Influenza viruses continue to cause epidemics around the world every year. Approximately 290,000–650,000 people die of respiratory diseases linked to influenza each year in the world, with mortality rates of 4.0–8.8 per 100,000 individuals.1,2 Influenza severity varies in every epidemic, which is characteristic of seasonal epidemics, and occasionally results in nationwide epidemics or worldwide pandemics. Five worldwide pandemics have been documented in the past 100 years since the Spanish influenza pandemic in 1918. Accurately predicting the timing, size and severity of influenza epidemics is very important for preplanning preventative measures.3
After the influenza A (H1N1) pandemic in 2009–2010, influenza has been a seasonal epidemic with relatively low activity. However, according to the World Health Organization (WHO) surveillance, influenza activity had significantly risen among several countries in the past few months.4 Until February 18, 2018, the majority of countries experienced high levels of influenza activity, and some countries had reported levels of hospitalization and intensive care unit admissions that exceeded the peak levels of previous influenza seasons.4 The influenza epidemic in the United States began in early November 2017 and continued for several more weeks.5 Based on a new methodology of intensity thresholds,6 the severity of the 2017–2018 season was classified as high for each age group (children and adolescents, adults, and older adults), and the mortality attributed to pneumonia and influenza, peaking at 10.8%, was the highest percentage reported since the 2014–2015 season.7 In the European Union/European Economic Area countries, influenza activity has triggered epidemics lasting for several months since October 2017, which was shown to have a severe impact.8 In China, the number of influenza patients from the end of 2017 to early 2018 was greater than that in previous years, resulting in an increase in influenza-related hospitalization, severe illness, and death.9
However, a lack of effective assessment for the timing, size, and severity of influenza epidemics had made the preplanning of preventative measures extremely difficult.3 Though influenza vaccine and quarantine considered well-known and common measures for the prevention of influenza epidemics, the effects of interventions are not yet known in the population. In particular, an unfavorable situation for controlling influenza is not clear, given the low provision rate of the population-based influenza vaccine. To effectively respond to the influenza epidemic and prepare for the epidemic, the parameters and the characteristics of the influenza epidemic from September 2017 to February 2018 and the intervention measures against this epidemic were explored in this study.
The reported incidence and fatality of influenza and its trend characteristics since January 2011
The reported incidence and fatality of influenza were 1,913,698 and 395, respectively, with an average yearly reported incidence rate of 19.21 per 100,000 and an average reported case fatality ratio (CFR) of 0.21 per 1000 from January 2011 to February 2018. These findings indicated that both the number of influenza cases and the number of deaths had peaked between the end of 2017 and early 2018 (Figure 1).
Through the joinpoint regression model, we also found that the incidence of influenza showed an increasing trend from January 2011 to February 2018, with an annual percentage changes (APC) of 3.7 (3.0–4.4) (P < 0.01). However, the trend of influenza incidence over the past 7 years was divided into two distinctly different epidemiological phases. Compared with the increasing trend from January 2011 to August 2017, with an APC of 1.9 (95% confidence interval: 1.2–2.5) (P < 0.01), the influenza incidence rose sharply from September 2017 to February 2018, with an APC of 48.1 (95% confidence interval: 17.7–86.2) (P < 0.01) [Supplemental Digital Content (SDC), Table 3]. This finding indicated that since 2011, there had been a significant severe epidemic of influenza during the period of September 2017 to February 2018 (influenza 2017/2018).
From September 2017 to February 2018, the reported incidence and fatality of influenza were 606,734 and 117, respectively, with an average yearly reported incidence rate of 87.29 per 100,000 and an averaged reported CFR of 0.19 per 1000. The average yearly reported incidence rate was significantly higher than that from January 2011 to August 2017; the latter was 14.10 per 100,000 (Table 1).
