Burden of paediatric influenza in Western Europe: a systematic review

Background Influenza illness in children causes significant clinical and economic burden. Although some European countries have adopted influenza immunisation policies for healthy children, the debate about paediatric influenza vaccination in most countries of the European Union is ongoing. Our aim was to summarise influenza burden (in terms of health outcomes and economic burden) in children in Western Europe via a systematic literature review. Methods We conducted a systematic literature search of PubMed, EMBASE, and the Cochrane Library (1970-April 2011) and extracted data on influenza burden in children (defined as aged ≤ 18 years) from 50 publications (13 reporting laboratory-confirmed influenza; 37 reporting influenza-like illness). Results Children with laboratory-confirmed influenza experienced hospitalisations (0.3%-20%), medical visits (1.7-2.8 visits per case), antibiotic prescriptions (7%-55%), and antipyretic or other medications for symptomatic relief (76%-99%); young children and those with severe illness had the highest rates of health care use. Influenza in children also led to absenteeism from day care, school, or work for the children, their siblings, and their parents. Average (mean or median) length of absence from school or day care associated with confirmed influenza ranged from 2.8 to 12.0 days for the children, from 1.3 to 6.0 days for their siblings, and from 1.3 to 6.3 days for their parents. Influenza negatively affected health-related quality of life in children with asthma, including symptoms and activities; this negative effect was smaller in vaccinated children than in non-vaccinated children. Conclusions Influenza burden in children is substantial and has a significant direct impact on the ill children and an indirect impact on their siblings and parents. The identified evidence regarding the burden of influenza may help inform both influenza antiviral use in children and paediatric immunisation policies in European countries.


Background
Influenza is a highly contagious infectious disease that is responsible for between 3 and 5 million cases of severe influenza illness each year [1]. Many countries' immunisation guidelines recommend influenza vaccination of the elderly and individuals with underlying chronic medical conditions because of their increased risk of complications due to influenza [2]. However, it is often forgotten that children have the highest influenza attack rates, with annual incidence rates of up to 30% [3]. This translates into significant illness and health care resource use, particularly related to outpatient consultations and hospitalisations [4][5][6][7][8]. Consequently, paediatric influenza leads to substantial economic and societal burdens [4,7,[9][10][11].
Children not only incur significant morbidity associated with influenza but also are primary vectors of influenza transmission in the community [12][13][14][15] because they have limited pre-existing immunity, shed virus at higher viral titers and for a longer period than adults [16], are in close contact with one another in schools and other settings, and can have poor hygiene habits [17]. Dynamic transmission modelling (a tool that can compute the effect of infectious disease transmission) and several communitybased studies have demonstrated that paediatric influenza vaccination, in addition to direct benefit to the vaccinated children, can also indirectly protect other members of the community [15,[18][19][20][21][22][23].
Multiple countries recognise the burden of paediatric influenza and the role of children in disease transmission and recommend paediatric vaccination. The United States Advisory Committee on Immunization Practices has recommended annual vaccination of all children aged 6 months and older since 2008 [9]. Canada, Hong Kong, several Latin American countries, China, Singapore, Korea, and Taiwan also recommend paediatric vaccination, although most policies in these countries are restricted to young children (i.e., those aged ≤ 2 years or aged ≤ 5 years, depending on the country) [24][25][26][27][28][29][30][31]. Finland recommends vaccination of all children aged 6 months to 35 months and has implemented a fully-funded childhood immunisation programme since 2007. Latvia and Slovenia recommend vaccination of children aged 6 months to 2 years. Austria, Estonia, Slovakia, and Saxony in Germany recommend vaccination of children aged 6 months to 18 years [32,33]. In the United Kingdom, the Joint Commission on Vaccination and Immunisation has recently recommended that the influenza vaccination program should be extended to low risk children aged 2 to less than 17 years [34].
Although influenza burden has been studied in many countries, including some European countries [35][36][37][38], the overall burden of seasonal influenza in multiple countries in Western Europe has not been summarised in a systematic manner. Furthermore, although ample data on influenza burden are available for adultsparticularly the elderly and individuals with comorbidities such data for children are sparse. A systematic summary of influenza burden in Western Europe could help inform the paediatric immunisation policy debate [2]. Our aim was to summarise influenza burden (in terms of health outcomes and economic burden) in children in Western Europe via a systematic literature review.

