Burden of medically attended influenza in Norway 2008-2017.

Background The burden of influenza in Norway remains uncertain, and data on seasonal variations and differences by age groups are needed. Objective To describe number of patients diagnosed with influenza in Norway each season and the number treated in primary or specialist health care by age. Further, to compare the burden of seasonal influenza with the 2009‐2010 pandemic outbreak. Methods We used Norwegian national health registries and identified all patients diagnosed with influenza from 2008 to 2017. We calculated seasonal rates, compared hospitalized patients with patients in primary care and compared seasonal influenza with the 2009‐2010 pandemic outbreak. Results Each season, on average 1.7% of the population were diagnosed with influenza in primary care, the average rate of hospitalization was 48 per 100 000 population while the average number of hospitalized patients each season was nearly 2500. The number of hospitalized influenza patients ranged from 579 in 2008‐2009 to 4973 in 2016‐2017. Rates in primary care were highest among young adults while hospitalization rates were highest in patients 80 years and older and in children below 5 years. The majority of in‐hospital deaths were in patients 70 years and older. Fewer patients were hospitalized during the 2009‐2010 pandemic than in seasonal outbreaks, but during the pandemic, more people in the younger age groups were hospitalized and fatal cases were younger. Conclusion Influenza causes a substantial burden in primary care and hospitals. In non‐pandemic seasons, people above 80 years have the highest risk of influenza hospitalization and death.


| INTRODUC TI ON
Even though infectious diseases have become less important as cause of morbidity and mortality in middle-and high-income countries, yearly influenza outbreaks remain a major cause of sick leave, doctor visits, hospitalizations and deaths worldwide. [1][2][3][4][5][6][7] Influenza outbreaks occur every winter in temperate climate zones. A UK study found that 18% of the unvaccinated population was infected each season. 8 Studies from several countries support that hospitalizations and complications of influenza are substantial. [9][10][11][12][13] For children under 5 years of age, it has been estimated that influenza contributes to around 870 000 hospitalizations each year globally. 3 Among the elderly, hospitalization rates and deaths are higher than in younger age groups. 4,6 A recent study suggested that seasonal influenza outbreaks cause up to 645 000 deaths each year globally. 6 The yearly number of fatalities in Norway has been estimated to be around 900. 14 In pandemic outbreaks, younger age groups are more affected and have higher hospitalization rate than the elderly. 15 The yearly outbreak of influenza in Norway, with a population size of around 5 million people, costs more than 250 million US $ (2005 US$), including costs for sick leave, medications, doctors' visits and hospitalizations. 13 Information on the disease burden of influenza is of value for national authorities when making policies on disease prevention, vaccine recommendations and hospitals in planning and scaling their capacity. Better data on severe influenza will inform physicians on the risk associated with influenza infection, which is useful when advising patients on preventive measures such as vaccination. Vaccination coverage, even in groups recommended seasonal influenza vaccination, is low in Europe, including Norway. [16][17][18] Many countries still lack good surveillance systems and have little information on the burden of influenza. 19 In Norway, national health registries provide data from the complete population, which makes it possible to study the burden of influenza on the healthcare system at a national level.
We aimed to describe the burden of medically attended influenza in Norway in the period 2008-2017. We utilized primary care data, emergency visits data and hospital data to investigate age effects and the overall impact of influenza in primary and hospital care. We also compared the 2009-10 pandemic outbreak to seasonal influenza outbreaks in terms of the age distribution of patients affected.

| Study population and study period
Our study population consisted of the total Norwegian population, and we included all persons registered as residents between 2008 and 2017. We retrieved the total population number per January 1 each year from Statistics Norway. The study period was from January

| Data sources
Our study is based on prospectively collected data from primary care and hospitals. Data were linked using unique personal identification numbers ensuring correct linkage at the individual level.
Population data were retrieved from Statistics Norway (SSB). The Regional Ethical Committee approved the study.

| Primary care data
All inhabitants in Norway are designated to a general practitioner (GP), and 99.3% of the population is enrolled in this public primary care system. 21 The Directorate of Health's system for control and payment of health reimbursements receives claims from all publicly funded GPs and primary care emergency clinics in Norway. From this database, we collected data on age, sex and date of diagnosis for patients who were diagnosed with influenza using the registry code R80 in the International Classification of Primary Care (ICPC-2) diagnostic system. This diagnosis is based on clinical criteria and does not require laboratory confirmation. 22 We included both face-to-face consultations and consultations by phone.

| Hospital data
The Norwegian Patient Registry (NPR) holds clinical data from all hospitals in Norway. 23 Reporting to the NPR is mandatory and includes diagnoses according to the International Classification of Diseases (ICD-10). We retrieved date of hospitalization and vital status at discharge on all patients diagnosed with the ICD-10 codes J09-J11 (influenza) in the years 2008-2017. While most of the J09-J11 codes require laboratory confirmation, information on laboratory testing is not reported to the NPR.

| Influenza seasons
We defined the influenza seasons according to the surveillance period which follow the standard of the European Centre for Disease Control (ECDC). 24 Thus, an influenza season was defined as the pe-

| Analysis
We list the number and proportions of the population diagnosed with influenza each season in primary care and hospitals. We calculated seasonal rates by dividing number of cases in the defined period by the total population number for the corresponding year.
For age-specific rates, we used number of persons in each age group.
We used negative binomial regression analyses to compare risks between age groups for influenza diagnosis in primary care or hospitals, using the age groups 40-59 years as reference.   (Table 1).

