We conducted a nationwide prospective observational study in Germany by using the German Survey Center for Rare Pediatric Diseases (ESPED; an established children’s hospitals network comprising all 375 pediatric hospitals in Germany) to identify children <15 years of age admitted to PICUs with confirmed A(H1N1)pdm09 infection and related deaths. Cases and related deaths during August 3, 2009–July 29, 2011 were reported by using a standardized form. In pandemic season 2009–10 (August 3, 2009–August 9, 2010), the case definition included only A(H1N1)pdm09 infection; in postpandemic season 2010–11 (August 10, 2010–July 29, 2011), the case definition included all influenza virus infections.

On notification by treating physicians of patients with A(H1N1)pdm09 infection, the ESPED study center distributed and subsequently collected a structured questionnaire, which had been adapted by the authors from an earlier study on seasonal influenza (11). Of 284 distributed questionnaires requested by 186 hospitals, 95% (271/284) were returned to the study center (Figure 1). After excluding 3 questionnaires that had been notified twice and 101 questionnaires for patients who did not meet the case definition, 62% (167/271) of the questionnaires from 83 hospitals remained. Reasons for not meeting the case definition included patient age >15 years and patient not admitted to PICU. In accordance with the case definition, only cases of A(H1N1)pdm09 infection were reported during the pandemic season (2009–10), and 9 cases of influenza A (not further subtyped) and 2 cases of influenza B infection were reported in the postpandemic season (2010–11). Therefore, further analyses were restricted to A(H1N1)pdm09 cases.

The structured questionnaire covered anonymous patient information and information regarding the hospital stay, clinical signs and symptoms, clinical and laboratory findings, specific treatments, status of influenza vaccination, disease complications, and underlying chronic medical conditions (chronic respiratory diseases; cardiac diseases; immunodeficiency; and neurodevelopmental disorders, including developmental delay, cerebral palsy, epilepsy, and other cognitive disorders). Answer categories were predetermined, but free space was designated for respondents to provide information about other diagnoses and coexisting illnesses/medical conditions. Hospital-acquired infection was defined by a date of symptom onset being >2 days after the date of hospital admission; 2 days corresponds with the median incubation time for A(H1N1)pdm09 according to Cao et al. (12). Data were double entered by using EpiData 3.1 software (www.epidata.dk/) in an electronic database.

Reported values are those for children with available information. Descriptive statistics comprised the calculation of median and interquartile ranges (IQRs) for continuous variables and absolute numbers and proportions (together with 95% binomial exact CIs, when appropriate) for categorical variables. Comparative analyses were based on the Wilcoxon rank-sum test for continuous variables and Fisher exact test for categorical variables. Odds ratios (ORs) and 95% CIs were calculated. Multivariable analysis was performed by using a logistic regression with a stepwise approach to compare cases of hospital-acquired infection with cases of community-acquired infection and survivors with nonsurvivors in PICUs. In doing so, risk factors with a p value <0.2 were considered for multivariable analysis, with the exception of age, sex, and season, which were included in all models. Reported p values are 2-sided, and p<0.05 was considered significant. Statistical analyses were performed by using Stata 11.0 (StataCorp LP, College Station, TX, USA).

Adherence to national data protection laws was approved by the Federal Commissioner for Data Protection and Freedom of Information of Germany. Ethical approval was granted by the Ethics Committee, Charité-Universitätsmedizin, Berlin, Germany.

We identified 156 critically ill children with confirmed A(H1N1)pdm09 infection: 112 in 2009–10 and 44 in 2010–11 (Figure 1). Dates of symptom onset ranged from September 21, 2009, to March 20, 2010, for 2009–10 and from December 20, 2010, to February 22, 2011, for 2010–11 (Figure 2). Cases were reported from 15 of the 16 federal states in Germany during 2009–10 and from 10 during 2010–11.

