Periods
We selected two seasonal influenza epidemics (H1N1 and H3N2) in the US and France, which were due to single viral strains and for which nationwide mortality and morbidity data were available.
The seasonal H1N1 epidemics were those of 1978-79 (A/USSR/90/77) in the US [8] and 1988-89 (A/SING/1/87) in France, while the seasonal H3N2 epidemics were those of 1989-90 (A/Shanghai/11/87) in the two countries [10].
The periods of the 2009 H1N1 pandemic were April 2009 to January 2010 in the US [11] and from September 2009 to January 2010 in France [12].
Influenza data
We chose influenza-like illness (ILI) as the sole indicator of morbidity. The dates of the US seasonal epidemics and the age distribution of ILI were extracted from published studies [8, 13]. The French data were obtained from the Sentinel system [12], created in 1984. The latter is a nationwide network of general practitioners who report, in real time, the number of medical visits for ILI.
The US ILI case definition consisted of fever and cough or sore throat. The French consisted of fever >39°C, aches, and sore throat. Influenza was not virologically confirmed in the French dataset, while it was virologically confirmed in the US for the H1N1 epidemic in 1978-79 and the H3N2 epidemic in 1989-90.
The age distributions of ILI during the 2009 pandemic in the US were collected from the Centers for Disease Control and Prevention [11, 14]. In France, the age distribution of ILI during the pandemic was collected from the Sentinel system [12].
The age distributions of influenza-related mortality were obtained from national death registries: the National Bureau of Economic Research for the US [15] and the Centre d'épidémiologie des causes médicales de décès for France [16]. Influenza was identified using codes 470 to 474 of the International Classification of Diseases (ICD) 8th revision before 1979, and code 487 of the 9th revision thereafter.
The age distributions of influenza-related mortality during the 2009 H1N1 pandemic were collected from the Centers for Disease Control and Prevention (CDC) [17] in the US and from the French equivalent of the CDC (Institut de Veille Sanitaire) in France [18].
Demographic data
We used yearly population data obtained from regular censuses. The age distribution of the national populations was obtained from the National Cancer Institute Surveillance Epidemiology and End Results for the US [19] and the Institut National de la Statistique et des Etudes Economiques for France [20].
The age distributions of all-cause deaths were obtained from the national death registries for seasonal epidemic periods [15, 16]. As the 2009 data were not yet available at the time of the study, we used 2004/2005 mortality data - the most recent period in the WHO database [21].
Ratios
We calculated age-standardized ratios (Relative Illness Ratio and Relative Mortality Ratio) in order to compare the age distribution of morbidity and mortality due to influenza between the two countries and between seasonal influenza epidemic and influenza pandemic. We first calculated the relative illness ratio (RIR), as the ratio of the percentage of sick persons in a given age group to the percentage of the general population belonging to the same age group. We then calculated the ratio between the percentage of influenza deaths in a given age group and the percentage of all-cause deaths in the same age group, yielding the relative mortality ratio (RMR). For both these ratios, a ratio above 1 indicates an excess risk.
The ratios were calculated as follows:
Relative Illness Ratio (RIR): (Ci/Σ Ci)/(Ni/Σ Ni)
Ci: number of cases of influenza-like illness in a given age group
Σ Ci: Sum of cases of influenza-like illness in all age groups
Ni: Population in a given age group
Σ Ni: Sum of populations in all age groups.
Relative Mortality Ratio (RMR): (Ii/Σ Ii)/(Di/Σ Di)
Ii: number of influenza deaths in a given age group
Σ Ii: Sum of influenza deaths in all age groups
Di: Number of all-cause deaths in a given age group
Σ Di: Sum of all-cause deaths in all age groups.
Statistical analysis
Confidence intervals for the RIR and RMR were calculated with an exact method based on the Poisson distribution [22]. Non parametric analysis of variance (Kruskall-Wallis test) was used to compare RIR and RMR values between epidemic and pandemic influenza. All statistical tests were two-tailed, with a type I error of 5%.