The economic burden of influenza among adults aged 18 to 64: A systematic literature review

Abstract While the economic burden of influenza infection is well described among adults aged 65 and older, less is known about younger adults. A systematic literature review was conducted to describe the economic burden of seasonal influenza in adults aged 18 to 64 years, to identify the main determinants of direct and indirect costs, and to highlight any gaps in the existing published evidence. MEDLINE and Embase were searched from 2007 to February 7, 2020, for studies reporting primary influenza‐related cost data (direct or indirect) or absenteeism data. Of the 2613 publications screened, 51 studies were included in this review. Half of them were conducted in the United States, and 71% of them described patients with influenza‐like illness rather than laboratory‐confirmed disease. Only 12 studies reported cost data specifically for at‐risk populations. Extracted data highlighted that within the 18‐ to 64‐year‐old group, up to 88% of the economic burden of influenza was attributable to indirect costs, and up to 75% of overall direct costs were attributable to hospitalizations. Furthermore, within the 18‐ to 64‐year‐old group, influenza‐related costs increased with age and underlying medical conditions. The reported cost of influenza‐related hospitalizations was found to be up to 2.5 times higher among at‐risk populations compared with not‐at‐risk populations. This review documents the considerable economic impact of influenza among adults aged 18 to 64. In this age group, most of the influenza costs are indirect, which are generally not recognized by decision makers. Future studies should focus on at‐risk subgroups, lab‐confirmed cases, and European countries.


| INTRODUCTION
Seasonal influenza outbreaks can occur every year worldwide, causing substantial morbidity and mortality. The World Health Organization estimates that worldwide each year, influenza epidemics result in up to 5 million severe cases of disease and up to 650,000 respiratory deaths across all ages. 1  Although influenza can affect any person and age group, certain populations are at greater risk of exposure to infection or of developing severe disease. Population groups at higher risk of developing severe outcomes are pregnant women, young children, adults aged 65 years and older, and those with underlying conditions. 3 For example, in the United States, patients aged over 65 years account for 70% to 85% of influenza-related deaths. 4 In addition to the aforementioned at-risk populations, healthcare workers (HCWs) are also a key population recommended for seasonal influenza vaccination. HCWs may be highly exposed to seasonal influenza viruses through contact with patients, which increases their own risk of illness and their potential to spread the disease to others. 5 Influenza imposes a large economic burden to healthcare systems and to society. Notably, for the adult American population, a study published before the COVID-19 pandemic has demonstrated that influenza accounted for 65% of the total economic burden caused by vaccine-preventable diseases. 6 The proportion of costs attributable to influenza was particularly high among the 19-to 49-year-old age group, in which influenza accounted for 85% of the vaccinepreventable disease economic burden, compared with 67% for 50-to 64-year-olds and 55% among those aged 65 and older. 6 Influenza costs originate from inpatient and outpatient care settings and substantial indirect costs related to lost productivity. 7 In the United States, total annual direct medical costs have been estimated to be US$3.2 billion, whereas indirect costs accounted for US$8.0 billion. For the latter, 67% were engendered by the 18-to 64-year-old age group. 8 In the European Union, costs of seasonal influenza are estimated at €6 billion to €14 billion annually. 9 While the economic burden of influenza infection among adults aged 65 and older has been well described, [10][11][12]  to 64 years) and at-risk status (at-risk/general population) was completed to assess whether specific groups are associated with a higher economic burden.

| METHODS
The SLR followed an a priori protocol that specified search terms, inclusion and exclusion criteria, and a methodological approach to the review aligned with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 13 The protocol was not publicly registered. Research questions and inclusion and exclusion criteria were defined in terms of population, intervention, comparator, outcomes, study design, and time frame (PICOS-T), as seen in For consistency, this publication uses the term "at-risk" to refer to studies that used the terms "at-risk" or "high risk" or identified one of the at-risk groups of interest as defined in the review methods (Table 1). The most reported outcomes were hospitalization costs (25 studies), overall direct costs (22 studies), and workdays lost (20 studies).
Costs of workplace absenteeism were reported in 10 studies, total costs in five studies, and overall indirect costs in two studies. All extracted cost data are available in the supplemental tables.
Finally, five studies looked at influenza costs specifically among HCWs and primarily focused on influenza-related absenteeism.

