. 2021 Jul 3;26:69. doi: 10.1186/s12199-021-00992-8

Table 2.

Relationship between annual mean temperature (°C) and seasonality estimates^*, expressed as the percent changes in temperature-unadjusted peak-to-trough ratio (95% confidence interval) for 1 °C increase in annual mean temperature

Models	All-cause mortality	Circulatory mortality	Respiratory mortality
Unadjusted for confounders	− 1.62 (− 2.10 to − 1.19)	− 2.07 (− 2.70 to − 1.53)	− 2.25 (− 3.49 to − 1.00)
Adjusting for confounders^§	− 0.98 (− 1.42 to − 0.54)	− 1.39 (− 1.97 to − 0.82)	− 0.13 (− 1.4 to 1.24)

Models

All-cause mortality

Circulatory mortality

Respiratory mortality

Unadjusted for confounders

− 1.62

(− 2.10 to − 1.19)

− 2.07

(− 2.70 to − 1.53)

− 2.25

(− 3.49 to − 1.00)

Adjusting for confounders^§

− 0.98

(− 1.42 to − 0.54)

− 1.39

(− 1.97 to − 0.82)

− 0.13

(− 1.4 to 1.24)

^§Confounders include relative humidity, proportion of population aged ≥ 65 years, consumer price index, and prevalence of air conditioning

^*Meta-regression models were used to investigate the relationships between annual mean temperature and seasonality estimates. Temperature-unadjusted seasonality estimates for each year in each prefecture (with natural logarithm transformation of temperature-unadjusted PTR) were the outcome. The percent changes in temperature-unadjusted PTR were calculated as 100(exp(β)-1), where β is the regression coefficient for log (PTR) on annual temperature (°C)