We conducted a retrospective cohort study of all patients admitted to the University Hospital Heidelberg, Germany, with laboratory-confirmed influenza A(H1N1)pdm09 infection from May 2009 through April 2011. We defined a case-patient as a hospitalized person with influenza-like illness and influenza A(H1N1)pdm09 virus infection confirmed by real-time PCR. We extracted positive influenza results from our laboratory information system and reviewed the patient charts for clinical and laboratory characteristics as described below. Additionally, we differentiated between pandemic and postpandemic season and severe and nonsevere infections. Severe disease was defined as either admission to the intensive care unit (ICU) or in-hospital death. Microbiologic studies, hospital and ICU admission criteria, and treatment decisions were not standardized but made at the discretion of the attending physicians.

We collected data on demographic characteristics, coexisting conditions, clinical signs and symptoms, biochemical analyses, chest radiograph findings, antiviral and antibacterial therapy, concomitant and secondary bacterial infections as well as outcome, including death. Pneumonia was defined as the presence of a new infiltrate shown on a chest radiograph plus fever (temperature >38°C) and respiratory symptoms. Bacterial infections were diagnosed by means of blood cultures and analysis of sputum or bronchoalveolar lavage specimens. Routine laboratory analyses included C-reactive protein level and leukocyte count.

Nasopharyngeal samples and bronchoalveolar lavage specimens were collected from the patients and either processed within 2 hours or refrigerated at −20°C. PCR-based detection of influenza A(H1N1)pdm09 and seasonal influenza A and B strains was performed by 1-step real-time reverse transcription PCR (16). Total RNA from respiratory samples was isolated with the QIAamp Viral RNA Mini Kit (QIAGEN, Hilden, Germany), according to the manufacturer’s protocol and reverse transcribed by using random hexamers from the Transcriptor First Strand cDNA Synthesis Kit (Roche, Mannheim, Germany). Reverse transcription PCR analysis was performed with 5.0 μL RNA by using the RNA Virus Master Kit (Roche,) and the LightCycler 480 System (Roche) under the following conditions: 8 min at 50°C; 30 s at 95°C; 50 cycles of 5 s at 95°C, 20 s at 60°C and 1 s at 72°C. The primer pairs were used at a concentration of 10 μM.

To describe the temporal distribution of admitted influenza cases, we aggregated cases by calendar week and compared our data with data from the German surveillance system of mandatory notifiable, laboratory-confirmed influenza cases (17). We summarized demographic and clinical data for time and severity. Frequencies were compared by using χ² test or Fisher exact test for categorical variables and the Student t test or Mann-Whitney U-test for continuous variables, as appropriate. Identified risk factors with a p value <0.2 in the univariate analysis were included in a multivariate logistic regression model to assess independent association with severity. In a stepwise backward procedure, exposures with p>0.05 were excluded from the model. All comparisons with p<0.05 were considered statistically significant. We used Stata version 11 SE (StataCorp. LP, College Station, TX, USA) for all statistical analyses.

We identified 178 hospitalized patients that fulfilled our case definition and included them in the study group. In the 2009 influenza pandemic season, August 2009 through April 2010, a total of 102 patients with influenza A(H1N1)pdm09 infection were admitted to the Heidelberg University Hospital. In the first postpandemic season (December 2010–March 2011), the number decreased to 76 patients. During the pandemic season, no influenza cases other than influenza A(H1N1)pdm09 infection were observed, whereas in the first postpandemic season, 4 patients received a diagnosis of influenza B virus infection. However, we did not include these influenza B–infected patients in this study.

We charted the temporal distribution of influenza case-patients admitted to the University Hospital Heidelberg by date of first positive laboratory result, compared with influenza cases notified to the German national surveillance system by date of notification (Figure 1). The first observed admission occurred in August 2009 when only sporadic cases of influenza were observed in Germany. The number of admissions increased in October during the first wave with measurable effects of disease at the population level, preceding the usual beginning of the seasonal epidemic by 3 months. The distribution patterns of admissions to Heidelberg University Hospital and notified cases in Germany are similar, both in the pandemic and postpandemic year. In the 2010–11 season, the first cases of influenza were diagnosed in late December in Heidelberg, which is consistent with the usual start of the influenza season.

The distribution of influenza cases among age groups is shown in Figure 2. In the pandemic season >50% of cases from the German national surveillance system were in adolescents (5–19 years of age). In the subsequent season, this pattern shifted to younger children with >50% of cases in patients <14 years of age. The age distribution of patients admitted to the University Hospital of Heidelberg differed from the German national data with higher rates of children <10 years of age in 2009–10 and higher rates in persons >50 years of age in the 2010–11 season.

The demographic and clinical characteristics of patients by influenza season are compared in Table 1. In both seasons, the sex distribution was balanced (55% male patients in 2009–10 vs. 47% in 2010-11). However, patients in the first postpandemic season tended to be older (mean age 38 vs. 18 years, p<0.05) than patients in the pandemic season and had a higher proportion of concomitant medical conditions (75% vs. 51%, p<0.05). Furthermore, an increased severity was observed in 2010–11 with a doubled mean duration of hospitalization (13.9 days vs. 7.6 days, p<0.05) and a tripled rate of patients with severe disease (46% vs. 14%, p<0.05). The mortality rate was 12% in 2010–11 compared with 5% in 2009–10 (p = 0.05).

Of the 178 patients with influenza A(H1N1)pdm09 infection hospitalized in both seasons, 50 patients (28.1%) had a severe course of infection; of these, 14 died. In univariate analysis, severity of disease was significantly associated with age, presence of pneumonia as well as with a history of an underlying medical condition, in particular, immunosuppression, chronic lung disease, cardiovascular disease, renal impairment, and diabetes (Table 2). In multivariate analysis, 4 independent risk factors for a severe clinical outcome were identified (Table 3): age, male sex, chronic lung disease, and infection in the postpandemic season were significantly associated with severity of disease. The rate of bacterial co-infection increased from 10% to ≈30% in the postpandemic season, but this rate may have been underestimated because microbiologic investigations were not performed routinely for all patients (data not shown).

Demographic and clinical characteristics of influenza patients with a fatal outcome are shown in Table 4. Notably, all 14 case-patients who died had an underlying medical condition; 4 of 14 had multiple myeloma.

Our multivariable logistic regression model identified 4 independent risk factors for severe clinical disease (Table 3). Older, male patients with any chronic lung disease were more affected in the postpandemic season. The other variables mentioned in Table 2 did not reach statistical significance and were therefore not included in our final model.