The phylogenetic characteristics of influenza 2017/2018
Phylogenetic analysis on the hemagglutinin (HA) showed that the homology of most influenza A (H1N1) viruses from September 2017 to February 2018 were significantly different from that of viruses from January to August of 2017, as well as those from January 2011 to December 2016 (Figure 2A). The HA of most influenza A (H3N2) viruses from September 2017 to February 2018 was also located on an independent branch in the phylogenetic tree (Figure 2B). The HA of the recommended A (H1N1) and A (H3N2) viruses for influenza vaccines 2017/2018 had high homology with the influenza A(H1N1) and A(H3N2) viruses from January to August of 2017, respectively (Figure 2A and 2B). However, both had low matches with the influenza viruses from September 2017 to February 2018. Furthermore, the recommended B (Victoria) viruses for influenza vaccines 2017/2018 had low matches not only with the viruses from September 2017 to February 2018 but also with those from January to August of 2017 (SDC, Figure 4). Most of the influenza A (H1N1) virus strains, as well as A (H3N2) and B (Victoria) viruses, from September 2017 to February 2018, had high homology in NA with those from January to August of 2017, as well as those from January 2011 to August of 2017. However, the recommended A (H3N2) and B (Victoria) viruses for influenza vaccines 2017/2018 had low homology in NA from January to August of 2017 and from September 2017 to February 2018, respectively (SDC, Figure 4).
The epidemiological characteristics of influenza 2017/2018 and assessment of its different intervention measures
An examination of the model fit for the expected and actual incidence rates of influenza (SDC, Table 5) indicated that the average latent period and the average infection period were 2.83 days and 4.67 days, respectively, during influenza 2017/2018 (Figure 3). The average basic reproduction number (R0) was 1.53, and the whole expected incidence rate was 9.13% (Figure 3).
To evaluate the impact of the influenza vaccine and its coverage during the influenza epidemic, vaccination-based susceptible-exposed-infectious-recovered (SEIR) models were established. Based on an influenza vaccine effectiveness (VE) of 60%, the expected reductions in the number of influenza cases are 65 million and 130 million cases, with an influenza vaccine coverage rate of 40% and 80%, respectively. The incidence reduction percentage was linearly related to the influenza vaccine coverage rate. According to the 0.77% influenza vaccine coverage rate from our surveillance, the incidence had been reduced to 1.25 million (Figure 4A and 4C). However, as indicated in the quarantine-based SEIR models, the incidence of influenza would decrease by 0%, 0.1%, 2.2%, and 57% when the quarantined population was covered at 20%, 40%, 60%, and 80%, respectively. The influenza incidence would decrease very little – less than 60% of the quarantined population – but it had significantly decreased in more than 80% of the quarantined population (Figure 4B). Unlike the benefit of vaccination, which has a linear relationship with vaccination coverage, the quarantine measure had no significantly benefited under 60% of the quarantined population (Figure 4C). Despite the insignificant benefit of less than 60% for the quarantined population, the generation time of the epidemic had been extended and had a positive correlation with the percentage of the quarantined population (Figure 4B and 4D).
Influenza is closely associated with the global burden of morbidity and mortality in human populations, and its intensity varies in every epidemic. Our results supported the trend that there had been a substantial nationwide epidemic of influenza during the period of September 2017 to February 2018 in China, which was the most severe epidemic after the 2009 pandemic influenza A (H1N1).10 In this epidemic, the average yearly reported incidence rate was 87.29 per 100,000, which was six times higher than that from January 2011 to August 2017, with the latter rate being 14.10 per 100,000. Influenza stronger activity had also clarified by the report that since the winter of 2017 in China than that in recent years.11 Surveillance findings in China indicated that the 2017–2018 influenza season was a high-severity season with high levels of outpatient clinic and emergency department visits for influenza-like illness.12 With the continuous improvement of disease diagnosis in China, the intensification of infectious disease screening and the implementation of online reporting, the incidence rate is “overestimated,” but actually closer to the actual level. Accurate predictions of the timing, size, and severity of influenza epidemics are difficult; however, relatively low influenza activity over a long period of time is more likely to lead to a large-scale and severe epidemic. In this study, it was further indicated by model that the expected incidence rates were 9.13%, which was significantly higher than the reported incidence rate. There are several possible reasons for this finding. First, not all influenza cases were tested for etiology in influenza outbreaks. Then, primary hospitals had a weak ability to detect influenza pathogens, lacking the necessary equipment for detection. Next, during the outbreak, many patients chose to go to the pharmacy directly to buy the medicine rather than go to the hospital first for diagnosis.
