Influenza surveillance indicates that influenza viruses circulate year-round in the Delhi (North India) area and that influenza contributes significantly to the number of ILI patients seeking care in government facilities in urban and peri-urban areas of Delhi, especially during the rainy and winter seasons. Although patients with ILI were seen throughout the year, there were distinct seasonal peaks in influenza virus isolations from ILI cases in July-August which coincided with peak rainfall in most years. Overall, both influenza A and B viruses co-circulated throughout the surveillance period. Seasonal influenza A(H1N1) and A(H3N2) viruses, however, predominated during distinct periods with little overlap, i.e. seasonal A(H1N1) was the predominant circulating virus in 2007, whereas A(H3N2) showed some circulation throughout the study period, with a marked increase in circulation in the first half of 2009. Our observation of rapid displacement of circulating A(H3N2) by pandemic virus during the initial phase of 2009 pandemic has also been observed in other parts of the world 
. The dynamic patterns and displacement of one subtype with another may partly be explained by limited immunological cross-reactivity between influenza subtypes. Pandemic influenza A(H1N1)pdm09 did not show typical seasonality in 2009 with bimodal peaks observed in September and December 2009, but a distinct peak in rainy season (August-September) was seen in 2010 in Delhi. We observed a consistent pattern of peak influenza activity in both urban and peri-urban surveillance sites, with some differences in circulating strains in urban and peri-urban areas.
The temporal peaks in influenza virus circulation coincided with rainy and winter seasons in and around Delhi. Our findings of seasonal peaks of influenza activity in India are consistent with data reported from surrounding countries in the region, where peaks of influenza activities coincide with rainy seasons 
. In some tropical regions, there is high background influenza activity throughout the year with distinct peaks appearing during monsoon or cooler months 
. While the exact mechanisms leading to variation in influenza seasonality are not clear, attempts to correlate fluctuation in meteorological variables have shown relationship with influenza positivity during the rainy season in the tropics 
. Multi-site influenza surveillance from different geographic regions in India has also revealed a positive correlation between the rainy season and rates of influenza virus isolation 
. The seasonality of influenza in Delhi area has policy implications such as vaccination timing and the use of northern vs southern hemisphere vaccine formulations in this region.
A strong surveillance system which encompassed diverse settings enabled us to monitor the emergence of Influenza A(H1N1)pdm09 in and around Delhi, North India. During the initial pandemic phase, we observed a higher prevalence of seasonal influenza A/H3N2 than Influenza A(H1N1)pdm09. This observation is different from some cities in Mexico and United States, where >90% of the positive samples were due to 2009(H1N1) influenza during the first wave of pandemic 
. The trend in Delhi changed in September, when Influenza A(H1N1)pdm09 became predominant, and a distinct bimodal pattern of peak activity was observed in September and December 2009, possibly due to colder weather where influenza viruses transmit more readily 
. These observations are similar to situations in other parts of the world where Influenza A(H1N1)pdm09 either completely replaced seasonal influenza 
or co-circulated with seasonal influenza viruses 
. During previous pandemics, the novel virus subtype has replaced previously circulating viruses 
. For example, H1N1 was replaced by H2N2 in 1957, subsequently H3N2 emerged in 1968, and H1N1 reemerged in 1977 
. The latter two have co-circulated since then, although in one season only one or the other tends to dominate 
. Now the Influenza A(H1N1)pdm09 virus has replaced the seasonal influenza A(H1N1) virus circulating in preceding years.
We observed influenza B circulation in the early part of 2010, followed by second wave of Influenza A(H1N1)pdm09 in August 2010, similar to what is observed in present times in most countries around the globe 
. While the exact mechanism causing such a discrete peak in August 2010 is not clear, molecular studies suggest that large populations in the tropics can serve as reservoirs of influenza infection throughout the year, with reseeding of drifted viruses possibly leading to outbreaks 
. Whether such a drift in Influenza A(H1N1)pdm09 could account for the peak remains to be determined, though to date minimal divergence among the limited sequences of Influenza A(H1N1)pdm09 makes this explanation unlikely 
. Alternatively, some environmental factors may play a role in virus survival and it is plausible that drastic changes in meteorological parameters, such as lowering of vapor pressure 
, may account for a sharp peak of Influenza A(H1N1)pdm09 in August 2010.
Another important aspect of our study is the comparative analysis of influenza seasonality and circulating virus strains in urban and peri-urban settings. Little is known about how demographic, nutritional and other environmental factors impact circulating influenza viruses, although vaccine studies have shown considerable differences in humoral and cell mediated immune responses to influenza vaccination among semi-urban and rural school children in Gabon 
. In the present study, peaks of influenza circulation were comparable in urban and peri-urban areas, but circulating strains varied somewhat in some years. For instance, a higher proportion of circulating A(H3N2) viruses was observed in the peri-urban site in July-August 2009, when compared to urban site. Whether a new antigenic variant of H3N2 had emerged, to which there was only partial immunity in the peri-urban site, remains to be determined. Studies are underway to better understand the evolution of H3N2 viruses from these surveillance studies.
Age group analysis revealed that the highest rate of influenza A(H1N1)pdm09 positivity was observed among children in the age group >5–18 yrs, which could be due to high exposure rates among school aged-children, who have highest numbers of contacts among all age groups, and therefore, appear to be at the leading edge of pandemic 
. The attack rates of Influenza A(H1N1)pdm09 have varied among various age groups in different parts of the world, however, most have found children to represent the highest risk group 
. Recent modeling study has suggested that while school-age children typically experience the highest attack rates during early pandemic phase, the burden likely shifts to adults during the subsequent season 
. It remains to be seen if Influenza A(H1N1)pdm09 strains of influenza may shift to adults in the next wave.
Phylogenetic analysis of HA1sequences revealed that all belonged to clade 7 with minimal diversity in sequences whether derived from urban or peri-urban area, and all had signature sequence S203T in the HA1 subunit 
. Our data are in concordance with published data where clade 7 of Influenza A(H1N1)pdm09 viruses remain the predominant circulating strains globally 
Taken together, our data provides evidence that the seasonality of influenza in Delhi area is related to rainfall, although the Influenza A(H1N1)pdm09 emerged after the rains stopped in 2009. Further, we demonstrate that seasonal influenza types and subtypes change constantly and unpredictably over time. In summary, continued surveillance globally will help to better define seasonal patterns in the circulation of influenza A and B viruses, regional differences in influenza seasonality, as well as to determine optimal periods to implement influenza vaccination programs among priority populations.