Mortality risk factors in children with severe influenza virus infection admitted to the pediatric intensive care unit

Abstract

Some children hospitalized for severe influenza virus infection require intensive care or die because of disease progression, which may be combined with other complications. The objective of this study was to identify the mortality risk factors in the patients with severe influenza virus infection admitted to the pediatric intensive care unit (PICU).

Seventy-seven pediatric patients with severe influenza virus infection who were admitted in the PICU at Guangzhou Women and Children's Medical Center between 2013 and 2017 were evaluated. Data were transcribed and analyzed.

The patients’ median age was 3.0 years (interquartile range, 1.0–4.0 years), with 59.7% of the patients aged <3 years. The mortality was 16.9%, and patients aged >3 years accounted for 69.2% of the cases. Influenza A virus infection was found in 83.1% of the patients. Coinfection was detected in 58.7% of the patients. Haemophilus influenzae (11.7%) and adenovirus (9.1%) were the predominant bacterial and viral pathogens isolated, respectively. Older age, oxygen saturation level of <90% at admission, acute respiratory distress syndrome, pneumorrhagia, influenza-associated encephalopathy (IEA), septic shock, low ratio of partial pressure of oxygen in arterial blood (PaO₂, <60 mm Hg) to the fraction concentration of oxygen in inspired air (FiO₂; P/F), higher oxygenation index, increased alanine aminotransferase level (>100 IU/L), increased aspartate aminotransferase level (>100 IU/L), increased lactate dehydrogenase level (>500 IU/L), high fraction concentration of oxygen in inspired air (FiO₂ > 60%), and positive end-expiratory pressure (>8 cmH₂O) were associated with poor outcome. The deceased patients were more likely to have oxygen saturation levels of <90% at admission and IEA than those who survived. Higher P/F ratio was a protective factor against death in patients.

The children with severe influenza virus infection who were admitted in the PICU were mainly aged <3 years. The presence of an oxygen saturation level of <90% at admission and IEA were the prognostic variables independently associated with mortality. Higher P/F ratio was a protective factor against death in patients.

1. Introduction

Influenza virus infection is common in children. Rates of influenza virus infection are consistently highest in children in different countries,^[1,2] and a systematic review published in the Lancet in 2011 estimated that 90 million cases of influenza occur each year in children aged <5 years worldwide.^[3] Approximately 870,000 preschool children are hospitalized around the world due to influenza virus infection each year.^[4] While influenza virus infection may often be self-limited, it can also lead to severe clinical outcomes or even death because of its complications. In a meta-analysis, Nair et al estimated that in 2008, >100,000 deaths in children aged <5 years were due to influenza-related causes, 99% of which occurred in developing countries.^[3] Some previous studies^[5–7] found a few of risk factors for severe influenza virus infection in children such as ages under 2 years old, premature, presence of chronic, severe underlying medical conditions, and immaturity of the immune system.

In this study, we analyzed 77 pediatric patients with severe influenza infection who were admitted to the pediatric intensive care unit (PICU). The purpose of this study was to identify the mortality risk factors in children with severe influenza virus infections who were admitted to the PICU. This information may be useful in the optimal utilization of scarce resources for the most effective preventive and early management strategies.

2. Methods

2.1. Setting

This study included 77 patients with severe influenza virus infections who were admitted to the PICU at Guangzhou Women and Children's Medical Center between January 2013 and December 2017 and had a respiratory specimen positive for seasonal influenza virus A or B. A list of patients was generated by identifying respiratory samples positive for influenza in real-time reverse transcriptase-polymerase chain reaction (RT-PCR). The respiratory specimens consisted primarily of nasopharyngeal secretions. The patients with incomplete data (discharged against medical advice) were excluded. Chest radiography was performed in all the patients. Thirty patients underwent high-resolution computed tomography (HRCT) according to the wide range of lesions found on chest radiography. All the patients underwent indirect immunofluorescence viral testing of nasopharyngeal secretions during the acute phase of the disease to identify other viral infections. Blood and/or bronchoalveolar lavage (BAL) cultures were obtained for suspected bacterial infection with Mycoplasma pneumoniae or fungal infection. Flexible bronchoscopy was performed in the 28 patients who exhibited consolidation on chest radiography or HRCT. This study was approved by the ethics committee of Guangzhou Women and Children's Medical Center, Guangzhou Medical University.