Though the antigenic characterization of the 2017 circulating influenza viruses did not show significant antigenic drift.11,13 The 2017 influenza epidemic showed that viruses had HA N121K substitution, which was rare before 2017.13 Our phylogenetic analysis on the HA of influenza A (H1N1) and influenza A (H3N2) viruses from September 2017 to February 2018 indicated lower homology to those from January to August of 2017, as well as to those from January 2011 to August of 2017, but the neuraminidase (NA) sequence of these two virus strains had high homology among them. Therefore, these strains were likely still susceptible to NA inhibitors. The phylogenetic analysis of the HA and NA genes may prove their importance with regard to the comprehensive process of vaccine virus selection.14 However, our phylogenetic analysis indicated that all the HA of the recommended A (H3N2), A (H1N1) and B (Victoria) viruses for influenza vaccines 2017/2018 had low matches with the viruses of influenza 2017/2018 from September 2017 to February 2018, which may have led to the severity of influenza in 2017/2018 with the reduction in VE. VE depends, in part, on the match between the vaccine virus and circulating influenza viruses.15 The dominant influenza virus of influenza 2017/2018 in many areas of China was the influenza B virus (the Yamagata strains), and both influenza A H3N2 and H1N1 were also present.11 However, the recommended influenza viruses for influenza vaccines 2017/2018 in China did not contain the Yamagata strain of influenza B.11
The R0, quantifying the transmissibility of any pathogen, is one of the important parameters for assessing the epidemic intensity and effectiveness of intervention. Study by Ferguson et al. shows that the quarantine strategies can greatly increase the effectiveness of 80% at R0 = 1.8.16 However, effectiveness of vaccination and quarantine strategy varied with the variation of R0 in different epidemic or at different stages. In addition, the vaccination and quarantine strategies are hardly to be fully implemented before or during the real epidemic. Effectiveness of vaccination and quarantine strategies were also affected by different coverage rates of vaccination and quarantine. In this study, the R0 for influenza 2017/2018 was 1.53. In terms of influenza vaccination, the incidence reduction percentage was linearly related with the influenza vaccine coverage rate at the observed R0. Based on the influenza VE of 60%, a reduction of 65 million and 130 million cases would expect with an influenza vaccine coverage rate of 40% and 80%, respectively. Some countries with influenza vaccination programs had estimated reductions in the prevalence of severe disease and the number of deaths.2 Even when the vaccine does not exactly match these viruses, and the VE is reduced, it may still provide some substantial protection.7 VE was estimated to be 67% against A (H1N1) pdm09 viruses, 25% against illness caused by influenza A (H3N2) virus, and 42% against influenza B viruses.17 However, the influenza vaccination coverage rate is fairly low, at less than 2% in China, which is far lower than the number of people who should be vaccinated. The fact that the vaccination coverage is very low also makes it less useful to change to a quadrivalent vaccine, as people are not taking the vaccine anyway. Therefore, taking comprehensive policies to improve the vaccine coverage in high-risk groups, as young children, elderly, people with underlying medical conditions is necessary in China.18
Apart from influenza vaccination, we also assess effectiveness of quarantine measure for controlling influenza at the observed R0 in this epidemic. However, unlike vaccination benefit, which has a linear relationship with vaccination coverage, the quarantine measure only substantially benefits over 60% of the quarantined population. Though less impactful benefits are seen in less than 60% of the quarantined population, the generation time of the epidemic had been extended and had a positive correlation with the percentage of the quarantined population. Influenza is different from severe acute respiratory syndrome, and mass transmission occurs before the onset of case-defining symptoms.19 Increasing the quarantined population may be the strongest preservation against a pandemic until sufficient vaccine and antiviral medicines can be created, at which point, mass vaccination and prophylaxis may be more effective than targeted approaches.20 Although the quarantine is difficult to achieve in the general population, it may be better in some places like schools and the sooner the quarantine measures are implemented, the better the control effect on the epidemic will be.21,22 In our study, we only compared the effectiveness of two interventions: vaccination and quarantine. In addition, frequent hand-washing may have effectively prevented the transmission of respiratory tract infections during a pandemic.23
There are a few limitations of this study. First, as large regional difference in a large country like China and difference of severity types or subtypes, data extracted from the reports of notifiable infectious diseases could not be stratified by demographic and subtype, which could influence assessment of its trend characteristics and influenza control measures. Second, China is a vast country, and different regions, such as the South and the North, have different influenza seasons and different dominant strains. Further study should be taken to explore the regional epidemiological characteristics and parameters of influenza epidemic for vaccination strategies.
In conclusion, it was the first severe seasonal influenza epidemic during the period of September 2017 to February 2018 in China since the 2009–2010 pandemic with the R0 of 1.53. The most effective way to prevent influenza is still vaccination, and its effectiveness was the linear with vaccination coverage, compared to the quarantine measure without substantially benefited under 60% of the quarantined population. Strategies to increase vaccination coverage should be taken for the prevention of severe epidemics of influenza, especially in high-risk groups (the elderly, children, pregnant women, people with chronic diseases, and health care workers).
