Epidemiology of influenza B in Australia: 2001‐2014 influenza seasons

Background Influenza B is characterised by two antigenic lineages: B/Victoria and B/Yamagata. These lineages circulate together with influenza A during influenza seasons, with varying incidence from year to year and by geographic region. Objective To determine the epidemiology of influenza B relative to influenza A in Australia. Methods Laboratory‐confirmed influenza notifications between 2001 and 2014 in Australia were obtained from the Australian National Notifiable Diseases Surveillance System. Results A total of 278 485 laboratory‐confirmed influenza cases were notified during the study period, comprising influenza A (82.2%), B (17.1%) and ‘other and untyped’ (0.7%). The proportion of notifications that were influenza B was highest in five‐ to nine‐year‐olds (27.5%) and lowest in persons aged 85 years and over (11.5%). Of all B notifications with lineage determined, 77.1% were B/Victoria and 22.9% were B/Yamagata infections. Mismatches between the dominant B lineage in a season and the trivalent vaccine B lineage occurred in over one‐third of seasons during the study years. In general, influenza B notifications peaked later than influenza A notifications. Conclusion The proportion of circulating influenza B in Australia during 2001‐2014 was slightly lower than the global average and was dominated by B/Victoria. Compared with influenza A, influenza B infection was more common among older children and young adults and less common in the very elderly. Influenza B lineage mismatch with the trivalent vaccine occurred about one‐third of the time.

Influenza B virus circulates in the population alongside influenza A, to varying degrees from season to season and by geographic region. 13,14 Influenza B was first isolated in the 1940s. The B/Victoria-like virus strain predominated during the 1980s and the B/Yamagata-like virus strain emerged in the late 1990s. 15 These two lineages of influenza B are antigenically and genetically different and studies in ferrets found no cross-reactivity between the two strains. 16 In recent decades, both lineages have cocirculated with varying relative intensity in the same season in many parts of the world including Australia. 17,18 While influenza B is thought to evolve more slowly than A, the interaction between the two virus types played a key role in the viral evolution process. 6,17 It has been suggested that influenza type B and the influenza A (H1N1) subtype have slower antigenic evolution and fewer epidemic episodes, resulting in fewer global movements, compared with influenza A/H3N2. A complex network of interacting factors such as viral evolution, epidemiology and human behaviour may describe different patterns of global circulation among influenza viruses. 6,19 It is not clear whether influenza B infection has similar clinical symptoms to influenza A infection including pandemic strain A/ H1N1pdm09. 20 Studies have shown controversial findings of clinical features in cases infected with A and B infections. [21][22][23] In both paediatric and adult populations, patients with influenza B infection showed similar clinical features compared with influenza A. 22,[24][25][26] Like influenza A, influenza B infection can lead to severe complications and death, in both children and adults. [27][28][29] However, another study found that abdominal pain, vomiting/diarrhoea, headache, general weakness, rhinorrhoea, pharyngitis and otitis were more common with influenza B infection compared with A/H3N2 in adults. 30 Pathological evidence of bacterial pneumonia and myocardial injury was reported from the autopsy tissue samples from cases with fatal influenza B infection in both children and adults. 31 Similar to influenza A, influenza B infection can present with neurological manifestations, such as febrile seizure and encephalitis/encephalopathy in hospitalised children. 28 Inactivated, trivalent influenza vaccine (TIV) was first introduced in the late 1970s and since then it has been widely used for prevention of influenza infection in many countries. 15 33 It is not fully understood whether vaccination with one B lineage confers significant protection against the other. However, vaccine cross-reactivity has been reported in adults. 34,35 Reduced vaccine effectiveness has been observed during seasons in which the vaccine B lineage does not match the circulating B lineage. 36,37 Since 2012, following recommendations by WHO, a quadrivalent influenza vaccine (QIV) (with two A strains and two B lineages) became available to avoid the risk of B lineage vaccine mismatches. The QIV offers improved protection against unmatched influenza B infections and has comparable safety measures to TIV in both children and adults. [38][39][40] The incidence of influenza B infection varies considerably by influenza season and by global region. 14 Here, the aim of our study was to describe the epidemiology and contribution of influenza B relative to influenza A in Australia.

