Point‐of‐care testing for influenza in a university emergency department: A prospective study

Abstract Background Seasonal influenza is a burden for emergency departments (ED). The aim of this study was to investigate whether point‐of‐care (POC) PCR testing can be used to reduce staff sick days and improve diagnostic and therapeutic procedures. Objectives The aim of this study was to investigate whether point‐of‐care (POC) PCR testing can be used to reduce staff sick days and improve diagnostic and therapeutic procedures. Methods Using a cross‐over design, the cobas® Liat® Influenza A/B POC PCR test (Liat) was compared with standard clinical practice during the 2019/2020 influenza season. All adult patients (aged ≥ 18 years) with fever (≥38°C) and respiratory symptoms were included. Primary end points were the prevalence of influenza infections in the ED and staff sick days. Secondary end points were frequency of antiviral and antibacterial therapy, time between admission and test result or treatment initiation, patient disposition, ED length of stay (LOS), and for inpatients mortality and LOS. Nurses were interviewed about handling and integration of POC testing. The occurrence of SARS‐CoV‐2 infections coincided with the second half of the study. Results A total of 828 patients were enrolled in the study. All 375 patients of the intervention group were tested with Liat, and 103 patients of them (27.6%) tested positive. During the intervention period, staff sick days were reduced by 34.4% (P = .023). Significantly, more patients in the intervention group received antiviral therapy with neuraminidase inhibitors (7.2% vs 3.8%, P = .028) and tested patients received antibiotics more frequently (40.0% vs 31.6%, P = .033). Patients with POC test were transferred to external hospitals significantly more often (5.6% vs 1.3%, P = .01). Conclusion We conclude that POC testing for influenza is useful in the ED, especially if it is heavily frequented by patients with respiratory symptoms.


| INTRODUC TI ON
Influenza virus infections are caused by RNA orthomyxoviruses and occur in seasonal epidemics with onset in the winter months and a strong increase in infection numbers after the turn of the year. [1][2][3] During the 2018/2019 epidemic in Germany, an estimated 5%-20% of the population was infected, resulting in an estimated 3.8 million influenza-associated physician consultations, 40 000 hospitalizations and 5000-25 000 deaths. 1,3 Thus, influenza represents a resource-intensive burden for the healthcare system, the associated costs approximate 145 million euros. 4,5 In the current COVID-19 pandemic, public measures like social distancing and wearing face masks may also influence the incidence of influenza, at least in upcoming seasons when these respiratory viruses may be co-circulating. 2 Typical influenza symptoms are fever, cough, sore throat, rhinitis, muscle or limb pain, headache, and fatigue. However, only onethird of patients present with these symptoms. 1,3,6,7 The course of the disease varies from mild respiratory symptoms to severe and lethal pneumonia. As the symptoms are not specific, it is difficult to clinically distinguish influenza infections from other respiratory tract infections. Testing is necessary to confirm the diagnosis. Patients are infectious for 4-5 days from the onset of symptoms and transmit the virus mainly by droplet infection and through aerosols. Rapid isolation of suspected cases is, therefore, necessary to protect other patients and medical staff. 1,8 Elderly patients, pregnant women, and patients with comorbidities (chronic heart or lung disease, metabolic diseases, immunodeficiencies, neurological or neuromuscular diseases, and obesity) are at higher risk for a severe course of the disease, so that their protection is particularly important. 1 Neuraminidase inhibitors (NAI) are a class of antivirals used to treat influenza when there is a risk of severe complications such as pneumonia, bacterial superinfections, or damage to other organs. 1,[9][10][11] Antiviral therapy should at best be initiated within 48 hours, but no later than 5 days after the onset of symptoms and is only partially effective. 1,10,[12][13][14] Diagnostic gold standard for influenza is a polymerase chain reaction (PCR) test, which is usually performed in a central laboratory. 1,15 The turnaround time (TAT) depends on several factors including transport, time of the day, and speed of communication of results. Especially after hours and at weekends, the TAT often exceeds 24 hours with centralized analysis. For the emergency department (ED) setting, a long TAT is associated with a prolonged stay of potentially infectious patients and thus an increased risk of infection for other patients and staff, as well as a possibly delayed start of therapy. According to current studies, a POC test performing a RT-PCR is a promising method with high sensitivity and specificity to enable a faster availability of test results directly in the ED. [16][17][18][19] Previous studies mainly focused on TAT and have shown strong effects on the length of stay (LOS). [20][21][22][23] The effects on frequency of antibiotic or antiviral therapy with NAI varied from study to study. 13,[19][20][21][22] However, the effects and the patient population depend on the role of the ED in the respective healthcare system and the established test procedure, a direct comparison with other countries is not easily possible. This study investigates the effects of an influenza point-of-care (POC) PCR test at a tertiary care facility in Germany, for the first time. Primary end points were the prevalence of influenza infections among ED patients presenting with respiratory symptoms and the duration of sick days of ED staff. Secondary end points studied were the frequency of antiviral and antibacterial therapy, as well as the time from patient admission to test results and initiation of therapy, the disposition of patients, and the LOS in the ED. In addition, mortality, LOS in-hospital, and intensive care unit stays (ICU) of inpatients were investigated.