2.2. Data collection

For each patient, data on demographics, clinical presentation, laboratory findings, microbiologic and radiologic findings, treatment, response to therapy, and outcome were collected. We also collected mechanical ventilation (MV) data. Oxygenation index (OI [fraction of inspired oxygen × mean airway pressure]/partial pressure of arterial oxygen) and the ratio of the partial pressure of oxygen in arterial blood (PaO₂) to the fraction concentration of oxygen in inspired air (FiO₂; P/F) were calculated on the 1st day of PICU admission. Acute respiratory distress syndrome (ARDS) was defined on the basis of the Pediatric Acute Lung Injury Consensus Conference criteria.^[8] Septic shock was defined on the basis of the International Pediatric Sepsis Consensus Conference criteria.^[9] Influenza-associated encephalopathy (IEA: neurologic complications including seizures, encephalitis, encephalopathy, Reye syndrome, and other neurologic disorders associated with influenza virus infection) was defined on the basis of the Center of Disease Control and Prevention criteria.^[10]

2.3. Data analyses

All statistical analyses were performed using the SPSS 24.0 software. Categorical data were presented as frequency with the corresponding percentage, and continuous data were presented as median with the interquartile range (IQR). The Chi-squared or Fisher exact test was used to compare categorical variables between the survivors and nonsurvivors. Univariate analyses were performed to determine the risk factors significantly associated with seasonal influenza-associated pediatric deaths of children admitted in the PICU. To determine the independent contribution of each factor to the case outcomes, multiple logistic regression analysis was performed. Two-tailed P-values of <.05 were considered statistically significant.

3. Results

3.1. Demographics and comorbid conditions

Over the 5-year period, 1770 hospitalized patients had a respiratory specimen positive for seasonal influenza A or B, of whom 80 were admitted to the PICU and 13 died. Of those who were admitted to the PICU, 3 were not enrolled in this study because of discharge against medical advice. Of the 77 PICU patients enrolled, 75.3% (58/77) were boys and 24.7% (19/77) were girls, with ages ranging from 1 month to 12 years. The median age was 3.0 years (IQR, 1.0–4.0 years), with 83.1% (64/77) of the patients aged <5 years and 59.7% (46/77) aged <3 years as shown in Figure 1. The most common period with the largest number of severe influenza infection cases in our study was April to June in the recent 5 years as shown in Figure 2. Of the 77 patients, 28 (36.4%, 28/77) had comorbid conditions. Renal and cardiovascular diseases were most common (10.4%, 8/77), followed by asthma (9.1%, 7/77) and prematurity (7.8%, 6/77; Table 1).

Figure 1.

Age distribution of the children with severe influenza virus infection.

Figure 2.

Seasonal distribution of the children with severe influenza virus infection.

Table 1.

Demographics and underlying medical conditions of the children with severe influenza infection who were admitted in the pediatric intensive care unit.

3.2. Clinical characteristics and laboratory findings

Cough (96.1%,74/77) and fever (90.9%,70/77) were the most common symptoms in our study, followed by shortness of breath or increased work of breathing (88.3%, 68/77) and altered sensorium (58.4%, 45/88). The median duration of cough was 4.0 days (IQR, 2.0–6.0 days), and that of fever was 4.0 days (2.0–5.0 days). Complications occurred in all 77 the patients, with pneumonia occurring in all the patients (100%, 77/77), followed by respiratory failure (96.1%, 74/77), ARDS (22.1%, 17/77), septic shock (15.6%, 12/77), and IEA (13.0%, 10/77). The abnormal clinical symptoms, physical examination findings, complications, laboratory findings are presented in Table 2.

Table 2.

Clinical characteristics and laboratory findings of the children with severe influenza infection who were admitted in the pediatric intensive care unit.

