Children Hospitalized With 2009 Novel Influenza A(H1N1) in California FREE

Following the detection of the first cases of 2009 novel influenza A(H1N1) in California in April 2009, the virus has rapidly spread throughout the world.^1- 2 In the United States, observations suggest that most cases of 2009 novel influenza A(H1N1) are in children or young adults, with 40% of cases in children between the ages of 10 and 18 years; the attack rate in children aged 5 to 14 years (147 per 100 000) has been higher than that of adults 60 years and older (10.5 per 100 000).^3- 8

Data are still emerging on the impact of 2009 novel influenza A(H1N1) in children. Early in the pandemic, reviews of hospitalized cases from the United States and Canada found a high proportion of patients with chronic comorbidity, predominantly asthma, with 20% to 30% requiring intensive care.^8- 11 In these studies, the overall case-fatality proportions were low (about 1%); in contrast, fatality rates of children hospitalized with 2009 novel influenza A(H1N1) in Argentina were 10-fold higher than prior rates from seasonal influenza.¹² In this report, we describe the first 345 children hospitalized with 2009 novel influenza A(H1N1) reported to the California Department of Public Health (CDPH) and analyze clinical and epidemiologic risk factors associated with severe illness and death.

Since April 23, 2009, CDPH and 61 local health jurisdictions have requested that clinicians report hospitalized or fatal cases of 2009 novel influenza A(H1N1) to their local health jurisdiction. A hospitalized case is defined as a person hospitalized for 24 hours or more with influenza-like symptoms (eg, fever plus cough, sore throat, or clinical suspicion for influenza infection) and a polymerase chain reaction test result indicative of 2009 novel influenza A(H1N1).¹³ Fatal cases had influenza-like symptoms, confirmation by polymerase chain reaction test results, and influenza listed in the death certificate as a cause of death. Cases were first reported by clinicians to local health jurisdictions and subsequently reported to CDPH using a standardized form. Data on height and weight, demographics, clinical presentation and course, comorbid conditions, and laboratory and radiographic findings were abstracted from medical records by local health jurisdiction or CDPH staff. A comorbid condition was considered to be any chronic medical condition recorded in the medical record, or in the case of obesity, height and weight data that indicated a body mass index (calculated as weight in kilograms divided by height in meters squared) in the 95th percentile or higher for age in persons older than 2 years.¹⁴ Secondary infection was defined by isolation of bacteria from a (1) sterile site or (2) lower respiratory tract specimen in conjunction with a new infiltrate on chest radiography.

Hospitalized children who required intensive care unit (ICU) admission and/or died were compared with hospitalized children who did not require ICU admission and/or die with respect to demographics, clinical characteristics, and underlying risk factors using bivariate analysis. Comorbid conditions were not treated as mutually exclusive; if a child had multiple comorbid conditions, each condition was included in the analysis. Individuals who did not have underlying medical conditions coded in the medical records were excluded in the bivariate analysis. In multivariate analysis, a “missing” category was created for all variables. Variables that were considered preexisting risk factors (eg, demographic data or comorbidities) that had a P value ≤.05 in the bivariate analysis were included in a multivariate logistic regression model using SAS 9.2 (SAS Institute, Cary, North Carolina); variables that were not considered preexisting risk factors (eg, clinical signs and symptoms, laboratory or radiographic findings, and events occurring during the hospital course) were not included in the model. Calculations of confidence intervals (CIs) for age-specific rates were conducted using the epitools package of R (http://cran.r-project.org/web/packages/epitools/index.html).

This activity was reviewed by the California Committee for the Protection of Human Subjects and deemed a public health response that did not require institutional review board approval (http://www.oshpd.ca.gov/Boards/CPHS/researchers.html).

Between April 23 and August 11, 2009, 345 of 1090 hospitalized and fatal 2009 novel influenza A(H1N1) cases (32%) reported in California were children younger than 18 years (Figure 1), including 68 infants (6%) younger than 12 months. The median age was 6 years (range, 1 week-17 years; with 30% younger than 2 years), and 59% were male. In the 297 children in whom data were available, the most common reported race/ethnicity was Hispanic (152; 51%), followed by non-Hispanic Asian/Pacific Islander (55; 19%) and non-Hispanic white (48; 16%) (Table 1). Nine children (3%) died. The rate of hospitalization and/or fatality was 3.5 per 100 000 population per 110 days of observation (or 0.97 per 100 000 person-months), ranging from 2.4 for 12- to 17-year-olds (0.67 per 100 000 person-months) to 13.9 for infants younger than 6 months (3.86 per 100 000 person-months) during the same period (Figure 2).

