Real-time surveillance systems: Applicability for the control of influenza in acute care.

BACKGROUND
The high morbidity and mortality caused by influenza viruses translate into a great impact on specialized health care. Apart from the annual vaccination, the relevance of other measures to prevent and control this infection is unknown. The objective of our research was to determine the importance of a real-time surveillance system to establish early extended transmission precautions.


METHODS
Quasi-experimental before-and-after study comparing the influenza cases detected in hospitalized adults during the 2016/2017 season (264 patients) with those detected after the implementation of a real-time surveillance system in the 2017/2018 season (519 patients). The improvements included early microbiological diagnosis, immediate communication of results, constant updating of patient information, coordination among professionals, periodic surveillance of the adequacy of preventive measures, and greater control of roommates. The effectiveness of the intervention was determined from the nosocomial infection rate in each season.


RESULTS
After the real-time surveillance system for influenza was implemented, patients with early microbiological diagnosis and immediate isolation increased significantly (13.7% vs 68.2%; P < .001). In addition, nosocomial infections decreased from 17% to 9.2% (P = .001) and overall hospital stay was significantly reduced. Assuming that the entire effect was due to the intervention, the absolute risk reduction was 7.8% and number needed to treat was 12.8.


CONCLUSION
The results in our study reveal the impact of nosocomial transmission of influenza virus in a tertiary hospital and highlight the need to supplement traditional strategies with novel methodologies such as modern surveillance systems based on early diagnosis, close case monitoring, and coordination among professionals.

special care isolating patients, and applying a multidisciplinary approach and constant coordination. 19 Although each measure (patient vaccination, professional vaccination, hand hygiene, patient isolation, early antiviral treatment, and use of masks) would limit the transmission of the virus on its own, 20 the use of combined strategies would reduce nosocomial transmission by half. 21 These estimates obtained by predictive mathematical models would have to be corroborated by evaluating multicomponent interventions in real environments. As far as we know, the effectiveness of an intervention that integrates the aforementioned strategies into a single surveillance system encompassing and coordinating multiple professionals from different specialties has not been evaluated.
Therefore, the objective of this study was to determine the impact of the implementation of a real-time surveillance system, similar to that used in the monitoring of multidrug-resistant microorganisms, on nosocomial transmission of influenza viruses as evaluated by the nosocomial infection rate.

| Study design
Quasi-experimental before-and-after study, in which we compared the cases of influenza detected in adults hospitalized during S1617 and those identified after the implementation of a real-time surveillance system in S1718.

| Setting
The study took place in a tertiary hospital with 907 beds in the province of Salamanca, Castilla y León (Spain). The total number of discharges during 2016 was 31 366, with an average stay of 7.12 days and 157 758 emergencies, of which 12.63% required admission.
During 2017, the total number of discharges was 33 336, with an overall average stay of 6.85 days and a total of 155 288 emergencies, of which 12.58% required admission.
In addition to special epidemiological situations, vaccination against influenza is recommended in the population over the age of 60 in Spain. 22 Castilla y León presented one of the highest influenza vaccination coverages in Spain in this group (over 60% in both seasons). 23 The vaccination rate against influenza virus among health professionals in our center was 29% in S1617 and 28% in S1718.
The specific epidemiological surveillance system of the center, used in S1617 and previous campaigns, consisted in evaluating patients who come to the emergency department with respiratory symptoms, identifying those potentially infected and putting on a surgical mask. When the patient required hospitalization, extended precautions for droplet transmission were used, isolating the patient in an individual room. If these rooms were not available, the cohort was isolated in double rooms with 2 patients with the same type of virus, guaranteeing more than 1 meter of distance between beds.
Additionally, the main measures for preventing the transmission of pathogenic microorganisms were intensified: extreme hand hygiene of people in contact with the patient and their environment, limiting the number of workers, visitors and relatives exposed to these patients, and reducing, as far as possible, the movement of patients within the hospital. The treatment of waste and the cleaning and disinfection of the environment were carried out following the usual protocol in our center.

