Retrospective review of factors associated with severe hospitalised community‐acquired influenza in a tertiary paediatric hospital in South Australia

Background Influenza infection can result in severe disease with debilitating complications. Young children have the highest rate of influenza hospitalisations with various factors influencing influenza susceptibility and severity. Objectives This study aimed to determine the disease burden and assess risk factors for severe hospitalised influenza in South Australian children under 5 years of age. Methods Influenza admissions to the tertiary paediatric hospital in South Australia from 2008 to 2012 were identified. Data from laboratory‐confirmed influenza cases were collected, including infecting influenza strain, co‐infections, prematurity, pre‐existing medical comorbidities and other potential risk factors. Predictors of high‐level care were assessed using logistic regression. Results A total of 267 children with laboratory‐confirmed influenza were hospitalised. Of these, 147 admissions (53%) occurred in children without underlying medical risk factors. Eighteen children (7%) required high‐level care, of which 11 (61%) had no underlying medical risk factors. No deaths were reported. The majority of children were unimmunised against influenza. Co‐infections were identified in 40% of children (n = 107). Influenza B was associated with a requirement for higher care (OR 3.7, CI 1.3–10.9, P = .02) as was a history of food allergies (OR 9.7, CI 1.5–61.4, P = .02) and iron deficiency anaemia (OR 4.8, CI 1.4–16.1, P = .01). Conclusions Influenza can be a severe illness, even in children without underlying medical conditions. The identification of Influenza B strain, history of food allergies and iron deficiency anaemia as predictors of severity in hospitalised cases warrants further investigation and may have important implications for preventative strategies to reduce the burden of childhood influenza.


| INTRODUCTION
Influenza is a highly contagious infection that can result in debilitating complications. There are many factors which may influence the susceptibility and severity of disease and disease outcomes. These include the influenza strain and subtype, 1-3 current methods of treatment and prevention as well as factors associated with the infected individual such as age, 4 the presence of co-infections and comorbidities. 5 Underlying chronic medical conditions including cardiac disease, neurological conditions, 2 impaired immunity, 6 obesity 5 and chronic respiratory conditions including asthma 3 may affect the severity of influenza infection in children. Many previous papers have identified risk factors during the H1N1 Influenza pandemic, with less known about predictors of severe disease during other non-pandemic influenza seasons.
Children under 5 years of age have increased susceptibility to becoming infected with influenza and are at risk of more severe disease. 7 They have the highest rate of influenza notifications compared with all other age groups and the highest hospitalisation rates. 8 In 2012, there were 44 571 cases of laboratory-confirmed influenza in Australia with 6286 of these cases reported in South Australia. 9 The proportion of influenza notifications occurring in children aged less than 5 years between 2008 and 2014 in Australia varied between 10% (in 2009) and 15% (2008 and 2012). Children are the major viral reservoir for influenza, spreading it to their families and the community. 10,11 A US study of childhood mortality from seasonal influenza indicated that the majority of childhood deaths attributed to influenza occur in those less than 5 years of age. 12 Influenza infection can range from asymptomatic infection to caus- The aims of this study were to determine the burden of laboratoryconfirmed influenza requiring hospitalisation in children under 5 years of age in South Australia from 2008 to 2012 and investigate factors associated with more severe disease (defined as requiring intensive or high dependency care). In particular, the study aimed to identify the different types and strains of influenza virus causing disease in hospitalised children to assess the impact of strain variation and known and potential risk factors such as pre-existing comorbidities, co-infections and prematurity on severity of disease.

| MATERIALS AND METHODS
We performed a retrospective observational study of children under 5 years of age who were admitted with a diagnosis of influenza to the tertiary paediatric hospital in South Australia, between January 2008 and December 2012. We used ICD-10-AM (International classification of diseases, version 10, Australian modified) codes J09, J10 and J11 to identify potential hospital admissions (admission for influenza due to a certain identified influenza virus, influenza due to other identified influenza virus and influenza with an unidentified virus).
Eligible cases required confirmed laboratory evidence of influenza (e.g. polymerase chain reaction or positive culture) as per Australian Government Department of Health and Ageing case definitions for confirmed influenza infection. Nosocomial cases were captured but excluded from analysis as disease severity may be different.
Nosocomial cases were defined as patients admitted to hospital for other primary medical reasons who subsequently contracted an influenza infection during hospital stay (no symptoms of influenza at time of admission and influenza diagnosis >48 hours following admission).
The study was reviewed and approved by the Women's and Children's Health Network Human Research Ethics Committee. in the year of or prior to admission or receiving the vaccine less than 2 weeks prior to presentation. Definitions for laboratory results above or below normal range were based on normal ranges for age.