Materials and methods
Influenza incidence and fatality data from 2011 to 2018 were extracted from reports of notifiable infectious diseases, which were open and available from the official website of National Health Commission of the People's Republic of China, and the number of reported cases came from all health facilities in China.9 All influenza cases from outpatients, inpatients, or intensive care units, which met the diagnostic criteria of laboratory-confirmed cases of influenza A or B virus,24 were reported. All sequences for HA and NA of human influenza virus from China during the period of 2011–2018 were obtained from the Global Initiative on Sharing All Influenza Data Database.25 The sequences for HA and NA of the WHO-recommended viruses for influenza vaccines 2017/2018 were also obtained from the Global Initiative on Sharing All Influenza Data Database. Nonsense sequences and incomplete sequences were removed, and replicate sequences within each subset were also excluded to eliminate oversampling. A maximum of 50 sequences per year were randomly sampled to avoid the trees being too large. The research ethics board at the First Affiliated Hospital, School of Medicine, Zhejiang University, approved the study.
Phylogenetic analysis of the HA and NA of the influenza virus
A total of 264 HA and 239 NA cases for the influenza A (H1N1) virus, 286 HA and 275 NA cases for the influenza A (H3N2) virus, and 190 HA and 204 cases NA for the influenza B (Victoria) virus were selected to reconstruct the phylogenetic tree. The sequences were divided into three groups of virus sequences from January 2011 to December 2016, from January to August of 2017 and from September 2017 to February 2018. The corresponding WHO-recommended viruses for influenza vaccines 2017/2018 were also combined. The phylogenetic tree was constructed using the neighbor-joining method, based on the bootstrap model with 1000 bootstrap replications.
Construction of joinpoint regression models and SEIR models
We first used joinpoint regression models to examine influenza incidence trends from 2011 to 2018.26 Trends were expressed as APCs, which were calculated as APCi = [(Exp(bi) − 1)] × 100, where bi represents the slope of the period segment. The Z test was used to assess whether an APC was significantly different from zero. When the slope (APC) of the trend was significantly greater than zero (P < 0.05), it indicated that compared with a perennially stable level of disease, there was a nationwide epidemic of influenza, and serial compartmental SEIR models were further used for exploring the epidemiological characteristics of the influenza epidemic.27 Using SEIR models, we calculated R0, which was defined as the average number of secondary cases generated by an index case in a large, susceptible population. The parameters of the mean latent period, mean infectious duration, and expected incidence rate were also estimated in models. The quarantine measure means keeping person apart from others for a period of time if they meet the diagnostic criteria of laboratory-confirmed cases of influenza A or B virus. The supplementary material provides more details about joinpoint regression (SDC 1, https://links.lww.com/IMD/A0) and SEIR models (SDC 2, https://links.lww.com/IMD/A0).
The incidence rate (per 100,000) was defined as the number of annual incident cases divided by the population size, and the CFR (per 1000) was defined as the number of annual deaths divided by the number of annual incident cases. For categorical variables, the percentages of patients in each category were calculated. IBM SPSS (Modeler version 14.1 and Statistics version 21) was used for data extraction, cleaning, and analysis. Joinpoint (version 4.3.1), Mega 7 (version 7.0.26), and Berkeley Madonna (version 8.3.18) for further modeling analysis.
Role of the funding source
The sponsor had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit this report for publication.
. Iuliano AD, Roguski KM, Chang HH, et al. Estimates of global seasonal influenza-associated respiratory mortality: a modelling study. Lancet 2018;391(10127):1285–1300.
. Barr IG, Cheng AC. Difficulties of predicting the timing, size and severity of influenza seasons. Respirology 2018;23(6):562–563.
. Budd AP, Wentworth DE, Blanton L, et al. Update: influenza activity – United States, October 1, 2017-February 3, 2018. MMWR Morb Mortal Wkly Rep 2018;67(6):169–179.
. Biggerstaff M, Kniss K, Jernigan DB, et al. Systematic assessment of multiple routine and near real-time indicators to classify the severity of influenza seasons and pandemics in the United States, 2003–2004 through 2015–2016. Am J Epidemiol 2018;187(5):1040–1050.
. Garten R, Blanton L, Elal AIA, et al. Update: influenza activity in the United States during the 2017–18 season and composition of the 2018–19 influenza vaccine. MMWR Morb Mortal Wkly Rep 2018;67(22):634–642.
