Introduction

Influenza infections occur in seasonal outbreaks during the winter period related to cold or humid weather conditions and are associated with an excess hospitalization and an increase in overall mortality [1, 9]. Due to a lack of accurate diagnostic tools in the past the cause of death might often have been inaccurate and the influenza-associated mortality underestimated [1, 2, 8]. While studies show pneumonia and influenza excess mortality during influenza seasons, the all-cause mortality in the same period is also 3.8 times higher [1]. High virulence and insufficient vaccination response result in up to 109 cases of influenza virus infections yearly [3,4,5]. Until recently laboratory tests for influenza diagnosis have been reserved for monitoring epidemics and clinical trials. This lack of confirmation renders the diagnosis relying on clinical parameters often inaccurate since the manifestations of influenza may vary greatly depending on patient characteristics, especially age and comorbidities [2, 6]. Particularly in high-risk hospitalized patients the diagnosis may be missed due to exacerbation of symptoms of pre-existing diseases or atypical presentation bearing the risk of inadequate treatment [7,8,9].

The increased availability of point-of-care testing (POCT) assists with the diagnosis of influenza virus infections during epidemic seasons. While rapid diagnostic tests (RDT) based on immunoassays have limited use due to low sensitivity [10, 11] polymerase chain reaction (PCR) POCT devices allow a simple, rapid, cost-effective and accurate means of diagnosis [12,13,14,15]. This ensures correct management of influenza virus infections including prompt initiation of antiviral treatment, preventing onward transmission and avoiding overuse of antibiotics [16,17,18,19,20]. With POCT being widely available, recent studies showed similar influenza virus positive samples for patients presenting with various typical as well as atypical clinical manifestations at emergency departments [8, 9].

Since real world information of molecularly confirmed influenza infections in hospitalized patients is missing a retrospective analysis of patient history, symptoms on admission, complications and mortality was conducted to identify any differences in hospitalized influenza A and B patients after implementing routine PCR testing prior to admission to influenza wards.

Methods

Study design and influenza management

This retrospective observational single center study was carried out at the flu isolation ward of the infectious diseases department of the Kaiser-Franz-Josef-Hospital in Vienna, Austria. The department consists of 2 wards with 28 beds each and an intensive care unit with 10 beds. The hospital has an isolation policy for influenza positive patients. Patients presenting at the emergency department with influenza-like illness undergo POCT for influenza virus infections with the AlereTM i Influenza A & B assay (Alere, Waltham, MA, USA) performed by trained professionals. The AlereTM i Influenza A & B assay uses rapid nucleic acid amplification and was chosen as it produces reliable results within 15 min [12,13,14,15]. The time until the test result is available is an essential factor in this organizational setting. Influenza virus positive patients who needed in-patient treatment were transferred directly to the isolation ward. Whether a patient had to be admitted to this ward or could be treated on an outpatient basis had to be decided by the physician at the emergency department. The study was approved by the local ethics committee (Approval number EK 18-106-VK) and due to the retrospective nature of this study informed consent was not necessary.

Data gathering and statistical analysis

Data were collected from patient medical files. The information was double entered into a MS Excel sheet (Microsoft, Redmond, WA, USA), anonymized and verified for accuracy. The patient medical files included demographic information, laboratory results, symptoms, information about treatment and underlying medical conditions. All patients older than 18 years who tested positive for influenza virus infections with the AlereTM i Influenza A & B assay and needed in-patient care at the isolation ward within the influenza season 2017/2018 (October 2017–April 2018) were included in this retrospective evaluation. Patients who required immunosuppression due to organ transplantation or had undergone stem cell transplantation were not included. The 30-day and 90-day follow-up evaluations were carried out by telephone calls. Baseline characteristics were tabulated with proportions for categorical variables or median (Md) and interquartile range (IQR) for not normally distributed metric variables. Cross-tables, χ2-test and Fisher’s exact test (where applicable) were carried out to identify differences between influenza type and dichotomous categorical variables. For non-normally distributed variables Mann-Whitney U‑tests were calculated to test for differences in metric variables. The study is exploratory and retrospective, therefore the alpha-level was not corrected for multiple tests. A two-sided p < 0.05 was considered statistically significant. After importing the MS Excel sheet all analyses were made with SPSS 25 (IBM, Armonk, NY, USA) for Mac OS (Apple, Cupertino, CA, USA).