| Intervention: improvements in the surveillance system
Before the start of S1718, the action protocol against influenza in adults was modified, with prospective collection of information and tested. If a patient had been hospitalized for more than 48 hours next to a confirmed case, a PCR was requested. In case of a positive result, given the same type of virus, the precautions for droplet transmission were put in place for both patients. If the result was negative, the patient was changed to a different room immediately.

| Study population
All adults admitted with confirmed influenza were included (264 cases during S1617 and 519 cases during S1718; total = 783 patients), with a positive test for viral RNA in respiratory samples, identified through the center's epidemiological surveillance program of influenza cases. We excluded positive PCRs of pediatric patients (22 in S1617 and 15 in S1718).

| Definition of confirmed influenza case
We defined case as a patient with influenza syndrome (at least one of the following four general symptoms: fever or high-grade fever, malaise, headache, myalgia; and at least one of the three respiratory symptoms: cough, sore throat, dyspnea; and absence of differential diagnosis), confirmed through the PCR detection of viral RNA in respiratory samples processed in real time in the microbiology laboratory.
Influenza of nosocomial origin was defined in the cases that developed influenza syndrome after 72 hours after admission or who re-entered with influenza symptoms within 72 hours after discharge.

| Study variables
The variables included in our study were sex, age, flu season, vaccination for S1617, vaccination for S1718, time elapsed since vaccination until the positive PCR result, type of infection (nosocomial/ community), admission in intensive care, situation at discharge (improvement/success), and days of stay from the microbiological diagnosis of the infection to discharge.

| Statistical analysis
The data were analyzed based on the computerized record included in our center's global strategy for surveillance and control of influenza.
After checking the normality of the distribution of the values in the sample through the Shapiro-Wilk test, we conducted the association study in two phases. First, we examined the relationship between variables within the same season. We then examined whether the values changed between the two seasons, before and after the implementation of the real-time surveillance system. The association between categorical variables was studied with the chi-square test and Fisher's exact test, while for quantitative variables, we used Student's t test. 14.0%; community: 4.2%, P < .001), mortality (nosocomial: 17.2%; community: 9.4%, P = .03), hospital stay after microbiological diagnosis (nosocomial: 11 ± 16 days; community: 7 ± 6 days, P < .001), and vaccination rate (nosocomial: 34.4%; community: 53.0%, P = .001).
The overall vaccination coverage was 50.8%, with 55.7% of adults vaccinated among those who had specific recommendations due to age. The specific characteristics of adults hospitalized in both seasons are shown in Table 1.
Between November 1, 2016, and March 31, 2017 (S1617), 63 492 patients were treated, of whom 51 380 were adults and 12 112 pediatric patients. The total number of PCR tests for influenza viruses that were performed in the hospital was 1359. In the same period of the following season (S1718), 65 959 patients (52 463 adults and 13 496 children) were attended to by the emergency department and 3054 PCR were requested. The proportion of positive PCR differed between seasons (S1617: 21%; S1718: 17.5%, P = .006). As Table 2 shows, the percentage of community cases with microbiological diagnosis from the emergency department increased significantly in the intervention season, with the consequent early isolation in a single room. The waiting time from arrival to hospitalization for these patients was also longer in S1718 (5.3 vs 6.4 hours; P = .03).
After the implementation of the real-time influenza surveillance system, nosocomial virus transmission was reduced significantly by 7.8%, from 17.0% in S1617 to 9.2% in S1718. Likewise, the duration of hospital stays decreased, with no differences in the vaccination rate between seasons. These results, together with the impact on nosocomial cases, are shown in Table 2. Assuming that the entire effect was due to the intervention, we obtained an ARR of 7.8%, meaning that for every 100 patients in whom the real-time surveillance system was implemented, almost 8 cases of nosocomial influenza were averted. The NNT was 12.8, meaning we would have to implement the system in 13 patients to avoid 1 case of nosocomial transmission.

