|Home | About | Journals | Submit | Contact Us | Français|
To characterize the health care burden of influenza from 2004 through 2009, years when influenza vaccine recommendations were expanded to all children aged ≥6 months.
Population-based surveillance for laboratory-confirmed influenza was performed among children aged <5 years presenting with fever and/or acute respiratory illness to inpatient and outpatient settings during 5 influenza seasons in 3 US counties. Enrolled children had nasal/throat swabs tested for influenza by reverse transcriptase-polymerase chain reaction and their medical records reviewed. Rates of influenza hospitalizations per 1000 population and proportions of outpatients (emergency department and clinic) with influenza were computed.
The study population comprised 2970, 2698, and 2920 children from inpatient, emergency department, and clinic settings, respectively. The single-season influenza hospitalization rates were 0.4 to 1.0 per 1000 children aged <5 years and highest for infants <6 months. The proportion of outpatient children with influenza ranged from 10% to 25% annually. Among children hospitalized with influenza, 58% had physician-ordered influenza testing, 35% had discharge diagnoses of influenza, and 2% received antiviral medication. Among outpatients with influenza, 7% were tested for influenza, 7% were diagnosed with influenza, and <1% had antiviral treatment. Throughout the 5 study seasons, <45% of influenza-negative children ≥6 months were fully vaccinated against influenza.
Despite expanded vaccination recommendations, many children are insufficiently vaccinated, and substantial influenza burden remains. Antiviral use was low. Future studies need to evaluate trends in use of vaccine and antiviral agents and their impact on disease burden and identify strategies to prevent influenza in young infants.
Influenza is an important cause of medically attended illness in young children before expanding influenza vaccine recommendations for children.
This study characterizes the health care burden of influenza in young children over 5 years (2004–2009) when influenza vaccine recommendations were expanded to all children aged ≥6 months.
Influenza remains a major cause of medically attended respiratory illness in young children. This has led to the Advisory Committee on Immunization Practices of the Centers for Disease Control and Prevention (CDC) expanding influenza vaccination recommendations from children ≥6 months old with underlying medical conditions, to all children 6 to 23 months old in 2004,1 to all children 6 to 59 months old in 2006,2 and to all children ≥6 months in 2008.3
It is hoped that expanding vaccine recommendations would result in rising immunization coverage and declining disease burden, yet it is challenging to evaluate trends in disease burden of influenza. The circulating influenza virus strains and their match to vaccine strains and the timing and severity of the influenza season vary annually. Therefore, assessment of the disease burden over multiple years is needed to estimate the impact of new strategies on the burden of influenza. Previously, we evaluated the overall burden of influenza in children enrolled in the New Vaccine Surveillance Network (NVSN) from 2000 through 2004 and found an average of 0.9 (range 0.4–1.5) influenza hospitalizations per 1000 children aged <5 years annually. Among outpatients, influenza caused 8% and 22% of acute respiratory illness and/or fever visits during the 2002–2003 and 2003–2004 influenza seasons, respectively.4
Since 2004, vaccination recommendations have expanded, and improved diagnostic tests and antiviral therapies have become available.1–3 Thus, an evaluation of the population-based burden of influenza is essential for assessing these control strategies. This report uses NVSN data from 2004 through 2009 to determine the burden of laboratory-confirmed influenza, vaccination coverage, and the frequency of physician-ordered influenza-specific tests, influenza-specific discharge diagnoses, and antiviral prescriptions.
We prospectively enrolled children aged <5 years who resided in Hamilton County, Ohio (Cincinnati); Monroe County, New York (Rochester); and Davidson County, Tennessee (Nashville) from 2004–2005 through 2008–2009, all before the 2009 influenza A (H1N1) pandemic in this network. County children who were hospitalized during influenza season (generally November–April) with fever and/or acute respiratory symptoms or who presented to the emergency department (ED) or outpatient clinic with these symptoms were eligible for enrollment. Surveillance was population-based in the inpatient and ED settings.4,5 After obtaining informed consent, a standardized parental questionnaire was administered, combined nasal turbinate and throat swabs were collected and stored for each child to identify laboratory-confirmed influenza illness,6 and the child’s medical record was reviewed.
