Tweet
BMC Infect Dis
. 2025 Oct 28;25(1):1427.
doi: 10.1186/s12879-025-11755-z. Impact of prefecture-level intensive care unit congestion on mortality in severe COVID-19 patients: a retrospective observational study in Japan
Yudai Iwasaki 1 2 , Takayuki Ogura 3 4 , Hiroyuki Ohbe 5 , Satoru Hashimoto 6 7 , Shigeki Kushimoto 6 5 , Shinichiro Ohshimo 8 , Nobuaki Shime 8 , Shinhiro Takeda 6 9
Affiliations
Background: The COVID-19 pandemic has placed unprecedented pressure on global healthcare systems, severely affecting the intensive care unit (ICU) capacity. Therefore, this study explored the association between prefecture-level ICU congestion and COVID-19 mortality in each prefecture of Japan.
Methods: This retrospective study analyzed data from the CRoss Icu Searchable Information System, covering all patients with COVID-19 who underwent mechanical ventilation with clearly documented initiation and cessation dates between January 1, 2020, and March 31, 2023. Prefecture-level ICU congestion was calculated as the total ventilator days over 2 weeks for severely ill patients with COVID-19, divided by the maximum potential ventilator days in the prefecture. Prefecture-level ICU congestion in each region was visualized by plotting time-series graphs capturing the temporal progression of congestion levels. Continuous variables were standardized, and missing data were imputed into 20 datasets by chained equations, assuming data missing at random. A mixed-effects logistic regression model was separately applied to each of the 20 multiply-imputed datasets generated during the imputation procedure to evaluate the association between prefecture-level ICU congestion and mortality. Estimates and standard errors from each model were pooled using Rubin's rules to obtain final combined results. A sensitivity analysis was performed using the same mixed-effects model, replacing the continuous congestion level with a categorical variable.
Results: A total number of 10,427 patients were included in this study, and the mortality rate was 22.7% in the original cohort, with 226 (1.9%) cases having missing data. Congestion levels varied by time and prefecture, ranging from 0 to over 1.5, indicating a strain beyond capacity. Median congestion level (interquartile range [IQR]) at admission was 0.3 (0.1, 0.7), and increased congestion level was statistically associated with increased mortality (odds ratio: 1.14; 95% confidential interval: 1.07-1.22). Sensitivity analysis showed consistent results, indicating the robustness of our findings.
Conclusions: Increased prefecture-level ICU congestion may be associated with increased COVID-19 mortality, warranting further investigation.
Keywords: COVID-19 mortality; ECMO; ICU congestion; Japan prefectures; Mixed effect model.
. 2025 Oct 28;25(1):1427.
doi: 10.1186/s12879-025-11755-z. Impact of prefecture-level intensive care unit congestion on mortality in severe COVID-19 patients: a retrospective observational study in Japan
Yudai Iwasaki 1 2 , Takayuki Ogura 3 4 , Hiroyuki Ohbe 5 , Satoru Hashimoto 6 7 , Shigeki Kushimoto 6 5 , Shinichiro Ohshimo 8 , Nobuaki Shime 8 , Shinhiro Takeda 6 9
Affiliations
- PMID: 41152746
- PMCID: PMC12570406
- DOI: 10.1186/s12879-025-11755-z
Background: The COVID-19 pandemic has placed unprecedented pressure on global healthcare systems, severely affecting the intensive care unit (ICU) capacity. Therefore, this study explored the association between prefecture-level ICU congestion and COVID-19 mortality in each prefecture of Japan.
Methods: This retrospective study analyzed data from the CRoss Icu Searchable Information System, covering all patients with COVID-19 who underwent mechanical ventilation with clearly documented initiation and cessation dates between January 1, 2020, and March 31, 2023. Prefecture-level ICU congestion was calculated as the total ventilator days over 2 weeks for severely ill patients with COVID-19, divided by the maximum potential ventilator days in the prefecture. Prefecture-level ICU congestion in each region was visualized by plotting time-series graphs capturing the temporal progression of congestion levels. Continuous variables were standardized, and missing data were imputed into 20 datasets by chained equations, assuming data missing at random. A mixed-effects logistic regression model was separately applied to each of the 20 multiply-imputed datasets generated during the imputation procedure to evaluate the association between prefecture-level ICU congestion and mortality. Estimates and standard errors from each model were pooled using Rubin's rules to obtain final combined results. A sensitivity analysis was performed using the same mixed-effects model, replacing the continuous congestion level with a categorical variable.
Results: A total number of 10,427 patients were included in this study, and the mortality rate was 22.7% in the original cohort, with 226 (1.9%) cases having missing data. Congestion levels varied by time and prefecture, ranging from 0 to over 1.5, indicating a strain beyond capacity. Median congestion level (interquartile range [IQR]) at admission was 0.3 (0.1, 0.7), and increased congestion level was statistically associated with increased mortality (odds ratio: 1.14; 95% confidential interval: 1.07-1.22). Sensitivity analysis showed consistent results, indicating the robustness of our findings.
Conclusions: Increased prefecture-level ICU congestion may be associated with increased COVID-19 mortality, warranting further investigation.
Keywords: COVID-19 mortality; ECMO; ICU congestion; Japan prefectures; Mixed effect model.