| Participants
In the study, all adult patients (aged ≥ 18 years) were included, in whom a body temperature ≥38°C was measured in the ED or reported within 24 hours prior to ED consultation. Additionally, at least one of the following symptoms had to be present: cough, rhinitis, hoarseness/sore throat, fatigue, headache, muscle pain, aching limbs, or chills. Data of all matching patients were collected in an electronic case report form. Central elements of data collection were, besides a thorough patient characterization, data on the stay at the ED and, in the case of admitted patients, on inpatient therapy.

| Influenza-testing
Roche cobas® Liat® System is a real-time PCR (RT-PCR) analyzer that provides a differentiated result for Influenza A and Influenza B within 20 minutes. 19 Sample material is an oro-nasopharyngeal swab (BD universal viral transport, 3mL, Flock Flex Mini), which was taken by the nursing staff. The POC PCR device was placed on site in the ED and was operated by the nurses. As part of standard clinical practice, patient samples, from the control group for whom an influenza test was ordered, were tested in the central laboratory. In the central laboratory, the Cepheid Xpert® Xpress Flu/RSV Kit was PERLITZ ET aL. | 609 used to performing a RT-PCR with a TAT of 20 minutes for positive and 30 minutes for negative results. 15 In order to determine how well the POC PCR device was implementable into clinical routine, nursing staff of one site was interviewed about the device by means of a questionnaire. They were asked about their satisfaction with sample handling, the integration into clinical routine, the display of results, and the usability of the results.

| Outcomes
ED staff sick days were recorded on an anonymized aggregated weekly basis and compared between POC intervention and control periods.

| Statistical methods
Data analysis was performed using ibm spss version 27 for Microsoft Windows. The distribution of quantitative data was checked and, due to a lack of symmetry, median and interquartile ranges (IQR) were compared. Due to the unfulfilled normal distribution assumption, statistical significance for quantitative characteristics was calculated using the Mann-Whitney U test. For categorical variables, absolute and relative frequencies were compared using Chi-Square test. A P-value of <.05 was considered statistically significant.

| Ethics
The Charité ethics committee had no reservations about the conduct of the study and approved it (EA2/204/19). The study was registered in the German Clinical Trials Registry (DRKS00019207).

| Description
In total, 1865 patients were screened (   | muscle pain (P = .030) or rhinitis (P = .029), and less likely with dyspnea (P < .001) compared to those who tested negative (

| Sick days of ED staff
The amount of sick days of ED nurses was significantly reduced in the POC intervention period: In the control period, there was a total of 697 sick days, of which 91 were recorded among physicians and 606 among nursing staff ( Figure 2). During the intervention period, there was an overall 34.4% reduction to 457 sick days (P = .023); although there was a slight increase in sick days for physicians to 103 (+13.2%, P = .506), there was a significant reduction in sick days for nurses to 354 (−41.6%, P = .005).

| Inpatient admission and ED disposition
312 patients (38.8%) were admitted from the ED to the Charité hospital as inpatients. Further 27 patients (3.3%) were transferred to external hospitals and 480 patients (58%) were discharged home ( Figure 4). Although there were no relevant differences between the study groups regarding discharged home and admission to hospital in general, differences in disposition between the groups were observed. Patients of the POC intervention group were more often transferred to external hospitals than patients of the control group (5.6% vs 1.3%, P = .010). This difference between both groups could also be observed in patients tested positive (3.9% vs 0.0%, P = .005).