Population-based inpatient surveillance included all 3 sites during the 2004–2005 and 2008–2009 seasons, and 2 sites (Rochester and Nashville) during the 3 intervening seasons. A protocol deviation affected 70% of all hospitalized children enrolled at the Cincinnati site from 2005–2008; thus all hospitalized children enrolled in Cincinnati during these 3 years, including those not affected by this protocol deviation, were excluded (Fig 1). Surveillance was performed annually from November 1 through April 30 but was extended if influenza was detected earlier (beginning with ≥2 influenza-positive samples per site per week detected in local research or hospital laboratories) or ended later (continuing until <2 positive samples were detected in 2 consecutive weeks). In 2004–2005 inpatient surveillance increased from 4 to 7 days/week when influenza was detected. From 2005–2006 through 2008–2009, inpatient surveillance increased from 2 to 5 days per week when influenza was detected.
In 2004–2005, samples were collected in the pediatric ED in Cincinnati; in 3 pediatric clinics (1 hospital-based and 2 private practices) in Nashville; and in the pediatric ED, 5 pediatric clinics (2 hospital-based and 3 private practices), and 1 family medicine clinic in Rochester. In all other years, all sites collected samples in the pediatric ED and 1 hospital-based, pediatric clinic. Surveillance in the EDs occurred every fourth day in 2004–2005 and 2005–2006. Surveillance was increased from 1 to 2 days/week when influenza was detected locally from 2006–2007 through 2008–2009 in the ED and for all years in the clinic.
Race/ethnicity and high-risk medical conditions for which influenza vaccine are specifically recommended were determined by parental report and/or chart review.1–3,7,8 The presence of asthma included any history of physician diagnosed asthma or reactive airway disease. Discharge diagnoses were determined by medical chart review not billing codes.
The child’s influenza vaccination status was obtained at enrollment by parental report and confirmed by vaccination records from health care providers in all sites and also through the Ohio State Immunization Registry, which served as the vaccination record for some practices in Cincinnati.9 Fully vaccinated children had received either 2 doses of influenza vaccine ≥24 days apart and ≥14 days before illness onset in that season or 1 dose of influenza vaccine that season and had been previously vaccinated. Partially vaccinated children had received only 1 of 2 recommended doses ≥14 days before illness onset and had not been previously vaccinated. Unvaccinated children had not received any influenza vaccine that season or had received 1 dose <14 days before illness onset.
All study children had nasal/throat swabs tested for influenza A and B by reverse transcriptase-polymerase chain reaction (RT-PCR), and many had viral cultures.4,10,11 Study-protocol-confirmed influenza A or B virus infections were defined as either 2 consecutive positive RT-PCRs or positive culture.6 Indeterminate RT-PCR had no detectable human RNA, indicating an inadequate sample. Because viral subtyping was performed on only a few samples, subtyping data were not analyzed. Study results were not shared with families or physicians providing patient care.
The proportions of enrolled children by laboratory-confirmed influenza status and demographic or clinical characteristics were assessed by using χ2 analyses, Fisher exact tests, or Cochran-Armitage test for trend. The 95% confidence intervals for proportions were generated by using the Wilson confidence intervals for binomial distribution.12
To compute rates of influenza hospitalizations, the numerator was weighted for both the days of surveillance and the proportion of eligible children enrolled. The denominators were derived from the US post-2000 census estimates by age for each county and year.13 The estimated rate per 1000 children equaled the weighted number of influenza hospitalizations divided by the estimated population in surveillance counties for each year multiplied by 1000. The 95% confidence intervals for rates were generated by using 2000 bootstrap samples.14 Stata 8.1 (Stata Corp, College Station, TX) and SAS 9.2 (SAS Institute Inc, Cary, NC) were used to analyze data.
This study was approved by the institutional review boards of each study site and the CDC.
Among 3589 eligible hospitalized children, 2970 (83%) children were enrolled with adequate nasal/throat samples (Fig 1). During the 5 surveillance seasons, study protocol-confirmed influenza was detected for a median of 18 weeks (range 15–22 weeks) with the influenza-positive proportion during each peak season being similar for all sampled days of the week. Almost half (48%) of hospitalized children were <6 months of age. No significant differences existed among influenza-positive and influenza-negative children by age group, race/ethnicity, or high-risk conditions (Table 1). Significantly more influenza-positive than influenza-negative children had public insurance, were unvaccinated, and had symptoms for ≤2 days.
Among 2970 enrolled children, 7% had study protocol-confirmed influenza, ranging from 4% to 11% over 5 seasons. This proportion increased to 10% (range 4%–15%) during the peak 13 weeks of influenza activity. Among hospitalized children, 159 (80%) had influenza A and 40 (20%) had influenza B. Overall, the proportion of influenza A-positive children was 80% (range 64%–88%).
Average annual influenza hospitalization rates for children aged <5 years were 0.58 per 1000 (range 0.36–0.97, Table 2 and Fig 2). Rates of influenza hospitalizations were highest among children <6 months of age and lowest among children aged 2 to <5 years (Table 2 and Supplemental Table 4). Influenza A caused ~85% and 50% of all influenza hospitalizations among children <2 years and 2 to <5 years, respectively.