| Influenza burden in primary care
During the nine influenza seasons, a total of 1 099 723 influenza  (Table 1). On average, 1.7% of the population were diagnosed with influenza during an outbreak; however, it varied from 1.0% during the 2013-2014 season to 3.9% during the 2009-2010 pandemic (Table 1).
Primary care rates varied considerably also by age, with the highest rates in the age group 20-39 ( Figure 2, Table S1). When comparing age-specific risks of being diagnosed with influenza in primary care, this age group also had the highest risk ratio in most influenza seasons ( Figure 3, Table S3).

| Influenza burden in hospitals
In total, 23 415 patients were hospitalized with influenza during the nine studied influenza seasons. After excluding multiple hospital registrations in the same patients within a 120-day interval, we included 22 228 patients for further analyses. Among the hospitalized patients, 52% were female. The average age was 56 years (median: 63). On average, 2470 patients were hospitalized per season (range: 579-4973), and the average seasonal cumulative rate was 48 hospitalizations per 100 000 population ( Table 1, Table S2). The In most seasons, influenza rates were highest for people above 70 years and above ( Figure 4, Table S2). Children under 5 years of age had a higher risk of hospitalization compared to all other age groups up to 59 years ( Figure 5, Table S3). The highest age-specific cumulative incidence rate was seen in those above 80 years during the 2016-17 season with 825 hospitalizations per 100 000 population (Table S2). In absolute numbers, however, the majority of hospitalized patients were in the age groups 5-79 years, as this group accounts for larger part of the population.

| Mortality
During the nine seasons in the study period, there were 665 fatalities registered among patients hospitalized with influenza. The average age at death in these patients was 76 years (median: 81).  (Table 1).

| D ISCUSS I ON
Our results show that influenza outbreaks cause a substantial dis-  or "respiratory infection." We found on average 48 hospitalizations per 100 000 population during an influenza season, which is within the range found in other countries, but among the higher country-specific rates. 4,12,[27][28][29] The increase in hospitalizations in the elderly could be a result of improved diagnostics in this group. From 2008 to 2017, the number of tests performed for influenza increased from approximately ten thousand a year to around 140 000. 18 It is reasonable to believe that awareness of influenza increased after the 2009 pandemic. This may partly explain the increasing number of diagnosis over time. Another explanation could be that the threshold for hospitalizing the elderly has changed during the study period. The distribution of influenza viruses varies by season ( Figure S1) and will most likely affect number and age distribution of patients.
Several influenza burden studies, like the one from Widgren and colleagues, 30 include cases of pneumonia and other complications of influenza and find even higher incidence rates. Bacterial pneumonia is a well-known complication of influenza infection, which contributes to the complete picture of the burden of severe influenza.
We know from previous studies that chronic obstructive pulmonary disease, cardiovascular disorders and pregnancy are risk factors for complications due to an influenza infection and that influenza infection itself can cause cardiovascular and neurological disease. [31][32][33][34][35] We did not include these consequences of influenza, and our estimation is a therefore lower than the total burden of influenza.  The low consultation rates in primary care and high hospitalization rates among the elderly could be explained by actual lower infection rate due to higher immunity, or that the elderly experience fewer

| Fatalities
As we included only deaths among hospitalized patients diagnosed with influenza, our study did not include all fatalities associated with influenza in Norway. The number of fatal cases found in our study is therefore lower than previous Norwegian estimates. 14 Also, individual-based methods are found to underestimate influenza mortality, 39,40 and models using all-cause mortality in combination with influenza data are used in many European countries to make better estimations on total influenza-related mortality. 41 Patients dying from exacerbations of chronic disease or other complications after an influenza infection may not be registered as influenzarelated death. Patients might not be tested for influenza, or if they were tested, it was done too late in the course of disease to find a positive result. In addition, our study did not include data from nursing homes, where fatalities associated with influenza occur every season. A UK study estimated that only half of the fatal cases in those above 75 years happen in hospitals. 28 When taking these factors into consideration, we assume that the actual mortality rates associated with influenza are considerable higher than numbers based on in-hospital influenza diagnosis only.

| Strengths and limitations
Our study includes complete follow-up information on the entire Norwegian population for 10 years. We included data from all  For general practitioners, laboratory confirmation is not required when using the ICPC diagnosis "R 80 Influenza." This might lead to both under-and overestimation of numbers of the number of patients. In addition to medically attended influenza, there will be a large number of influenza patients who never seek medical care. A Belgian study found that around half of those experiencing influenza-like illness seeks medical counselling. 45 Simulation studies from the pandemic in Norway support this finding, as an estimated 30% of the population was infected with the pandemic influenza virus, while less than 5% of the population consulted a doctor. 46 Considering this, the impact on society caused by influenza infections is higher than observed in our study.

| CON CLUS ION
Annual influenza outbreaks cause substantial morbidity and mortality and have a high impact on healthcare services. We found that the youngest and oldest are at highest risk of hospitalization during seasonal influenza outbreaks. Even though the elderly have the highest risk of dying, fatal outcome in young influenza patients is seen almost every season. The 2009-2010 pandemic caused similar number of hospitalizations as seasonal outbreaks, but the hospitalized and fatal cases were younger compared to seasonal outbreaks.

ACK N OWLED G EM ENTS
This work was partly supported by the Research Council of Norway Centres of Excellence funding scheme (project number 262700).
We would like to thank Karoline Bragstad for supplying us with virological surveillance data. Also, thanks to Preben Aavitsland and Registry is intended nor should be inferred.