3.3. Radiologic and microbiologic findings

Chest radiographic examination results mostly exhibited diffuse infiltration of both lungs (Fig. 3). Segmental or lobar infection distribution which predominantly involved the lower and middle lobes. The patients who died tended to present more severe infection on radiography. The most common finding on HRCT was consolidation (Fig. 4). The other main chest imaging findings were pleural effusion (28.6%, 22/77) and pneumothorax (6.5%, 5/77). Among the 77 patients, 83.1% (64/77) and 16.9% (13/77) had influenza A and B virus infections, respectively, with an influenza A to B ratio of 4.9:1.0. Besides influenza A/B, another causative agent (defined as coinfection) was detected in 58.7% (46/77) of the cases, with bacterial coinfections accounting for 36.4% (28/77), M pneumoniae coinfections for 11.7% (9/77), and viral coinfection for 11.7% (9/77). Of the bacterial coinfection cases, Haemophilus influenzae (11.7%, 9/77) and Streptococcus pneumoniae (7.8%, 6/77) were the predominant typical bacteria isolated in children with severe influenza virus infection. The most common viral pathogen isolated was adenovirus (9.1%, 7/77). The abnormal radiologic and microbiologic findings are shown in Table 3.

Figure 3.

Chest radiograph showing bilateral diffuse infiltration indicative of acute respiratory distress syndrome.

Figure 4.

High-resolution computed tomography scan of the chest, showing areas of airspace consolidation in the left lower lobes and increased bilateral transparency in a 2-year-old child with severe influenza virus infection.

Table 3.

Radiologic and microbiologic findings of the children with severe influenza infection who were admitted in the in pediatric intensive care unit.

3.4. Treatment and outcome

Of the 77 patients with severe influenza infection, 13 died (16.9%), with the patients aged >3 years accounting for 69.2% (9/13) of them. All 77 patients needed assisted ventilation, including MV in 93.5% (72/77) and noninvasive ventilation in 6.5% (5/77), of whom 46.8% (36/77) needed assisted ventilation for >7 days. The median length of hospitalization was 15.0 days (IQR, 10.0–20.0 days). The median length of PICU stay was 8.0 days (IQR, 7.0–12.3 days). On the 77 patients, 71.4% (55/77) stayed in the PICU for >7 days. Table 4 shows the patients’ treatments as follows: all the patients received oseltamivir antiviral drug treatment before hospitalization in 13 patients and after hospitalization in 64. The median time from onset of symptoms to initiation of antiviral therapy was 5.0 days (IQR, 3.0–6.0 days). Only 11.7% (9/77) and 77.9% (60/77) of the patients received antiviral therapy in 48 and >72 hours from the onset of symptoms, respectively. Of the patients, 84.4% (65/77) received antibiotics therapy and 72.7% (56/77) received corticosteroid therapy. On the 1st day of PICU MV therapy, the median FiO₂ was 55.5% (45–76.3%); median peak inspiratory pressure, 29.5 cmH₂O (26.0–32.0 cmH₂O); and median positive end-expiratory pressure (PEEP), 5.5 cmH₂O (5.0–7.75 cmH₂O).

Table 4.

Treatments of the children with severe influenza infection who were admitted in the in pediatric intensive care unit.

3.5. Mortality risk factors in children with severe influenza virus infection

Older age, oxygen saturation level of <90% at admission, ARDS, pneumorrhagia, IEA, and septic shock were associated with mortality risk (all P < .05). The laboratory findings associated with death were lower PaO₂ (<60 mm Hg) and P/F, higher OI, and increased alanine aminotransferase (>100 IU/L), aspartate aminotransferase (>100 IU/L), and lactate dehydrogenase levels (>500 IU/L; all P < .05). The treatment associated with death was MV therapy, with higher FiO₂ (>60%) and PEEP (>8 cmH₂O), and length of PICU stay of >7 days (all P < .05). In the multivariate analysis, the independent risk factors of mortality were oxygen saturation level of <90% at admission (odds ratio [OR], 7.80; 95% confidence interval [CI], 1.02–50.54; P = .048), IEA (OR, 20.35; 95% CI, 1.62–256.13; P = .020). Higher P/F ratio (P = .011) was a protective factor against mortality in patients with severe influenza virus infection (Table 5). No relationship was found between death and comorbid conditions, specific pathogens, or radiological findings in the univariate and multivariate analyses.

