Torner et al. BMC Public Health (2019) 19:1089 https://doi.org/10.1186/s12889-019-7414-9 RESEARCH ARTICLE Open Access Assessment of two complementary influenza surveillance systems: sentinel primary care influenza-like illness versus severe hospitalized laboratory-confirmed influenza using the moving epidemic method Núria Torner1,2,3* , Luca Basile1, Ana Martínez1,2, Cristina Rius2,4, Pere Godoy1,2, Mireia Jané1,2, Ángela Domínguez2,3 and The Working Group on PIDIRAC Sentinel Surveillance of Catalonia Abstract Background: Monitoring seasonal influenza epidemics is the corner stone to epidemiological surveillance of acute respiratory virus infections worldwide. This work aims to compare two sentinel surveillance systems within the Daily Acute Respiratory Infection Information System of Catalonia (PIDIRAC), the primary care ILI and Influenza confirmed samples from primary care (PIDIRAC-ILI and PIDIRAC-FLU) and the severe hospitalized laboratory confirmed influenza system (SHLCI), in regard to how they behave in the forecasting of epidemic onset and severity allowing for healthcare preparedness. Methods: Epidemiological study carried out during seven influenza seasons (2010–2017) in Catalonia, with data from influenza sentinel surveillance of primary care physicians reporting ILI along with laboratory confirmation of influenza from systematic sampling of ILI cases and 12 hospitals that provided data on severe hospitalized cases with laboratory- confirmed influenza (SHLCI-FLU). Epidemic thresholds for ILI and SHLCI-FLU (overall) as well as influenza A (SHLCI-FLUA) and influenza B (SHLCI-FLUB) incidence rates were assessed by the Moving Epidemics Method. Results: Epidemic thresholds for primary care sentinel surveillance influenza-like illness (PIDIRAC-ILI) incidence rates ranged from 83.65 to 503.92 per 100.000 h. Paired incidence rate curves for SHLCI –FLU / PIDIRAC-ILI and SHLCI–FLUA/ PIDIRAC-FLUA showed best correlation index’ (0.805 and 0.724 respectively). Assessing delay in reaching epidemic level, PIDIRAC-ILI source forecasts an average of 1.6 weeks before the rest of sources paired. Differences are higher when SHLCI cases are paired to PIDIRAC-ILI and PIDIRAC-FLUB although statistical significance was observed only for SHLCI- FLU/PIDIRAC-ILI (p-value Wilcoxon test = 0.039). Conclusions: The combined ILI and confirmed influenza from primary care along with the severe hospitalized laboratory confirmed influenza data from PIDIRAC sentinel surveillance system provides timely and accurate syndromic and virological surveillance of influenza from the community level to hospitalization of severe cases. Keywords: Sentinel surveillance, Influenza, Epidemic, Threshold, Primary health care, Influenza like illness, Hospitalization * Correspondence: nuria.torner@gencat.cat 1 Department of Health, Public Health Agency of Catalonia, Generalitat of Catalonia, Salvany Building, Roc Boronat 81-95, 08005 Barcelona, Catalonia, Spain 2 CIBER Epidemiología y Salud Pública (CIBERESP) Institute Carlos III, Madrid, Spain Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Torner et al. BMC Public Health (2019) 19:1089 Page 2 of 10 Background detected by the surveillance systems and when these Influenza seasonal epidemics occur yearly during each changes occur with respect to onset of seasonal epidemics. hemisphere’s winter season, whether it be in the Northern Furthermore, immune response varies depending on age, or Southern hemisphere. Acute respiratory infection caused underlying diseases, and immunosuppression [7] . by influenza virus ranges from mild to severe and can even In all, vaccine effectiveness is hampered and derives cause death in at risk population such as the elderly and into low vaccine uptake. These facts underscore the im- young infants. Annual epidemics are estimated to result in portance of any surveillance data available whatsoever, about 3 to 5 million cases of severe illness, and about making them essential for the detection and prepared- 250,000 to 500,000 deaths worldwide. Hospitalization ness for any changes or unusual behavior of seasonal and death occur mainly among these high risk groups, and non-seasonal influenza. However, a limitation to which include also pregnant women and anyone with sentinel surveillance is the limited coverage of popula- underlying medical conditions such as diabetes, obesity, car- tion within a geographical area and the bias towards diovascular or chronic obstructive pulmonary disease [1, 2]. young children who are more prone to seek medical as- Influenza epidemics can cause an overload in the health- sistance. During the 2009 influenza pandemic, the need care system because of high incidence rates of affected to assess influenza severity became evident and from population, especially during peak illness periods. Influenza then on a number of European such as Spain, as well as surveillance is made up information on morbidity and mor- other, countries introduced surveillance