Among 3496 eligible children seen in the ED from November through April each year, 2698 (77%) children comprised the ED study population. Among 4009 eligible children seen in the outpatient clinics, 2920 (73%) comprised the clinic study population (Fig 1). All but 2 influenza-positive outpatient samples were detected from November through April each year with the median duration of influenza detection being 18 weeks (range 15–23 weeks).
Overall, 11% (range 8%–17%) of the enrolled outpatient children had study protocol-confirmed influenza with the proportion being slightly higher in the ED than the clinic (12% vs 10%, P = .05). A higher proportion of influenza-positive than influenza-negative children were ≥6 months of age, were black, resided in Rochester, were unvaccinated, and had symptoms for ≤2 days (Table 1). During the peak 13-week influenza seasons, the proportion of outpatient children <5 years who were influenza-positive was 16% (range 10%–25%, Fig 3 and Supplemental Table 4).
Among 616 influenza-positive children seen in outpatient settings, 70% had influenza A infection, 30% had influenza B infection, and <1% (n = 2) had both influenza A and B infections concurrently. The proportion of influenza A–positive children varied annually, ranging from 60% to 92% in the ED and from 53% to 91% in the clinic.
The influenza vaccination status among influenza-negative children ≥6 months of age varied by year (Fig 4 A and B), but overall, <45% of children were fully vaccinated in any setting. Partially vaccinated children were common in all settings. Among clinic children, a significant increase in fully and partially vaccinated children occurred over time.
Among inpatients with and without study protocol-confirmed influenza, treating physicians ordered influenza diagnostic testing for 35% of enrolled children, ranging from 22% in 2004–2005 to 53% in 2007–2008. Among all hospitalized children who had physician-ordered influenza tests, the annual proportion of physician-ordered influenza tests that were positive varied (range 5%–17%).
Among children with study protocol-confirmed influenza, 93% of inpatients and 98% of outpatients with physician-ordered influenza testing had rapid influenza testing performed and were positive for 66 (62%) of 107 hospitalized children and for 23 (56%) of 41 outpatient children with study protocol-confirmed influenza.
Among hospitalized children with study protocol-confirmed influenza, physician-ordered influenza testing was highest among infants <6 months (Table 3) and was more common among children with study protocol-confirmed influenza in the inpatient setting (58%) than in the ED (9%) or clinic (1%, P < .001).
Discharge diagnoses specified influenza in 35% of children hospitalized with study protocol-confirmed influenza, ranging annually from 30% to 44% (P = .65). A specific influenza diagnosis was significantly more common among infants <6 months of age than older children (Table 3) and among study protocol-confirmed influenza-positive children who also had positive rather than negative rapid influenza tests (91% vs 2%, P < .001).
Overall, discharge diagnoses specified influenza in 7% of outpatient visits with study protocol-confirmed influenza and were similar in the ED and clinic settings and across age groups. Eighteen (42%) of 43 children with a specific discharge diagnosis of influenza had a positive rapid influenza test. Children with study protocol-confirmed influenza but with negative rapid influenza tests infrequently (<1%) had a discharge diagnosis of influenza.
Antiviral medications were prescribed for 3 hospitalized influenza-positive children aged 1 to 23 months and for 3 influenza-positive children aged 16 to 29 months in the outpatient setting. One child in each setting had high-risk medical conditions. Antivirals were infrequently prescribed among all sites and years, even though 43% of influenza-positive children were hospitalized within 2 days of symptom onset.
Over 5 influenza seasons from 2004 through 2009, the overall influenza vaccination coverage changed little, whereas the rates of influenza hospitalization and prevalence of influenza among outpatients varied annually. Influenza testing among children with study protocol-confirmed influenza was highest among hospitalized infants aged <6 months. Although influenza-specific diagnoses were more common in inpatient than outpatient settings, a minority of hospitalized children with study protocol-confirmed influenza had physician-ordered testing, and <2% were prescribed antiviral therapy.
In our earlier study conducted in 2000–2004, influenza-specific discharge diagnoses among all hospitalized infants <6 months with study protocol-confirmed influenza were 28%.4 In this study, the proportion of infants <6 months with study protocol-confirmed influenza who were discharged from the hospital with an influenza-specific diagnosis increased to 48%. However, for children 6 months to <5 years of age, the proportion diagnosed with influenza remained similar in the 2 time periods (28% in 2000–2004 and 23% in 2004–2009). These data suggest that physician awareness of influenza has increased among infants <6 months of age but not among older children. Thus, it should be possible to increase the diagnosis of influenza among hospitalized children ≥6 months of age who represent half of all study children <5 years of age admitted with influenza.