. Adlhoch C, Snacken R, Melidou A, Ionescu S, Penttinen P. The European Influenza Surveillance Network. Dominant influenza A(H3N2) and B/Yamagata virus circulation in EU/EEA, 2016/17 and 2017/18 seasons, respectively. Euro Surveill 2018;23(13):18-00146.
. China NHCotPsRo. National Statutory Epidemic Situation of Communicable Diseases in February 2018; 2018. http://www.nhc.gov.cn/jkj/s3578/201803/e07e57169d544fd09337b569738aa594.shtml
. Accessed April 3, 2018.
. Li WJ;, Wang DY. Research progress on the disease burden of influenza in China [J/OL]. Chinese Journal of Zoonoses.1–6 [August 3, 2019]. http://kns.cnki.net/kcms/detail/35.1284.R.20190723.1447.004.html
. Accessed February 14, 2019.
. Prevention CCfDCa. Recent Seasonal Influenza Epidemiology in China and Knowledge of Prevention and Control; 2018. http://www.chinacdc.cn/yyrdgz/201801/t20180108_158017.html
. Accessed March 1, 2018.
. Prevention CsCfDCa. Answers on the Issues Related to Recent Influenza Outbreaks from the China's Center for Disease Control and Prevention; 2018. https://mp.weixin.qq.com/s/hSWkd7Dz6UyTd8iPxoYaaw
. Accessed March 1, 2018.
. Zhu H, Lee ACY, Li C, et al. Low population serum microneutralization antibody titer against the predominating influenza A(H3N2) N121K virus during the severe influenza summer peak of Hong Kong in 2017. Emerg Microbes Infect 2018;7(1):23.
. Jin IK, Lee I, Park S, et al. Phylogenetic relationships of the HA and NA genes between vaccine and seasonal influenza A(H3N2) strains in Korea. Plos One 2017;12(3):e0172059.
. Prevention CfDCa. What's New for the 2017-2018 Flu Season? 2017. https://www.cdc.gov/flu/pdf/freeresources/general/whats-new-2017-2018-factsheet.pdf
. Accessed October 30, 2017.
. Ferguson NM, Cummings DAT, Simon C, et al. Strategies for containing an emerging influenza pandemic in Southeast Asia. Nature 2005;437(7056):209–214.
. Flannery B, Chung JR, Belongia EA, et al. Interim estimates of 2017-18 seasonal influenza vaccine effectiveness – United States, February 2018. Mmwr Morb Mortal Wkly Rep 2018;67(6):180–185.
. Peng ZBWD, Yang J, Yang P, et al. Current situation and related policies on the implementation and promotion of influenza vaccination, in China. Chin J Epidemiol 2018;39(8):1045–1050.
. Fraser C, Riley S, Anderson RM, Ferguson NM, May R. Factors that make an infectious disease outbreak controllable. Proc Natl Acad Sci U S A 2004;101(16):6146–6151.
. Mills CE, Robins JM, Lipsitch M. Transmissibility of 1918 pandemic influenza. Nature 2004;432(7019):904–906.
. Ting H, Xiuwei C, Lijuan Z, et al. Epidemiological analysis of influenza outbreaks in sichuan province from 2004 to 2016. Mod Prevent Med 2018;45(4):592–599.
. Jianming Z, Jing L, Xueyun L, Kaijie D, Congrui G. Effect of dynamic model of prevention and control measures for influenza outbreak in schools. Mod Prevent Med 2019;46(11):1946–1950.
. Little P, Stuart B, Hobbs FDR, et al. An internet-delivered handwashing intervention to modify influenza-like illness and respiratory infection transmission (PRIMIT): a primary care randomised trial. Lancet 2015;386(10004):1631–1639.
. China NHCotPsRo. The Notification of the Prevention and Treatment of the Flu in 2018; 2018. http://www.nhc.gov.cn/cms-search/xxgk/getManuscriptXxgk.htm?id=5737c258bb0c4a0493cb4f65fcf11be1
. Accessed March 1, 2018.
. Shu Y, Mccauley J. GISAID: global initiative on sharing all influenza data – from vision to reality. Eurosurveillance 2017;22(13):30494.
. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med 2000;19(3):335–351.
. Brown GD, Oleson JJ, Porter AT. An empirically adjusted approach to reproductive number estimation for stochastic compartmental models: a case study of two Ebola outbreaks. Biometrics 2016;72(2):335–343.