Surveillance of medically‐attended influenza in elderly patients from Romania—data from three consecutive influenza seasons (2015/16, 2016/17, and 2017/18)

Abstract Background Influenza is an acute infection affecting all age groups; however, elderly patients are at an increased risk. We aim to describe the clinical characteristics and the circulation of influenza virus types in elderly patients admitted for severe acute respiratory infection (SARI) to a tertiary care hospital in Bucharest, Romania, part of the I‐MOVE+ hospital network. Methods We conducted an active surveillance study at the National Institute for Infectious Diseases “Prof. Dr Matei Balș,” Bucharest, Romania, during three consecutive influenza seasons: 2015/16, 2016/17, and 2017/18. All patients aged 65 and older admitted to our hospital for SARI were tested for influenza by PCR. Results A total of 349 eligible patients were tested during the study period, and 149 (42.7%) were confirmed with influenza. Most patients, 321 (92.5%) presented at least one underlying condition at the time of hospital admission, the most frequent being cardiovascular disease, 270 (78.3%). The main influenza viral subtype circulating in 2015/16 was A(H1N1)pdm09, followed by A(H3N2) in 2016/17 and B influenza in 2017/18. Case fatality was highest in the 2015/16 season (3.7%), 0% in 2016/17, and 1.0% in 2017/18. Vaccination coverage in elderly patients with SARI from our study population was 22 (6.3%) over the three seasons. Conclusions Our study has highlighted a high burden of comorbidities in elderly patients presenting with SARI during winter season in Romania. The influenza vaccine coverage rate needs to be substantially increased in the elderly population, through targeted interventions.


| BACKG ROU N D
Influenza is an acute infection affecting all age groups; however, elderly patients are at an increased risk, particularly due to a clustering of comorbidities which puts them at risk of severe influenza, complicated influenza, or influenza-related decompensation of underlying conditions. 1 On a global level, in 2017, influenza has been estimated to be responsible for 54 481 000 episodes of medically-diagnosed lower respiratory tract infections leading to 9 459 000 hospitalizations, and specifically in Romania, with a population in 2017 of 19.6 million people, 174 000 episodes of lower respiratory tract infections with 65 000 hospitalizations, as calculated by the Global Burden of Disease Study 2017. 2 Continued surveillance of influenza is warranted in each country, in order to better inform public health policies and to fill the existing information gaps, particularly for specific influenza risk groups.
In Romania, influenza surveillance is performed at national level by two surveillance systems, one for severe acute respiratory infections (SARI: testing one patient hospitalized with SARI per week from 20 hospitals throughout six counties, during the influenza surveillance season), and one for influenza-like illness (ILI: testing all patients with ILI attending 192 sentinel general practitioners from 16 counties every Tuesday). 3 The National Institute for Infectious Diseases "Prof. Dr Matei Balș," a tertiary care hospital in Bucharest, Romania, was part of the I-MOVE+ hospital network (http://www.i-movep lus.eu/) from 2015 to 2018 and implemented a protocol for systematic screening for influenza in elderly patients admitted to the hospital for SARI. 4 Here, we aim to describe the clinical features of the elderly patients admitted with SARI in our hospital and the circulation of influenza virus types in these patients during three consecutive influenza seasons in Bucharest, Romania, in order to characterize the epidemiology of influenza in this particular patient population, at high risk for influenza-related morbidity.

| ME THODS
An active epidemiologic surveillance study was implemented at the National Institute for Infectious Diseases "Prof. Dr Matei Balș," Bucharest, Romania, during three consecutive influenza seasons: 2015/16, 2016/17, and 2017/18. The study consisted of systematic daily screening of all consecutive admissions in patients aged 65 and older admitted to our hospital with an acute (onset <7 days) illness meeting the following SARI case definition: one or more general signs or symptoms, defined as: fever/feverishness, malaise, headache, myalgia, and altered clinical state (asthenia, anorexia, confusion, or weight loss) associated with one or more respiratory signs or symptoms defined as: cough, odynophagia, and dyspnea. 5 Patients were excluded from the study if they met any of the following exclusion criteria: had a contraindication for influenza vaccine, had SARI onset ≥48 hours after hospital admission, were unwilling to participate or unable to communicate and give consent (either by the patient or by her/his legal representative), were institutionalized, had a respiratory specimen taken ≥8 days after SARI onset, and had tested positive for any influenza virus in the current season before the onset of symptoms leading to the current hospitalization. No information was collected on the number of exclusions overall and by each criterion, and therefore, this information is not available and will not be reported in Results.
