Hypothesis 1: the overall carbon footprint of NUH has increased over the last 10 years

Figure 2a represents the overall carbon emissions, with and without waste, at the NUH campus. The highest carbon emissions (75,200 tCO2e) were observed in 2019. The carbon footprint showed an increasing trend over the study period, except in 2020. Additional Table 2 presents the annual median carbon footprints from 2010 to 2020. Moreover, the carbon footprint decreased from 2019 to 2020, with a monthly median carbon footprint for the NUH campus of 6,209.19 tCO2e [interquartile range: 6,139.21, 6,294.47] in 2019 and 6,057.13 tCO2e [interquartile range: 5,975.74, 6,277.47] in 2020.

Fig. 2

Total annual carbon footprint of Nagoya University Tsurumai Campus from 2010 to 2020 (a) with and without waste and (b) for each emission scope

Comparison of carbon emissions among the three scopes

Figure 2b presents the annual carbon emissions of each scope. Most carbon emissions fall under scope 3. The carbon emissions under scope 3 increased over the study period, reaching 50,000 tCO2e in 2019. The carbon emissions under the other two scopes were stable over the 10-year period, at approximately 5,000 tCO2e in scope 1 and less than 25,000 tCO2e in scope 2. However, the carbon emissions under scope 2 exhibited a slight decrease since 2017. Additional Table 2 shows the median monthly carbon footprint of each scope. In scope 1, minor changes were observed over the study period, but the carbon footprint remained relatively stable within a range from 345 tCO2e to 425 tCO2e (Additional Fig. 1 a). The carbon footprint of scope 2 (Additional Fig. 1 b) increased from 2010 to 2013 and in 2017, then gradually decreased after 2017. As waste data were not available until 2014, scope 3 was divided into the carbon footprint with waste and that without waste. In scope 3 (Additional Fig. 1 c), the monthly median carbon footprint both with and without waste increased over the study period, reaching up to 4,200 tCO2e in 2019 (with waste).

Carbon emissions within each scope

No significant differences were observed in the monthly median electricity and gas usage from 2010 to 2020 (p = 0.178 and 0.570, respectively, Additional Table 2). However, annual electricity usage increased after 2016 (Additional Fig. 2 ), which was attributed to use of the new NUH clinical building from approximately 2017. Annual gas usage gradually decreased from 2010 to 2020 (Additional Fig. 3 ). The monthly median carbon emissions resulting from electricity use showed a significant difference (p = 0.024) over the study period, with a decreasing trend from 2017. However, no significant differences were observed in the carbon footprint of monthly median gas usage from 2010 to 2020 (p = 0.570, Additional Table 2). The usage and carbon footprint of clean water and sewage exhibited similar trends over the study period, except in 2020. The annual usage of clean water and sewage decreased from 2010 to 2013, was stable from 2014 to 2017, increased until 2019, then decreased again in 2020 (Additional Fig. 4 a), whereas the monthly median carbon footprint of clean water and sewage gradually decreased until 2017, increased in 2018, then decreased again in 2020 (Additional Fig. 4 b). The annual usage, monthly median usage, and monthly median carbon footprint of solid waste were stable from 2014 to 2017, increased in 2018, then decreased again after 2018 (Additional Fig. 5a; Additional Table 2); the same values for scrap metal exhibited a gradually decrease over the study period (Additional Fig. 5b and Additional Table 2). Medical waste showed different trends. First, the amount of non-infectious medical waste gradually decreased over the study period, whereas the amount of infectious medical waste gradually increased, except for an apparent spike in 2020 (p < 0.001; Additional Fig. 6c and Additional Table 2). Second, the amount of both solid waste and scrap metal decreased in 2020 (Additional Fig. 6a and b).

Fig. 3

Correlation between monthly average external temperature and monthly carbon footprint from April 2014 to March 2021. Autumn: September, October, November. Spring: March, April, May. Summer: June, July, August. Winter: December, January, February

Fig. 4

Kruskal–Wallis test plots from 2018 to 2020 showing (a) the number of hospital admissions per month, (b) the number of occupied beds per month, (c) the monthly carbon footprint per hospital admission, and (d) the monthly carbon footprint per occupied bed

Hypothesis 2: monthly average external temperature is correlated to the monthly carbon footprint

Figure 3 shows the correlation between monthly average temperature and monthly carbon emissions. The R2 value was greater than 0.70 (R2 = 0.7452, p < 0.001), indicating a positive correlation between monthly average temperature and monthly carbon emissions. A similar trend was observed between monthly average temperature and electricity usage, emissions from electricity usage, and emissions from gas usage, with R2 values of 0.873, 0.8073, and 0.8186, respectively (Additional Figs. 7 and 8, and 9). However, the monthly average temperature exhibited no correlated with the emissions from clean water usage, sewage, or waste (Additional Figs. 10 and 11, and 12).

Hypothesis 3: the carbon footprint decreased during the COVID-19 pandemic

Total carbon emissions for the NUH campus were lower in 2020 than in 2019 (2019: 75,192 tCO2e. 2020: 73,546 tCO2e) (Additional Fig. 13a). A comparison of carbon emissions in each month showed no significant differences between 2019 and 2020, except in scope 3 (total carbon emissions, p = 0.18; scope 1, p = 0.95; scope 2, p = 0.13; scope 3, p < 0.0001) (Additional Fig. 13b and c). In scope 3, the amount of solid waste decreased significantly from 2019 to 2020 (46,219.50 [43,784.25, 47,937.50] kg and 42,379.00 [40,210.25, 44,453.25] kg, respectively, p = 0.015) (Table 1). However, we observed no significant difference in the amount of sewage, scrap metal, or non-infectious medical waste, or in the emissions from non-infectious medical waste between 2019 and 2020 (Table 1). The amount of infectious medical waste increased significantly from 2019 to 2020 (44,890.00 [42,332.50, 46,235.00] kg and 57,890.00 [55,120.00, 59,150.00] kg, respectively, p < 0.001) (Table 1).

Table 1 Comparison of scope 3 factors between 2019 and 2020 Sensitivity analysis

Figure 4 shows the Kruskal–Wallis plots of the number of admissions per month, the number of occupied beds per month, the monthly carbon footprint per admission, and the monthly carbon footprint per occupied bed. A significant difference was observed in both the number of admissions per month and the monthly carbon footprint per admission between 2018 and 2020 (Fig. 4a and c) (admission number: 25,907.00 [25,532.50, 26,735.25] in 2018 and 23,534.00 [23,023.00, 24,804.75] in 2020, p = 0.0021; carbon footprint per admission: 0.24 [0.23, 0.24] tCO2e in 2018 and 0.26 [0.25, 0.27] eCO2e in 2020, p = 0.001). Notably, although the number of admissions per month decreased, the monthly carbon footprint increased. Conversely, we observed no significant difference in either the number of occupied beds per month or the monthly carbon footprint per occupied bed (Fig. 4b and d) (p = 1.00 and p = 0.18, respectively).

Hospital data

The annual number of admissions exhibited a general decrease over the study period (Additional Fig. 14). Moreover, the total monthly number of outpatients in 2020 was typically lower than that in other years (Additional Fig. 15). Although the length of the average hospital stay typically decreased over the study period (Additional Fig. 16), the average hospital stay increased in 2020 (12.2 days). Finally, the number of occupied beds per year decreased throughout the study period, except in 2015, then decreased in 2020 by almost 13% from the peak value in 2015 (Additional Fig. 17).