Reduced Hospital Stay in Influenza Patients after Mass Vaccination during the 2009 Influenza Pandemic in Norway

Background: Norway had a preorder of pandemic vaccine, when the pandemic influenza A (H1N1)pdm09 in 2009 was declared. Mass vaccination occurred 1-3 weeks prior to the peak of the pandemic. Emergency plans were in place, but the predicted severe numbers of hospitalizations did not occur. Objective: To study the epidemiology and clinical presentation of adult patients hospitalized with Influenza A(H1N1)pdm09, and to evaluate the impact of vaccination on the course of the pandemic at a tertiary hospital. Methods: The low dose oil-in-water adjuvanted vaccine was used to vaccinate healthcare workers (HCWs) and at risk patients groups, and vaccination rates were recorded for the community and the hospital. Demographic and clinical information was obtained for 129 patients (>15 years), hospitalized with influenza A (H1N1)pdm09 between August 2009-January 2010. A confirmed case of influenza A (H1N1)pdm09 was defined as meeting a clinical case definition and/or laboratory confirmed disease (rt-PCR or serology). Hospital stay of more than 2 days was defined as a sign of severe illness. Results: 1/3 of at risk patients in the community and >90% of frontline HCWs at the hospital were vaccinated. The median length of hospital stay of infected patients was significantly reduced 7 days after the onset of mass vaccination (p=0.029). There was a predominance of female and moderately obese (BMI 25-30) patients. Infiltration on chest X-ray upon admission was significantly associated with a hospital stay of >2 days (p=0.001). Conclusion: Mass vaccination of frontline HCWs at the hospital and at risk patients in the community contributed to the observed significant reduction in hospital stay of patients infected with influenza. Almost no absenteeism enabled staff confidence and the ability for quick and safe patient turnover. This study highlights the importance of early influenza vaccination, to protect the high-risk patients and the integrity of the healthcare system. Journal of Vaccines & Vaccination J o ur al of Va ccines & Vainati o n


Introduction
The emergence of a new influenza pandemic had long been anticipated since the 1968 Hong Kong Influenza H3N2 pandemic. The focus of pandemic preparedness had in later years been on the zoonosis caused by the highly pathogenic avian influenza virus (H5N1), and on increased surveillance of man and poultry, particularly in South-East Asia. Hence the world was taken somewhat by surprise when Mexico in April 2009 alerted the world to a novel influenza A virus of swine origin, subsequently referred to as Influenza A(H1N1)pdm09, heralding the advent of the first influenza pandemic for 4 decades [1,2].
Influenza vaccination remains the most effective prophylactic measure to prevent infection and limit viral spread in the general population. With limited vaccine-manufacturing capacity, a clear global and national prioritized vaccination strategy was developed. The WHO Strategic Advisory Group of Experts (SAGE) on Immunization recommended two high priority groups for vaccination: frontline health care workers (HCW) and high risk populations, including pregnant women, individuals with BMI >40 kg/m², and people <65 years with asthma/ chronic obstructive pulmonary disease (COPD), diabetes, chronic heart, -kidney, -hepatic or neurologic disease or immunocompromised individuals. Vaccination of HCWs was recommended to ensure the integrity of the health care system, reduce absenteeism and prevent spread of the virus in the hospital [3][4][5][6].
Norway, with a population of 4.9 million people [7], was fortunate to be well informed of the on-going pandemic [8,[9][10][11][12][13] Preparedness plans were in place. Due to a pre order, the pandemic vaccine became rapidly available 1-3 weeks prior to the peak pandemic activity. Norway was one of the European countries with the highest vaccination coverage (45%). Haukeland University Hospital (HUH) provides specialized healthcare for approximately 260 000 inhabitants in Bergen, and is a tertiary hospital for Western-Norway, serving a population of 1 027 000 [7,14]. Vaccination among HCWs was voluntary, and willingness to become vaccinated increased with serious patient reports, and the death of a patient with no known risk factor at our hospital on October 18 th (week 42). Immunization of frontline HCWs and the population at risk in Bergen commenced on October 21 st 2009 (week 43), coinciding with a rapid increase in local hospital admissions due to Influenza A(H1N1)pdm09, which peaked three weeks later (week 46).
In this report, we studied adult patients (>15 years old) admitted to Haukeland University Hospital with confirmed Influenza A(H1N1) pdm09 disease from August 2009 to January 2010. Our aim was to describe the epidemiological characteristics, clinical features, treatment and outcome of these patients. The study period coincided with mass vaccination of both frontline HCWs at the hospital and patient groups at risk in the community, providing a unique situation to study the impact of the vaccine on the course of the pandemic. We hypothesize that early vaccination reduced the burden of the pandemic and protected the integrity of our healthcare system. The continued presence and lack of absenteeism among employees enabled confident and efficient patient care and increased handling capacity. In addition, immunization reduced the need for hospitalization in patients in the community at increased risk of severe influenza disease, thus reducing the burden of mass hospitalization.

