Rate ratios (RRs) were calculated from the number of
outpatients during the period of disaster (a), which was
collected from meteorological data, and the number of
outpatients in the reference periods (b). Formula used for
the RR is “RR Z a/b”, and 95% confidence level
(CI) was calculated by: “95% CI of
RRZexpðlogRR 1:96 pffi1ffiffi=ffiffiaffiffiffiffiþffiffiffiffiffi1ffiffi=ffiffibffiffiffiÞ” [14]. Persontime
of those two periods was assumed to be the same
and would not affect RR because the parent population
of the research period would not change [15,16]. Up to
7 days of delay effect were allowed (a delay of 0 means
the exact days of a disaster period). Reference periods
were selected as being days close to a disaster period