| D ISCUSS I ON
This study aimed to analyze the impact of implementing a real-time surveillance system on influenza control. The results indicate an important reduction of 7.8% in nosocomial transmission of the virus after the intervention, as well as in the length of hospital stay of all patients.
Our results are coherent with global data on influenza activity in Spain. 98.6% of sentinel detections in S1617 were influenza A viruses, 100% in our center. 5 During S1718, there was a predominance of influenza B (59.0% in national data compared to 57.4% of notifications of our microbiology department). 6 The different temporal pattern of influenza B with respect to A could be due to various synergistic causes. The difference in days from vaccination to PCR diagnosis between the two types of virus and the known divergence in multiple seasons between the vaccine strain and the lineage of the circulating B strain 24,25 could cause an initial containment of influenza A that was not produced for the B strain.
Nosocomial transmission of influenza viruses has a great impact on specialized healthcare. 26 The proportion in our center (11.9%) is consistent with figures published by hospitals of similar characteristics, varying from 4.3% to 17% in studies from Australia and the United Kingdom. 16,27,28 In line with previous research, when comparing nosocomial and community cases, we noted a higher probability of admission to intensive care and an increased mortality in the former. Our experience placed that difference in mortality at 7.8%. In addition, contrary to results presented in other publications, 16 we highlight the importance of vaccination in preventing this type of transmission: In our sample, 53.0% of cases of community-acquired influenza had been vaccinated in the previous campaign, compared to 34.4% of nosocomial cases. Despite the great variations between seasons in the effectiveness of the vaccine (ranging from 10% to 90% protection 25 ), its capacity is demonstrated by reducing the main complications associated with influenza, as well as the number of most serious cases. 29 Similarly, vaccination of health professionals is one of the most effective resources in reducing the spread of seasonal flu in hospitals. 30 Most European experiences place vaccination coverage in this group close to 30%, with a worrying decreasing trend. 13 Given this scenario, it is important to increase efforts and design new strategies that allow professionals to raise awareness about the importance of vaccination to ensure patient safety.
TA B L E 1 Characteristics of admitted patients with PCR-confirmed influenza in S1617 and S1718 However, and reiterating the commitment to this measure, we emphasize the specific impact of other fundamental actions, such as the development of surveillance systems, compliance with standard precautions and the establishment of extended precautions for droplet transmission. 15 As part of our real-time surveillance system, concrete strategies, such as the training of professionals,

TA B L E 2 Comparison of patients admitted with viral influenza during two seasons
Total patients Nosocomial cases S1617 S1718 P-value S1617 S1718 Note: In bold, statistically significant values (P < .05).
hand hygiene, or the use of masks, are protective factors for nosocomial transmission of influenza. 20,31 Similarly, the early microbiological diagnosis of the infection is effective both in the pediatric population 32 and in the general population. 33 In our case, more than two-thirds of the patients hospitalized in S1718 with community flu already entered through the emergency department with a positive Moreover, the intervention was implemented in an environment of demand and usual care pressure during the study period. 34,35 However, the intervention does require the determined commitment and coordination of the management team of the center and the different professional groups involved in the process. A relatively simple strategy can have a great impact on patient safety.
The economic impact of our results is undeniable. For every 100 patients attended when the real-time surveillance system was implemented, more than 7 cases of nosocomial influenza transmission were averted. Previous experience in similar care complexes placed the average cost for hospitalized patients at 6236 euros. 7 To this, we must add the indirect costs, since each case of influenza is associated with 5-6 days of limited activity and about 3 days of work absenteeism. 36 Cost analysis is required to determine the specific economic impact of real-time surveillance systems such as the one we propose.
Both the chosen design and the subject under study entail a series of limitations. Developing a before-and-after study in a single center for two seasons does not allow us to know whether the effectiveness of the intervention is maintained over time and makes it difficult to generalize the results. In addition, multiple factors can influence mortality and nosocomial transmission of influenza viruses, such as their pathogenicity, vaccine effectiveness or global and weekly incidence rates, that complicate the determination of the specific effect of a preventive measure. Despite knowing the proportion of patients who are isolated after a positive PCR in the emergency department, we do not have more data to assess the general adherence of professionals to the new surveillance system. Finally, focusing on patient-to-patient transmission has prevented us from analyzing other possible sources of infection, such as visitors and healthcare workers.
The results obtained reveal the impact of nosocomial transmission of influenza in a tertiary hospital, pointing out the need to continue actively promoting the vaccination of health professionals, but not forgetting that other prevention and control measures, integrated in surveillance systems, can contribute significantly to reducing virus transmission.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.