Influenza clinical testing and oseltamivir treatment in hospitalized children with acute respiratory illness, 2015–2016

Abstract Background Antiviral treatment is recommended for all hospitalized children with suspected or confirmed influenza, regardless of their risk profile. Few data exist on adherence to these recommendations, so we sought to determine factors associated with influenza testing and antiviral treatment in children. Methods Hospitalized children <18 years of age with acute respiratory illness (ARI) were enrolled through active surveillance at pediatric medical centers in seven cities between 11/1/2015 and 6/30/2016; clinical information was obtained from parent interview and chart review. We used generalized linear mixed‐effects models to identify factors associated with influenza testing and antiviral treatment. Results Of the 2299 hospitalized children with ARI enrolled during one influenza season, 51% (n = 1183) were tested clinically for influenza. Clinicians provided antiviral treatment for 61 of 117 (52%) patients with a positive influenza test versus 66 of 1066 (6%) with a negative or unknown test result. In multivariable analyses, factors associated with testing included neuromuscular disease (aOR = 5.35, 95% CI [3.58–8.01]), immunocompromised status (aOR = 2.88, 95% CI [1.66–5.01]), age (aOR = 0.93, 95% CI [0.91–0.96]), private only versus public only insurance (aOR = 0.78, 95% CI [0.63–0.98]), and chronic lung disease (aOR = 0.64, 95% CI [0.51–0.81]). Factors associated with antiviral treatment included neuromuscular disease (aOR = 1.86, 95% CI [1.04, 3.31]), immunocompromised state (aOR = 2.63, 95% CI [1.38, 4.99]), duration of illness (aOR = 0.92, 95% CI [0.84, 0.99]), and chronic lung disease (aOR = 0.60, 95% CI [0.38, 0.95]). Conclusion Approximately half of children hospitalized with influenza during the 2015–2016 influenza season were treated with antivirals. Because antiviral treatment for influenza is associated with better health outcomes, further studies of subsequent seasons would help evaluate current use of antivirals among children and better understand barriers for treatment.

interview and chart review. We used generalized linear mixed-effects models to identify factors associated with influenza testing and antiviral treatment.
Results: Of the 2299 hospitalized children with ARI enrolled during one influenza season, 51% (n = 1183) were tested clinically for influenza. Clinicians provided antiviral treatment for 61 of 117 (52%) patients with a positive influenza test versus 66 of 1066 (6%) with a negative or unknown test result. In multivariable analyses, factors associated with testing included neuromuscular disease (aOR = 5.35, 95% CI  [1][2][3][4] Influenza has a high attack rate in children with an estimated incidence of 19/1000 per year and an overall mortality rate of 15 deaths per 1000 influenzapositive children. 5 Children less than 5 years of age, and especially children less than 2 years of age, American Indians or Alaska Natives and those with underlying comorbidities are at higher risk for developing influenza-associated complications. 6 However, nearly 50% of hospitalized children with influenza do not have an underlying medical condition. 7 Influenza vaccination is the mainstay of prevention against influenza disease and can prevent influenza-associated complications in children 6 months of age and older. 8 If an infection is acquired, antiviral treatment for influenza disease has been shown to reduce complications, shorten the length of hospitalization, and reduce mortality; these benefits are more pronounced when treatment is initiated within 48 h of symptom onset. [9][10][11] However, variation in the prescribing patterns among clinicians exist, possibly due to concerns about effectiveness and reporting biases in industry funded trials. 12,13 Detailed recommendations by the Infectious Diseases Society of America (IDSA) for antiviral treatment for influenza disease were published in response to the 2009 pandemic and has been updated periodically. 14 The recommendations include antiviral treatment for all hospitalized individuals with confirmed or suspected influenza, regardless of underlying illness or vaccination status, and recommend initiation of treatment within 48 h of symptom onset. 14 In addition, the recommendations advise that persons hospitalized for confirmed influenza may also benefit from treatment even if initiated more than 48 h after the onset of illness. 14 The American Academy of Pediatrics (AAP) recommends treatment as soon as possible for children hospitalized with suspected influenza, hospitalized for severe, complicated, or progressive illness attributable to influenza regardless of duration of symptoms, and to children with suspected influenza and at increased risk of complications. It also recommends considering treatment for any healthy child with suspected influenza, and to healthy children with suspected influenza who live with a household contact who is <6 months old or has a medical condition that predisposes them to complications. 15 Despite these recommendations, controversy among physicians exists and suboptimal antiviral use has been reported in recent studies. 16,17 Clinical testing for influenza illness may influence antiviral treatment; however, based on updated 2018 IDSA recommendations, antiviral treatment decisions should not be delayed until laboratory confirmation of influenza. 6 Data are limited on factors associated with making clinical decisions whether to test for and/or to treat influenza. Children were excluded if they had a known nonrespiratory cause for hospitalization, had fever and neutropenia with malignancy, were discharged from a hospital in the prior 4 days, were transferred after admission at another hospital for 48 h, had never been discharged home after birth, or had previously enrolled in this study <14 days prior to their current admission. 18 For this study, we included children who were hospitalized during each site's influenza season defined as the date of first through the last influenza positive case for each site based on research testing results ( Figure 1A). We excluded 24 children who received influenza antiviral treatment prior to hospitalization ( Figure 1A).

