Our study is the first attempt to estimate burden of influenza among acute medical hospitalized patients in a large population based study in rural India. Adopting a broad surveillance case definition ensured that influenza cases presenting as acute exacerbations of chronic disease or having atypical presentations were not missed. However, the numbers of patients with underlying chronic disease is low; there is a possibility that chronic diseases go undiagnosed in the community. Population-based surveillance in a well-enumerated population and health utilization surveys enabling adjustment for patients admitted to non-participating hospitals allowed us to better estimate the annual incidence of all acute medical illness hospitalizations, respiratory hospitalizations, as well as influenza-associated hospitalizations. I Influenza accounted for a substantial proportion of all acute medical and respiratory illness hospitalizations during and after the emergence of the A(H1N1)pdm09 virus. Use of ILI or SARI case definitions, to estimate the burden of influenza-associated hospitalizations would have substantially underestimated the impact of influenza in the study population.
The average annualized incidence of influenza-associated hospitalizations in current study was 44.1 per 10,000 persons, which is substantially higher than 3.6–11.5/10,000 persons reported in the United States 
. We undertook this burden study when a pandemic virus was emerging, thus our estimated incidence rates of influenza-associated hospitalizations are higher than what has been observed for seasonal influenza elsewhere. We found the highest influenza-associated hospitalization incidences occurred among persons aged 5–29 years which differs markedly from findings for seasonal influenza in other countries. In the US and Canada, influenza-associated hospitalization incidences are highest among those >65 years with underlying medical conditions (40–56 per 10,000), infants aged <6 months (18–104 per 10,000) and adults >65 years without underlying conditions (9–23 per 10,000) 
. Similarly, in Thailand and Hong Kong, hospitalization incidences are highest among older adults, and incidences among children exceed those in the United States 
.In contrast, we found the lowest hospitalization incidences in persons aged >60 years, followed by infants aged <1 year. The difference in the relative age distribution of influenza-associated hospitalization incidence between our study and others might be due to the A(H1N1)pdm09 virus which resulted in increased morbidity and mortality among younger persons compared to seasonal influenza. However, persons aged 1–29 years had the highest hospitalization incidence for acute medical illness, respiratory illness, and influenza suggesting that our population is likely to have differed from populations in other studies in propensity to seek care. Despite these differences, the incidence estimates for infants aged <1 year and children aged 1 to 5 years in our study are similar to the estimates from the US for similar age groups 
, suggesting that these age groups be targeted for influenza prevention strategies.
In our population, 46–57% of the burden of influenza-associated hospitalizations was accounted for by A(H1N1)pdm09 influenza virus with seasonal influenza A virus accounting for 35% of the burden in 2009–10 and 4% in 2010–11. Children and young adults bore a disproportionate burden of A(H1N1)pdm09-associated hospitalizations, possibly due to lack of prior exposure to similar viruses. In a study carried out in Pune, high rates of influenza-associated hospitalizations and deaths among persons aged <35 were observed during the peak of A(H1N1)pdm09 activity in August-September 2009 
. Influenza B virus infection accounted for 19–39% of all influenza-associated hospitalizations in our study, despite conceptions that influenza B is typically associated with milder disease 
The seasonality of influenza virus circulation and seasonal patterns for excess hospitalizations for pneumonia and influenza during the winter are known for temperate countries 
. Some tropical and sub-tropical countries have documented significant transmission throughout the year while others have documented a biannual pattern 
. Although the A(H1N1)pdm09 virus resulted in influenza-associated hospitalizations continuously during July2009–April2011, we documented a clear seasonality to influenza-associated hospitalizations in our study with peaks during the monsoon season each year. In year 1, Influenza A(H3N2) predominantly contributed to the peak and A(H1N1)pdm09 in year 2. A previous multisite study conducted from 2004 to 2008 identified similar seasonal patterns for influenza in western India, similarly, influenza activity has been shown to peak in the monsoon season in other parts of India 
Our study has certain limitations. 6% patients enrolled were not included in the study due to poor quality of samples collected and approximately 10% patients were sampled more than seven days after onset of illness when virus shedding may have ceased by then, this could have led to missing some cases.
At the onset of this study to assess the burden of seasonal influenza-associated hospitalizations, the A(H1N1)pdm09 virus emerged. Thus, our estimates may not reflect influenza-associated hospitalization burden during typical seasonal influenza epidemics. Nevertheless, our study does document both seasonal and A(H1N1)pdm09-associated hospitalization burden during the 2 years following emergence of the A(H1N1)pdm09 virus. The study is ongoing and will produce estimates of influenza-associated hospitalization for more typical seasonal influenza epidemics in the future. Rates of hospitalization were high in comparison to other countries, it is possible that the media publicity, increased public awareness, and panic amongst the public following deaths due to the H1N1pdm influenza virus in 2009 could have unusually altered population's propensity to seek care so also heightened provider awareness and responsiveness to the pandemic. However, the rate of hospitalization continued to be high even in the 2nd year of our study when public and provider panic had subsided.
While we could have over-estimated the influenza burden in first year of study, we observed comparable rates of influenza burden in second year. India is a large country with diverse geography, demography, and climate, and our estimates of influenza-associated hospitalization do not necessarily reflect the impact of influenza in all regions of the country. However, the study area is representative of a phenomenon that is increasingly seen across India viz. fast-growing communities situated not far from urban area.
Rates of hospitalization, especially related to influenza-associated illness, are critical to measuring public health impact of influenza in order to enable countries to make informed evidence-based decisions while allocating scarce resources towards prevention and control. Reliable disease burden estimates will assist health care planners in prioritizing investments in health and research, improving access to health care and evaluating intervention strategies.