Definition of variables and diagnosis

Fever was defined as a body temperature ≥38 °C, measured by via ear thermometers. Respiratory insufficiency was defined as the need for oxygen by any kind of device (e.g. nasal cannula, masks). The need for oxygen was assessed by the treating physician. Pneumonia was defined as a consolidation and/or opacity on a radiological image (mostly X‑ray) accompanied by elevated inflammation markers and/or fever. Rhabdomyolysis was defined as a creatinine kinase (CK) level above 1000 U/l. Heart failure was defined by new onset or worsening of peripheral edema and/or congestion on X‑ray in patients with history of chronic heart failure and without any other cause. Acute kidney injury was defined as an increase of creatinine level by 0.3 mg/dl from the baseline kidney function within 48 h or an increase of ≥1.5 times the baseline presumed to have occurred with the previous 7 days due to the current episode of illness. In cases of a missing creatinine baseline the acute kidney injury was retrospectively assessed by comparing the creatinine level on admission and on the day of discharge, which was assumed to be the baseline function. Symptoms, such as diarrhea, vomiting, absence of fever or acute heart failure were considered as atypical because they are primarily not indicative of influenza virus infections.

Results

Patient characteristics

The study population consisted of 396 people. The distribution between influenza A and influenza B positive patients was unequal with influenza B being the more common infection. Gender distribution was equal in both groups. The vaccination status was not known for all patients. There were no differences in vaccination rates between the two influenza groups. Patients who tested positive for influenza A were significantly younger than influenza B positive patients. Significantly more patients in the influenza A group had a history of a chronic pulmonary disease and were current smokers. Patients with influenza B suffered significantly more often from chronic kidney disease. No differences were found in the proportion of patients with atrial fibrillation, myocardial infarction, peripheral artery disease, history of congestive heart failure, diabetes, stroke, dementia and tumors and no differences in the Charlson comorbidity score were found. For details see Table 1.

Table 1 Patient characteristics
Full size table

Symptoms on admission

The symptoms with which the patients presented at this hospital are listed in Table 2 and sorted from the most common to the least common. Fever, cough and malaise were the three most common symptoms with no differences between the two influenza types. Patients with influenza A had a higher body temperature and a higher number of patients suffered from high grade fever defined as a body temperature ≥39 °C (41.5% vs. 27.6%, p = 0.037). More patients with influenza A suffered from dyspnea, whereas vomiting was reported more frequently in influenza B patients. No differences were found in the symptoms on admission between influenza A and B in the proportion of cough, malaise, abrupt onset of symptoms, muscle ache, diarrhea, cognitive impairment, thoracic pain, coryza, rales, headache, incontinence, epigastric pain, chills and sore throat. For details see Table 2.

Table 2 Symptoms
Full size table

Laboratory parameters on admission

Laboratory parameters were analyzed on admission for every patient (Table 3). Patients with influenza A had a significantly higher leucocyte count and a higher C‑reactive protein (CRP) level on admission. In contrast patients with influenza B presented with a higher creatinine level and therefore a lower estimated glomerular filtration rate (eGFR) No differences were found in the level of creatinine kinase (CK), lactate dehydrogenase (LDH), alanine aminotransferase (ALAT), bilirubin and oxygen saturation (O2).

Table 3 Laboratory parameters of the patients on admission
Full size table

Complications and duration of stay

Table 4 Complications
Full size table

Patients with influenza A had a significantly longer length of hospital stay. No differences were found for the following complications: pneumonia, respiratory failure, acute kidney injury, rhabdomyolysis, intensive care unit (ICU) admission rate, acute heart failure and myocardial infarction. For details see Table 4.

Mortality

In total 22 out of 396 patients (5.6%) died during the stay at hospital. There was no effect of the influenza type on the in-hospital mortality or the impact of the influenza type on the 30-day and the 90-day mortality. The 90-day mortality increased further to 9.4%. For details see Table 5.

Table 5 In hospital, 30-day and 90-day mortality
Full size table

Treatment

In both groups similar numbers of patients received oseltamivir (influenza A vs. B, 55.5% vs. 59.5%, p = 0.476). The time from symptom onset to start of antiviral treatment between patients with influenza A (7.3% <24 h, 29.2% 24–48 h, 16.7% >48 h, 46.9% no treatment) and B (13.3% <24 h, 27.3% 24–48 h, 14% >48 h, 45.3% no treatment) was similar (p = 0.438). The decision to prescribe oseltamivir was made by the treating physician based on the time of symptom onset and the disease severity. In the season examined it was not common at this ward to treat every hospitalized patient with oseltamivir.

There was no relationship between treatment with oseltamivir and the rate of complications or in-hospital mortality. Patients who were treated with oseltamivir had a longer in-hospital stay (7 days, IQR 6–9 vs. 6 days, IQR 4.5–9, p = 0.011). Antibiotics were given to 44.8% of influenza A patients and 36% of influenza B patients (p = 0.123).