Comorbid conditions reported in 345 children (67%) are shown in Table 2. Two-hundred twelve (61%) had Advisory Committee on Immunization Practices (ACIP) risk factors for severe influenza^15- 16; the most common were chronic pulmonary disease (including asthma), underlying neurologic disorders (including cerebral palsy/developmental delay and seizure disorder), and immunosuppression. Of 80 children with asthma in which admitting diagnosis was recorded, 19 (24%) were admitted for asthma exacerbation. Ninety-four children (28%) had other conditions not currently considered by ACIP as risk factors, including gastrointestinal disorders and history of prematurity (birth at gestational age ≤36 weeks). In addition, body mass index data were available in 124 hospitalized children; 23 (19%) were obese. Only 19 infants (28%) younger than 1 year had underlying conditions, most commonly chronic lung disease associated with prematurity (9; 47%) or prematurity (8; 42%).

The symptoms or signs present in the study subjects are shown in Table 3. The median time from symptom onset to hospitalization was 2 days (range, 0-21 days). The most frequent symptoms or signs reported included fever, cough, shortness of breath or increased work of breathing, nausea or vomiting, diarrhea, and muscle aches. Of 27 infants hospitalized younger than 2 months, 21 (78%) were admitted with fever to evaluate potential invasive bacterial infection (“rule out sepsis”).

Central nervous system manifestations observed in 30 children (8.7%) included seizures (n = 17), altered mental status/delirium (n = 18), or both (n = 12). Presentation with seizures (P = .01) or altered mental status (P < .001) was significantly associated with ICU admission or death (Table 3). Eleven of these 30 children (37%) had preexisting cerebral palsy/developmental delay or seizure disorder. Seventeen of these 30 children (57%) were hospitalized in the ICU. The median length of stay for these children with neurologic symptoms was 4 days (range, 1-24 days).

Two hundred twenty-one of 261 children (85%) tested had positive enzyme immunoassay or direct fluorescent assay results. Of 278 children with chest radiographs or computed tomography, 163 (59%) had infiltrates suggestive of pneumonia or acute respiratory distress syndrome. Ninety-four of the 345 children (27%) required intensive care, of whom 35 (37%) required mechanical ventilation. There was no difference in the median duration between onset of symptoms and hospitalization among ICU and/or fatal cases (2 days) as compared with non-ICU, nonfatal cases (2 days) (P value = .87). Fifteen patients (4%) had evidence of a secondary bacterial infection. The median length of stay for all hospitalized cases was 3 days (range, 1-73 days) (Table 3).

For 319 children for whom data were available, 221 (69%) received antiviral treatment; more children with ACIP risk factors than those without were treated (74% [145 of 196] vs 62% [76 of 123]; P = .03). Eighty-eight of 202 children (44%) with available data received antiviral treatment within 48 hours of symptom onset; this proportion was similar for those with and without ACIP risk factors (42% vs 46%; P = .66). Seventy-three children were seen by a primary care provider for influenza-like symptoms prior to hospitalization, including 53 (73%) within 48 hours after onset of symptoms. Thirty-eight (72%) of these 53 children had available data on antiviral treatment; 11 (29%) were treated within 48 hours of symptom onset. Of the 116 children hospitalized within 48 hours of onset of symptoms and for whom dates of antiviral treatment are known, 79 (69%) received antiviral treatment within 48 hours of symptom onset.

Nine children (3%) died; their median age was 5 years (range, 6 months-14 years). Two of these children were declared dead on arrival in the hospital emergency department and had not been evaluated as outpatients prior to death. In 8 fatal cases (89%), children had comorbid illness, including neurologic disorders (n = 5, including 4 with cerebral palsy/developmental delay), chronic pulmonary disease (n = 4), immunosuppression (n = 3), chronic heart disease (n = 3), prematurity (birth at ≤32 weeks) (n = 3), and gastrointestinal disorders (n = 2). Three of the deaths were in children younger than 2 years and therefore body mass index could not be calculated; of the 6 remaining children, body mass index data were available in 4, and 3 of these 4 (75%) were obese. The mean number of comorbid conditions among fatal cases was 3. The remaining fatal case was a child reported as previously healthy. Three (33%) had evidence of secondary bacterial infection with group A Streptococcus (n = 2) or Staphylococcus aureus (n = 1). Six children (67%) received antiviral treatment, but only 1 within 48 hours of symptom onset. In addition to 2009 novel influenza A(H1N1), other contributory causes of death listed for the fatal cases included pneumonia or acute respiratory distress syndrome (n = 5), sepsis, (n = 3) and scarlet fever (n = 1). The median length of stay for fatal cases was 22 days (range, 1-73 days).

