Each year mortality increases in winter, especially in the elderly. Much of the excess is believed to be related to respiratory viral infections but robust data are scarce. Three recent papers cover this topic and provide further insights into the role of influenza.

Thompson et al provide data on national estimates of influenza associated hospital admissions in the US. This is important because previously only data on influenza associated mortality were available, which may incompletely reflect the severity of influenza infections. The authors used data from National Hospital Discharge Survey and WHO collaborating laboratories influenza surveillance to estimate the annual numbers of hospital admissions associated with influenza from 1979–80 to 2000–01. 8.6% of all primary admissions and 8.0% of any listed pneumonia and influenza hospital admissions were associated with influenza viruses. The total numbers of primary and influenza related hospital admissions increased in a linear fashion during the period investigated. Influenza associated hospital admissions rates and length of stay increased dramatically with age. Rates of hospital admission were lower during seasons in which A(H1N1) viruses predominated compared with A(H3N2). The authors conclude that the observed increase in influenza associated hospital admissions was due to aging of the population, the predominance of A(H3N2) viruses in many recent seasons, and the prolonged circulation of viruses in recent years. This study is, however, limited by the absence of previous health information such that it was not possible to determine which individuals were at risk for influenza complications and, since no information was available on vaccination status, assessment of vaccine effectiveness was not possible.

Vaccine effectiveness was investigated by Armstrong et al. They performed a prospective cohort follow up in 24 535 subjects aged over 75 years, supplemented by weekly counts of influenza isolations, to estimate the protection against death provided by influenza vaccination. Overall mortality was lower in vaccinated people than in unvaccinated people. Vaccination significantly reduced deaths due to all cause and respiratory diseases. During the influenza season the proportion of deaths attributable to influenza was 13.4% in unvaccinated and 2.2% in vaccinated people, resulting in a derived estimate of vaccine effectiveness of 83%. Vulnerability to confounding was substantially reduced in this study by a new approach avoiding direct comparison of mortality in vaccinated and unvaccinated groups, but comparing vulnerability within each group to increasing mortality associated with high circulation of influenza. However, the robustness and improved outcome specificity obtained by this method was at the cost of low precision, highlighted by the findings that only protection from all cause and respiratory mortality were statistically significant at conventional levels.

The overall determinants of vulnerability to winter mortality were examined by Wilkinson et al using a population based cohort study investigating people aged over 75 years recruited from primary care. Assessments included medical, sociodemographic, and socioeconomic data. Mortality was related to information on circulating influenza. Overall mortality in the winter months (December to March) was higher (100.1 deaths per 1000 person years) than in the other months (76.4 deaths per 1000 person years). Only female sex and a prior history of respiratory illness were associated with excess winter deaths.

Taken together, these studies show that significant numbers of influenza associated hospital admissions occur among the elderly and that influenza vaccination has a protective effect on mortality in these patients. The increase in winter mortality in elderly people seems particularly related to pre-existing respiratory illness and, surprisingly, female sex.