Discussion

This study showed various differences in the clinical presentation and complication rates in hospitalized patients with influenza A and B. Patients with influenza A tended to be younger and were more likely to have dyspnea and higher fever. Patients with influenza B had atypical presentations more often including vomiting and a partial absence of respiratory symptoms, which are classically associated with influenza. Mortality rates tended to be numerically higher in influenza A versus B. In previous studies comparing demographic attributes of patients with influenza A and B, a higher age for patients with influenza A was described although this may vary according to influenza A strains. Patients with H3N2 were shown to be older than influenza B patients while patients with H1N1 tended to be younger than patients with influenza B [21,22,23]. A French study by Chagvardieff et al. analyzed patients presenting at an emergency department and similarly to the present results influenza B patients were significantly older [24]. The age difference therefore seems to depend on the study population as well as the dominant influenza A subtype. Unfortunately, no information was available about influenza subtypes in the current population; however, during the season concerned influenza A/H1N1 and influenza B/Yamagata were the dominant subtypes in Austria [25]. It has been shown that patients with influenza A suffered more often from chronic diseases, such as asthma, chronic obstructive pulmonary disease, diabetes and cardiovascular disorders [22, 27]. Furthermore, patients with influenza A were more likely to be smokers. In patients suffering from influenza B infections chronic kidney disease was more frequent. A non-statistically significant trend towards a higher proportion of atrial fibrillation, myocardial infarction and peripheral artery disease in the history of patients with influenza B was observed in this study population, which could be explained by the higher mean age of this group. In this population influenza A patients presented with a higher body temperature and with dyspnea statistically more often. Influenza B infections were associated with vomiting. A Japanese study with adults who consulted a medical center with influenza infections combined gastrointestinal symptom (e.g. nausea, diarrhea, epigastric pain) with one parameter and showed that they are more common in influenza B. Matching the present findings they also showed a higher body temperature in influenza A infected patients [26]. Other studies with a younger population and outpatient treatment did not show any differences in clinical presentation [21, 23].

A study of 2791 hospitalized patients reported a higher rate of ICU admission, ARDS, mechanical ventilation and pneumonia in influenza pdmH1N1 patients compared to H3N2 and influenza B patients although this study included children as well and represented a much younger study population [27]. In this study the rate of ICU admission and respiratory failure was higher in influenza A but this did not reach statistical significance. Pneumonia was the most common complication (24%) and equal in both groups. There was a trend towards a higher proportion of acute kidney injury in the influenza B group, which probably could be attributed to the higher age, higher amount of chronic kidney disease and vomiting as a common symptom in this population. Despite the fact that the influenza A patients were younger, they had a longer median in-hospital stay (8 days vs. 7 days). This might reflect the fact that patients with influenza A were sicker and had a prolonged recovery.

The overall in-hospital mortality rate for both groups was 5.6% with a numerically but not statistically significantly higher mortality rate in the influenza A group (8.3% for influenza A vs. 4.7% for influenza B, p = 0.172). Other studies of hospitalized patients described a similar mortality rate (3.7–5.2%) without differences between influenza A and B virus infections [27, 28, 32]. Of the studies one analyzed community and healthcare-acquired influenza infections and described an in-hospital mortality rate of 10%. The reason for this discrepancy is not known but may be due to the dominance of H3N2, which caused approximately 60% of the influenza infections in this study [33]. In the present population influenza B accounted for 75% of the infections and most of the influenza A infections were most likely caused by H1N1 [25]. In this study the 90-day mortality increased further to 9.4% with no difference between influenza A and influenza B, which represented a twofold increase in mortality. This was not surprising as there is an increased risk for complications leading to death after influenza infections as previously described [29].

The strength of this study is that it focused on hospitalized patients, so any immediate complications could be tracked, especially pneumonia, acute kidney injury, acute heart failure and death which are of great interest and a potential healthcare burden. In addition, efforts were made to assess mortality in the months after influenza infection. There are several limitations of this study. Almost 75% were infected with influenza B therefore the results may not be generalized to influenza A dominant seasons. The study did not differentiate between H1N1 and H3N2 and analyzed the two strains as one group. The symptoms on admission were not collected via a standardized questionnaire which may explain why mild common influenza-like symptoms, such as sore throat and coryza were underreported, especially in patients presenting with severe symptoms, such as dyspnea to the emergency department. There was no standardized testing for specific laboratory results which is why some parameters were missing.

While the three most common symptoms (fever, malaise/prostration, cough) were according to typical manifestations described in textbooks [30] the frequently described abrupt onset was not nearly as common in the setting of hospitalized patients with influenza. It is remarkable that gastrointestinal complaints, acute kidney injury and acute heart failure were the leading symptoms.

In conclusion, this study identified various differences in the presentation and medical history between patients infected with influenza A and B; however, symptoms were overlapping and definite differentiation on mere clinical parameters was not reliable. The use of PCR-POCT for accurate diagnosis seems to be necessary.