| Overall direct costs
A total of 22 studies reported overall direct costs attributable to influenza, of which eight studies investigated costs in LCI and only two provided data for an at-risk population among those looking at ILI.
Thirteen studies reported individual costs (cost per influenza case), and 10 studies estimated aggregated costs at the population level across the specified age group.
Reporting of individual costs was heterogenous and included the following denominators: influenza case, influenza case medically attended, case hospitalized or presented to an ED, or hospitalized case.
Moreover, some studies did not report all the direct costs items, and others did not specify how the costs were attributed to their categories.
On average, in Europe, overall direct costs ranged from €56 in Italy per self-reported influenza case 14 to €90 in Germany per medically attended case. 15 In the United States, overall direct costs ranged from US$161 to US$363 per medically attended case. 16 Moreover, an Italian study showed that the overall direct cost per case who presented to an ED or who were hospitalized was 1.6 times higher among those >50 years old with at least one risk factor compared to those without risk factors. 19 Concordantly with individual results, aggregated cost data showed that overall direct cost attributable to influenza increases with age and was significantly higher in the elderly compared to adults less than 65 years old. 25 The proportion of patients taking sick leave increased with age within the 18-to 64-year-old age group and for at-risk populations compared with a general population in a US study. 18 Moreover, as highlighted in a Belgian study, increased levels of healthcare resource utilization during an influenza episode are associated with more absenteeism. More patients presenting to an ambulatory clinic had an interruption of daily activity than did patients remaining in the community; all hospitalized patients experienced absenteeism. 21 The reported average duration of workplace absence per influenza patient with sick leave ranged from 6.0 to 8.2 days in European studies 15,22,23,34,37 and from 14 to approximately 24 working hours in studies from North America. [38][39][40] In studies of influenza patients regardless of whether they had access to sick leave, reported durations of absenteeism were consistently lower, ranging from 0.6 days in the United States to 3.3 days in Germany. 15,17,18,41,42 Across all populations of influenza patients considered, the average duration of absenteeism increased with age within the 18-to 64-year-old age group. 18,23,25,37,39 Moreover, in two studies reporting workdays lost specifically for hospitalized patients, length of absence was substantially higher compared with outpatients. 21,25 Finally, data from US studies showed that the average duration of sick leave among the at-risk population was up to 1.8 times higher compared with the general population. 18,25 Among the studies that focused on HCWs, three provided average lengths of influenza-related sick leave. In Italy, where influenza is responsible for 11.9% of all-cause work absences in HCWs, they had an average of 4.6 days of absence when they had influenza. 43 Two additional European studies showed this duration varied substantially across categories of HCWs from less than 1 day to over a week. 44,45 Furthermore, a US study reported that each influenza season causes a median of 12 hours of working time lost per HCW aged ≥18 years, whether or not they contracted influenza. 46 Only one study incorporated presenteeism as an outcome. While the length of workplace absence was generally lower in the United States compared with European countries, this study suggested that lost productivity is substantial. At 7 to 17 working days post-disease onset, adults aged ≥18 years reported 67% of their working hours were lost due to absenteeism or presenteeism. 47  59-year-olds) versus older adults (≥65 years or ≥60), largely driven by indirect costs. 14,15 The German study showed that 82% of the average total cost per episode is attributable to indirect cost in adults. 15 Aggregated total costs increased with patient age within the 18 to 64 age group. 25,26 In the general US population, costs continued to increase for those aged ≥65, 25

| DISCUSSION
The economic burden of seasonal influenza is substantial among adults. Despite methodological heterogeneity observed across studies identified by this SLR, indirect costs have been reported to be the primary driver of total costs of influenza in adults, at both individual and population levels. Indeed, whatever the scope of indirect costs considered, they accounted for up to 82% of the total cost per influenza case 15 and 70% to 95% of the total costs of influenza when measured at population level. 23,25,26 Three studies provided data allowing to calculate this outcome by age group and showed that 79% and 83 to 99% of the economic burden of influenza is attributable to indirect costs respectively in populations aged 18 to 59 years and 18 to 64 years. 23,25,26 These values are consistent with the 83% estimated in the 18-to 64-year-old population in a recent US study outside the scope of this review. 8 When considering direct costs only, hospitalization costs were the main driver, with 73 to 75% of direct costs attributable to hospitalization. 26,27 Within the target age range of this review, older patients (i.e., those aged 50 to 64) generally had higher influenza-associated costs compared with younger adults (i.e., those aged 18 to 49). Additionally, patients with one or more risk factors for influenza complications had higher costs than those patients not at risk. Healthcare resource utilization data were opportunistically extracted alongside costs and showed similar trends with age groups and the presence of risk factors. These data should reflect the fact that severe forms of influenza are more frequent with age and the presence of comorbidities.
While this review focused on adults aged 18 to 64 years, many of the included publications also reported data for elderly (≥65 years) patients, which were also opportunistically extracted. Across studies that provided age-stratified data, the average total cost per case was higher in adults aged <65 than in the elderly. This could be related to the major role played by indirect costs, which are more substantial in those aged 18 to 64 partly due to higher employment rates. Conversely, studies have shown that the overall direct costs per case were higher in the elderly than in younger adults. This may be explained by the more severe impact influenza can have generally on these patients, which would result in increased healthcare use and therefore higher costs. Nevertheless, this trend was not observed when considering only hospitalization costs. Most of the studies reported higher hospitalization costs per case in patients aged 50 to 64 than in the elderly. One reason for this could be a kind of selection bias with the elderly more easily hospitalized due to the greater risk for complications in this age group. Hospitalized young adults may then have more severe disease than the elderly and require more medical intervention.
It would be interesting to confirm this trend and hypothesis in future studies. However, when looking at hospitalization costs at population level, they were higher in elderly patients than in younger adults because of the increase of hospitalization incidence with age. 26,27 The ability to draw robust conclusions regarding the economic burden of influenza in adults is constrained by many limitations, first those inherent to systematic reviews and second those related to the included studies in this review.
While studies reporting costs due to influenza in patients aged 18  This review upholds the complexity of assessing the full economic burden of a disease. 49 Even if measuring direct costs could be relatively straightforward when using insurance claims databases, estimating indirect costs is much trickier and is associated with significant uncertainty, particularly for absenteeism-related data. Indeed, in most studies, absenteeism data were not collected from insurance databases or from employers' records, but from patient surveys or analyses of doctors' certificates to collect durations of sick leaves. Patient surveys are associated with at least response and recall biases, while doctors' certificates assess the prescribed duration of sick leave rather than the actual sick leave duration and underestimates absenteeism because only patients who visited a GP are counted. Given that indirect costs account for nearly 80% of the economic burden in adults aged 18 to 64 years, 8 which represents most of the workforce, efforts should be made to better document influenza-related absenteeism.

| CONCLUSIONS
This review identified indirect costs as the main driver of economic burden due to influenza in the 18-to 64-year-old age group and hos- resources; supervision; validation; writing-review and editing.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12963.