of severe disease tality as well as virological assessment of circulating influ- and death due to influenza into their seasonal influenza enza viruses. The resulting assessment and interpretation of surveillance systems [8–10]. these data can support public health authorities to imple- In Catalonia severe hospitalized confirmed influenza data ment special measure to strengthen capacity and prepared- from sentinel hospitals are informed on a weekly basis to ness which will in turn translate into action a reduction of the Daily Acute Respiratory Infection Sentinel Surveillance morbidity and mortality. In most developed countries a System of Catalonia (PIDIRAC) [9]. Identification of admis- network of sentinel physicians report cases attended for in- sions due to pneumonia, to intensive care units and deaths fluenza-like illness (ILI), which is used as a proxy to esti- among hospitalized laboratory confirmed influenza cases is mate influenza virus circulation, and on the other hand available and Information on hospitalization, clinical pres- collect samples for virological confirmation, identification entation, underlying diseases, vaccination and antiviral of causative virus and describe predominant circulating treatment as well as and mortality is added to notification type and subtype of influenza virus [3]. records for influenza cases. In some countries community-based ILI reporting sys- Hospital admissions for influenza and pneumonia tems and indicators such as school and workplace ab- through emergency departments provide some infor- senteeism data are recorded weekly and have been used mation on community cases, yet it is currently not easy to estimate a proxy influenza intensity in combination to track a patient’s journey from community care (pri- with routine outpatient physician consultations and hos- mary care consultations) into the hospital and the lack pital influenza sentinel surveillance data [4, 5]. of denominator (source population) data is a major Sentinel surveillance networks in primary care are a limitation of hospital-based surveillance [8]. readily available and basic source of data that estimates In October 2010, the Public Health Agency of Catalonia disease burden of ILI cases as well as laboratory-con- of the Department of Health in Catalonia, a region in the firmation of respiratory viruses, including influenza, by northeast of Spain with 7.5 million inhabitants, imple- systematic sampling of ILI cases. From these data mented the surveillance of severe hospitalized cases of thresholds can be established to pinpoint the start and confirmed influenza as a tool to complement information termination of influenza seasonal epidemics. Other rele- provided by the influenza sentinel system based on pri- vant information gathered, such as patients’ vaccination mary healthcare physicians implemented since 1998 [9]. status, will allow for vaccine effectiveness estimation [6]. In- The aim of this study is to assess and compare which of fluenza causes high rates of consultations, hospitalization the two networks, primary care and hospital based severe and, in severe cases, death every year and vaccination is the cases, grants a more timely description of influenza main preventive measure undertaken globally, but despite epidemics. policy recommendations, influenza vaccination uptake remains suboptimal in most countries. Low levels of vac- Methods cination coverage among at high risk groups is a missed Data sources opportunity for preventing influenza infection and compli- The PIDIRAC primary care network was composed by cations that can derive into hospitalization and severe out- 60 sentinel physicians, general practitioners and pediatri- come [6]. In addition, influenza vaccine manufacturing is cians, covering 1% of the total population of Catalonia. conditioned to antigenic changes in the circulating strains Representativeness in terms of age, sex, and urban or Torner et al. BMC Public Health (2019) 19:1089 Page 3 of 10 rural setting was considered. Sentinel physicians reported to emergency room and transferred to ward or admitted ILI cases detected in their reference populations on a daily to ward on initial consultation. basis, following the European ILI case definition [9, 11]. Date of hospitalization and type and subtype of influ- For virological influenza surveillance, systematic swab sam- enza virus were collected, other clinical and sociodemo- pling (nasal or nasopharyngeal) of the first 2 ILI patients graphic variables and outcomes, although collected, each week which are sent to the network-affiliated labora- were not relevant for this study. We used data obtained tory (Hospital Clinic, Barcelona) for detection of a batch of from the surveillance of SHLCI for the same study respiratory viruses, including influenza. Samples are con- period as in primary care surveillance for ILI, to calcu- firmed by polymerase chain reaction (PCR) and/or culture late the cumulative hospitalization