Table 1. Parameter estimates and model fit of dose–response model for heat-related ambulance transport as a function of WBGT in Tokyo, 2015–19.

Each model was fitted independently to the data by age group. We used heat-related ambulance transported data from 2015–19 retrieved from the Fire and Disaster Management Agency share. T_w represents threshold WBGT and its unit is Celsius degree. AIC stands for the Akaike Information Criterion, calculated assuming it follows a Poisson distribution. The hockey stick model assumes that the risk of developing heatstroke increases linearly when the WBGT exceeds the threshold T_w. Two-step hockey stick model assumes that the risk of developing heatstroke increases in a two-step manner, and the boundary thresholds are T_w1 and T_w2. The exponential formula assumes that the risk of developing heatstroke increases non-linearly when the WBGT exceeds the threshold T_w. CI stands for the confidence interval as calculated from bootstrapping method.

	The hockey stick model				Two-step hockey stick model						Exponential model
Age	AIC	T w	β 1	β 2	AIC	T w1	T w2	β 1	β 2	β 3	AIC	T w	r	β 1
0–17 years	2493.8	28.8	0.6	1.8	2467.3	28.0	30.8	0.6	0.8	3.0	2456.0	25.7	0.4	0.5
95% CI		(28.6–29.0)	(0.55–0.67)	(1.62–1.97)		(27.6–28.4)	(30.4–31.1)	(0.53–0.64)	(0.58–1.04)	(2.40–3.54)		(25.1–26.3)	(0.37–0.45)	(0.38–0.54)
18–64 years	1747.0	28.4	0.3	1.6	1650.8	26.1	30.6	0.2	0.5	3.1	1630.3	23.7	0.5	0.1
95% CI		(28.2–28.5)	(0.26–0.34)	(1.49–1.76)		(25.7–26.5)	(30.3–30.9)	(0.16–0.24)	(0.36–0.56)	(2.63–3.62)		(22.8–24.6)	(0.44–0.48)	(0.09–0.17)
65 years and older	3999.9	27.9	1.0	4.4	3671.2	25.5	30.5	0.6	1.4	9.0	3620.2	22.7	0.4	0.4
95% CI		(27.8–28.0)	(0.88–1.04)	(4.23–4.62)		(25.3–25.7)	(30.3–30.7)	(0.52–0.67)	(1.22–1.50)	(8.16–9.79)		(22.1–23.3)	(0.42–0.44)	(0.28–0.44)