| Study design
Following written informed consent from a parent or guardian and assent when applicable, demographic data, history of current illness, social history and treatment received before presentation were collected through parent/guardian interviews. Standardized medical chart reviews were performed, and clinical interventions and outcome data were collected including chronic comorbid conditions, types and results of clinical influenza diagnostic studies performed, antiviral treatment, intensive care unit (ICU) admissions, and oxygen requirement.
Institutional Review Board approval was obtained from the CDC and at each individual site.

| Influenza season
Influenza season was defined as the period between the dates of the first through last influenza positive case for each specific site, based upon research laboratory testing. Research laboratory diagnostic influenza assays varied by site, but all were nucleic acid amplification tests (NAATs) for which CDC-generated influenza proficiency panels were successfully completed. 19 Influenza clinical testing was defined as any influenza testing that was ordered by providers. Clinical testing was available for those subjects whose treating provider ordered testing from the clinical laboratory of their respective hospital as part of standard care; the method of clinical laboratory testing was either rapid influenza diagnostic testing (RIDT) or NAATs. Positive test results were defined according to results documented in medical charts for influenza A, influenza A/(H1N1)pdm09, influenza A(H3N2), and influenza B lineage viruses.

| Antiviral use
Influenza antiviral use was defined as in-hospital receipt of a neuraminidase inhibitor (oseltamivir or zanamivir) or adamantane (amantadine or rimantadine) documented by chart review.

| Influenza vaccine reporting
Receipt of influenza vaccine was determined by parental report of receiving influenza vaccination for the current season for children who were 6 months or older. Demographic and clinical characteristics were evaluated using descriptive statistics (frequency and percentage for categorical variables, or mean and standard deviation for continuous variables).

| Data analysis
Between-group comparisons were performed using Pearson's chi-squared test for categorical variables and two-sample t tests of mean differences for continuous variables.
We used a generalized linear mixed-effects model on the logodds scale to evaluate factors associated with influenza testing and antiviral treatment, separately. 20 To address missing data, we used multiple imputation via chained equations with M = 500 iterations, aggregating results using Rubin's rules. 21,22 The following predictors were included a priori in each of the two models: continuous age (years), sex, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other), fever, cough, fever & cough, duration of illness prior to admission, chronic lung disease, neuromuscular disease, immunocompromised status, congenital heart disease, influenza vaccination, and insurance status (public, private, both, and selfpay). 2,12,23,24 We included a random intercept for each study site.
From these models, we estimated adjusted odds ratios for each predictor and derived corresponding Wald-based 95% confidence F I G U R E 1 (A) Study cohort including total admissions due to acute respiratory illnesses and excluded subjects. (B) Acute respiratory illness admissions: total and during influenza season intervals and p-values. Statistical significance was determined at the nominal α = 0.05 level (two-sided).

| Patient characteristics
Among the 3926 enrolled children who were hospitalized with ARI or febrile illness between July 2015 and June 2016, 2299 (58%) met eligibility criteria for this analysis ( Figure 1A). The total duration of influenza season by each site is represented in Figure 1B had symptoms for ≤2 days prior to hospitalization.
Moreover, children who were tested for influenza were more likely to receive antiviral treatment (tested vs. not tested: 127/1183 (11%) vs. 22/1183 (2%), p < 0.001). Additionally, children who tested positive were more likely to be treated (positive vs. nonpositive 61/117 (52%) vs. 66/1066 (6%), p < 0.001). were treated with an antiviral. 24 In contrast, during the 2009-H1N1 pandemic, a study noted that 77% of children hospitalized with influenza received an antiviral but a 27% decline was reported the following year. 17 The FluSurv-NET subsequently reported that 72% of 6469 hospitalized children with confirmed influenza received antivirals between 2010 and 2015. 16 Only evaluating populations with positive clinical testing compared to broader populations that might be eligible for antiviral treatment may overestimate the antiviral coverage. 27 Our findings indicate that additional efforts are needed to increase awareness of antiviral effectiveness and current empiric influenza treatment recommendations in hospitalized children with suspected influenza without delay for testing results.

| DISCUSSION
Approximately half of the enrolled children had a providerinitiated influenza test with the majority having a NAAT performed, which is the recommended test for this population. 28 Also our study showed an association between testing, testing results, and receiving treatment. Historically, RIDT was the most commonly used diagnostic test. 2,24,29 While RIDT sensitivity is higher in children than adults, sensitivity in children is estimated to be 67%. 30 During the 2009 H1N1 pandemic, RIDT was associated with a higher false negative rate and clinicians were directed to start antiviral treatment if influenza was highly suspected despite a negative result. 14 NAAT for influenza is more sensitive compared to RIDT and is currently more widely used in point of care testing. 30,31 Although testing for influenza is not required to initiate antiviral treatment, availability of accurate and timely diagnostic tests represents a potential challenge to antiviral use. One study in adults hospitalized with influenza found that 26% of those testing positive for influenza by a provider were treated compared to 5% of those testing negative and to less than 1% of those not tested; 24% of those treated with antivirals were not tested. 27 Despite the limited sensitivity of RIDT, a study among children seeking care in the Emergency Department showed that a positive RIDT was associated with increased antiviral use. 32