The demographic characteristics, comorbid conditions, and clinical characteristics identified in bivariate analyses as significantly associated with intensive care and/or death are shown in Table 1, Table 2, and Table 3, respectively. Hispanic (odds ratio [OR], 0.4; 95% CI, 0.2-0.8) and black (OR, 0.3; 95% CI, 0.1-1.0) children were less likely to require ICU admission and/or die. In bivariate analysis, the presence of either an ACIP-established risk factor (OR, 2.0; 95% CI, 1.2-3.4) or a non–ACIP established risk factor (OR, 2.1; 95% CI, 1.2-3.4) was associated with ICU admission and/or death. Among these, chronic lung disease of prematurity (OR, 4.1; 95% CI, 1.4-12.0), chronic cardiac disease (OR, 4.3; 95% CI, 1.9-9.5), chronic neurologic disease (OR, 2.8; 95% CI, 1.6-5.0), chronic gastrointestinal disorder (OR, 2.4; 95% CI, 1.3-4.5), and prematurity (birth at ≤32 weeks) (OR, 2.8; 95% CI, 1.1-6.9) were most strongly associated with ICU admission and/or death. An abnormal chest radiograph or computed tomography consistent with pneumonia (P = .009), presence of secondary bacterial infection (P < .001), and diagnosis of sepsis (P < .001) were all associated with ICU admission and/or death, while a positive rapid test result (P = .004) and early antiviral treatment (P = .01) were not. In a multivariate regression model that examined age, race/ethnicity, and chronic medical conditions, factors that were significantly associated with intensive care and/or death included congenital heart disease (OR, 5.0; 95% CI, 1.9-13.5) and cerebral palsy/developmental delay (OR, 3.5; 95% CI, 1.7-7.4). Hispanic (OR, 0.4; 95% CI, 0.2-0.8) and black (OR, 0.3; 95% CI, 0.1-0.9) race/ethnicity were associated with a decreased likelihood of intensive care and/or death (Table 4).

We report on the first 345 children known to be hospitalized with and/or die of 2009 novel influenza A(H1N1) in California. Hospitalization rates in the first 110 days of the pandemic were comparable with seasonal influenza and were highest in young infants, although these rates may be an underestimate given limited availability of sensitive laboratory testing, among other reasons. Most children presented with acute respiratory illness. As with seasonal influenza,¹⁶ gastrointestinal symptoms were frequent in children infected with 2009 novel influenza A(H1N1). Similar to other reports,^8- 11 we found a high proportion of underlying medical conditions in hospitalized children but not infants, of whom more than 70% were previously healthy. More than one-quarter of cases required intensive care. The case-fatality proportion was low; 33% of fatal cases had a secondary bacterial infection detected. In multivariate analysis, congenital heart disease and cerebral palsy/developmental delay were associated with increased likelihood of ICU admission and/or death, while Hispanic and black children had a decreased likelihood. Most hospitalized children had multiple underlying medical conditions, which may explain why most of the clinically important conditions noted as significant in bivariate analysis were not retained in the multivariate analysis.

Infants younger than 6 months were most likely to be hospitalized. Similarly, 60% of hospitalized children in Argentina were younger than 1 year.¹² Hospitalization rates per 100 000 population for 2009 novel influenza A(H1N1) in Chicago, Illinois, were 25 for children aged 0 to 4 years vs 11 for older children.⁴ This is also comparable with seasonal influenza, where children younger than 6 months have annual rates per 100 000 of 240 to 720, compared with 17 to 45 at ages 2 to 5 years and 8 to 41 at ages 5 to 17 years.^15- 21 Increased hospitalization rates may in part reflect a lower threshold for clinicians to admit young infants for evaluation of febrile illness.

