[Effect of influenza epidemics on mortality in Santa Fe, Argentina, during 1992-1999].

Objectives: To define the effect of influenza epidemics on mortality and to establish the best criterion for predicting mortality so as to provide a method for advance warning of the severity of an influenza epidemic.

Methods: The study was carried out in La Capital, a department in Santa Fe province, Argentina, during 1992-1999. In order to fulfill the first objective, a retrospective analysis was performed with mortality data for pneumonia and influenza in persons over 65 years of age, using the auto-regressive integrated moving averages (ARIMA). The latter were used to determine the excess mortality attributable to influenza epidemics. In order to attain the second objective, a regression analysis was performed so as to study the correlation between weekly morbidity from influenza and monthly mortality from pneumonia or influenza in personas over 65. Morbidity was expressed in terms of three summary measures which were derived from the number of cases of influenza that were reported during the first 35 weeks of the year: the sum total of all cases reported weekly, their standard deviation, and the maximum number of cases in any given week. We included in the analysis the type and subtype of influenza. These four parameters (type and subtype of influenza, along with one of the three summary measures) were compared among themselves in terms of their ability to explain the mortality observed during the first eight months of the year.

Results: Epidemics occurred during the winters of 1993, 1995, and 1999 and in the spring of 1997. During those seasons, excess deaths were observed in connection with the circulation of a predominant strain of influenza virus, type A (H3N2). There were no epidemics in the winter months of 1994, 1996, and 1998, despite the circulation of this viral strain. During the winters in which influenza virus strains A (H1N1) and B were in circulation (1992 and 1997, respectively) - both are associated with low mortality figures - no excess deaths were detected.

Conclusions: The number of weekly cases of influenza reported during the peak of the winter season is the best criterion for predicting how much excess mortality can be attributed to the epidemic.