Methods
To retain a manageable volume of literature, this study included eight Western European countries: Austria, Finland, France, Germany, Italy, the Netherlands, Spain, and the United Kingdom (UK). We selected countries known for their influence on decision-making in the European Union (EU), paediatric immunisation policies, or previous epidemiologic investigations of paediatric influenza. We conducted a systematic and comprehensive search of medical literature electronically indexed in PubMed, EMBASE, and the Cochrane Library. We used a detailed search strategy and combined free-text search terms with Medical Subject Headings. The search terms were related to influenza, cost or burden analyses, health-related quality of life (HRQoL), absenteeism, and productivity (see Additional file 1: Table S1 in the Online Appendix for more details). Our search methodology complied with Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) requirements [39].
Our search included studies conducted in children (defined as aged ≤ 18 years) and published since January 1970. We retained only studies that reported data on costs, resource use, absenteeism, and/or impact on HRQoL associated with influenza specifically for children; however, data on absenteeism covered both enrolled children and their parents in some studies. We excluded publications related to the H1N1 pandemic in 2009-2010 because pandemic influenza is substantially different from seasonal influenza. We also excluded publications relating to influenza-like illness (ILI) and retained only those publications reporting laboratoryconfirmed influenza, as the latter assessment method is more specific than all-cause ILI. For the purposes of this article, we refer to "seasonal influenza" as "influenza". We limited publications by excluding any editorials, letters, meta-analyses, practice guidelines, and comments. We examined references of included articles to find additional relevant publications for the review. We also searched Web sites of the health technology assessment bodies of the eight countries of interest for relevant influenza vaccine appraisals.
Two independent reviewers (CR and one other researcher at RTI Health Solutions) screened the identified publications according to pre-defined inclusion and exclusion criteria described above. We next conducted our review in two phases. In Phase 1, the reviewers screened all the titles and abstracts and selected publications for a full-text review. In Phase 2, the reviewers examined the full text of the selected articles and determined the articles' concordance with the study's research aims. To control quality, both reviewers compared their results and resolved all discrepancies at each phase. We fully documented the inclusion and exclusion processes using a PRISMA flow chart that detailed the number of included and excluded articles [40] (Figure 1).
We summarised data in evidence tables by the following domains: (1) health care resource use, (2) costs (including direct medical costs, direct non-medical costs, and indirect [societal] costs), (3) HRQoL, and (4) absenteeism and loss of productivity in patients and their families. We excluded studies not specific to children. We also extracted data on the incidence of influenza and of complications of influenza (e.g., acute otitis media, pneumonia, and lower respiratory tract infections) as secondary outcomes from the articles on economic burden. We recorded the following influenza assessment methods: via laboratory methods (virology or polymerase chain reaction) or via a list of ILI symptoms. However, we report only studies of laboratory-confirmed influenza, and the ILI studies are summarised in the supplementary appendix (see Additional file 1: Table S2 in the Online Appendix).
We assessed the quality of published studies according to the Oxford Centre for Evidence-based Medicine Levels of Evidence [41]. Finally, we assessed each study's risk of bias, particularly focusing on the study population (community members or a particular subset of patients). An independent quality-control specialist confirmed data-extraction accuracy.

Results
The results below represent studies of laboratory-confirmed influenza. Results from studies of ILI can be found in Additional file 1: Table S2 in the Online Appendix. Figure 1 presents the details of the literature search and review process. From the initial literature search (conducted in 2009), we identified 747 unique, potentially relevant articles. After Phase 1 screening of titles and abstracts and searches of relevant Web sites, 171 articles were considered for more detailed evaluation. Of these, three articles were duplicates, three articles could not be sourced, and a further 129 articles were excluded at Phase 2 screening. Another three articles were identified in the reference lists of the included articles. Therefore, a total of 39 articles were eligible for inclusion.