Study participants
Adult patients admitted with influenza-like illness (ILI) and hospitalized for more than 24 hours at the Medical and Thoracic Departments, HUH, were eligible to be included in the study. Of the 233 eligible patients, 104 were excluded due to lack of traceability, death from non-related illness, failing to provide informed consent, or not meeting the case definition.
A clinical case definition for Influenza A(H1N1)pdm09 disease (modified CDC case definition) included: temperature >38°C and one of the following criteria; I: ILI symptoms (two of the following symptoms): dry cough, sore throat, runny nose, muscle pain, joint pain, headache, malaise, dyspnea, vomiting/diarrhea; II: Pneumonia; III: Organ failure or collapse. Laboratory confirmed Influenza was defined as a positive (H1N1)pdm09 rt-PCR and/or positive serological antibody titers. All patients provided written, informed consent before inclusion and The Regional Ethical Committee of Western-Norway approved the study.

Study design
Vaccine: The arrival of the oil-in-water adjuvanated Influenza A (H1N1)pdm09 vaccine (Pandemrix ® GlaxoSmithKline, Belgium) to Norway coincided with the increase in Influenza A(H1N1)pdm09 infection rate. Prioritized groups for vaccination were frontline HCWs at the hospital and patients in risk groups in the community. The Centre of Infection Control, HUH organized vaccination of HCWs, and the distribution of vaccine doses was initially strictly limited to frontline HCWs. Mass vaccination of the at risk population and general population was organized by the local health authorities in Bergen municipality. The city's soccer stadium and an elderly center were chosen as venues. Volunteers, retired doctors and nurses performed the vaccination with assistance from the Civil defense. The National Vaccine Register supplied data on the number of vaccinated individuals.
Subjects: Hospitalized patients were asked to provide demographic and clinical information, in addition to access to their medical charts and permission to store blood samples in a bio bank. The data were collected prospectively for each patient, including body-mass index (BMI), smoking and alcohol habits, possible Influenza A(H1N1)pdm09 exposure, influenza vaccination status, co-morbidities, pregnancy, symptoms at onset and upon admission, CRB-65 score at presentation (clinical score predicting mortality in community-acquired pneumonia (scale 0-4)), laboratory and radiological findings including Influenza A(H1N1)pdm09 rt-PCR and influenza A serology, antiviral and antibiotic treatment, supportive treatment, disease complications and outcome. Acute organ failures were specifically defined [14][15][16]. Data were collected through interviews with cases, and/or their parent/ guardian or HCW. Physicians and two study nurses performed medical chart abstractions. Each patient was allocated a unique identification number and the data were plotted into SPSS Data entry (version 4.0).

Laboratory methods
Laboratory confirmation of Influenza A(H1N1)pdm09 was performed using swabs taken from the nasopharynx, tonsils, or lower respiratory tract (sterile brush from endotracheal tube or bronchoalveolar lavage) and collected into virus transport medium. All samples were tested at the Department of Microbiology, HUH, using rt-PCR according to CDC (Centre for Disease Control & Prevention) protocol on a light cycler 2.0 (Roche) [17][18][19]. Serum samples were collected from some patients and assayed in the complement fixation test (CFT) for influenza A or by haemagglutination inhibition (HI) assay for H1N1pdm09. A positive result was defined as CFT>32 or HI ≥ 40. Due to the low pre-existing immunity against Influenza A(H1N1)pdm09 in the population, one serum sample with a positive result was considered sufficient for laboratory confirmation of Influenza A(H1N1)pdm09 disease [19][20][21].