The proportion of 2009 novel influenza A(H1N1)–infected children with comorbid conditions hospitalized in California is similar to Montreal, Canada, and Toronto, Canada, where up to 80% of hospitalized children had underlying conditions.^8- 11 Nearly one-third of hospitalized children in our series had asthma (32%), a frequency similar to observations of others,^8- 11 and more than double the estimated prevalence of asthma (15.4%) in the California population younger than 18 years.²² Pediatric asthma is also associated with a higher rate of hospitalization for seasonal influenza.^{16- 19,23- 24} One-quarter of children with a history of asthma were admitted for asthma exacerbation and fever; in such cases, clinicians should maintain high clinical suspicion and initiate prompt antiviral treatment while awaiting test results for 2009 novel influenza A(H1N1) infection.

Comorbidity was associated with increased likelihood of ICU admission and/or death, including novel associations with gastrointestinal disease (often coexistent with underlying neurologic disorders such as children with cerebral palsy receiving gastrostomy feedings) and prematurity (birth at ≤32 weeks) (which was often coexistent with chronic lung disease). We found that cerebral palsy/developmental delay and congenital heart disease were independently associated with ICU admission and/or death. In Toronto, 17% of children hospitalized for 2009 novel influenza A(H1N1) had a history of neurologic impairment.⁹ Of 36 pediatric fatalities due to 2009 novel influenza A(H1N1) reported to the Centers for Disease Control and Prevention, 92% with high-risk conditions had a neurodevelopmental diagnosis.²⁵ Similarly, approximately 8% to 12% of children requiring intensive care or dying with seasonal influenza had an underlying cardiac condition.^15,26- 27 Children with either chronic cardiac or neuromuscular disease hospitalized with seasonal influenza are more likely to require intensive care, prolonged hospitalization, or mechanical ventilation.^27- 30 The fragility of children with these conditions warrants prompt immunization before illness and empirical therapy if 2009 novel influenza A(H1N1) is suspected.

In our case series, neurologic manifestations, such as seizures and encephalopathy/altered mental status, were reported frequently. Although many of these children were admitted to the ICU, their length of stay was relatively short, and most recovered to their neurologic baseline at discharge. Similar frequencies of neurologic complications in 2009 novel influenza A(H1N1) have been reported from many countries.^9,31- 32,32 Although the mechanism is not well understood, seasonal influenza is also associated with seizures and encephalopathy and post–infectious encephalopathy.^31- 33 In contrast to recent reports of novel influenza A(H1N1), in more than 50% of Japanese cases of encephalitis associated with seasonal influenza, patients had severe neurologic sequelae or died.³³

We note an association with severe illness and race/ethnicity, with Hispanic and black children less likely to have 2009 novel influenza A(H1N1) requiring ICU admission and/or dying than other race/ethnicity groups. Early in the surveillance period, CDPH guidance recommended testing of persons traveling from Mexico, which may have accounted for a larger proportion of Hispanic individuals having been tested and hospitalized with less severe illness. Reasons why the proportion of black children was less likely to have a severe outcome are unclear, although the numbers in our series are small. Data on 2009 novel influenza A(H1N1)–associated hospitalizations and deaths and race/ethnicity in the United States are limited. In Chicago, hospitalization rates in children 14 years and younger were higher for non-Hispanic black, Asian/Pacific Islander, and Hispanic vs non-Hispanic white children.⁴ Further data are needed on whether there are racial and ethnic disparities in 2009 novel influenza A(H1N1) and the causes of such disparities.

In contrast to seasonal influenza, bacterial coinfection was infrequently reported in our series and may have been underdiagnosed because of infrequent testing, testing performed late in hospitalization, or the receipt of antibiotics prior to microbiological testing. Of note, the presence of secondary bacterial infection was significantly associated with admission to the ICU and/or death, with one-third of the fatal cases having secondary bacterial infections. When postmortem tissue has been evaluated using specific immunohistochemistry and polymerase chain reaction methods, higher (29%-43%) rates of secondary bacterial infection due to Streptococcus pneumoniae, S aureus, and group A Streptococcus have been identified.^25,34 Studies of fatal pediatric cases due to seasonal influenza have also suggested high rates of bacterial coinfection.^26,35

Possibly consistent with the known higher sensitivity of rapid tests for seasonal influenza in children, who may shed higher viral loads for prolonged periods compared with adults,¹⁵ sensitivity of rapid tests in our case series was 85%. However, other studies have demonstrated poor sensitivity of rapid tests for 2009 novel influenza A(H1N1) in both adults and children.^36- 39 A negative rapid test result should not be the sole basis for excluding a diagnosis of 2009 novel influenza A(H1N1) when clinical suspicion is high. Almost 70% of cases received treatment with antiviral drugs, but often more than 48 hours after onset of symptoms, in contrast to recommendations that patients at high risk for severe disease with 2009 novel influenza A(H1N1) infection, including hospitalized patients, be treated with a neuraminidase inhibitor as soon as possible.⁴⁰ In our case series, children with a positive rapid test result or who received early antiviral treatment were less likely to require ICU admission or die.