Articles included
From the literature search update (conducted in 2011), we identified a total of 406 unique, potentially relevant articles. After Phase 1 screening and searches of relevant Web sites, 48 articles were considered for detailed evaluation, and of these, we excluded 41 articles at Phase 2 screening. Another four articles were identified from prior knowledge of one of the authors; thus, 11 additional articles were eligible for inclusion. In total, 50 articles were eligible for inclusion from the two stages of the review. Of those, 13 articles reported relevant data for children with laboratory-confirmed influenza, and the remaining 37 articles reported data on ILI. Table 1 describes the study details, quality, and the risk of bias for the 13 articles that included data based on laboratory-confirmed influenza in children. Most studies investigated the burden of confirmed influenza in children 15 years of age or younger, and children's ages varied among studies.
Study quality and types varied. The 13 articles were classified as follows: two were randomised, controlled trials that had placebo groups (therefore providing data on influenza burden); seven were prospective cohort studies with good follow-up; three were prospective cohort studies with poor follow-up; and one was an ecological study.

Incidence of influenza
Studies reported influenza rates across a range of seasons and settings. Table 2 summarises influenza incidence and prevalence rates in various study settings. Annual incidence of influenza was 62 per 1,000 in fullterm infants aged 1 year or younger [50]. Annual incidence of lower respiratory tract infection due to influenza was 1.1 per 100 children-years in children aged 36 months or younger [45]. During fall, winter, and spring, when infection risk is highest in European countries, influenza incidence rates were 167 per 1,000 in children aged 13 years or younger ( [37]).
Influenza was a common culture-confirmed pathogen in children seeking health care due to ILI. In a randomised, controlled trial of children with ILI, influenza was confirmed in 25% of children aged 1 through 3 years during an active epidemic period (14 January -9 April 2008) in Finland [42]. In children presenting to the emergency department (ED) with ILI, influenza was identified in 33% (0-11 months old) and 49% (≤ 36 months old) during a 4-week local influenza peak (weeks 3 -6 in 2002) in France [36,43]. In Italy, influenza was identified in children presenting to the ED during 5 winter months: 15% in those younger than 15 [38,48]). Among children hospitalised with ILI during 37 days of a local influenza epidemic 1 January -6 February 2002), influenza was confirmed in 51.8% of those who tested negative for respiratory syncytial virus [44]. Among hospitalised children, the prevalence of influenza was highest in the toddler group (2-3 years; 68.9%), followed by infants (≤ 1 year; 42.4%), pre-school children (3-5 years; 41.2%), and school-aged children (6-15 years; 36.8%) [44].
Febrile seizures [43,44,47], respiratory distress [44], and croup [36,43,46,47] were reported only in ED-or hospital-based studies. Wheezing and pneumonia were less prevalent in community-or primary care-based studies than in studies conducted in ED or hospitals ( Figure 2). Respiratory distress was more common in children younger than 3 years than in those younger than 15 years [44].

Economic burden
Health care resource use Studies reported health care use primarily through hospitalisations, medical visits, and medications (antibiotic, antipyretic, or analgesic). Hospitalisation rates of children with laboratory-confirmed influenza ranged  . b Studies that investigated the impact of influenza at the whole population level rather than the impact in a particular subset of patients (e.g., children admitted to hospital with fever). c Resource use, absenteeism, and HRQoL also reported but associated with ILI only (not associated with confirmed influenza).  from 0% to 20% (Figure 3), and the mean length of stay ranged from 1.8 to 7.9 days (Figure 4). As would be expected, hospitalisation rates reported in physician office-based studies (0.7% [37]) and community studies (0.3% [35]) were lower than in ED-based studies (20% [43], 10% [36], 5.6% [47]) due to more severe influenza illness in children presenting to the ED than in those treated at a physician's office. Two studies with identical country, influenza season, and health care settings demonstrated that infants (aged 0-11 months) presenting to the ED were more likely to be hospitalised (20%) and stayed in the hospital longer (mean: 7.9 days) [43] than children 0 through 3 years of age (10%; mean: 1.8 days) [36].
Medication use was high in children with influenza and varied by the study setting. Antibiotics were prescribed to 28% of children in community-based studies [35]; to 43% of children in primary care-based studies [37]; to 52% [50], 42% [36,43], or 55% [38] in the ED studies; and to 34% in hospital-based studies [44] (see Additional file 1: Table S3 in the Online Appendix).

