We investigated the short-term effects of synoptic and mesoscale atmospheri
c circulation types on mortality in Athens, Greece. The synoptic patterns i
n the lower troposphere were classified in 8 a priori defined categories. T
he mesoscale weather types were classified into 11 categories, using meteor
ologic parameters from the Athens area surface monitoring network; the dail
y number of deaths was available For 1987-1991. We applied generalized addi
tive models (GAM), extending Poisson regression, using a LOESS smoother to
control for tile confounding effects of seasonal patterns. We adjusted for
long-term trends, day of the week, ambient particle concentrations, and add
itional temperature effects. Both classifications, synoptic and mesoscale,
explain the daily variation of mortality to a statistically significant deg
ree. The highest daily mortality was observed on days characterized by sout
heasterly flow (increase 10%; 95% confidence interval (CI), 6.1-13.9% compa
red to the high-low pressure system), followed by zonal flow (5.8%; 95% CI,
1.8-10%). The high-low pressure system and the northwesterly dow are assoc
iated with the lowest mortality. The seasonal patterns are consistent with
the annual pattern. For mesoscale categories, in the cold period the highes
t mortality is observed during days characterized by the easterly flow cate
gory (increase 9.4%; 95% CI, 1.0-18.5% compared to flow without the main co
mponent). In the warm period, the highest mortality occurs during the stron
g southerly flow category (8.5% increase; 95% CI, 2.0-15.4% compared again
to flow without the main component). Adjusting for ambient particle levels
leaves the estimated associations unchanged for the synoptic categories and
slightly increases the effects of mesoscale categories. In conclusion, syn
optic and mesoscale weather classification is a useful tool for studying th
e weather-health associations in a warm Mediterranean climate situation, sy
noptic classification.