1–13 days before the index date (n = 63), having a positive COVID-19 test result (n = 53), and having a negative influenza test result but an influenza disease/pneumonia or COVID-19 pneumonia diagnosis (n = 37).From 19 October 2022 to 18 February 2023, 1862 ARI-associated hospitalizations were eligible for the analysis, including 263 (14.1%) influenza A case-patients and 1599 (85.9%) influenza-negative patients (Table 2). Of influenza-associated hospitalizations, 31.1% occurred among vaccinated patients (17.5% of influenza case-patients vs 33.4% of influenza-negative patients had been vaccinated), with a median time since vaccination of 54 days (IQR, 36–80 days). The proportion of vaccinated influenza-negative hospitalized pediatric patients ranged across sites from 11.7% to 22.3% during the first week of data capture and expanded to 36.4–66.7% during the last week (Supplementary Table 5). Among 397 (68.4%) patients with a documented influenza vaccine product, 95.7% received egg-based standard dose IIV4, 0.5% received LAIV, and 3.8% received cell culture–based IIV4. The median age was 2 years (IQR, 1–5 years), with the highest proportion vaccinated among children aged 6 months–4 years (34.2%) and lowest among those aged 9–17 years (22.3%). Of hospitalizations, 53.1% occurred in males, 59.2% were White, 75.1% were non-Hispanic, and a majority had 1 or more underlying medical conditions (57.4%). Among 67 (3.6%) hospitalized patients who received invasive mechanical ventilation, 11 (16.4%) tested positive for influenza and 21 (31.3%) were vaccinated. Among 207 (11.1%) hospitalized patients who were admitted to an intensive care unit, 28 (13.5%) tested positive for influenza and 76 (36.7%) were vaccinated. Among 7 (<1%) patients who died in the hospital, 1 patient (14.3%) tested positive for influenza and 3 patients (42.9%) were vaccinated.The overall adjusted VE against influenza-A–associated pediatric hospitalizations was 40% (95% CI, 6–61%) (Figure 2). In models stratified by age group, the VE point estimate was higher among patients aged 6 months–4 years (56%; 95% CI, 23–75%) versus patients aged 5–17 years (46%; 95% CI, 2–70%), although estimate precision was limited and CIs over-lapped across age groups. Both sensitivity analyses produced similar estimates to the primary analysis but with limited precision (Supplementary Figures 6 and 7).

    DISCUSSION

    During the 2022–2023 season with predominant vaccine-matched influenza A/H3N2 virus circulation, seasonal influenza vaccination reduced the risk of influenza-A–associated ED/UC visits and hospitalizations by almost half across ED/UC (48% VE) and inpatient (40% VE) settings. VISION Network VE estimates against both influenza-A–associated ED/UC encounters and hospitalizations in pediatric age groups were similar to early-season estimates from other US-based networks for the 2022–2023 influenza season [21, 22]. VISION Network estimates also fall within the range of estimates against ED and/or UC encounters (46–56%) and hospitalizations and critical illness (41–63%) observed during recent seasons [14, 23-26]. However, despite the protection offered by seasonal influenza vaccines, vaccination coverage among children and adolescents in our analysis was relatively low. These findings align with other data finding a low proportion of children and adolescents vaccinated during the 2022–2023 season compared with prior seasons [27, 28]. To maximize the clinical and public health benefits of vaccination, continued efforts to optimize influenza vaccine uptake among children and adolescents are needed.Vaccine effectiveness against influenza-A–associated ED/UC encounters tended to be higher among younger children (6 months–4 years) than older children and adolescents (9–17 years). These results are notable considering our analysis of a single dose of current season influenza vaccine and low circulation of influenza during the past 2 seasons due to the COVID-19 pandemic. Prior exposure to influenza viruses among very young children may have been limited and reduced adaptive immune responses or influenced VE by other mechanisms [29-31]. While point estimates of VE against influenza-associated hospitalizations were higher in younger versus older age groups, this comparison was limited due to a smaller sample size.Both influenza VE and coverage directly impact illnesses, medical visits, and hospitalizations averted by vaccination programs [15]. Lower vaccine coverage diminishes the potential vaccine impact on disease burden. Children and adolescents also experience high influenza attack rates and can accelerate disease transmission among households and close contacts [1, 2]. Pediatric influenza can additionally result in substantial burden through school absenteeism, economic and productivity loss for caregivers, and direct medical expenses (eg, medical visits, over-the-counter and prescription medications) [32-35]. The atypically early start of the 2022–2023 influenza season may have resulted in lower observed vaccine coverage, as activity started and peaked before many children and adolescents may have had an opportunity to get vaccinated. Vaccination during July and August can be considered for children needing 1 dose of vaccine; children aged 6 months–8 years requiring 2 doses to be fully vaccinated should receive the first dose as soon as possible, including July and August [10, 36].This analysis is subject to several limitations. First, we were unable to assess VE by influenza A virus subtype (A/H3N2 or A/H1N1pdm09) because VISION relies on clinician-driven testing and influenza subtype was not included in most molecular assays. Second, as influenza testing is clinician-driven and may be influenced by clinical presentation, illness severity, and prior testing, variation in testing practices across VISION sites may have occurred. To mitigate this bias, we used a specific definition of ARI for eligibility, and all included patients had both influenza and SARS-CoV-2 testing performed. Among pediatric ARI encounters, 87.2% of ED/UC encounters and 98.8% of hospitalizations included molecular influenza testing, suggesting adequate capture of influenza using our ARI definition. Furthermore, our sensitivity analyses using different inclusion criteria produced estimates similar to our primary analysis. Third, selection bias may have been introduced if care-seeking differs between those with influenza-associated ARI and those with non-influenza ARI. However, others have found that this potential bias in influenza VE test-negative studies using an ARI definition is unlikely to be meaningful [37]. Fourth, influenza vaccination history across seasons was not routinely collected for the determination of partially or fully vaccinated children. However, our VE estimates for younger age groups are similar to other studies of known fully vaccinated pediatric patients [26, 38]. Additionally, the 2022–2023 influenza season began in early October 2022, reducing the opportunity for young children to receive 2 doses of vaccine 4 weeks apart and be fully vaccinated 2 weeks past the second dose. Fifth, although we captured encounters from geographically diverse health systems across 4 states, patients included in this analysis may not represent the overall US population and patients seeking care in other health systems.

    CONCLUSIONS

    In this analysis, receipt of the 2022–2023 seasonal influenza vaccine reduced the risk of influenza-A–associated ED or UC visits and hospitalizations by almost half. Influenza vaccination remains a critical public health tool for the prevention and mitigation of serious illness among children and adolescents and can reduce strain on healthcare resources, especially during periods of high co-circulation of influenza, SARS-CoV-2, RSV, and other respiratory viruses.
    SG

    Dr. Shikha Garg

    Infectious Disease · General Infectious Disease
    Atlanta VA Medical Center
    shikhag