During 5 study seasons (2004–2009), rates of hospitalization for influenza ranged from 0.4 to 1.0 per 1000 children aged <5 years with an annual average of 0.58 (95% confidence interval 0.50–0.67) per 1000 children. We had previously reported rates ranging from 0.4 to 1.5 with an annual average of 0.9 (95% confidence interval 0.8–1.1) per 1000 children in 2000–2004. Although the average hospitalization rate was 36% lower in 2004–2009 than in 2000–2004, there was much variability with overlapping annual rates during some years.
Although influenza vaccination coverage among hospitalized children 6 months to <5 years was higher in 2004–2009 than in earlier years,15 complete vaccination still remained <50% and did not increase appreciably among hospitalized children over the later study years. Only in the clinic population did full vaccination coverage increase significantly. National estimates using different methodologies demonstrated an increase in partial and full influenza vaccination coverage in children from 2004–2005 through 2008–2009 with wide variability. Our results fall within these reported ranges.15–27
Thus, it is difficult to determine the impact of expanded influenza vaccination recommendations on influenza disease burden beyond estimates based on vaccine effectiveness.5,28,29 Other secular trends could account for the differences in disease rates between 2000–2004 and 2004–2009. For example, the large burden of influenza in 2003–2004, predominantly influenza A H3N2 with a suboptimal vaccine match, was associated with higher disease rates and could at least partially explain this difference. Also, heptavalent pneumococcal conjugate vaccine was introduced in 2000 and has been associated with a decrease in all-cause and influenza-associated pneumonia hospitalizations among children <5 years of age during these time periods and likely affected rates of influenza disease over time.30,31 Additional study is needed to assess the impact of influenza vaccine on disease burden.
Similar to the inpatient setting, the proportions of outpatient children with influenza-positive samples varied annually for all age groups. During five 13-week peak influenza seasons in this study, the proportion of children with study protocol-confirmed influenza in either the ED or clinic ranged from 10% to 25% and was similar to the previously reported proportions of 8% in 2002–2003 and 22% in 2003–2004.4
Our results of the sensitivity of physician-ordered rapid influenza tests (66% among inpatients and 56% among outpatients) fall within the previously reported range.32–35 Rapid influenza tests can provide clinically useful results in a timely fashion. However, many factors can influence the accuracy of rapid influenza test results including the pretest probability of influenza based on presenting clinical signs and symptoms, the local prevalence of influenza, the duration of illness, the quality and type of specimen collected, and the criterion standard to which the rapid test is compared.36 Clinical practice often differs from current guidelines, which encourage presumptive treatment of hospitalized or children at high risk with influenza-like illnesses without requiring influenza confirmation.37 We found that many clinicians appeared to rely on results of rapid influenza testing to determine influenza diagnoses as most children with influenza-specific discharge diagnoses had a positive rapid test. This reliance on positive rapid influenza tests is consistent with the literature but not with the current guidelines for treatment.37 In 2006–2007, ED physicians reported being more likely to treat children and adults who presented within 48 hours of illness and who had a positive rapid influenza test than those who presented later and/or did not have a rapid test performed.38 The increased availability of molecular methods to diagnose influenza has the potential to significantly enhance the ability to reliably diagnose and treat children with influenza and will need to be evaluated in the post-2009 pandemic era. Only 3 hospitalized children with study protocol-confirmed influenza were treated with antiviral medications, including children who were admitted to the hospital within 2 days of illness onset. Surveys of physicians have indicated that antiviral medications are often not prescribed for multiple reasons including late presentation, self-limited disease, uncertain diagnosis of influenza, and expense.39 However, timely administration of oseltamivir has been associated with shorter hospitalizations among critically ill children,40 faster resolution of illness (1.0–1.5 days in healthy children), and less parental work absenteeism.41–43 Also, 43% to 44% fewer outpatient children who received oseltamivir within 48 hours of influenza symptoms developed acute otitis media.43–45 When the 2009 pandemic influenza began in the United States in mid-April 2009, recommendations were made for expanded use of antiviral medications and were associated with increased antiviral usage46; thus, usage patterns from before versus after the 2009 pandemic may differ. In 2011, the CDC recommended antiviral treatment as early as possible for any patient with confirmed or suspected influenza who (1) is hospitalized; (2) has severe, complicated, or progressive illness; or (3) is at higher risk for complications.47 More efforts are needed to increase awareness of these antiviral treatment guidelines in children.