Costs
Multiple studies reported medical and societal costs associated with paediatric influenza, but only two reported results based on laboratory-confirmed illness (Table 3) [37,45]. In the first study, the annual total cost of paediatric community-acquired influenza in Germany was estimated at €7,530,105. This study considered children between 0 and 36 months of age with lower respiratory tract infection and estimated 2,913 influenza cases per year, with a median hospitalisation cost of €2,585 per case. Hospitalised paediatric influenza cases were more expensive than the officebased ones (€2, 597 vs. €223, year 2002 costs [45]).
The second study was conducted in Italy and considered both direct (medical) and indirect (absenteeism) costs. The care for children with laboratory-confirmed influenza (averaging €132 per patient) was 32% more expensive than that for children with influenza-negative ILI [37]. An influenza hospitalisation cost approximately €3,000, and associated costs averaged €22.40 among all influenza cases (the total cost of hospitalization divided by the total number of children with influenza, regardless of their hospitalization status). The average indirect cost of working days lost by parents (€70) had the greatest impact on the average total cost of an influenza case. Furthermore, the cost of care for young children was greater than the cost of care for older children: €153 for children younger than 2 years, €148 for children aged 2 to 5 years, and €74 for children aged 5 to 13 years.

Absenteeism
Childhood influenza resulted in absenteeism for sick children, their siblings, and parents (Additional file 1: Table S3 in the Online Appendix and Figure 5). Absenteeism in ill children was assessed through absence from school or day care [35,38,42,[46][47][48] or lost work or school days for the parents or siblings of infected children [35][36][37][38]42,43,[46][47][48]. The rates of child absenteeism in the community-based study were 76% (in children < 3 years), 73% (3-6 years), and 77% (7-13 years) [35]; absenteeism was not assessed in other studies. The duration of absenteeism depended on influenza severity (Additional file 1: Table S3 in the Online Appendix). For example, children in ED-based studies were absent from   school for a median of 12.0 days [46,47] and 5.1 days [38], children in physician office-based studies were absent from school for a median of 4.0 days [42], and children in community-based studies were absent from school for a mean of 3.6 days (calculated as weighted mean from Heikkinen et al. [35]). One study reported that some children missed up to 15 school days [46].
Parental absenteeism was common: between 11.2% and 61% of parents of children with laboratory-confirmed influenza took absence from work for their own influenza illness or to care for their children (Additional file 1:  [36,43]). The mean duration of work absenteeism for parents ranged from 1.3 to 6.3 days [35][36][37][38], and the median work absenteeism for parents ranged from 2 to 4 days [42,46,47].
Siblings of children with influenza also missed school days due to subsequent ILI. For example, siblings of children with influenza lost a mean of 1.3 days [38,48] or a median of 5.0 to 6.0 days [46,47] of school or work.

Health-related quality of life
A single study reported HRQoL associated with confirmed influenza specifically in children [49]. That study (a prospective, randomised, double-blind, placebocontrolled trial) compared trivalent-inactivated vaccine with no vaccine. It assessed HRQoL in children with asthma through the Paediatric Asthma Quality of Life Questionnaire (range of possible scores: 1-7, with 7 indicating the highest HRQoL) [49]. Both the vaccinated and the placebo groups experienced worsening of asthma-related HRQoL. However, the worsening was less severe in children in the vaccinated group than in children in the placebo group, as measured by the total score (-0.40 vs.