| METHODS
We requested weekly, age-specific counts of laboratory-confirmed influenza notifications by type, subtype/lineage and age group across all states and territories in Australia between 2001 and 2014, from the National Notifiable Diseases Surveillance System (NNDSS).
In 2001, laboratory-confirmed influenza infections became notifiable in Australia. A scheduled reporting of laboratory-confirmed influenza infections is the primary source of influenza data in Australia and influenza notifications are reported initially from laboratories and some practices to regional, state or territory health jurisdictions, which report on to the NNDSS daily. The Australian case definition for notifiable influenza does not specify criteria for testing patients, so selection of patients tested may vary by individual clinician and by regional policy or clinical practice norms. The surveillance case definition for influenza is described below. Only confirmed cases are notified and a confirmed case requires laboratory definitive evidence: The threshold for a high titre could vary over time and between states and territories.
In addition, a selection of influenza isolates is referred by local laboratories to the WHO Collaborating Centre for Reference and Research on Influenza (WHOCC) which conducts molecular strain typing to inform the WHO for vaccine formulation policy as well as for monitoring genetic changes and antiviral resistance of the influenza viruses. The WHOCC information on circulating strains is reported in national surveillance reports. 42,43 However, selection of influenza isolates for referral to WHOCC is not nationally standardised and thus may not be a random sample of the case population.

| Study data and analysis
In the data set received, the recorded date of onset was used for the disease onset date if available. Otherwise, the earliest of the specimen collection date, or the notification received date was applied. Although there are variations between states and territories, laboratory testing methods used for diagnosis include at least one of the followings: nucleic acid testing, antigen detection, serology and culture. Except

| Mismatch with TIV vaccine B lineage
We also determined the number of years where mismatch occurred between the TIV vaccine B lineage and the circulating B lineage in Australia for the study period. The NNDSS also provided subtype and lineage data as part of the data request in our study, but few notifications included the information. Thus, type B lineage mismatch information analysed here was obtained from published influenza surveillance reports. 42,43,45 Influenza B lineage mismatch was defined as a season when >60% of circulating B lineage virus was different to the lineage included in the TIV for that season. In the season, if both B lineages cocirculated at equal or almost equal proportions (40%-59%), then a partial mismatch was defined for the year.
In Australia, the influenza season typically begins in May, peaks between July and August and finishes in October, although this can vary. 1,43 To determine differences in peak seasonal activity between influenza A and B, we determined the week in which the maximum number of notifications was received in the season for each type in each of the 14 years, except 2009. The year 2009 was excluded from this analysis as it was a pandemic year. We then compared the timing of the annual epidemic peak between influenza A and B.

| Ethics approval
Ethical approval was received for this study from the University of   19.8% in Western Australia (Figure 1). Figure 2 shows the proportion of influenza notifications in each age group, which were due to influenza B for the entire study period. The age groups with the highest proportions of influenza B were school-age children and young adults, For seasons in which influenza B represented at least 20% of all notifications, age-specific notification rates per 100 000 population are shown in Figure 3A (Figure 4). In addition, majority of influenza notifications had limited information regarding further subtypes or lineages in the data set. For influenza B, a small proportion of notifications had lineage information available by year, which is presented in the Table S1.  Table 2.

| RESULTS
Timing of peak influenza activity also varied between influenza A and B in the study. The peak influenza B activity appeared later than influenza A in seven seasons (54%), and the epidemic peak of both influenza A and B was overlapped in two seasons [ Figures S2 (A,B)]. The difference in peak epidemic activity between influenza A and B was also determined. The peak week of influenza B fell a median 2 weeks later, than that of influenza A over the study period (ignoring pandemic year 2009). The median peak week for influenza A was week 32 (range 32-39) and for influenza B was week 34 (range 23-41).

| DISCUSSION
We described the epidemiology and seasonal pattern of influ- In general, peak influenza B activity arrived later in the season than peak influenza A, and both types cocirculated in each season to varying amounts. However, this relative incidence pattern can vary across different regions of the world, 52 although our finding is compa-

| CONCLUSION
Our findings indicated that proportion of influenza B circulating in