rates for each season of nasopharyngeal swabs. Respiratory tract samples were and threshold for epidemic onset. processed within 24 h of receipt at the laboratory. Subse- Indicators for this data source were the incidence rate quently, two specific one-step multiplex real-time PCR was of severe hospitalized influenza laboratory confirmed carried out for typing A/B influenza virus and subtyping cases (SHLCI -FLU) and separated as to influenza virus influenza A virus [12]. type A (SHLCI-FLUA) and type B (SHLCI-FLUB). For The information collected in the PIDIRAC includes SHLCI-FLU and SHCLI-FLUA and FLUB the same data on demographic, clinical and virological character- method was applied to calculate epidemic thresholds istics, seasonal vaccination status, chronic conditions, using MEM system [13, 14]. and pregnancy. Indicators provided by the PIDIRAC sentinel system for seven influenza seasons (2010–2011 Statistical analysis to 2016–2017) was analyzed in the study. Indicators for Epidemic thresholds for incidence rates by the Mobile this data source were the incidence rate of ILI syn- Epidemics Method (MEM) were calculated and catego- dromes (PIDIRAC-ILI) calculated according to weekly rized as to intensity of epidemic (baseline, low, moder- population under surveillance by each sentinel physician ate, high and very high) for the PIDIRAC-ILI and of the network that is reporting. Laboratory influenza PIDIRAC-FLU indicators taking into account the seven confirmed isolates from reported ILI cases from primary epidemic seasons included in the study [14, 15]. care (PIDIRAC-FLU) were registered as weekly inci- Epidemic curves obtained from both data sources were dence rate positive to influenza. The latter indicator was paired as PIDIRAC-ILI vs SHLCI-FLU, PIDIRAC-FLU separated as to influenza virus type A (PIDIRAC-FLUA) vs SHLCI-FLU, PIDIRAC-FLUA vs SHLCI-FLUA and and type B (PIDIRAC-FLUB). PIDIRAC-FLUB vs SHLCI-FLUB. For each pair, normal The surveillance of Severe Hospitalized Laboratory partial correlation were calculated at log scale. The week Confirmed Influenza (SHLCI) was made up by twelve at which incidence rate exceeds estimated epidemic tertiary care hospitals representing a mean of 63.6% of threshold level was used as an indicator to identify the population under surveillance. Coverage increased whether one source forecasts the onset of the epidemic from 53.3 to 94% of the total Catalan population in the activity with respect to the other. To calculate this factor 2015–2016 season. The catchment population for by pairs the Wilcoxon signed rank test was carried out. SHLCI surveillance was the population assigned to each To compare intensity of influenza epidemic observed by hospital by the health system as a referral facility for any both data sources, correlation of intensity was calculated for condition leading to hospitalization. each pair and the percentage of agreement within each epi- The system is based on the notification of those la- demic threshold level. Correlation was defined as the pro- boratory confirmed influenza hospitalized cases admitted portion of weeks in which the observed intensity of activity to the sentinel facility and who met the SHLCI case def- for one source matches with the observed by the other inition: Any case with clinical features compatible with source. To identify whether one source forecasts the onset influenza (sudden onset of symptoms and at least one of of the epidemic with respect to the other for each season the following four systemic symptoms: fever or feverish- and for each indicator, the week at which incidence rate ex- ness,malaise,headache .myalgia and at least one of the ceeds the estimated epidemic threshold level, this factor was following three respiratory symptoms: cough,sore throat, studied by pairs according to Wilcoxon signed rank test. shortness of breath) [11], requiring hospitalization for The analysis was performed using the SPSS v.18 statis- clinical severity (at least one of the following criteria: tical package and the R v3.3.0 statistical software (http:// pneumonia, septic shock, acute respiratory distress syn- cran.r-project.org). drome, multiple organ dysfunction syndrome, or admis- sion to intensive care unit (ICU) [9]. Results Only cases that were laboratory confirmed for influ- Estimated epidemic threshold levels obtained by the MEM enza and met the SHLCI case definition were reported, method for each source of data are presented in Table 1. Of regardless of whether they had been previously admitted the seven seasons under study, there was co-circulation of Torner et al. BMC Public Health (2019) 19:1089 Page 4 of 10 Table 1 Moving Epidemic Method (MEM) threshold levels for influenza virus type A and B in 3 of seasons (2010–2011, influenza epidemic activity 2014–2015 y 2015–2016), in the 2011–2012 season influ- Threshhold level (× 100.000 h) baseline low medium high enza virus A was predominant and some sporadic circula- SHLCI 1.82 9.18 