Burden of severe illness associated with laboratory confirmed influenza in adults aged 50–64 years: A rapid review

Abstract Background While the high burden of illness caused by seasonal influenza in children and the elderly is well recognize, less is known about the burden in adults 50–64 years of age. The lack of data for this age group is a key challenge in evaluating the cost‐effectiveness of immunization programs. We aimed to assess influenza‐associated hospitalization and mortality rates and case fatality rates for hospitalized cases among adults aged 50–64 years. Methods This rapid review was conducted according to the PRISMA; we searched MEDLINE, EMBASE, Cochrane, Web of Science, and grey literature for articles and reports published since 2010. Studies reporting rates of hospitalization and/or mortality associated with laboratory‐confirmed influenza among adults 50–64 or 45–64 years of age for the 2010–11 through 2019–20 seasons were included. Results Twenty studies from 13 countries were reviewed. Reported rates of hospitalization associated with laboratory‐confirmed influenza were 5.7 to 112.8 per 100,000. Rates tended to be higher in the 2015–2019 compared with the 2010–2014 seasons and were higher in studies reporting data from high‐income versus low and middle‐income countries. Mortality rates were reported in only one study, with rates ranging from 0.8 to 3.5 per 100,000 in four different seasons. The case fatality rate among those hospitalized with influenza, as reported by population‐based studies, ranged from 1.3% to 5.6%. Conclusions Seasonal influenza imposes a significant burden of morbidity on adults 50–64 years of age but with high heterogeneity across seasons and geographic regions. Ongoing surveillance is required to improve estimates of burden to better inform influenza vaccination and other public health policies.


| INTRODUCTION
Annual seasonal influenza epidemics exact considerable morbidity and mortality worldwide with an estimated 3-5 million cases of severe illness and 290,000-500,000 deaths each year. 1 The highest burden is in young children and the elderly, 2-6 and many countries' immunization recommendations emphasize vaccinating children aged 6-59 months and adults aged ≥65 years. [7][8][9] Less is known about the burden of disease in adults aged 50-64 years and the vaccination coverage in this age group remains relatively low. In the 2015-2016 season, 37.9% and 43.6% of adults aged 50-64 years were vaccinated in Canada and the United States (US), respectively, compared with 64.6% and 63.4% of adults ≥65 years. 10,11 Similarly in South Korea, the influenza vaccination coverage in 2016 was 31.9% in adults aged 50-64 years and 81.7% in those aged ≥65 years. 12 There is increasing evidence that the burden of influenza in adults aged 50-64 years is sufficient to make influenza vaccination programs cost-effective for this age group. [13][14][15][16][17][18] For instance, in Australia, a policy to recommend and pay for influenza vaccines for all 50-64 year olds was considered likely to be cost-beneficial for both health care payers and government, with an incremental cost-effectiveness ratio of $22,408 per quality-adjusted life year gained. 14 However, these and other authors have noted that the lack of adequate epidemiological data regarding the burden of disease was a key challenge in assessing the cost-effectiveness of influenza vaccination programs for this age group. 19 The objective of this literature review was to describe the burden of laboratory-confirmed influenza among adults aged 50-64 years for the 2010-2011 through 2019-2020 influenza seasons. We aimed to assess influenza-associated hospitalization and mortality rates, as well as case-fatality rates for hospitalized cases.

| METHODS
We performed a rapid literature review following the guide by Tricco et al. 20

and the Preferred Reporting Items for Systematic reviews and
Meta-Analyses (PRISMA). 21 We used database subject terms and text words for influenza, hospitalization or intensive care or ventilation or mortality, and middle-aged adults (50 to 64 years). We limited the search to articles published since 2010. We refined and expanded search terms as required to ensure the retrieval of sentinel references. Detailed search strategies are available in section 4 of the Supporting Information.
The titles of all citations returned by the searches were scanned for relevance by one reviewer (PK). Titles were reviewed for the presence of any of the following key words: epidemiology, surveillance, burden, mortality, deaths, pneumonia, influenza-like illness, hospitalization, or hospital admission. Articles with titles that included at least one of these words and the term influenza were retained and the abstract was reviewed. If an abstract contained information on disease burden, it was considered relevant and the full-text was obtained for further screening.
Articles and reports were eligible for full-text review if: i. influenza diagnosis was laboratory-confirmed (i.e., by testing of respiratory specimens via reverse transcriptase or real-time polymerase chain reaction [rt-PCR], culture, enzyme-linked immunosorbent assay, direct fluorescent antibody, or other rapid antigen test); ii. at least one of the following outcomes was assessed: incidence of influenza-associated hospitalization, mortality, or case fatality rate in a population-based sample; and iii. data could be extracted for adults aged 50-64 or 45-64 years.
Studies were not eligible if they presented data only for the 2009-2010 influenza season or earlier, for nosocomial influenza only, or for types/subtypes of influenza other than seasonal influenza (e.g., influenza H7N9). Three types of articles with relevant data were identified. They were categorized as (a) population-based surveillance studies including those in which the population of the catchment area was estimated; (b) ecologic studies that provided burden estimates based on excess deaths or hospitalization during periods of influenza activity; and (c) surveillance studies that applied age group-specific influenza positivity to the rate of people hospitalized with a severe acute respiratory illness (SARI).