For each consenting patient, a standardized medical questionnaire was filled out by study investigators through review of medical (hospital or general practitioner) records and patient/relative interview, collecting demographic variables, a complete medical history, influenza vaccination status in the respective season, and data related to the current SARI episode's onset, characteristics and outcome, as previously described. 6  MGB probes were designed for both B virus lineages that can be detected and discriminated simultaneously, as only one of the two probes will give a fluorescent signal. 8 All real-time RT-PCR reactions were performed using commercial kits-SuperScript ® One-Step qRT-PCR System (Invitrogen). We report descriptive data as number and percentage for categorical variables, and as median and interquartile range (IQR) or range for non-parametric continuous variables. Statistical associations were tested using the chi-squared test for categorical variables and Mann-Whitney's U test for continuous non-parametric variables. Two-tailed P values <.05 were interpreted as statistically significant. IBM SPSS Statistics for Windows, version 20 (IBM Corp.) was used for the statistical analysis.

| RE SULTS
A total of 349 eligible patients were tested in this study, ranging between 53 and 191 by season included in the study. The baseline characteristics overall and by each of the three influenza seasons are presented in Table 1. The median (IQR) age was 74 (68, 80) years, and 43.6% of patients were men.
Most patients (92.5%) had at least one underlying condition at the time of hospital admission, the most frequent being cardiovascular disease, present in 78.3% of cases, followed by diabetes (32.5%), obesity (29.2%), and chronic lung disease (22.6%). Other types of comorbidities were less frequent, being reported in less than 15% of patients (Table 1) During the study period, 20 (13.4%) strains of influenza virus were tested for antiviral resistance; all strains showed susceptibility to oseltamivir.
Among the SARI criteria, fever (94.6%), malaise (90.6%), and myalgia (71.7%) were the most frequently encountered signs or symptoms in patients testing positive for influenza, and specifically fever (OR = 3.0) and odynophagia (OR = 1.8) were significantly predictive for testing positive for influenza ( Table 2). The distribution of comorbidities was comparable between patients with and without influenza ( Table 2). We observed no significant differences in clinical signs or symptoms, or in the distribution of comorbidities between patients with influenza A and influenza B (Table 3).
Overall, 65.6% of the patients included in the study received antiviral treatment with oseltamivir during hospital admission, and the proportion was higher in patients with laboratory-confirmed influenza (81.2%). The clinical course of disease was generally favorable. However, eight patients died during hospital admission, all having at least one underlying condition; of them, four had been confirmed with influenza, and four were negative for influenza. All four deaths that occurred in patients with laboratory-confirmed influenza were caused by influenza A [three A(H1N1)pdm09 and one A(H3N2)]; none of these four patients had been vaccinated against influenza, three were females, all were aged 76 years old, and had a range of 1-3 comorbidities, as follows: three had diabetes, two had chronic lung disease, one had cardiovascular disease and one had cancer. The overall calculated case fatality in this study was 2.3%: 2.7% in patients with laboratory-confirmed influenza and 2.0% in the patients testing negative (Table 2). Case fatality was higher in the 2015/16 influenza season than in the subsequent seasons (3.7%, 0.0%, 1.0%) ( Table 1)   Death during hospitalization g 4 (4.3%, 95% CI: 1.2%-10.6%) 0 (0.0%, 95% CI: 0.0%-6.4%) 4 (2.7%, 95% CI: 0.7%-6.7%) P = .297 Note: All data is presented as number (percentage), unless otherwise specified. Statistical analysis was performed with the two-tailed chi-squared test or Mann-Whitney's U test. Frailty index was calculated by dividing the number of active comorbidities by the total number of comorbidities assessed for each patient.
Abbreviations: 95% CI, 95% confidence interval; N/A, not applicable; OR, odds ratio; SARI, severe acute respiratory infection. a Missing data for 1 patient. b Missing data for 2 patients.  sensitivity and specificity for diagnosing influenza. 16 Falsey et al also suggested that in elderly patients, the threshold for defining fever could be as low as 37.3°C. 16 Overall, influenza A(H1N1)pdm09 was the main viral subtype circulating in 2015/16, followed by A(H3N2) in 2016/17 and B (supposedly mainly B/Yamagata) in 2017/18 in our study. These data are in line with that reported on a national level by the Romanian National Center for Surveillance and Control of Transmissible Diseases, with certain particularities, discussed below.
