Epidemiological and Virological Characteristics of Influenza in the Western Pacific Region of the World Health Organization, 2006–2010

Background Influenza causes yearly seasonal epidemics and periodic pandemics. Global systems have been established to monitor the evolution and impact of influenza viruses, yet regional analysis of surveillance findings has been limited. This study describes epidemiological and virological characteristics of influenza during 2006–2010 in the World Health Organization's Western Pacific Region. Methodology/Principal Findings Influenza-like illness (ILI) and influenza virus data were obtained from the 14 countries with National Influenza Centres. Data were obtained directly from countries and from FluNet, the web-based tool of the Global Influenza Surveillance and Response System. National influenza surveillance and participation in the global system increased over the five years. Peaks in ILI reporting appeared to be coincident with the proportion of influenza positive specimens. Temporal patterns of ILI activity and the proportion of influenza positive specimens were clearly observed in temperate countries: Mongolia, Japan and the Republic of Korea in the northern hemisphere, and Australia, New Zealand, Fiji and New Caledonia (France) in the southern hemisphere. Two annual peaks in activity were observed in China from 2006 through the first quarter of 2009. A temporal pattern was less evident in tropical countries, where influenza activity was observed year-round. Influenza A viruses accounted for the majority of viruses reported between 2006 and 2009, but an equal proportion of influenza A and influenza B viruses was detected in 2010. Conclusions/Significance Despite differences in surveillance methods and intensity, commonalities in ILI and influenza virus circulation patterns were identified. Patterns suggest that influenza circulation may be dependent on a multitude of factors including seasonality and population movement. Dominant strains in Southeast Asian countries were later detected in other countries. Thus, timely reporting and regional sharing of information about influenza may serve as an early warning, and may assist countries to anticipate the potential severity and burden associated with incoming strains.


Introduction
Influenza, an acute viral infection characterised by fever with cough and/or sore throat, occurs as annual seasonal epidemics in winter or early spring in countries with temperate climates [1]. These yearly epidemics pose a substantial health burden arising from related complications such as lower respiratory tract infections and exacerbation of cardiopulmonary and other chronic diseases. Influenza burden in tropical or subtropical countries is not well-defined, although there is mounting evidence that prevalence and excess mortality from influenza are comparable to those seen in temperate countries [2,3].
The Global Influenza Surveillance Network (GISN) was established by the World Health Organization (WHO) in 1952 and was re-named the Global Influenza Surveillance and Response System (GISRS) in 2011 [4]. The GISRS monitors the impact and evolution of influenza viruses and emergence of novel influenza viruses with pandemic potential. It also provides recommendations on suitable virus strains for inclusion in vaccines (twice yearly for northern and southern hemisphere seasons) and on diagnostic tests and antiviral drug sensitivity. The GISRS currently consists of six WHO Collaborating Centres (WHO CCs) for Reference and Research on Influenza (five for human influenza and one for animal influenza), four Essential Regulatory Laboratories (ERLs) and 136 WHO National Influenza Centres (NICs) in 106 WHO Member States.
The WHO's Western Pacific Region includes 37 countries and areas that span from the northern hemisphere through the tropics to the southern hemisphere ( Figure 1). This region covers nearly one-quarter of the world's population with approximately 1.6 billion people. The GISRS in the region currently consists of 21 NICs in 15 countries, three WHO CCs, one each in Australia, Japan and China, and two ERLs, in Australia and Japan.
Progress that has been made in the development of influenza surveillance capacity in the Western Pacific Region was demonstrated by the rapid availability of data from 34 countries and areas during the influenza A(H1N1)pdm09 pandemic [5].
However, there has been limited reporting of epidemiological and virological data on seasonal influenza in the region [6]. This study describes the epidemiological and virological characteristics of influenza for 2006-2010 in the Western Pacific Region, with a view to supporting the development of future regional strategies and workplans to further strengthen influenza programmes.