| Data extraction, analysis, and reporting
Data extraction was performed by one reviewer (PK) using a template that collected details on study characteristics including: study identifier, year of publication, study design, country (ies), region(s), season(s), and demographic characteristics of participants including age range and number of participants/patients by underlying chronic disease/immune compromised (see section 3 of the Supporting Information for more details).
Outcome information included study-specific outcome definitions, criteria for influenza testing, timing of relevant specimen collection, and cause of death. The outcome measures included were crude and adjusted rates of influenza-associated hospitalizations and mortality as well as case fatality rates for hospitalized cases. Data were collected on the variables used to adjust estimates, number of people at risk in each exposure group and stratum, and any risk factors assessed for influenza-associated death or hospitalization. When available, outcome data were extracted by influenza type/subtype and for overall rates. All incidence rates were reported per 100,000 persons.
One reviewer (PK) assessed each study for the risk of bias using the Risk of Bias in Non-randomized Studies of Intervention (ROBINS-I) tool, the preferred method for Cochrane review of nonrandomized studies. 22,23 Studies assessed as having critical biases were excluded from this review.

| Population-based surveillance for influenzaassociated hospitalization
Six publications/reports presented data from five population-based surveillance studies for hospitalization in the United States, 24,25,42 Canada, 43 Cambodia, 26 and China. 27 The three studies from the United States analysed population-based surveillance data from the  Table 1 shows the hospitalization rates reported in these studies by season.
Five of the six studies 24,25,27,42,43 specified that rt-PCR was used to diagnose influenza. Two studies 27,43 included patients whose specimen was collected at (or within 24 h of) hospital admission, two 24,42 included patients whose positive specimen was collected within the 14 days prior to admission, and two 25,27 did not specify the timing of influenza testing. Patients were tested for influenza as per clinician's discretion/ hospital policy in all but two studies 26,27 ; in these studies, laboratory testing for influenza was conducted only for patients who had a SARI.

The US Center for Disease Control and Prevention's (CDC)
FluSurv-NET encompassed >70 counties in 14 states (representing  Tokars and Hughes adjusted rates for the age group-specific percentage of hospital inpatients with respiratory disease that were tested for influenza and the sensitivity and specificity of laboratory methods used. b Fever or history of fever (≥38 C), cough or sore throat, and shortness of breath or difficulty breathing in a hospitalized person with onset of symptoms within 10 days before hospitalization.
c Rectal or axillary temperature ≥37.3 C and at least one of cough, sore throat, tachypnea, difficulty breathing, abnormal breath sounds on auscultation, sputum production, hemoptysis, chest pain, or chest radiograph consistent with pneumonia.
about 9% of the US population). 42 The crude hospitalization rates for laboratory-confirmed influenza among adults aged 50-64 years admissions that were tested for influenza and the sensitivity and specificity of laboratory tests resulted in substantially increased estimated rates (see Table 1). Hughes et al. 25

| Ecologic studies estimating hospitalization rates due to laboratory-confirmed influenza
We identified four ecologic studies 31-34 that estimated the incidence of influenza-attributable hospitalizations using excess hospitalizations during periods of influenza activity (Table 2A). Two studies 32,34 reported data from Singapore, the third was from the United States 31 and the fourth was from Portugal. 33 42 Similar trends were reported from Canada, 43 Singapore, 32 and Oman. 35 Rates reported by population-based surveillance studies ranged surveillance or ecologic studies in high-income countries were higher than those reported from SARI surveillance studies from low-and middle-income countries ( Figure 2B).
In a single circumstance, a population-based study and an ecologic study assessed populations in the same country for the same season. The population-based study 42

| Hospitalization rate compared with adults aged ≥65 years
All studies included data for adults aged ≥65 years; the median ratio between the incidence among adults aged 50-64 and those ≥65 years  35 We identified no other studies reporting population-based mortality rates.
The case fatality rate of patients hospitalized with laboratoryconfirmed influenza was available from four population-based studies from the United States, 28,29 Canada, 43 and Spain 30 (Table 3 old who were hospitalized with laboratory confirmed influenza. Arriola et al. 28 Figure 2A shows data by season over time; for studies providing only summary data over more than one season, 34,36,37,[39][40][41] data are attributed to first year/season (if data for 2 years/seasons), the 2nd if 3 or 4 years/seasons. Figure 2B shows data by study type and World Bank country income classification. The overall average rate was used for studies that provided data for more than one season 24,25,27,32,33,35,42,43 31 It is more difficult to assess how much of an effect this consideration may have for studies of laboratory-confirmed influenza as both clinician discretion and hospital policies may result in substantial variation in the extent to which patients being admitted for nonrespiratory causes (e.g., myocardial infarction with fever) were tested for influenza. 48 The significant increases in the estimated burden in US surveillance studies when cases were adjusted for undertesting supports the argument that data in this review remain an underestimate of the true burden of influenza. 24,25 This review has limitations. We considered only articles published in peer-reviewed English language journals or grey literature. Most of the articles we reviewed estimated, rather than measured, population denominators. Many studies were limited by the fact that influenza testing was at the discretion of the admitting practitioner while others limited their inclusion criteria to particularly severe cases or patients with SARI. Although our study eligibility criteria included only community-acquired infections, data from the US CDC FluSurv-NET 24,42 included nosocomial cases. However, it has been reported that hospital-acquired influenza accounts for between 4.3% and 11.4% of total hospitalizations with laboratory-confirmed influenza. 49,50 As such, their inclusion should not alter the estimates significantly. Also, the mortality and case fatality rates were based on analyses of in-hospital mortality only; the potential contribution of death that occurred without hospitalization or after hospital discharge was not included, which may result in a small underestimation of deaths attributable to influenza. Finally, we were unable to control for different rates of vaccine coverage and vaccine effectiveness in the populations and seasons studied.
In sum, this review provides an updated assessment of the burden of influenza in the 50-64 year old age group that can be used to inform future studies of burden as well as the parametrization of studies of cost-effectiveness of influenza vaccination in this age group. supervision.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12955.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.