20.91 30.1 tion of influenza type B virus, in the 2012–2013 season SHLCI-FLUA 1.33 7.07 23.36 39.61 there was predominant circulation of virus B with sporadic cases of A, and in seasons 2013–2014 and 2016–2017 there SHLCI-FLUB 0.5 1.56 4.27 6.25 was absolute predominance of influenza A virus (Fig. 1). PIDIRAC-ILI 74.00 293.45 425.12 500.8 Epidemic intensity reached moderate threshold PIDIRAC-FLU 13.35 59.48 84.98 99.5 levels for ILI incidence rates and influenza virus iso- PIDIRAC-FLUA 10.06 33.99 90.52 139.56 late incidence in all epidemic seasons except for the PIDIRAC-FLUB 5.95 24.97 57.66 83.47 2010–2011 post-pandemic season, when the duration SHLCI Severe hospitalized laboratory confirmed influenza, SHLCI-FLUA A Severe of the epidemic was longer in time but of lower in- hospitalized laboratory confirmed influença type A, SHLCI-FLUA B Severe tensity. Highest peak incidence levels were reached in hospitalized laboratory confirmed influença type B, PIDIRAC-ILI Incidence rate for Primary care inflenza like illness, PIDIRAC-FLU Incidence rate for Primary 4 seasons from 2011 to 2015. In seasons with co-cir- care influenza laboratory confirmed cases, PIDIRAC-FLUA Incidence rate for culation of both A and B influenza virus, individually Primary care influença type A laboratory confirmed cases, PIDIRAC-FLUB assessed intensity did not exceed moderate level Incidence rate for Primary care influenza type B laboratory confirmed cases (Fig. 2). Influenza A epidemic curves presented higher peaks of epidemic intensity than B influenza virus but with shorter duration. Assessment of the SHLCI source showed an increase in the rate of cases for the four last seasons (2013–2017) Fig. 1 Proportion of circulating influenza virus according to type/subtype A (H1N1) pdm09, A (H3N2) and B per season Torner et al. BMC Public Health (2019) 19:1089 Page 5 of 10 Fig. 2 Epidemic curve and MEM levels (baseline, low, moderate, high and very high) estimated for PIDIRAC data source. Influenza epidemic seasons 2010–11 to 2016–17 although threshold level reached did not surpass the SHLCI -FLUA and PIDIRAC-FLUA showed the best cor- moderate level and did not reach high intensity level. relation index (0.805 and 0.724 respectively). Incidence rates of SHLCI-FLUA cases reached peak Comparison of the curves according to MEM intensity values of 15–20 cases/100,000 inhabitants, which are set levels, paired curves that showed higher correspondence in within the moderate intensity level during epidemic their levels of intensity was the SHLCI -FLUB/PIDIRAC- weeks, while incidence rates of SHLCI-FLUB were less FLUB pair, being above 85% globally and 55% during frequent and displayed lower incidence rates of 3–5 epidemic weeks (Table 2). Assessing delay in reaching cases/100,000 inhabitants (Fig. 3), slightly exceeding epidemic level, the PIDIRAC-ILI source forecasts an baseline threshold level in those seasons with greater B average of 1.6 weeks before SHLCI. Differences are type virus circulation. higher when SHLCI cases are paired to PIDIRAC-ILI When comparing PIDIRAC–FLU and SHLCI, a de- and PIDIRAC-FLUB (SHLCI-FLU/PIDIRAC-ILI and creasing severity trend was observed for the first three SHLCI-FLUB/PIDIRAC-FLUB), although statistical sig- seasons, while for the latter four seasons the trend was nificance was observed only for SHLCI-FLU/PIDIRAC-ILI opposite showing a constant increase in SHLCI incidence (p-value Wilcoxon test = 0.039) (Table 2). with respect to PIDIRAC–FLU (Fig. 4). SHLCI-FLUB cases in seasons 2014/2015 and 2015/2016 showed a Discussion marked increase with respect to PIDIRAC-FLUB. Paired This study is based on surveillance data of ILI incidence incidence curves for SHLCI -FLU and PIDIRAC-ILI and rates from primary care sentinel surveillance, virology Torner et al. BMC Public Health (2019) 19:1089 Page 6 of 10 Fig. 3 Epidemic curves and MEM levels (Baseline, low, moderate, high, very high) estimated for data source SHLCI Influenza epidemic seasons 2010–11 to 2016–17 results from sampling by sentinel primary care physi- laboratory confirmation of primary care sentinel cians and severe laboratory-confirmed influenza (SHLCI) samples as well as severe confirmed influenza that required hospital admission during seven influenza hospitalization rates, a fact that has also been ob- epidemic seasons, from 2010 to 2017, in Catalonia. Data served by other researchers as well [19]. The comple- obtained from these seven epidemic seasons showed a mentary virological data available in the SHLCI good correlation between the two systems, primary care should be more accurate to confirm the start of the and SHLCI, although some differences in prompt detec- epidemic period. In the present study this fact was tion of epidemic onset were observed according to pre- observed only when there is Influenza B predominant dominant circulating influenza virus. circulation. Although the low incidence of influenza B Case definitions with a lower relative