Data Collection
Influenza-like illness (ILI) and influenza virus data for 2006-2010 were obtained from the 14 countries with National Influenza Centers (NICs) in the Western Pacific Region that participated in the Fifth Meeting of NICs in WHO's Western Pacific and South-East Asia Regions in June 2011. Participating countries included Australia, Cambodia, China (including Hong Kong), Fiji, New Caledonia (France), Japan, Lao People's Democratic Republic (PDR), Malaysia, Mongolia, New Zealand, the Philippines, the Republic of Korea, Singapore, and Viet Nam (Figure 1). Data were obtained from NICs and disease surveillance units with support of WHO country offices.
Data were collected using an email survey between April and October 2011. The survey included a questionnaire that ascertained qualitative data on ILI surveillance systems, case definitions applied and virological surveillance methods. The email survey also obtained data on number of weekly or monthly ILI cases reported and total number of consultations for any cause by age group. Age groups were categorized as 0 to ,2 years, 2 to ,5 years, 5 to ,18 years, 18 to ,65 years, and $65 years; however, alternative age groupings were accepted to allow for reporting differences by countries. These data were obtained using a reporting template developed in Microsoft ExcelH.
Virological data were initially extracted from FluNet, the webbased data collection and reporting tool of the GISRS, to which NICs report the number of specimens processed for influenza virus testing and the number testing positive for influenza A(H1N1)pdm09, seasonal A(H1), A(H3), A(H5), A(not subtyped), B(Victoria), B(Yamagata) or B(lineage not determined). The extracted data were sent electronically to each participating country in March 2011 for confirmation. Using the abovementioned Microsoft ExcelH template, virological data by week were also requested from countries that had not reported to  Table S1 for the type and period of data available from each country participating in this study.

Data Analysis
ILI case definitions and surveillance systems were tabulated for each reporting country.
For data presentation and comparison, the raw data were tabulated and graphed, and countries were grouped according to geographic location and apparent similarities in ILI and virus circulation patterns. The groups are: (A) the Republic of Korea and Mongolia, (B) China, (C) Cambodia, Lao PDR, Malaysia, Philippines, Singapore and Viet Nam, and (D) Australia, Fiji, New Caledonia (France) and New Zealand. Percent ILI was calculated as a proportion of ILI cases per total consultations, and percent specimens positive for influenza was calculated as a proportion of influenza positive specimens per total specimens submitted for influenza testing. For each group of countries, the denominators (total consultations or total specimens tested) were first summed and then used to calculate regional proportions. Ratios for ILI cases and ratios of specimens positive for influenza could not be calculated for Japan, as denominators were not reported in the data collection template, thus it was not included in any of the groups. Case and specimen numbers were described in the results text.

Results
Fourteen countries responded to the survey. ILI case definitions, surveillance approach and intensity varied across countries, as did the number of years of reporting.
The definition of ILI varied across sentinel surveillance schemes within countries and areas and across countries (Table 1). Seven countries and one area reported applying the WHO definition: a person with sudden onset of fever .38uC and cough or sore throat in the absence of other diagnoses. Other definitions applied by reporting countries used broader clinical symptoms, separated definitions for different age groups or limited the timeframe for onset of different symptoms. All 14 countries reported using a sentinel site approach for influenza surveillance ( Table 1). The number and type of sites varied widely across countries and areas, and some countries incrementally increased the number of surveillance sites during 2006-2010, especially in response to the A(H1N1)pdm09 pandemic. Confirmatory testing of a proportion of ILI cases was conducted through the surveillance systems. The number of specimens collected by surveillance schemes varied within and across countries and areas. Data were not collected on methods for selecting cases for specimen collection.
During the study period, 1,070,792 influenza virus detections were reported to FluNet globally, of which 301,195 (28%) were reported from NICs in the Western Pacific Region ( Figure 2). Within the Western Pacific Region, China contributed the most data for virus detections (57%), followed by Japan (19%) and the Republic of Korea (7%).
The number of Western Pacific Region countries submitting virus information to FluNet and the reported number of specimens tested for influenza increased over the five-year study period ( Table 2) ILI case numbers and total all-cause consultations were provided by 10 of the 14 countries in this study. Australia, China (including Hong Kong), the Republic of Korea, Malaysia, Mongolia and Viet Nam provided data for the five year study period. New Zealand provided aggregate data for the five years, and was analyzed separately. Japan provided the total number of ILI cases but not the total number of all-cause consultations for the study period, and was also analyzed separately. Lao PDR provided ILI data for three years (2008-2010) and Cambodia, the Philippines and Singapore provided ILI data for two years (2009-2010) ( Figure 1). Data for total specimens tested and number positive for influenza were available for all countries except Japan, where only the number of specimens positive for influenza was provided.
Over the five year period, peaks in ILI reporting appeared to be coincident with the proportion of specimens positive for influenza ( Figure 3). This was particularly evident for the northern and southern hemispheres (Panels A & D, Figure 3).
A yearly seasonal pattern typical of temperate northern hemisphere countries was observed in the Republic of Korea and Mongolia (Panel A, Figure 3), where peaks in ILI activity were detected in January-March each year. A similar seasonal pattern was observed in Japan, where the reported number of ILI cases and the number of specimens positive for influenza were greatest in the first three months of each year (data not shown). A seasonal pattern typical of temperate southern hemisphere countries was observed in Australia (Panel D, Figure 3), where ILI activity and specimens positive for influenza peaked in July-September each year. A similar pattern was observed in New Zealand where ILI activity peaked approximately five weeks earlier than it did in Australia each year (data not shown). Two annual peaks in activity were observed in China (Panel B, Figure 3 Figure 4).
In 2009, pandemic A(H1N1)pdm09 was predominant in the region. A(H3) viruses were reported from all countries in the region, and peaked earliest in Japan, Mongolia and the Republic of Korea, followed by all other countries except China which saw a later A(H3) peak in August through September. Australia, the Philippines and Singapore were the first countries to report detection of the pandemic A(H1N1)pdm09 subtype to FluNet in May 2009 (although New Zealand was the first country in the region to detect this virus). The dominance of this virus in the

