where ?_00k is the random intercept, ?_0tk is the regression coefficient for X_jk (each individual-level variable), and u_0jk is the random effect for participant j in household k. ?_10k-?_12k are coefficients to predict ?_1jk (the regression coefficient for Temp_{Inside ijk}).

where ? is the fixed effect, Z_k stands for each household-level variable and ?_00k is the random effect for household k.

_ijk) was used as an objective variable instead of HBP_ijk in the above equation. p_ijk is the probability of observed BP measurement increasing to ?135 mm Hg (diagnostic threshold for hypertension or normotension) for participant j in household k on day i.

Day-level variables were centered around each participants’ mean, while household-level and individual-level variables were centered around the overall mean. Regression coefficients were estimated using the maximum likelihood method. List-wise deletion was used for missing values, after confirming that the results on indoor temperature from list-wise deletion were almost the same as the results from multiple imputation (Table S3). All P values were 2 sided, and a 2-sided P value <0.05 was considered statistically significant. All analyses were performed using SPSS Version 24 (SPSS Inc, Chicago, IL).

Investigation Character off set up biggercity a baseline Questionnaire during the Wintertime

Figure 1 shows the flow for the selection of valid samples in the winter survey. A total of 2095 households and 3775 participants throughout Japan responded to the survey (the average age was 57 years, the ratio of men was 47%, and the average body mass index was 22.8 kg/m 2 ). The characteristics of 3514 valid participants (2007 households) included into the cross-sectional analysis were aggregated in Tables 1 and 2. The average systolic BP (SBP) in the morning (130 mm Hg) was significantly higher than that in the evening (123 mm Hg; P<0.001). The average indoor ambient temperature at the time of the morning and evening HBP measurements was 14.5°C (range, 3.3–25.2°C) and 17.8°C (range, 4.3–27.5°C), respectively. The indoor ambient temperature in the morning was significantly lower than that in the evening (P<0.001).

The clo is a unit that represents the thermal resistance of clothing. 1 clo=0.155 (m2K)/W. For example, trousers, long-sleeved shirt, long-sleeved sweater, T-shirts are equivalent to 1.01 clo. BMI indicates body mass index; HDBP, home diastolic blood pressure; HSBP, home systolic blood pressure; HR, heart rate; Temp_Br, bedroom temperature; Temp_Cr, changing room temperature; Temp_In, indoor ambient temperature; Temp_Lr, living room temperature; and Temp_Out, outdoor temperature.

Figure 1. Flow of the selection of valid samples. Insulation retrofitted indicates participants who had already finished insulation retrofitting before the baseline survey. We excluded these samples because we analyzed data from the baseline (before insulation retrofitting) survey in this article. The reasons for attrition are as follows: (1) after removing error data, data are <5 d either in the morning or evening; (2) unavailability of data of living room or bedroom or changing room temperature; (3) unavailability of data of questionnaire or diary; (4) mismatch between responses to an questionnaire and a dairy; and (5) under 20 years old.

Relationships Ranging from Interior Temperature and you will HBP in Wintertime

First, the model expressed as a linear function of the indoor ambient temperature was used to compare the relationship between indoor temperature and SBP and diastolic BP in the morning and evening (Table 3, detailed models are shown in Table S4). While there was no significant difference in the regression coefficients between diastolic BP in the morning and evening (P=0.075, the interaction term between morning and Temp_In in Table S5), SBP in the morning showed significantly higher sensitivity to changes in indoor ambient temperature than that in the evening (P<0.001, Table S5). Participants aged 57 years (mean age in this survey) had an average increase in morning/evening SBP of 8.2/6.5 mm Hg per 10°C bient temperature. The model of HR is also shown in Table S6. HR is positively correlated with indoor ambient temperature. This correlation might reflect a similarity between the circadian rhythm in HR 28 and diurnal rhythm of indoor temperature (Figure S2). Both HR and indoor temperature rise in the morning and drop in the evening with activity inside the house.