Numerous examples from different climates show adverse health effects from exposure to heat and cold. Although we fundamentally know the mechanism of heat exchange between the human body and its thermal environment the experience show that there are different appropriate approaches to describe oppressive thermal conditions. Another experience is that e.g. mortality data by causes are less correlated with the thermal environment than total mortality. So there is a need for a systematical consideration of disturbing effects in time series of both meteorological and health data.

The assessment of the thermal environment: Numerous examples were published all over the world where synoptic-climatological (so-called holistic) approaches (weather classifications, air mass types) have successfully been used to identify those meteorological conditions that lead to statistical significant increases in mortality. Also Heat Health Warning Systems (HHWSs) are based on such procedures. Unfortunately the holistic approach does not allow to get insight in cause-effect relationships. There is also some good experience in using simple thermal indices (such as Heat Index that is based on indoor Apparent Temperature). Compared to that thermophysiologically relevant assessment procedures (complete heat budget models) take into account all significant heat exchange conditions. Input variables include air temperature, water vapour pressure, wind velocity, short- and long-wave radiant fluxes, in addition to metabolic rate and clothing insulation. The German Weather Service (DWD) uses such a complete heat budget model: the Klima-Michel-model with the outcome Perceived Temperature PT.

Health data: As morbidity data are rarely available mortality data MD are taken to assess the relationships between thermal environment and health. These mortality data however only represent the tip of the iceberg. To gain insight into possible cause-effect relationships several methods of data processing can be applied, such as 1) discrimination according to age groups and gender (special emphasis to groups most at risk), 2) smoothing (to obtain estimates of the baseline or "normal" conditions), 3) lagging (e.g. annual mean conditions show a 21 d time lag between PT and MD), 4) use of parameters for describing changing thermal stress, 5) calculation of daily correlation coefficients of deviations of PT and MD from normal conditions (long time series presupposed), 6) identification of pseudocorrelations, 7) consideration of confounders (e.g. air pollution), 8) consideration of socio-economic influences, 9) separation of extreme thermal events, 10) calculation of excess mortality and harvesting effects in relation to extreme thermal events, 11) identification of the effects of the intensity of extreme thermal events, 12) identification of behavioural effects by evaluation of the impacts of heat wave forecasts and warnings, etc.