Statistical analysis
Epidemiological analysis describing the patient population were conducted using PASW (version 18.0), while the graphics were produced using Matlab R2010b. Data quality assurance was undertaken through implementing standard data entry checks and manual checking of entered data against the hard-copy case report forms (CRF). Assessment of risk factors for severe disease was conducted through single variable analysis using the Exact Chi-square test for categorical variables. The general significance level was set to 0.05. To handle multiple testing effects, we used a Bonferroni adjustment for testing risk factors versus hospital stay, leading to sign level 0.0045. The length of hospital stay was defined as the difference between discharge date and admission date, and defined as 1 if they were the same day. Severe disease was defined as a hospital stay >2 days. The moving median was computed for each day from the median length of hospital stay of all patients admitted during the subsequent 10 days. The moving median was used to highlight time trends.

The study population
This study reports on adult patients (aged 15-93 years old, median 44 years old) admitted to HUH with confirmed Influenza A(H1N1) pdm09 during the study period. All 129 patients included either met the clinical case definition (n=119) and/or had laboratory confirmed disease (n=69). The majority of patients were ≤65 years old (81.4%) and 24 patients (18.6%) were > 65 years old (Table 1) The female to male ratio was ~ 1:1 in all age groups, except for 20-50 year olds, where it was 1.6:1.

Vaccination and length of hospital stay
Mass vaccination of frontline HCWs and patients in risk groups commenced in week 43 ( Figure 1). Within two weeks approximately 3000 frontline HCWs at the hospital were vaccinated. In the Infectious Disease Unit, >95% of doctors and nurses were immunized by week 45  (Table 1). Despite an increase in number of hospitalizations, there was a significant reduction in length of hospital stay one week after vaccination commenced, in patients admitted with confirmed Influenza A(H1N1)pdm09 (p=0.029) ( Figure 2). This reduction was maintained throughout the rest of the study.
Chronic cardiovascular disease (including hypertension) was the most common underlying co-morbidity, followed by chronic pulmonary disease including COPD and immunosuppression 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 Table 2). The mean BMI was 24 kg/m². All three patients with BMI > 40 kg/m², needed intensive care treatment (ICU). Unilateral or bilateral infiltrates were present on chest X-ray in 34 patients (26.6%) ( Table 3). Of the 120 patients with fever >38°C, blood cultures were performed in 90 and 11 were positive. The most common bacteria was streptococci. There was no correlation between infiltrate shown on X-ray and positive bacteriology findings.
H1N1pdm09 rt-PCR was performed on 126 patients, and found positive in 51 patients (40.5%). Influenza A serology was performed on 47 patients, of whom 32 (68.1%) were positive. Nineteen patients with negative H1N1pdm09 rt -PCR had positive serology results. Two of the three ICU patients had three negative PCR tests before the fourth test, obtained from the lower respiratory tract, was found to be positive.

Treatment and outcome
The median length of hospital stay was three days (range 1-125 (cut-off)). A length of stay >2 days was chosen as a sign of severe disease which was associated with a significantly higher rate of infiltrates on the initial chest X-ray (29 vs. 7, p=0.001) and a tendency to more severe hypoxia (pO2<8.0 kPa) (Figure 3). Patients reporting an underlying medical condition (n=73), specifically, chronic cardiovascular disease or immunosuppression higher risk of increased hospital stay (Figure 3).   Altogether 95 patients received antiviral treatment (Tamiflu) (73.6%), with a median time from onset of illness to treatment of 2 days (Table 3). In our cohort 64.4% of patients received antibiotic treatment, most commonly Penicillin, followed by Cefotaxime. Sixty-three patients (48.8%) received Oxygen treatment, of which 7 (5.5%) received noninvasive respiratory support (CPAP/BIPAP). Hypotension requiring vasopressor treatment occurred in 4 (3.1%) patients.
The three patients (2.1%) requiring ICU treatment all had fever > 38°C upon admission, were severely hypoxic (lowest had pO2 5.10 kPa), and therefore put on a ventilator (including oscillator). They were hypotensive, required vasopressor treatment, and had initial CRP counts >200. Two of the 3 patients had been infected abroad and none of them were vaccinated. They all had co-morbidities (1): COPD and obesity; (2): asthma, IV drug abuse and BMI <15 kg/m² and (3): cardiovascular risk, diabetes and obesity. They had CRB scores of 1,2 and 3 respectively, thus this was not a predictor of severe disease.
None of the study patients died in hospital. The majority of patients were discharged, recorded as "healthy/getting better/unchanged function in daily life". Fourteen patients were discharged with "increased need of help/support in daily life".
One patient with confirmed Influenza A(H1N1)pdm09 disease and with BMI 47 kg/ m², survived despite multi organ failure. This patient was hospitalized for 18 months, but followed the study for 125 days, the maximum time frame in this study. The patient was discharged to a rehabilitation center with tetra paralysis as sequelae, but cognitively intact. Despite having COPD and obesity, this patient had chosen not to be vaccinated.

