Background
The World Health Organization recommends that seasonal and alert thresholds and average epidemic curves be established in countries as tools for early detection of influenza outbreaks to help control annual influenza epidemics [
1,
2]. Seasonal influenza epidemics occur all over the world and cause substantial economic burden through health care costs and absenteeism [
3]. Although surveillance is now routinely conducted in many countries providing seasonality data and recommendations for timely vaccination [
4,
5], not all countries have established influenza thresholds or epidemic curves. Developed countries like the United States (US), United Kingdom (UK), most European countries, Australia, and New Zealand [
6‐
10] are already using thresholds for influenza surveillance calculated through various methods [
11,
12]. In contrast, few countries in Asia have established their own thresholds and epidemic curves to determine if an influenza outbreak has started while concurrently monitoring influenza activity [
2]. The purpose of this article is to report the results of Philippine National Influenza Surveillance from 2006 to 2012, particularly on the establishment of seasonal and alert thresholds, and average epidemic curve according to methods in the WHO manual [
1], define the seasonality of influenza, and describe the circulating influenza strains.
Discussion
In this study we have established the seasonal and alert thresholds, and average epidemic curve based on 5 years of influenza surveillance data using the method described in the WHO surveillance manual [
1], described the seasonality of influenza in the Philippines, and presented the influenza strains circulating in the country from 2006 to 2011. While influenza viruses were detected year-round, increased activity was seen from June to November. We used the established thresholds and epidemic curve to assess the influenza season for 2012, its onset and end, and its severity in relation to these thresholds. We also showed that intensity of influenza activity was different for each year, and that different strains of influenza viruses circulated every year. Analysis of influenza activity using weekly detections showed more than one peak of activity within the influenza season from June to November of each year. Different dominant strains were associated with each peak. Two influenza B lineages circulated for five out of six surveillance years. Circulating strains matched the Southern Hemisphere vaccine more often than Northern Hemisphere vaccine strains.
The WHO surveillance manual has emphasized the need for standardized tools to establish thresholds and epidemic curves that could be used to compare influenza season between countries, and to assess the severity of the current season in a country to previous seasons of that same country. The method recommended in the manual [
1] is relatively simple and uses software, which is readily available, and can be implemented by NICs as long as five or more years of influenza surveillance data are available. Knowing the usual baseline level of disease and the seasonal pattern as a point of reference aids in determining whether the current season is atypical both in timing and relative severity compared to previous ones. This information can help improve the accuracy of clinical diagnosis, appropriate use of antiviral medication, and the uptake and timeliness of seasonal influenza vaccines. As of this writing, we have not seen studies in Southeast Asia reporting established alert and seasonal thresholds or epidemic curves using the WHO surveillance manual. Although there is one publication from Thailand wherein Early Aberration Reporting System (EARS) and Cumulative sum (CUSUM) were used, calculations were only for ILI [
22]. Routine influenza weekly monitoring reports from surveillance activities in Asia included ILI and influenza strains but compared these to historical influenza activity [
23]. We have not been able to obtain information from countries in Asia on the use of the WHO method or moving epidemic method (MEM) to establish influenza seasonal thresholds or epidemic curves. At this point in time, there are already many countries in Asia, which have more than 5 years of surveillance data. The WHO method for calculating thresholds and epidemic curves could very well be applied in these countries.
The Philippines, Bangladesh, Cambodia, India, the Lao People’s Democratic Republic, Thailand and Vietnam have similar influenza seasonality [
5]. Seasonality has been associated not only with climatological but also latitudinal variations [
5,
24‐
26]. Influenza virus infection was found throughout the year in these countries, but more than 60% of influenza positivity rates were observed during the months of June to November [
5]. The capital cities of these countries lie north of the equator (between 11.6 to 28.7° N). Thus, April to May is the most appropriate month for influenza vaccination for the Philippines and countries with similar seasonality [
5]. In areas in India north of 30° N however, there was also increased influenza activity during winter, so that a separate vaccination timing has been recommended [
5,
24,
25] for this country. In contrast, tropical countries in Asia below 11° N like Indonesia (Jakarta, 6.2° N), Malaysia (Kuala Lumpur, 3.2° N), and Singapore (1.3° N), influenza was found all throughout the year with no distinct influenza season [
5]. Specific analysis on the association between influenza activity and climate variables has not been done in this study but June to November coincides with the rainy season in the Philippines similar to that of Cambodia [
27], Myanmar [
28], North Thailand [
29], Bangladesh [
5,
30], Lao [
31], and some parts of India South of Srinagar [
24,
25]. Factors other than the rainy season might also play a role for the increased influenza virus activity starting the month of June in the Philippines. Schools open in June for all levels in the Philippines, suggesting the effect of crowding on influenza transmission.