| In-Hospital therapy and mortality
Compared with ED treatment, no differences between the study groups were shown in frequency of inpatient antiviral or antibiotic therapy. However, median LOS was 2 days longer in the intervention group (9 vs 7 days, P = .026). This difference is even more pronounced when compared to the tested subgroup (9 vs 6 days, P = .003). 23.1% of the hospitalized patients were in the ICU, with no significant difference between the study groups. Nevertheless, more patients in the intervention group were ventilated (n = 52, 20.1%, vs 13.4%, P = .105) and the mortality of inpatients was slightly higher than in the control group (6.7% vs 4.8%, P = .465).

| Employee satisfaction survey
The questionnaire was distributed to 40 nurses; 25 of whom replied.
52% of all respondents said they were "satisfied" with the handling of the samples and 36% said they were "neither satisfied nor dissatisfied" with it. 60% of respondents were at least "satisfied" with the integration of POC testing in clinical routine and 76% said it could be easily integrated into ED care. In addition, 56% said that the POC influenza results influenced their patient management. The main reason for dissatisfaction was that results could neither be printed nor be fed directly into the hospital or laboratory information system (HIS/LIS). Another point of criticism was that the sample handling was considered cumbersome and sometimes even unhygienic. A total of 92% of those surveyed stated that the topic of infectious diseases should gain more attention overall.  shown in other studies. 21,23 The reasons for this finding could be that in these studies the TAT of central laboratory testing was significantly higher than in this study and that patients of the intervention group had more comorbidities than patients of the control group, and therefore, antibiotics were more likely to be prescribed to prevent coinfections. In standard clinical practice, the influenza test result is entered into the HIS by the central laboratory, but the medical staff is not informed of an existing result. It is, therefore, obvious that the attending physician only learned of the result during a later review. This "time to brain" or time to "actionable result" is additionally shortened by POC testing. 26 Furthermore, the Liat can now be implemented in the LIS and HIS and it is possible to connect a printer, which has improved key points of criticism. Other studies showed that the POC test influences the decision to treat with NAI. 22,27 This could not be confirmed, as the proportion of influenza-positive patients treated with NAI was comparable between both groups, though significantly more patients overall were treated with NAI in the POC intervention group than in the control group. It can be assumed that therapy was not started until a test result was available, since no patient was empirically treated with NAI. Nevertheless, due to the easy availability of a POC test in the ED, more symptomatic patients were tested, and thus, more patients infected with influenza were identified. In particular, faster treatment with NAI led to lower mortality and LOS in several studies. 13,[21][22][23]27 This could not be confirmed. However, this is probably not related to faster testing, but more likely related to the disposition of patients and characteristics of the study groups. Although the number of inpatients was comparable between both groups, significantly more patients in the intervention group were transferred to external facilities. It can be assumed that mainly severely ill patients were treated at the study sites and that the intervention group had more comorbidities, which is a risk factor for severe disease progression. 1 It is, therefore, likely that mostly less severe ill patients were transferred. This finding is particularly relevant during severe waves of influenza, but also in view of the COVID-19 pandemic. It seems that it is well possible to optimize in-hospital flow by transferring infectious patients with mild courses to peripheral hospitals as soon as the infection has been confirmed by a test. Available POC testing could make it possible to reserve capacities of maximum care providers for severely ill patients. In addition, the same applies to confirmed negative patients who could also be transferred more easily.

| Strengths and limitations
Although this is a prospective study, a large part of the patient data was extracted retrospectively from the HIS. The advantage was a large, unselected study population because informed consent was not required but resulted in some missing data.

| CON CLUS ION
POC influenza PCR testing significantly reduced the sick days of staff in the ED. The POC testing was easily integrated in routine procedures and run by ED nurses. The indication for treatment with NAI (in positive cases) and antibiotics (in negative cases) was more precise. The transfer to external hospitals was enhanced by the early availability of the influenza status. We conclude that POC testing for influenza is useful in the ED, especially if it is heavily frequented by patients with respiratory symptoms. Note: ED therapy and disposition is shown for both study groups in general and for the distinguished subgroups named in the second column. The percentages refer to the subgroup named in the second column.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/irv.12857.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.