Table 5.

Results of the multivariate analysis of the mortality risk factors in children with severe influenza infection who were admitted in the pediatric intensive care unit.

4. Discussion

The outbreaks of influenza occur in winter in temperate zones and in rainy season in the tropic and subtropics.^[11] The incidence peaked from April to July in Singapore,^[12] but from June to September in India.^[11] The most common months with the largest number of severe influenza virus infection cases in our study were April to June. The case fatality rate of severe influenza virus infection in the PICU ranges from 10.5% to 47% in other reports.^[13–15] The mortality of severe influenza virus infections in patients in the PICU was 16.9% in our study, in line with previous reports. In the present study, severe influenza virus infection in the PICU mainly occurred in children aged <5 years, especially those aged <3 years. According to the surveillance data of the United States, patients with influenza virus infection aged <6 months and 6 to 23 months were >6 and >3 times more likely to die, respectively, than those aged 13 to 17 years.^[16] Younger children and those with Advisory Committee on Immunization Practices-defined high-risk comorbid conditions such as asthma, neurologic disorders, cardiovascular disease, and renal disease may be at increased risk of influenza-related death.^[16–18] However, it is an unexpected finding that the patients aged >3 years accounted for 69.2% of deaths and were significantly more likely to die than the younger patients (P = .045). In addition, the comorbid conditions were not associated with death in our study. The reason may be that the parents of these younger children with high-risk comorbid conditions paid more attention to them and took them to the hospital earlier. In the hospital, those children were thought to be at higher risk of death and targeted for more-aggressive treatments as early as possible, which may have altered the course of their illness.

According to data from Taiwan, among patients admitted to the ICU, more had influenza A virus infection, with an influenza A-to-influenza B ratio of 2.6.^[15]In Australia, influenza A was the predominant type (69%) in hospitalized children with influenza virus infection.^[19] In the United States, 65% of pediatric influenza-related deaths are associated with influenza A infection.^[16] In our study, 83.1% of the patients had influenza A infection, with a mortality of 84.6%. This result is consistent with previous reports. The incidence of bacterial coinfections caused by influenza viruses ranged from 7% to 43% in other reports,^[16,18] and the incidence of viral coinfection ranges from 15.1% to 21.6%.^[19] In our study, bacterial and viral coinfections accounted for 36.4% and 11.7%, respectively. The reason for the high incidence rate of bacterial coinfections in our study may be the long clinical courses of our patients and length of PICU stay, which could increase the risk of secondary bacterial infections. On the contrary, more than half (58.4%) of our patients underwent fibro-bronchoscopy, and BAL cultures were obtained for suspected bacterial infection; thereby, more bacterial coinfections may have been detected. According to another report,^[20] bacterial coinfection can result in disease progression and severe disease in children. Streptococcus and Staphylococcus aureus, which were mainly responsible for secondary severe bacterial pneumonia, were the most commonly reported bacterial coinfections.^[11,21] However, H influenzae was the predominant organism isolated in children with severe influenza virus infection in our patients. The reason may be that H influenzae vaccine was not free for children. In our study, most patients (84.4%) had received antibiotic therapy, including 48% (37/77) with bacterial coinfections and 66.2% (49/77) with a higher C-reactive protein level. This may be the reason we did not find any association between bacterial coinfection and mortality. Viral coinfections were infrequent in our study. The low incidence of other respiratory viruses may be explained by the fact that the incidence of the illness peaked in April to June before the winter respiratory viral season. Respiratory syncytial virus was the most frequently identified coinfection.^[16] However, adenovirus was the most common viral pathogen in this study. Our data may suggest an increased prevalence of adenovirus in severe influenza virus infection cases or may even lead to fatal ones.