For the 2015/16 season, influenza A was the main circulating type in Romania in all age groups, accounting for 90.6% of all influenza cases reported and analyzed on a national level, 17 compared to 97.1% in our study of elderly patients, and the main subtype was A(H1N1)pdm09, accounting for 93.5% of all subtyped A strains nationally, 17 compared with 89.4% in our study. The A(H3N2) subtype was relatively less frequently identified at the national level (6.5% 17 compared with 10.6% in our study), while the circulation of B strains, reported at 9.4% in the country, 17 was almost negligible in our study (2.9%) for the 2015/16 season. These slight differences might arise from a different sampling frame (one tertiary care hospital analyzing hospitalized SARI in our study, compared to an ILI-based national surveillance system 17 in the country as a whole), but they might also reflect to a certain degree a particularity of the patients admitted in our hospital that is a referral hospital for the south of Romania. 18 For the 2016/17 season, the data from our current study confirmed to some extent our previous report regarding the exclusive co-circulation of A(H3N2) and B/Victoria. 19 However, in our study of elderly patients the season was dominated by the circulation of A viruses (90.0% compared with the national estimate of 65.7% in all age groups 20 and compared to our previous report of 33.9% in all age groups with an emphasis on children 19 ), suggesting that A viruses predominated in elderly patients in 2016/17; A(H3N2) accounted for 100% of all circulating A viruses in 2016/17 in the current study and in our previous report. 19 Data from Bulgaria and Poland also report the predominance of influenza A virus (97.5% 21 and 95.5% 22 of all influenza cases, respectively), and specifically the A(H3N2) subtype in the 2016/17 season, but A(H1N1)pdm09 was also present in these two countries, to a lower extent. 21,22 Our data are also in agreement with a recent meta-regression analysis reporting that type B influenza viruses are somewhat less frequent in elderly patients compared to the other age groups. 23 For the 2017/18 season, the predominance of B viruses was evident in our SARI study, accounting for 86.9% of all influenza cases, but the rate was higher than the one reported for ILI-based surveillance at the national level (66.5%). 24  and that the overall severity of influenza was high that season. 26 Furthermore, based on the same national report, the general fatality rate for SARI in 2015/16 was reported at 20.7% and at 49.0% for SARI confirmed as influenza, 26 and another study has reported a case fatality proportion of 39.8% (95% CI: 29.5%-50.8%) for that same season for SARI confirmed as influenza. 27 Differences in the lower case fatality reported in our study compared to the national F I G U R E 2 Influenza vaccine coverage in Romania in the general population and in elderly patients, 2007/08 to 2018/19 estimate could potentially be due to selection testing bias of severe cases in the conventional surveillance system, leading to an artificial increase in the reported fatality ratio. These differences may also be partly explained by the specific profile of our institute, which is a major reference center for infectious diseases in Romania, and is well equipped to manage influenza cases in a timely manner, and to institute specific antiviral treatment promptly after hospital admission of SARI cases. In our study, 65.6% of patients admitted with SARI received antiviral treatment. In SARI cases confirmed as influenza, the percentage was higher, 81.2%, whereas in patients who tested negative for influenza, 54.0% received an antiviral, but treatment was stopped after influenza was ruled out and an alternative diagnosis was established.
In Romania, influenza vaccination is provided each year by health authorities, through the general practitioners, to priority risk groups as defined by the World Health Organization (WHO). However, vaccine coverage remains generally low for most risk groups, and for elderly patients in particular, with a marked decrease in the immediate post-2009-pandemic period (49.4% to 19.1% for elderly patients in 2008/09 vs 2010/11), as also reported for other countries, 28 and a slowly increasing trend in the past 3 years (Figure 2). [28][29][30][31][32] In our study population, the influenza vaccination uptake was low over the three seasons. In the 2015/16 season, the vaccination uptake was 7.9% (95% CI: 4.5%-12.7%) in our study population, and the national reported rate for elderly patients was 10.3%. 17 For the 2016/17 season, the influenza vaccine coverage in our study was 3.8% (95% CI: 0.5%-13.0%), and the national reported coverage was 8.2% in elderly patients. 20 For the 2017/18 season, the vaccine coverage in our study (4.8%, 95% CI: 1.6%-10.8%) was lower that the coverage in elderly patients on a national level, reported at 16.3%. 24 Given the overall low number of vaccinated patients included in our study, we cannot conclude whether there was a protective effect of vaccination against hospital admission for SARI. Stronger data from Australia have shown that an influenza vaccine coverage of 80.2% in elderly patients was able to prevent 49.5% of hospital admissions due to influenza in the 2015 influenza season. 33 We recorded a total number of seven cases of influenza in  36 Published studies have shown that standard-dose influenza vaccines may induce lower hemagglutination-inhibition HAI titers and seroprotection rates in elderly patients. 37,38 In Romania, high-dose influenza vaccines have so far not been available and therefore, the data we are reporting here refer to elderly patients who had received standard-dose inactivated trivalent vaccines.
Our study's main strength is that it employed systematic screening of all elderly patients admitted for SARI to one tertiary care hospital in Romania, applying the same standardized methodology throughout three consecutive influenza seasons, thus allowing an analysis of the differences from season to season in terms of clinical characteristics, viral type/subtype circulation, case fatality, and use of healthcare resources.
This study also had a number of limitations. The series is relatively small, 349 patients from one single tertiary care center over the course of three consecutive influenza seasons. The overall low vaccination uptake in our study population and in the country pre- Continued surveillance of influenza is needed in order to inform the best local practices. In elderly patients from our setting, given the high burden of comorbidities that we have characterized, interdisciplinary management and good control of underlying diseases are important, and should be coupled with targeted interventions to substantially increase vaccine uptake.