Discussion
The goal of pandemic vaccination is to elicit appropriate immunological effector mechanisms to reduce viral replication and shedding, thus reduce transmission and provide protection against serious illness and death from influenza. In Bergen, Norway, the pandemic virus spread rapidly from the beginning of November 2009 and hospitalizations peaked three weeks after the onset of mass vaccination. At risk patients in the community and frontline HCWs at the hospital received the first batches of vaccine (Pandemrix ® ) arriving mid October in week 43. This is a unique report on hospitalized patients with confirmed Influenza A(H1N1)pdm09 disease, where mass vaccination of the at risk population in the community and frontline HCWs at the hospital was performed prior to the peak in pandemic activity.
Despite the further increase in the number of hospitalizations, there was a significant reduction in the length of hospital stay in patients admitted with confirmed Influenza A(H1N1)pdm09 one week after mass vaccination commenced (Figure 2). This important finding we mainly attribute to the effect of vaccination. We experienced unparalleled immunization coverage of frontline HCWs at the hospital (>90%) and > a third of patients at increased risk of severe influenza in the community in weeks 43-45. We have earlier shown that protective immunity is elicited as early as 6-7 days post vaccination with adjuvant pandemic vaccine [20]. Clinical trials of candidate pandemic avian vaccines have shown 2 doses of an adjuvanted vaccine are needed to provide protective immunity. In contrast only one dose of the pandemic vaccine was required. Although no serological response was detectable in most of the population to the novel H1N1 pandemic virus, vaccines studies revealed that one dose of the adjuvanated pandemic vaccine elicited high antibody titers and hence protection within a week of vaccination. Despite the novel HA, vaccination elicited a rapid secondary response to cross reactive epitopes, hence the vaccine acted as a booster and not a priming dose of vaccine [22].
We hypothesize that the observed reduction in length of hospital stay, is due to a combined effect of lack of absenteeism of staff and reduced mass hospitalization (Figure 2). One man died prior to vaccine availability, but there were no further deaths in hospital after mass vaccination commenced.
The low HCW absenteeism possibly improved clinical confidence among medical staff and increased the ability for quick and safe patient turnover throughout the pandemic. This may also have contributed to the reduced hospital stay. It is there for likely that Norway experienced a reduced burden of the pandemic influenza disease. The burden could have been further eased, had the vaccine arrived just weeks earlier [23,24].
In our cohort, of hospitalized, infected patients, only 19% were vaccinated (compared to 45% in the general population), indicating  that vaccination prevented hospitalization. This is supported by findings from Canada, which found mass vaccination to be cost effective [25,26]. Twenty-five patients were vaccinated with Pandemrix ® prior to hospital admission of which 11 patients had confirmed Influenza A(H1N1)pdm09. This indicates that they probably were infected and vaccinated simultaneously, without suitable time to gain protective immunity. Modeling of the effect of mass vaccination on preventing viral spread in the Norwegian community, suggested that vaccination did not commence in time to strongly influence the pandemic [24]. However, the western part of Norway, where our study was performed, experienced a slightly later wave of pandemic influenza and importantly focuses on patients requiring hospitalization due to influenza infection.
The length of hospital stay may be seen as a parameter of severity of disease as well as the ability of the hospital to rapidly handle the mass influx of patients with milder disease within the hospital´s normal capacity. These findings suggest that the pandemic vaccine provided protection from influenza illness, avoiding hospitalization in patients at highest risk of severe disease who would have required longer hospital stay ( Figure 2) [5,[25][26][27].
Timely vaccination has been estimated to have significant effect on morbidity, ICU need and mortality [2,4,25,[28][29][30]. A study in Scotland concluded that the use of pandemic vaccine was associated with reduction in the burden of consultations with health care providers, emergency hospital admissions and most importantly mortality [30]. Our study and clinical experience supports this modeling. In Bergen, the early deployment of vaccine allowed rapid protection of front line HCWs, reduced absenteeism, despite experiencing substantial pandemic activity, thus allowing the integrity of the health care system to be maintained [20,25]. Our hospital did not experience the predicted severe mass hospitalization and avoided setting pandemic emergency plans in action such as cancelling elective surgery. A case control study from nine hospitals in Berlin suggested a protective effect of the pandemic vaccine for the prevention of hospitalization despite low vaccination coverage [31].
