H Uphoff1, N I Stilianakis. 1. Centre for Health Protection, State of Hesse, Germany. h.uphoff@suah-ldk.hessen.de
Abstract
OBJECTIVES: Death attributable to influenza is noted under various causes in the mortality statistics. Therefore, excess of total mortality is frequently used for the estimation of the entire impact of influenza on mortality. Various models for the estimation of the expected mortality are in use but are rather complex which hampers their routine use. A simple and hence transparent model was developed and applied to the total mortality in Germany from 1947 to 2000. METHODS: The method is based on the pattern of the distribution of the mortality over the months. Additional trends over the time could be included with simple factors. In this manner the model was applicable over the total observation period. RESULTS: The fit for the months where influenza was not epidemic was good and comparable to other models (R2 = 0.91). The estimated excess mortality is plausible and congruent with estimates based on other models. CONCLUSION: This method is applicable to long time series of any duration and obvious trends could be considered by simple factors in a readily identifiable and plausible way. Possible reductions in precision due to the consideration of a given monthly distribution pattern of the annual mortality seem tolerable with respect to the goodness of fit of the model. The estimation includes the pandemics of 1957/58 and 1968 to 1970.
OBJECTIVES: Death attributable to influenza is noted under various causes in the mortality statistics. Therefore, excess of total mortality is frequently used for the estimation of the entire impact of influenza on mortality. Various models for the estimation of the expected mortality are in use but are rather complex which hampers their routine use. A simple and hence transparent model was developed and applied to the total mortality in Germany from 1947 to 2000. METHODS: The method is based on the pattern of the distribution of the mortality over the months. Additional trends over the time could be included with simple factors. In this manner the model was applicable over the total observation period. RESULTS: The fit for the months where influenza was not epidemic was good and comparable to other models (R2 = 0.91). The estimated excess mortality is plausible and congruent with estimates based on other models. CONCLUSION: This method is applicable to long time series of any duration and obvious trends could be considered by simple factors in a readily identifiable and plausible way. Possible reductions in precision due to the consideration of a given monthly distribution pattern of the annual mortality seem tolerable with respect to the goodness of fit of the model. The estimation includes the pandemics of 1957/58 and 1968 to 1970.
Authors: Ivo M Foppa; Po-Yung Cheng; Sue B Reynolds; David K Shay; Cristina Carias; Joseph S Bresee; Inkyu K Kim; Manoj Gambhir; Alicia M Fry Journal: Vaccine Date: 2015-03-23 Impact factor: 3.641