specificity for influ- virus in the seasons studied does not allow to con- enza, such as ILI can decrease the specificity of the model clude about the benefit of combining the two systems to detect an influenza season [16]. This occurs because of of surveillance respect the surveillance made by the increasing circulation of other respiratory pathogens, PIDIRAC outpatient sentinel System. This fact differs such as respiratory syncytial virus that add up to ILI Inci- from the observations from Murray et al. that found dence rates in the primary care system, yet this is not so in timing to be the same in all influenza seasons where the SHLCI because these are all laboratory confirmed influ- influenza A (H3N2) was the most prevalent subtype enza cases [16, 17]. Other factors influencing outpatient and that using historical data from all influenza sea- consultation rates, such as public concern, or the circula- sons regardless of circulating subtype, weeks predicted tion of a novel recombinant influenza virus with pandemic to be above the epidemic threshold do not change. potential, can also act as amplifiers of ILI consultation rates Meaning that the MEM approach could be useful for and disease burden results [18]. This may result in patterns planning regardless influenza virus subtype [19]. with bimodal waves as observed in some seasons such The present study is an observational study with as the 2014–2015 or pre-epidemic peaks as observed strengths and limitations. As a strength we can highlight in the 2010–2011 season. Nevertheless increase in ILI that there are few studies investigating timeliness of sen- rates was consistent with increases in influenza tinel primary care vs SHLCI to predict the onset of the Torner et al. BMC Public Health (2019) 19:1089 Page 7 of 10 Fig. 4 Comparison of both surveillance systems by data source pairs: PIDIRAC and SHLCI epidemic allowing for on time preparedness to cope with pandemics, and to assess control measures, such as the ef- the influenza epidemic and will allow for timely assess- fectiveness of vaccination campaigns [20]. ment of intensity and severity of an influenza epidemic. As a limitation we consider the fact that a certain propor- This study reveals the importance of both systems to tion of the population will not seek medical attention. understand yearly influenza epidemic behavior and to Other complementary systems could possibly account for strengthen healthcare resource preparedness. those who do not seek health care attention for ILI have Although MEM was primarily developed for modeling been described such as Influweb or internet flu con- ILI data, it has been proven successful with SHLCI data sultation trends such as Google flu trends, although showing that influenza severe cases’ waves correlate well these systems have, by themselves alone, low repre- with ILI, with higher epidemic thresholds and shorter range sentability [3, 21]. among intensity thresholds. Intensity, understood as the An epidemic threshold was determined for the ILI level of the population consultation rates or the percentage consultation rates reported by the sentinel physicians of ILI primary care consultations which is not synonymous and for the SHLCI to assess the period of increased in- of intensity or severity specific to influenza, but has proven fluenza epidemic earlier. Yet on time complete reporting to be one of the most reliable indicators to describe the im- of hospitalized severe cases is, at the best scenario, on a pact of influenza on the population. On the other hand, week’s delay, making this system unable to predict an SHLCI can complement primary care indicators giving a overload in hospital resource demand with anticipation. picture of the severity of the epidemic. Comparing inten- Nevertheless the surveillance of SHLCI has the potential sities and severity across seasons is essential for the under- to reveal timely insights into epidemic severity as has standing of seasonal epidemic patterns and of future been also pointed out by other authors [22–24]. Torner et al. BMC Public Health (2019) 19:1089 Page 8 of 10 Table 2 Delay in achieving epidemic threshold and correlation between curves and MEM epidemic levels Influenza season SHLCI / PIDIRAC-ILI SHLCI / PIDIRAC-FLU SHLCI -FLUA / PIDIRAC-FLUA SHLCI -FLUB / PIDIRAC-FLUB 2010_11 1 1 1 na 2011_12 2 2 2 na 2012_13 6 3 na 3 2013_14 1 1 1 na 2014_15 1 2 2 1 2015_16 0 -1 -1 4 2016_17 0 0 0 na mean ± SD 1.6 ± 2,1 1.1 ± 1.3 0.8 ± 1.2 2.7 ± 1.5 P-value 0.039 0.071 0.129 0.109 Correlation (LN) 0.805 (0.814) 0.694 (0.749) 0.724 (0.743) 0.583 (0.666) MEM correlation 0.745 0.649 0.752 0.689 Global ILSR 81.0% 77.5% 80.8% 85.9% Epidemic ILSR 47.4% 38.2% 46.0% 58.8% Na not available, LN natural logarithm, ILSR Influenza Likelihood Signed Rank, MEM Moving Epidemics Method, SHLCI Severe hospitalized laboratory confirmed influenza, SHLCI-FLUA A Severe hospitalized laboratory confirmed influença type A, SHLCI-FLUA B Severe hospitalized