Several limitations exist for this multicenter, multiyear study. Enrolled children could have systematically differed from nonenrolled children and/or from children presenting at nonsurveillance times, and 3 years of inpatient surveillance were excluded from 1 site. Because influenza virus has infrequently (0%–2%) been detected by RT-PCR among asymptomatic children, it is possible that a few children could have had influenza detected without clinically relevant influenza infection.48–51 Circulating influenza viruses, vaccination trends, viral testing, discharge diagnoses, and antiviral use could vary geographically, and the NVSN results may not be reflective of the entire United States.
Our study shows that seasonal influenza remains an important cause of hospitalizations, ED visits, and outpatient visits among children and that the use of tools known to reduce the burden of influenza (namely, vaccination and antiviral medications) were substantially underused during the study period. Additional efforts are needed for greater dissemination and use of the existing recommendations for influenza vaccination of children ≥6 months and of pregnant women, which partially protects younger infants.52–55 Also needed are the development and dissemination of evidence-based guidelines for laboratory testing and therapeutic options, including antiviral medications.
We thank all the children and their families who generously contributed to this study. We thank the following members of the NVSN: Nashville: Carol Ann Clay, RN, Erin Keckley, RN, Diane Kent, RN, Nayleen Whitehead, Mariah Daly, RN, and John Williams, MD; Rochester: Christina Albertin, MPH, Linda Anderson, RN, Charlene Freundlich, Geraldine Lofthus, PhD, Andrea Marino, Kenneth Schnabel, MBA, and Lynne G. Shelley; Cincinnati: Vanessa Florian, Diana Henderson, MPH, MSW, Michol Holloway, MPH, Linda Jamison, RN, BSN, CIC, Monica Kishman, David Glazer, Jessica Scholemer, and Nancy Back, BSN, MPH; Centers for Disease Control and Prevention: Mila Prill, MSPH and Aaron Curns, MPH. We also thank the anonymous reviewers whose comments and suggestions enhanced this article.
Dr Poehling contributed to the conceptualization and design of the study, interpreted the data, drafted the initial manuscript, and approved the final manuscript as submitted. Drs Edwards and Griffin substantially contributed to conception and design, acquisition of data, and interpretation of data; revised the article critically for important intellectual content; and approved the final version as submitted. Drs Szilagyi and Weinberg contributed to the conceptualization of the study and review and edits of drafts and revisions, and approved the final manuscript as submitted. Drs Staat and Iwane made substantial contributions to the conception and design and interpretation of the data, reviewed and revised the manuscript, and approved the final manuscript as submitted. Dr Snively provided input on data analysis and presentation of results, critically reviewed and edited the manuscript, and approved the final manuscript as submitted. Ms Suerken contributed to the analysis and interpretation of data, critically reviewed the manuscript, and approved the final manuscript as submitted. Dr Hall helped coordinate and supervise personnel collecting data and supervised the laboratory testing at 1 site, reviewed and revised the manuscript, and approved the final manuscript as submitted. Dr Chaves contributed to the conception and design of the analysis and interpretation of results and has critically reviewed and approved the submitted manuscript. Dr Zhu contributed to the analysis and interpretation of data, critically reviewed the manuscript, and approved the final manuscript as submitted. Ms McNeal performed the laboratory assays to acquire the data for the Cincinnati site, critically reviewed the manuscript, and approved the final manuscript as submitted. Dr Bridges made substantial contribution(s) to conception and design of the study/surveillance network, interpretation of data, review and revision of drafts for important intellectual content, and final approval of the version to be submitted for publication.
The findings and conclusions of this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention or the National Institutes of Health.
FINANCIAL DISCLOSURE: Dr Edwards received grant support from Novartis (for meningococcus); Dr Hall was a consultant for MedImmune (for respiratory syncytial virus [RSV]) and a past consultant for GlaxoSmithKline (for RSV); Dr Staat received research support and was a consultant for MedImmune (for RSV); Dr Weinberg was a consultant for MedImmune and speaker’s bureaus of Merck & Co., GlaxoSmithKline, and Sanofi Pasteur; and Dr Poehling received research support from BD Diagnostics. The other authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: The primary support for the NVSN was the US Centers for Disease Control and Prevention (cooperative agreement numbers U38/CCU217969, U38/CCU417958, U38/CCU522352, U01/IP000017, U01/IP000022, and U01/IP000147). In addition, Dr Poehling received research support from Wachovia Research Fund and National Institute of Allergy and Infectious Diseases (grant R01 AI079226). Funded by the National Institutes of Health (NIH).