| Data processing and statistical analysis
Data were analysed using STATA ® IC11. Weight for age Z-Scores were calculated using EpiInfo™ 3.5.4. Children with a weight for age Z-score ≥95% percentile were considered obese for analysis as per Australian clinical practice guidelines. 18 Hospitalised influenza cases were classified as severe if they required high-level care (ICU/High Dependency Unit (HDU) admissions) and were compared with those that did not (short-stay ward or general admission). Associations with severity were investigated with logistic regression. P values <.05 were considered significant. Results were reported as odds ratios (OR) with 95% confidence intervals.
Univariate and multivariable logistic regression was used to determine which factors were associated with severe disease. Multivariable models included all variables that had complete data (>80%) and P values <.1 on univariate regression analyses for associations with severity. A restrictive P value of ≤.1 was chosen due to the small number of children who were classified as severe (n = 18). In addition, we included age category (<2 or ≥2 years of age) and pre-existing risk factors (asthma, cystic fibrosis, chronic lung disease, heart disease, diabetes, chronic metabolic disease, kidney disease and failure, immunocompromised, neurological disorder and blood disorder as identified in the Australian Immunisation Handbook 19 ) in the multivariable model, based on their clinical relevance and potential for confounding.

| RESULTS
A coding review identified 510 influenza admissions in children less than 5 years of age, of which 279 were laboratory-confirmed cases. Of these, 267 admissions were included in the analysis and 12 excluded as they were hospital-acquired (nosocomial) influenza cases. Almost all of the influenza cases (260/267, 97.4%) were confirmed by PCR.
There were no reported deaths among the 267 cases.

| Admissions due to influenza
The number of influenza admissions per year was highest in 2012 (n = 64) and lowest in 2008 (n = 37) ( Table 1)

| Comorbidities and Potential Risk Factors
Comorbidities were analysed as potential predictors of severity in a univariate analysis (  2 years are also a higher risk age group for iron deficiency due to factors such as inadequate intake and amplified iron requirements for rapid growth. 20 In previous literature, iron deficiency has been reported as a comorbidity associated with influenza A. 21 In addition, anaemia has been identified as a risk factor for respiratory failure in children admitted to hospital for community-acquired pneumonia. 22 Iron has many important functions in the body including regulation of the immune system, 23,24 and therefore, deficiency could compromise immune function and impact on acquisition and severity of infections including influenza.   to poor knowledge about the vaccine or concern about side effects or lack of immunisation provider recommendations, 33 all of which are potential avenues for further exploration.

| DISCUSSION
Another issue, not explored in this study, is the impact of host genetic factors on influenza disease susceptibility and/or severity.
Genetic polymorphisms in certain immune effector proteins such as the interferon-induced transmembrane protein-3 (IFITM3) 34 and the tumour necrosis factor-alpha (TNF-α) and interleukin genes 35 have been reported in the literature as being linked to poorer outcomes with influenza. An IFITM3 genetic variant, rs12252-C, which is more prevalent among East Asian populations, has been shown to increase influenza susceptibility and disease severity. 36 Nosocomial influenza infections, whilst infrequent, were also identified as an important group impacting on influenza disease burden as the majority required high-level care.
Our results may be subject to bias due to the small number of cases requiring high-level care, which is one of the major limitations of our study. This was a retrospective study, which limited information resources and is potentially subject to bias and conducted in a tertiary care hospital where more severe cases may be transferred. Nevertheless, further investigation of the association between increased disease severity in children with food allergies and iron deficiency anaemia should be considered. New quadrivalent vaccines containing two influenza B strains in addition to two influenza A strains may provide additional protection against severe Influenza B-related disease.
In summary, the associations observed between disease severity and influenza type B, iron deficiency anaemia and food allergies are newly identified factors that affect influenza severity in South Australian children under 5 years of age. These findings have important implications for preventive strategies and warrant further study.
Confirmed iron deficiency anaemia or food allergy could be included as additional conditions which meet eligibility for government-funded influenza vaccination. Furthermore, the association between Influenza B and increased disease severity support the recommendation of quadrivalent influenza vaccine for children.

CONFLICT OF INTEREST AND SOURCE OF FUNDING
There are no conflict of interests to declare by the authors that are directly related to this manuscript. However, HM, MC and SM's
institution has received funding for investigator-led research from bioCSL, Novartis and GSK.

Helen Marshall is supported by the National Health Medical Research
Council Career Development Fellowship (1084951).