Data availability by country
Data availability varied by country. Studies reporting burden of laboratory-confirmed influenza were conducted in Italy (5 articles), France (3 articles), Finland (2 articles), the Netherlands (2 articles), and Germany (1 article). No articles that reported burden data relating to confirmed influenza were identified from Austria, Spain, or the UK.
Data on health care resource use were most commonly reported in Italy and France, whereas costs of laboratoryconfirmed influenza were reported for Germany and Italy. Absenteeism data were most commonly reported for Italy, whereas HRQoL impact of confirmed influenza was reported only in the Netherlands.

Discussion
The current literature review provides a comprehensive summary of the available influenza burden data in children from selected Western European countries. The incidence and prevalence rates of ILI and laboratoryconfirmed influenza, as well as the rates of complications summarised by our review, were similar to those reported in the United States and Asia [51][52][53]. The current analysis demonstrated that, along with substantial clinical burden, influenza results in a significant economic burden, including hospital stays, physician office visits, and use of antibiotics and antipyretics. Consistent with other literature, our study found that influenza also resulted in significant illness-related absenteeism among the enrolled children and their siblings and parents. These results concurred with a 2008 review of the Working days lost by fathers: €26.60 (€90.40) a ; €29.40 (€111.40) a ; €4.20 (€39.10) CI = confidence interval; ED = emergency department; ILI = influenza-like illness; LRTI = lower respiratory tract infection; NA = not applicable; SD = standard deviation. Note: For further details on study design and patient characteristics, see Table 1 a P < 0.05 vs. age > 5 years.
literature from North America, Western Europe, Asia, and Australia: in studies from these countries, parental absenteeism ranged between 1.0 day and 5.9 days per influenza episode due to either a parent's own illness or the child's illness [54].
As expected, influenza burden depended on the severity of child's illness (as reflected by the health care settings) and the child's age. Children admitted to an ED or a hospital, in general, had more complications (wheezing, pneumonia, croup, respiratory distress, and febrile seizures or convulsions), were absent from school or day care for more days, and were more likely to receive an antibiotic prescription than their counterparts without these characteristics. Studies reporting health outcomes by children's ages demonstrated that younger children were more likely to experience respiratory distress, child and parent absenteeism, hospitalisation, and antibiotic use than older children.
The small number of studies made cross-country comparisons difficult. It is challenging to draw fair crosscountry comparisons between studies, because several other factors contributed to the studies' variations in the results: age, health care setting, and influenza season. Differences in the paediatric health care systems across the EU also may contribute to difficulties in comparing and interpreting burden-of-disease information from different countries. For example, the UK, the Netherlands, and Finland have a general practitioner-led system; Spain has a paediatrician-led system; Italy, France, Germany, and Austria have a combination of the two in which school-age children can choose to see either a general practitioner or a paediatrician [55]. These differences in the health care systems may lead to differences in health care-seeking behaviours and ultimately to variance in influenza-reporting patterns [56].
The published literature had a number of limitations, primarily related to the design of the reported studies. Heterogeneity existed among influenza seasons, ages of studied children, and study methods for evaluating influenza illness. Additionally, all cost estimates were not directly collected but imputed (the rate of health care resource use multiplied by the reimbursement or standard prices associated with these resources) [37,45]. Figure 5 Children's absence from day care or school and parents' absence from work associated with paediatric influenza. ED = emergency department. * Interquartile range. † This is the number of children with confirmed influenza in this article, so the value is the same for the row showing the whole parent population, the mothers only, and the fathers only. The article does not provide the number of mothers and the number of fathers. Notes: Each point represents a mean or median value from one of the identified studies: ♦ = mean value; ▪ = median value. Effect sizes around each point represent standard deviation if the value is a mean; range represents minimum-maximum range if the value is a median.