Discussion
Influenza was prevalent in all countries and areas that participated in this study, which included countries and areas in temperate parts of the northern and southern hemispheres and countries and areas in the tropics. Despite differences in surveillance approaches, methods and intensity, commonalities in disease and virus circulation patterns were identified.
The seasonal pattern of influenza activity observed in Mongolia and the Republic of Korea in the northern hemisphere is consistent with a previous report from Mongolia [7]. Similarities in seasonality were also observed amongst Australia, Fiji, New Caledonia (France) and New Zealand in the southern hemisphere. Other countries in the region had less discernable seasonal patterns with influenza activity occurring throughout the year, which is aligned with previous research findings and generally consistent with their location in the tropics [6,8,9,10]. Although Fiji and New Caledonia (France) are tropical, their influenza circulation patterns appeared to be more aligned with Australia and New Zealand than other tropical countries in the region. This highlights that influenza circulation may be dependent on a multitude of factors including seasonality and people movement pathways. Patterns of disease may also vary within countries, such as China where research suggests that seasonal patterns and peaks of influenza activity differ between the northern and the southern parts of the country [9].
Detection of A(H3) in Australia and New Zealand in mid-2007 was preceded by frequent detection of the virus in Cambodia, China, Lao PDR, Malaysia, the Philippines, Singapore and Viet Nam in early 2007. This supports the hypothesis that dominant strains in East and Southeast Asian countries spread to other countries in the region [11,12,13,14]. Similarly, the dominance of former seasonal A(H1) in Mongolia and the Republic of Korea in December 2008 to February 2009 was preceded by dominance of this subtype in China in mid-2008. In the future, timely reporting and regional sharing of influenza virus information may serve as an early warning, and may assist countries to anticipate the potential severity and burden associated with incoming influenza strains. One example of this in practice is in the Federated States of Micronesia which has utilized the influenza surveillance findings from Guam for early warning (personal communication). Efforts are ongoing within GISRS in the region to increase timely reporting and regional sharing of influenza virus information to maximize the utility of these surveillance systems.
This study included data on the influenza A(H1N1)pdm09 pandemic, providing the opportunity to assess changes in patterns of influenza activity during and after this outlier event. The novel A(H1N1)pdm09 virus was first detected in Mexico in April 2009. Despite rapid spread to the Western Pacific Region, pandemic virus activity peaked in the temperate countries during their typical annual influenza seasons: the winter months in Australia and New Zealand (July to September) and in Japan, Mongolia and the Republic of Korea (November to January). Even though this virus dominated epidemiological patterns in 2009 to 2010, other types/subtypes of influenza co-circulated at low levels during the pandemic period throughout the region [15]. The epidemiological characteristics of the pandemic in the region have been described previously [5].
Overall, temporal patterns of influenza activity appeared to be coincident between ILI syndromic surveillance and virological surveillance. This highlights the utility of combining data from ILI surveillance with virological surveillance for interpreting influenza activity. Virological surveillance adds specificity to the overall influenza surveillance program, informs vaccine strain selection at the global level and enables prompt detection of antigenic, genomic and drug-sensitivity changes in the virus. An example of the latter was the detection of increased antiviral drug resistance among seasonal A(H1) strains in 2008 [16]. This highlights the importance of timely reporting into these surveillance systems, including FluNet, so that virus detection is synchronized with disease activity and controls measures.
This study highlighted that regional participation in global virological surveillance systems has been high. Of the 34 different virus strains recommended by WHO for use in influenza vaccines between 1988 and 2010, 26 (76.4%) were developed from isolates from this region [17]. Participation increased during the study period as evidenced by the addition of three NICs, one each in Cambodia, Lao PDR and Viet Nam, and one WHO Collaborating Centre in China. There was also an increase in the number of sentinel surveillance sites in some countries.
This study had a number of limitations. A key limitation of the study is that the data are aggregated across countries, over time and age groups. This limits the type of analyses and inferences made about the data. Since data aggregation can result in reporting bias, we have tried to reduce the risk of bias through the involvement of authors from all countries participating in the study. Authors from each country were tasked with ensuring that the analyses and interpretation were in line with their nationallevel context and conclusions.
Reporting countries and areas applied different case definitions, as well as different methodologies for data collection and sampling, and some incrementally added surveillance sites or changed surveillance methods or testing methods. These differences limited the ability to make comparisons and draw conclusions, especially for the magnitude of disease activity and representativeness of the surveillance systems.
Use of different ILI case definitions impacts disease prevalence comparisons, where some definitions may be more sensitive than others in case-ascertainment. Even though the ILI case definitions used by the 14 reporting countries appear relatively similar, the impact of the differences on the magnitude of disease activity was not assessed. This is an area of future work. It may be argued that consistent use of different case definitions over time still enable comparison of disease trends. However, in this study, the data collection tool did not assess whether countries changed their ILI case definitions during the five year period (2006-2010). Again, this necessitates further research to determine whether case definitions have changed over time and the impact changes may have had on reported ILI trends.
Although the data collection tools included disaggregation by age groupings, the available data was too limited to do summary analyses. In addition, not all influenza surveillance information in the region was captured. For example, countries and areas without NICs in the Pacific also contribute virus data to the international influenza surveillance systems by submitting samples to reference  laboratories for testing, but their data were not included in this study.