laboratory confirmed influença type B, PIDIRAC-ILI Incidence rate for Primary care inflenza like illness, PIDIRAC-FLU Incidence rate for Primary care influenza laboratory confirmed cases, PIDIRAC-FLUA Incidence rate for Primary care influença type A laboratory confirmed cases, PIDIRAC-FLUB Incidence rate for Primary care influenza type B laboratory confirmed cases Besides, the voluntary participation of hospitals that could and were highly correlated with each other. The ILI con- account for a selection bias. However, although a selection sultation rates reported by the sentinel primary care bias cannot be ruled out, the 12 participating tertiary hospi- physicians remain the basis of surveillance in Catalonia, tals cover 62% of population of Catalonia meaning that the because the system integrates epidemiological and viro- effect can be assumed not to invalidate data. logical information. Hospital data and virological data Another limitation is the representativeness of the samples will remain very important to provide information on tested in primary care by the systematic sampling of two the severity, molecular and phenotypic characterization first weekly ILI patients attended. Other circulating respira- of the viruses, and whether the ILI incidence is truly re- tory virus can mimic influenza and contribute to increase in lated to influenza virus infections. ILI incidence rates, not so in SHLCI that are all influenza The combined PIDIRAC Influenza sentinel surveil- confirmed. Thus, increase in hospitalization of SHLCI re- lance system provides timely and accurate syndromic flects true influenza virus circulation, albeit a certain delay and virological surveillance of influenza from the com- depending on circulating virus predominance. The imple- munity level to hospitalization of severe cases. The mentation of a third hospital warning system based on data implementation of a hospital warning system based on from emergency department records for laboratory con- data from emergency department records would be de- firmed influenza would be desirable to further improve sirable as a complementary component of the combined timely information to health care for appropriate response. outpatient-SHLCI scheme, allowing to inform control Another limitation to determining burden of measures to lessen the burden of influenza in Catalonia hospitalization due to confirmed influenza is presented and ensure access to quality timely information to health by the lack of the inclusion in the sentinel surveillance care for accurate preparedness. system of a data source for Influenza hospitalization. Abbreviations The use of other data systems such as hospital discharge CI: Confidence interval; FLU: Laboratory confirmed influenza; FLU- data might spare the underestimation of the burden of A: Laboratory confirmed influenza A virus; FLU-B: Laboratory confirmed laboratory confirmed influenza hospitalization [25]. Im- influenza B virus; ILI: Influenza like illness; MEM: Moving Epidemics Method; PIDIRAC: Daily Acute Respiratory Infection Sentinel Surveillance System of provement to upgrade the assessment of influenza burden Catalonia; SHLCI: Severe Hospitalized Laboratory Confirmed Influenza could be achieved by adding an automated surveillance system based on ICD − 10 codes assigned by physicians Acknowledgements The authors also wish to acknowledge the support from all colleagues that after emergency room consultation [26]. make up the PIDIRAC Working Group on Sentinel Primary care and Severe Hospitalized Laboratory Confirmed Influenza in Catalonia: Conclusions Aizpurua J, Alonso J, Azemar J, Aizpurua P, Ardaya PM, Basas MD, Batalla J, Biendicho P, Bonet M, Callado M, Campos S,Casanovas JM, Ciurana E, Clapes Results of two different surveillance systems during M, Cots JM, De la Rica D, Domingo I, Elizalde G, Escapa P, Fajardo S, Fau E, seven influenza seasons studied showed similar trends Fernandez O, Fernandez M, Ferrer C, Forcada A, Fos E, Gadea G, Garcia J, Torner et al. BMC Public Health (2019) 19:1089 Page 9 of 10 Garcia R, Gatius C, Gelado MJ, Grau M, Grivé M, Guzman MC, Hernández R, Surveillance in Italy. PLoS One. 2017;12(1):e0169801 [cited 2017 Jul 7]. Jimenez G, Juscafresa A, LLussa AM, López C, Kristensen L, Macià E, Mainou Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5226807/. A, Marco E, Martínez M, Martínez JG, Marulanda K V, Masa R, Moncosí X, 4. Dalton CB, Carlson SJ, Butler MT, Feisa J, Elvidge E, Durrheim DN. Flutracking Naranjo MA, Navarro D, Ortolà E, París F, Pérez MM, Pozo C, Pujol R, weekly online community survey of influenza-like illness annual report, Ribatallada A, Ruiz G, Sabaté S, Sanchez R, Sarrà N, Tarragó E, Teixidó AM, 2010. Commun Dis Intell Q Rep. 2011;35(4):288–93. Torres A, Valén E, Van Esso D, Van Tarjcwick C, Vink Schoenholzer R, Zabala E. 5. Paterson B, Caddis R, Durrheim D. Use of workplace absenteeism surveillance Marcos MA, Mosquera MDM, de Molina P, Rubio E, Isanta R, Anton