The proportion of boys among case-patients was higher in 2009–10 than 2010–11 (59% vs. 37%, p = 0.02) (Table 1). The median age of case-patients was 5.3 and 3.2 years in 2009–10 and 2010–11, respectively, and differed statistically (p = 0.007) between the 2 seasons. The age distribution in 2010–11 compared with that in 2009–10 was characterized by a markedly higher proportion of children < 2 years of age and a lower proportion of children 10–14 years of age (Figure 3). In both seasons, infants <1 year of age represented the age group with the highest number of cases (Figure 4).

Of the 146 children with available information, 114 (78%) had >1 chronic underlying medical condition; the difference between seasons for these conditions was not statistically significant (Table 1). In both seasons, neurodevelopmental disorders were the most prevalent underlying medical condition. Of the 156 critically ill case-patients, 130 were >6 months age and thus eligible for vaccination against A(H1N1)pdm09 virus; however, for children with available information on vaccination status, only 5 (7%) of 67 vaccine-eligible case-patients had been vaccinated in the 2009–10 season, and none had been vaccinated in the 2010–11 season. Of the 69 total children in both seasons with underlying chronic medical conditions, 64 (93%) had not been vaccinated against A(H1N1)pdm09 virus.

More cases of sepsis were reported during the postpandemic season than during the pandemic season (21% vs. 8%; p = 0.048) (Table 1). Treatment with oseltamivir was used equally (in ≈62% of children) during both seasons. The time to oseltamivir administration after symptom onset (median 4 days) was similar throughout both seasons. The use of catecholamine and mechanical ventilation was more frequent in 2010–11 than in 2009–10, but the difference was not statistically different.

Hospital-acquired infections accounted for 11% (11 of 101) of the cases in 2009–10 and for 23% (8/35) in 2010–11 (p = 0.0931) (Table 1). Of the total study cohort, 14% (19/136) of the patients (9 in a general ward and 10 in a PICU) most likely had hospital-acquired infection. For these case-patients, the median time from hospital admission to symptom onset was 29 days (IQR 12–73 days). The median age for patients with hospital-acquired infection was 1.1 years, and 56% (10/18) were boys (difference not statistically significant between seasons).

The overall case-fatality ratios were 26% (5/19) among patients with hospital-acquired infection and 20% (23/117) among those with community-acquired infection (p = 0.543). Compared with patients with community-acquired infection, those with hospital-acquired infection had more complications, including acute respiratory distress syndrome (ARDS) (OR 2.7, p = 0.054) and sepsis (OR 3.1, p = 0.064), but the differences were not statistically significant (Table 2). In the multivariable model, immunodeficiency (OR 5.9, 95% CI 1.5–23.9; p = 0.013) and mechanical ventilation (OR 8.9, 95% CI 1.1–74.7; p = 0.043) were significantly associated with hospital-acquired infection after adjusting for age, sex, and season.

The case fatality ratio in PICUs did not differ between seasons: 15% (16/106) and 21% (9/44) of PICU case-patients died in 2009–10 and 2010–11, respectively (p = 0.473) (Table 1). For the 2 seasons, 25 of 150 PICU case-patients died, corresponding to a case-fatality ratio of 17% (95% CI 11%–24%). On hospital discharge, 26% (27/104) of the survivors were reported to have possible sequelae or worsening of a pre-existing medical condition.

No statistical differences were found between survivors and nonsurvivors in underlying chronic medical conditions and vaccination status. ARDS (OR 3.2, 95% CI 1.1–9.2, p = 0.029), myocarditis (OR 30.9, 95% CI 2.6–360.7,; p = 0.006), and mechanical ventilation (OR 18.3; 95% CI 1.3–251.6, p = 0.030) were independently associated with a fatal outcome in the multivariable model after adjusting for age, sex, and season (Table 3).

Compared with survivors, nonsurvivors more frequently required mechanical ventilation (p = 0.001) and treatment with catecholamine (p = 0.002); no differences were found in oseltamivir administration (65% vs. 62%, p = 0.8185). Time from symptom onset to oseltamivir uptake did not differ between survivors (median 4 days, IQR 1–6 days) and nonsurvivors (median 4 days, IQR 2–8 days).