Severe influenza virus infections that resulted in death were accompanied by a wide range of complications, with pneumonia, septic shock, and ARDS being the most commonly reported.^[17,21] In our study, the presence of septic shock, ARDS, and IEA were associated with death. Meanwhile, an oxygen saturation level of <90% at admission was associated with death from severe influenza virus infection in our univariable and multivariable analyses, similarly to previous studies in Taiwan and the United States in which shortness of breath or dyspnea were indicators of poor outcome.^[15,20] Pneumonia is a common complication, and the influenza virus itself increases susceptibility to pneumonia. All our severe cases were diagnosed as pneumonia by chest radiography, and the patients who died tended to present with a more severe pneumonia patch or effusion, which coincided with prior evidence.^[14,15] Severe pneumonia with more patch or effusion on radiography were more likely to develop into ARDS or even pneumorrhagia. In the present study, 46.2% of the patients who died had ARDS and 38.5% had pneumorrhagia. All of them needed higher FiO₂ and PEEP in MV therapy. ARDS and pneumorrhagia were associated with death from severe influenza virus infection in the univariable analysis. Similarly, lower PO₂ and P/F, and higher OI were also significantly associated with death (both P < .05). Higher P/F (P = .011) was identified as a protective factor against mortality in patients with severe influenza virus infection in the multivariate analysis. Neurologic complications such as IEA were present in a few patients, but it was a strong predictor of ICU admission or even death.^[19] In the United State, IEA was only reported in 6% of influenza-associated deaths among children during one influenza season, while it was reported in 31.8% of influenza-associated deaths in Japan.^[17,22] In our study, 30.8% of the patients who died had IEA with the presence of fever, coma, seizure, or even brain failure, higher than that in the United States but similar to that in Japan. In addition, IEA was associated with death and was an independent risk factor of death in our study. This may support the speculation of an East-Asian ethnic susceptibility to IEA.^[23] Future prospective studies and more samples are needed to evaluate its contribution to mortality in severe influenza virus infections.

Previous studies showed that early oseltamivir treatment can decrease the risks of ICU admission and mortality.^[24] Oseltamivir treatment should be started within 48 hours of illness onset, and patients with good outcome can benefit from the early use of oseltamivir treatment.^[20] Although all the children in our study received oseltamivir treatment, only 11.7% received oseltamivir treatment within 48 hours from illness onset. The reason may be that the clinical signs and symptoms of influenza virus infection in children were nonspecific and influenza is difficult to differentiate from other respiratory infections. In addition, the primary care physicians in our country did not pay enough attention to these patients and realize the importance of early treatment. Therefore, pediatrician should be more likely to initiate antiviral medication earlier for patients suspected with influenza virus infection or use influenza rapid tests to confirm the infection during the influenza season. This may help to reduce the risk of PICU admission and mortality from severe influenza virus infection.

5. Conclusion

Our results showed that severe influenza virus infection occurred mainly in children aged <5 years, especially those aged <3 years, among those admitted in the PICU. The mortality from severe influenza virus infections was 16.9%; most of the patients who died were aged >3 years. In the patients with coinfections, H influenzae and adenovirus were the most common bacterial and viral pathogen isolated, respectively. The presence of an oxygen saturation level of <90% at admission and IEA were the prognostic variables independently associated with death. Higher P/F ratio was a protective factor against mortality in the patients with severe influenza infection. Early recognition and intervention of severe cases can help to decrease PICU admission and mortality.

Author contributions

Data curation: Huifeng Fan.

Formal analysis: Diyuan Yang.

Investigation: Li Huang.

Project administration: Huifeng Fan.

Resources: Gen Lu, Li Huang.

Software: Diyuan Yang.

Supervision: Gen Lu.

Validation: Dongwei Zhang.

Visualization: Dongwei Zhang.

Writing – original draft: Tingting Shi, Zhiqiang Nie.

Writing – review & editing: Tingting Shi, Zhiqiang Nie, Gen Lu.