A, data for outbreak detection [letter]. Emerg Infect Dis. 2011;17(10):1963–4 Pumarola T. Vilella A, Gorrindo P, Espejo E, M Andrés, Barcenilla F, Navarro G, Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3310669/. Barrabeig I, Pou J, Alvarez P, Plasencia E, Rebull J, Sala MR, Riera M, Camps N, 6. Jorgensen P, Mereckiene J, Cotter S, Johansen K, Tsolova S, Brown C. How Follia N, Oller A, Godoy P, Bach P, Rius C, Hernández R,Perez R,Torra R, Carol close are countries of the WHO European region to achieving the goal of M, Minguell S, Marce R, Garcia-Pardo G, Olona M, Alvarez A, Ramon JM, vaccinating 75% of key risk groups against influenza? Results from national Mòdol JM, G Mena, Campins M, Massuet C. Tora G, Ferràs J, Ferrús G. surveys on seasonal influenza vaccination programmes, 2008/2009 to 2014/ 2015. Vaccine. 2018;36(4):442–52 [cited 2018 Jun 27]. Available from: https:// Authors’ contributions www.ncbi.nlm.nih.gov/pmc/articles/PMC5777640/. All authors were involved at the initial start of the study. NT and LB 7. Atsmon J, Caraco Y, Ziv-Sefer S, Shaikevich D, Abramov E, Volokhov I, et al. conceived the idea of the study, conducted data analyses and prepared the Priming by a novel universal influenza vaccine (Multimeric-001)-a gateway manuscript. AM, MJ, PG, CR, and AD advised on data analyses and for improving immune response in the elderly population. Vaccine. 2014; contributed to subsequent revisions of the manuscript. All authors read and 32(44):5816–23 Available from: https://doi.org/10.1016/j.vaccine.2014.08.031. approved the final manuscript. 8. European Centre for Disease Prevention and Control. Severe influenza surveillance in Europe. Stockholm; 2012. Available from: https://ecdc.europa. Funding eu/sites/portal/files/media/en/publications/Publications/20120614_SUR_ This study was funded by the Program of Surveillance, Prevention and severe_influenza.pdf Control of Transmissible Diseases (PREVICET) of CIBER de Epidemiología y 9. Department of Health PHA of. Pla d ’ informació de les infeccions Salud Pública. (CIBERESP), Instituto de Salud Carlos III, Madrid and the Catalan respiratòries agudes a Catalunya ( PIDIRAC ) 2017–2018 Estrategia de Agency for the Management of Grants for University Research (AGAUR Grant Vigilància dels casos greus produits per la infecció pel virus de la grip Number 2017/SGR. 1342) the funders had no role in study design, data [Internet]. Barcelona; 2017. Available from: http://canalsalut.gencat.cat/ca/ collection and analysis, decision to publish or preparation of the manuscript. professionals/vigilancia-epidemiologica/pla-dinformacio-de-les-infeccions- respiratories-agudes-a-catalunya-pidirac/. Accessed 15 Sept 2018. Availability of data and materials 10. Sullivan SG, Franklin LJ, Raupach J, Pennington K, Bareja C, De Kluyver R. A Data is available upon request by the corresponding author of the brief overview of influenza surveillance systems in Australia, 2015. Commun manuscript. Dis Intell Q Rep. 2016;40(3):E351–E355. 11. European Commission. Commission Implementing Decision (EU) 2018/945 Ethics approval and consent to participate of 22 June 2018 on the communicable diseases and related special health The study does not need formal ethical approval according to national issues to be covered by epidemiological surveillance as well as relevant guidelines with respect to public health surveillance activities according to case definitions. 2018;(2119):1–74. Available from: https://eur-lex.europa.eu/ the actual legislation on Public Health activities (Public Health Law 18/2009, legal-content/EN/TXT/PDF/?uri=CELEX:32018D0945&from=en October 22,; Diari oficial DOGC 30/10/2009,Núm. 5495 Available at: https:// 12. Nair H, Brooks WA, Katz M, Roca A, Berkley JA, Madhi SA, et al. Global portaljuridic.gencat.cat/ca/pjur_ocults/pjur_resultats_fitxa/?documentId=532 burden of respiratory infections due to seasonal influenza in young 871&action=fitxa). All data are anonymous and derive from sentinel children: a systematic review and meta-analysis. Lancet. 2011;378(9807): surveillance. Verbal participation consent in the surveillance system was 1917–30. Available from: https://www.thelancet.com/journals/lancet/article/ obtained by each attending clinician as set in practice by patients’ rights PIIS0140-6736(11)61051-9/fulltext. legislation (Law 41/2002, Nov 14 (art 8, 9 i 10 Chapter IV). 13. Jimenez-Jorge S, de Mateo S, Delgado-Sanz C, Pozo F, Casas I, Garcia-Cenoz No sensitive information was used and data were anonymized. M, et al. Effectiveness of influenza vaccine against laboratory-confirmed influenza, in the late 2011-2012 season in Spain, among population Consent for publication targeted for vaccination. BMC Infect Dis. 2013;13:441 Available from: http:// Not applicable www.ncbi.nlm.nih.gov/pmc/articles/PMC3848794/pdf/1471-2334-13-441.pdf. 14. Vega T, Lozano JE, Meerhoff T, Snacken R, Beauté J, Jorgensen P, et al. Competing interests Influenza surveillance in Europe: comparing intensity levels calculated using The authors declare that they have no competing interests. the moving epidemic method. Influenza Other Respir Viruses. 