Conclusions and Way Forward
This collaborative effort to summarize epidemiological and virological characteristics of influenza adds to the body of knowledge on influenza in the Western Pacific Region. This study highlights the progress made in influenza surveillance capacity in countries of the Western Pacific Region between 2006 and 2010. Epidemiological and virological characteristics of influenza appeared to be interrelated across countries, underscoring the importance of information-sharing and collaboration. However, it is clear that variations in data collection, case definitions and testing procedures currently limit comparisons.
Regional strategies have been developed to further strengthen influenza surveillance and research in the region and address the current limitations in cross-region analysis [18]. Countries and areas are encouraged to develop detailed workplans and monitor progress in three areas of work: (1) defining the epidemiology and burden of influenza; (2) improving virological testing capacity; and (3) improving communication and reporting through the development or strengthening of regional and global networks.
Through implementation of workplans to strengthen influenza surveillance, conduct of the research agenda and regular information-sharing about epidemiological findings, policy makers in the Western Pacific Region will be able to refine their policies and enhance influenza control based on a solid foundation of knowledge of the burden and characteristics of influenza virus infection in their own and neighbouring countries.

Supporting Information
Table S1 Type and period of available data for this study from the participating Western Pacific Region countries. *These data were used as denominators to calculate proportions of ILI and influenza positive specimens, respectively. ' Data on total consultations and total specimens tested were not provided, therefore data were analyzed separately. # Data on ILI cases and total consultations were provided aggregated as rates, therefore data were analyzed separately. (DOC)