More than one peak of influenza activity was observed during the surveillance years but these peaks were confined to the defined influenza season from June to November. The presence of the different dominant influenza strains for each of the peaks may be through the following mechanisms: virus interaction leading to interference between viruses, break in transmission due to school vacation, development of immunity to the dominant virus, or a combination of these factors. Goldstein et al. [
32] have noted that there was a negative association between strains’ incidences indicating that high infection rates with one strain can interfere with the transmission of other strains. Raoult [
33] has postulated that the dynamics of influenza virus subtypes against those of other subtypes and even other respiratory viruses is complex and interference between these viruses might impact on their transmissibility in humans. School closure has been associated with reduced transmission of influenza virus because of decreased contact rate [
34,
35]. School vacation or semestral break occurs at around the end of October to the first week of November. We have observed that influenza activity decreases during this time of school disruption. A break in the transmission of the first peak dominant virus causes co-circulating influenza strains to increase in activity leading to the development of a second peak. Increased transmission of the first dominant influenza strain may later lead to development of sufficient immunity among the population, which in turn will stop or decrease the spread of the virus. Consequently increased activity of co-circulating strains [
36,
37] occurs.
The influenza strains circulating in the Philippines from 2006 to 2011 frequently matched the Southern rather than the Northern Hemisphere vaccine strains. The vaccine currently used in the Philippines is the Southern Hemisphere vaccine. Influenza B strains both of the Yamagata and the Victoria lineages were found during the years of surveillance. With these findings, the appropriate timing of vaccination should be from April to May, the current choice of Southern Hemisphere influenza vaccine should be continued, and most importantly, the vaccine should contain two influenza B lineages. Therefore, quadrivalent rather than trivalent vaccines would be a better choice.
There are several limitations to this study. The first limitation is that we used collated data from all sentinel sites to analyze seasonality. Sixty-four percent of influenza positive cases came from Luzon where Manila is located, 15% from the Visayas islands, and 21% from Mindanao. Since majority of the cases came from the Luzon islands, the results may be representative for only the north of the Philippines. The actual latitudinal location of the country extends from 5 to 20° N. Mindanao islands are situated below 9.7° N up to 5° N. The second limitation is that we were not able to compare the WHO method to other analytic methods. The third limitation is that data from this study mostly came from the pediatric age group (<15 years of age). There were very few adult ILI consultations probably because adults are able to handle ILI much better than children. The fourth limitation is that we could only match vaccine and circulating strains through names of the strains detected and not by genomic sequencing or antigenic testing. Although the Philippine National Surveillance followed the CDC definition for ILI, the definition of ILI was different from other countries and this may be considered a fifth limitation. Despite these limitations, this study has given a valuable insight into the measurement of seasonal and alert thresholds, and epidemic curves that will help DOH influenza program managers in the management of annual influenza epidemics in the country.
Acknowledgements
Dr. Eric Tayag, Director of the National Epidemiology Center (NEC)-DOH, Regional Directors of the Center for Health Development and Head of the Regional Epidemiology Surveillance Units in Regions I, II, III, V, VI, VII, IX, X, XI, XII and CAR, Chief of Hospitals, Municipal Health Officers, health center physicians, nurses, and other health workers in the sentinel sites, the Influenza Surveillance Officers (ISOs) in the sentinel sites, the staff of the Data Management Unit (DMU) of RITM. Acknowledgement is also given to Dr. Anthony Mounts of CDC for the initial discussion on the WHO method of calculating thresholds and epidemic curves using the first version of the WHO manual [
38]. We acknowledge the contribution of the RITM-Tohoku team for allowing us to use the data collected in the BOD study.