2015;9(5):234–46. 15. Vega T, Lozano JE, Meerhoff T, Snacken R, Mott J, Ortiz de Lejarazu R, et al. Author details Influenza surveillance in Europe: establishing epidemic thresholds by the 1 Department of Health, Public Health Agency of Catalonia, Generalitat of Moving Epidemic Method. Influenza Other Respi Viruses. 2013;7(4):546–58 Catalonia, Salvany Building, Roc Boronat 81-95, 08005 Barcelona, Catalonia, [cited 2018 Jun 27]. Available from: http://doi.wiley.com/10.1111/j.1750-265 Spain. 2CIBER Epidemiología y Salud Pública (CIBERESP) Institute Carlos III, Madrid, 9.2012.00422.x. Spain. 3Medicine Department, University of Barcelona, Barcelona, Spain. 4Public 16. Green HK, Charlett A, Moran-Gilad J, Fleming D, Durnall H, Thomas DR, et al. Health Agency of Barcelona, Barcelona, Spain. Harmonizing influenza primary-care surveillance in the United Kingdom: piloting two methods to assess the timing and intensity of the seasonal Received: 2 November 2018 Accepted: 31 July 2019 epidemic across several general practice-based surveillance schemes. Epidemiol Infect. 2015;143(1):1–12. 17. Opatowski L, Baguelin M, Eggo RM. Influenza interaction with cocirculating References pathogens and its impact on surveillance, pathogenesis, and epidemic 1. Iuliano AD, Roguski KM, Chang HH, Muscatello DJ, Palekar R, Tempia S, et al. profile: A key role for mathematical modelling. PloS Pathog. 2018;14(2): Estimates of global seasonal influenza-associated respiratory mortality: a e1006770 [cited 2018 Aug 21]. Available from: https://journals.plos.org/ modelling study. Lancet. 2018;391:10127. plospathogens/article?id=10.1371/journal.ppat.1006770. 2. Thompson WW, Weintraub E, Dhankhar P, Cheng P-Y, Brammer L, Meltzer 18. Zheng X, Song Z, Li Y, Zhang J, Wang X-L. Possible interference between MI, et al. Estimates of US influenza-associated deaths made using four seasonal epidemics of influenza and other respiratory viruses in Hong different methods. Influenza Other Respir Viruses. 2009;3(1):37–49 [cited Kong, 2014–2017. BMC Infect Dis. 2017;17(1):772 [cited 2018 Aug 21]. 2018 Dec 28]. Available from: http://www.ncbi.nlm.nih.gov/pubmed/1 Available from: https://bmcinfectdis.biomedcentral.com/articles/10.1186/ 9453440. s12879-017-2888-5. 3. Perrotta D, Bella A, Rizzo C, Paolotti D. Participatory Online Surveillance as a 19. Murray JLK, Marques DFP, Cameron RL, Potts A, Bishop J, von Wissmann B, Supplementary Tool to Sentinel Doctors for Influenza-Like Illness et al. Moving epidemic method (MEM) applied to virology data as a novel Torner et al. BMC Public Health (2019) 19:1089 Page 10 of 10 real time tool to predict peak in seasonal influenza healthcare utilisation. The Scottish experience of the 2017/18 season to date. Euro Surveill. 2018; 23(11). https://doi.org/10.2807/1560-7917.ES.2018.23.11.18-00079. 20. Rakocevic B, Grgurevic A, Trajkovic G, Mugosa B, Sipetic Grujicic S, Medenica S, et al. Influenza surveillance: determining the epidemic threshold for influenza by using the moving epidemic method influenza seasons. Eurosurveillance. 2019;24(12) Available from: https://www.eurosurveillance. org/content/10.2807/1560-7917.ES.2019.24.12.1800042. 21. Thursky K, Cordova SP, Smith D, Kelly H. Working towards a simple case definition for influenza surveillance. J Clin Virol. 2003;27(2):170–9. 22. Pelat C, Lasserre A, Xavier A, Turbelin C, Blanchon T, Hanslik T. Hospitalization of influenza-like illness patients recommended by general practitioners in France between 1997 and 2010. Influenza Other Respi Viruses. 2013;7(1):74–84 [cited 2017 Jul 7]. Available from: https://www.ncbi. nlm.nih.gov/pmc/articles/PMC5780733/. 23. Chaves SS, Lynfield R, Lindegren ML, Bresee J, Finelli L. The US influenza hospitalization surveillance network. Emerg Infect Dis. 2015.21(9);1543–50. 24. Ayscue P, Murray E, Uyeki T, Zipprich J, Harriman K, Salibay C, et al. Influenza-associated intensive-care unit admissions and deaths - California, September 29, 2013-January 18, 2014. MMWR Morb Mortal Wkly Rep. 2014; 63(7):143–7 [cited 2017 Jul 7]. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/24553197. 25. Williamson DA, Huang QS, Roberts SA, Grant CC, McArthur C, Baker MG. Surveillance for influenza using hospital discharge data may underestimate the burden of influenza-related hospitalization. Infect Control Hosp Epidemiol. 2012;33(10):1064–6 [cited 2017 Jul 7]. Available from: https:// www.cambridge.org/core/product/identifier/S019594170003188X/type/ journal_article. 26. Gerbier-Colomban S, Potinet-Pagliaroli V, Metzger M-H. Can epidemic detection systems at the hospital level complement regional surveillance networks: case study with the influenza epidemic? BMC Infect Dis. 2014; 14(1):381 Available from: http://www.biomedcentral.com/1471-2334/14/381. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.