There is an on-going discussion in the USA regarding mandatory influenza vaccination of HCWs, and it would be interesting to compare absenteeism and patients' length of hospital stay in institutions that conducted mandatory or voluntary vaccination of HCWs, respectively [4,5,28,29,32].
This study has several limitations. The cohort represented 129 of the 233 adults admitted with suspected pandemic influenza, as we only included patients with confirmed disease, and were obliged to include only patients who provided written informed consent. Some information may therefore have been lost. Some patients may have gone undetected if initially admitted to another ward, or if they did not have fever or respiratory symptoms. In the later stages of the pandemic, the proportion of prospectively included cases decreased due to capacity limitations. Inclusion was performed retrospectively for approximately 25% of patients. The study was conducted during the heat of the pandemic without the capacity to collect convalescence serological samples. Although the study design was not primarily intended to evaluate the effect of vaccination per se, our study indicates that the Influenza A(H1N1)pdm09 pandemic peaked at a lower level than anticipated, possibly due to mass vaccination.
To lessen the burden on health care services, pharmacists could temporarily prescribe anti-neuraminidase drugs with effect from November 3rd 2009 (week 43). Furthermore, people with suspected pandemic Influenza infection were advised to stay at home for 7 days on paid sick leave, not requiring a doctor's note. Both measures contributed to lower levels of viral shedding in the community, however, the direct impact on reduced disease burden has not been evaluated in this study.
Data from the national Institute of Publlic Health (NIPH) in Norway show that approximately 1300 patients were hospitalized, nearly 200 were treated in ICU and 32 patients died from virologically confirmed Influenza A(H1N1) pdm09 only 1 of which had been vaccinated [33]. Although potentially lethal, Influenza A(H1N1)pdm09 generally caused mild disease, reflected in the short length of hospital stay in this study population. Patients were often young, otherwise healthy adults, who for a limited period were in need of observation, oxygen, antiviral treatment, and/or antibiotics.
Individuals ≥ 50 years showed a lower incidence of Influenza A(H1N1)pdm09 infection, probably due to partial pre-existing immunity from previous exposure to H1N1 infection [19,21] The present study included hospitalized adults (mean age 44.1 yrs.), adding to the growing body of literature showing a marked lower age distribution for Influenza A(H1N1)pdm09 infected patients than seasonal influenza, where the vast majority of infected patients are >65 years. The lower age distribution has been calculated to increase the estimated numbers of years of life lost (YLL) by 3-5 times compared to seasonal influenza, to 9.7 million years. Globally, South East Asia was the region with the highest YLL [34].
The study population consisted of more women than men and thus differs from the nationally reported equal gender distributions in Norway and reports from other countries where male patients dominated suspected Influenza A(H1N1)pdm09 cases [11,[35][36][37]. The difference may be due to Norway having one of the highest percentages worldwide of working women (84%), hence influencing women to seek medical care when ill, with a female/ male ratio of 4.7:3.0. Furthermore, the female to male ratio in our cohort was almost 1 in all age groups, except for the 20-50 years old where this ratio doubled, reflecting that women are more often the main caretakers of young children, and hence more exposed to infection.
The clinical findings mirror reports from other countries that fever, dry cough and general malaise were the predominant symptoms upon hospital admission [35,[39][40][41][42][43]. Furthermore, the majority of patients had underlying co-morbidities, most commonly cardiovascular disease.
In agreement with other reports, we found that the incidence of laboratory confirmed cases greatly underestimate the impact of the disease [43]. At the peak of the pandemic, rt-PCR test results were delayed due to overwhelming demand and limited laboratory capacity, resulting in the low number of positive PCR tests. The WHO global mortality numbers for the pandemic are based on reported laboratory confirmed pandemic-associated deaths. Less than 12% were reported to the WHO from the most densely populated regions [34]. Recent modeling has indicated that the WHO estimated global mortality from pandemic Influenza could be underestimated by more than a tenfold (15 times) and 51% of pandemic deaths estimated in Southeast Asia and Africa [34]. The disproportionate burden of pandemic disease born