There are different ways to explore relationships between weather and mortality. In the most cases statistical methods were used to find some sort of temperature threshold; that is a temperature beyond which mortality significantly increases. This method is particularly adaptive in areas, where the climate is not that variable. The climate of Vienna instead is characterized through changing weather patterns and distinct seasons. Weather conditions may cause a permanent stress for humans, especially older and/or ill people have a reduced capacity to adapt to external stimuli. Death can be seen as a signal of a collapse of adaptation-systems, which can be partly influenced through changing weather conditions. In environments with fast changing external stimuli, like in the so called west wind dominated temperate climate, there is a time lag between weather and mortality. It is unknown how long each ones adaptation capacity could buffer external triggers, how big the trigger has to be in each individual and what the actual threshold of which trigger is for an individual.
In a first step of this study a statistical significant relationship of mortality fluctuations in heat as well as a threshold morning temperature for the summer months was found in Vienna. A sudden large decrease in air temperature is not always associated with an immediate change in mortality rate. There is often a lag between the mortality response and a given weather event. One until seven days prior to the day of the deaths were also analysed to find out if a lag time exists between the weather event and the associated mortality. To identify heat and cold periods the sum of the different temperatures from 2 until 7 days were also used. Pressure changes were defined as a difference between the morning and evening measurements as well as from the evening to the following morning and were used additionally as predictor variables in regression equations.
A detailed exploration of the data set has uncovered many relationships. Depending on the classification variables, thresholds and time lags, different meteorological variables and combinations of variables seem to influence mortality variability. Once different classification variables were established, multiple stepwise regression analysis was used to identify further meteorological predictors of mortality. Synoptic patterns expressed as the predominating wind direction at 850 hPa were used as one classification variable. The meteorological variables involved in the relationship of mortality and weather vary through the course of the year, depending on the weather class and the month. There is no month without problematic weather situations and there is no synoptic class without any influence on mortality. During a predominating wind direction of south west in August (13 days), for example, about 60% of the mortality variability can be explained with the previous day temperature amplitude. A higher temperature amplitude corresponds with a higher mortality rate. During the same synoptic situation in April (21 days), about 45% of the mortality variability can be explained with the minimum temperature four days before that day. As the minimum temperature decreases, the mortality rate increases. Conclusively, many statistical relationships are found which can be useful in some sort of prediction. However, the analyses do not reveal any single variable influencing mortality significantly and no single equation can be suggested explaining the complex link between weather and mortality in Vienna.