Our data are subject to several limitations. Data were extracted from nonstandardized medical records. Case ascertainment was based on passive reporting by clinicians, and underreporting may have occurred because of poor recognition, nonspecificity of influenza-like symptoms, and limited availability of confirmatory laboratory testing, leading to underestimation of rates of hospitalization. Likewise, clinicians may have been more inclined to suspect and test for 2009 novel influenza A(H1N1) in patients with complex comorbid conditions or more severe illness, to admit those children to an ICU for closer monitoring, and to report more severely ill cases to their local health jurisdiction, resulting in possible overestimation of the proportion of children admitted to the ICU. Finally, ascertainment of race/ethnicity data was higher in severe cases than nonsevere cases, possibly reflecting a bias in completeness of reporting data according to severity.

In conclusion, clinicians should be aware of when and what type of influenza and other respiratory viruses, including 2009 novel influenza A(H1N1), are circulating in their community. When 2009 novel influenza A(H1N1) activity is present, clinicians should maintain high suspicion for infection in pediatric patients presenting with febrile respiratory illness and initiate prompt treatment in infants and children with underlying risk factors regardless of rapid antigen test results. The epidemiology of severe pediatric 2009 novel influenza A(H1N1) in California supports national recommendations for targeting immunization, when vaccine supplies are scarce, to those caring for infants younger than 6 months, pregnant women who may convey immunity to infants younger than 6 months, children 6 to 59 months of age (especially those 6-18 months of age), and children with conditions associated with severe influenza.

Correspondence: Janice K. Louie, MD, MPH, California Department of Public Health, 850 Marina Bay Pkwy, Richmond, CA 94804 (janice.louie@cdph.ca.gov).

Accepted for Publication: May 6, 2010.

Author Contributions: Dr Louie had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Louie, Acosta, and Matyas. Acquisition of data: Louie, Gavali, Acosta, Samuel, Winter, and Jean. Analysis and interpretation of data: Louie, Gavali, Acosta, Samuel, Winter, Jean, Glaser, Matyas, and Schechter. Drafting of the manuscript: Louie, Gavali, Winter, Glaser, Matyas, and Schechter. Critical revision of the manuscript for important intellectual content: Louie, Gavali, Acosta, Samuel, Jean, Matyas, and Schechter. Statistical analysis: Gavali, Acosta, Samuel, Winter, and Schechter. Administrative, technical, and material support: Acosta, Jean, Glaser, and Matyas. Study supervision: Louie and Samuel.

Financial Disclosure: None reported.

Funding/Support: This work was supported by the CDPH.

Role of the Sponsor: Design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript was fully sponsored by the CDPH.

Additional Contributions: All members of the California Pandemic (H1N1) Working Group work at local health departments or the CDPH and helped collect information on the reported cases. They also reviewed multiple versions of the manuscript. We are especially indebted to Christopher Anderson, BS, Jaynia Anderson, BS, Nina Huynh, BS, Daniel Zhuang, BA, Samuel Yang, BA, Melanie Stewart, BA, Tong Kong, BS, Maria Nevares, Erica Boston, BA, Sabrina Gilliam, MPH, Edward Powers, DVM, Huyen Cao, MD, David Schnurr, PhD, Sharon Messenger, PhD, Kathleen Harriman, PhD, Duc Vugia, MD, Jon Rosenberg, MD, John Talarico, DO, Carol Pertowski, MD, Douglas Hatch, MD, and Gilberto Chavez, MD, for their support and dedication. Thanks to the Centers for Disease Control and Prevention staff that support and advise us, including Tim Uyeki, MD, and Scott Harper, MD. Many thanks to all dedicated staff at the CDPH, epidemiologists and laboratorians alike, who contributed to surveillance and laboratory investigation of novel influenza A(H1N1) cases. In addition, while we cannot name them all, we gratefully acknowledge the contributions of the clinicians throughout California and all the staff at California local health departments who diligently worked to help acquire the epidemiological and clinical information and ensured that these cases were reported to CDPH.