Table 1.
Demographic data of the deceased homeless.
Table 2.
Causes of death (n≥5).
Fig 1.
Number of characteristic days in 2010–2016 by thermal categories.
Table 3.
Frequency (%) of thermal stress categories determined on the basis of UTCI.
Fig 2.
Number of deaths in categories of thermal stress according to UTCI (n = 174).
Table 4.
Relative risks (RR) of mortality among the homeless population due to cold and heat stress (exposed group) in comparison to thermoneutral conditions (control group).
The RR and 95% confidence interval (CI) are calculated according to Altman 1991 [29].
Table 5.
Assessment of meteorological parameters used in the Generalized Linear Model (GLM) assuming the Poisson distribution with log link.
Fig 3.
Two-way cluster (bi-cluster) diagram of the monthly UTC indices and a number of deaths among the homeless as a result of group A causes excluding smoking (A), group A causes with smoking (B) and group B causes (C).
Abbreviations referring to causes of death correspond to those in Table 2. UTCI and cause clusters are based on Euclidean distance measures and Ward’s linking method. The shading of each box represents a relative contribution of the number of deaths in a column. Diseases abbreviations: see Table 2.