Multiple organ dysfunction and the mortality in sepsis: A multicenter retrospective study

To the Editor: Sepsis is the leading cause of death in hospitalized patients.^[1] Organ dysfunction is a central feature of sepsis. The sequential organ failure assessment (SOFA) score, which is most commonly used in the diagnosis and treatment of sepsis, is based on the assessment of organ function.^[2] The study of organ injury is currently a hot topic in the field of sepsis research.^[3] Clarifying the distribution status of multiple organ dysfunction and mortality in sepsis is highly important for further research in these fields. We designed this study to analyze the distribution of multiple organ dysfunction syndrome (MODS) in sepsis patients based on SOFA scores. The diagnosis of sepsis was based on the third international consensus definition of sepsis and septic shock.^[4] Each patient was reevaluated and given a SOFA score based on the raw data. In this survey, patients with sepsis in eICU database, Medical Information Mart for Intensive Care (MIMIC)-IV database and admitted to Peking Union Medical College Hospital (PUMCH) were enrolled. The exclusion criteria were as follows: patients aged less than 18 years, patients not at first admission, and patients who were unable to have SOFA scores. The data in the PUMCH database were collected from June 8, 2013 to October 12, 2022. MIMIC is the largest free clinical database in critical care medicine. MIMIC-IV (version 1.0) is the latest version and contains data from 2008 to 2019. The eICU database included more than 200,000 admissions to 335 ICUs from 208 hospitals in 2014 and 2015. We obtained the corresponding certificate to use the above databases. The protocol was approved by the Central Institutional Review Board at Peking Union Medical College Hospital (No. I-23PJ1416). The requirement for individual informed consent for this retrospective analysis was waived because all protected health information was anonymized. Patients with sepsis (2984 in the PUMCH database, 12,943 in the eICU database, and 10,150 in the MIMIC-IV database) were ultimately enrolled. The basic information and comorbidities are shown in Supplementary Table 1, http://links.lww.com/CM9/C490. A SOFA score of a single system ≥1 is defined as impairment of organ function, and a score of ≥3 is defined as severe impairment of organ function. All the analyses were conducted via R software (version 3.4.3, R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were summarized as medians and interquartile ranges, while categorical variables were expressed as frequencies and percentages.

First, we analyzed patients with two-organ dysfunction. In the eICU database, coagulation + renal impairment was the most common (1015/12,943, 7.84%). In the MIMIC database, coagulation + cardiovascular impairment had the highest proportion (915/10,150, 9.01%). In the PUMCH database, respiratory + central nervous system (CNS) impairment was the most numerous (185/2984, 6.20%). According to the eICU database, patients with respiratory + cardiovascular impairment had the highest mortality of 23.08% (6/26). According to the MIMIC database, patients with liver + renal impairment had the highest mortality of 20.72% (23/111). According to the PUMCH database, patients with respiratory + renal impairment had the highest mortality (3/15) [Figure 1].

Figure 1.

Number and mortality of sepsis patients with two organ dysfunction. CNS: Central nervous system; MIMIC: Medical Information Mart for Intensive Care-IV database; PUMCH: Peking Union Medical College Hospital.

Second, we analyzed patients with three-organ dysfunction. In the eICU database, coagulation + CNS + renal impairment was the most common (659/12,943, 5.09%). In the MIMIC database, coagulation + liver + renal impairment was the most dominant in terms of numbers (308/10,150, 3.03%). In the PUMCH database, respiratory + coagulation + CNS impairment was proportionally the most dominant (376/2984, 12.60%). According to the eICU database, patients with respiratory + cardiovascular + CNS impairment had the highest mortality of 36.00% (18/50). According to the MIMIC database, patients with respiratory + cardiovascular + renal impairment had the highest mortality of 60.87% (28/46). According to the PUMCH database, patients with coagulation + liver + renal impairment had the highest mortality of (1/4) [Supplementary Table 2, http://links.lww.com/CM9/C490].

Third, we analyzed patients with four-organ dysfunction. In the eICU database, respiratory + coagulation + CNS + renal impairment had the highest proportion (386/12,943, 2.98%). In the MIMIC database, coagulation + liver + cardiovascular + renal impairment was the most numerous (212/10,150, 2.09%). In the PUMCH database, respiratory + coagulation + liver + CNS impairment was the most common. According to the eICU database, patients with respiratory + coagulation + cardiovascular + renal impairment had the highest mortality of 52.54% (31/59). In the MIMIC database, patients with respiratory + liver + cardiovascular + renal, respiratory + cardiovascular + CNS + renal, and liver + cardiovascular + CNS + renal impairments had the highest mortality (12/16, 3/4, 6/8) [Supplementary Table 3, http://links.lww.com/CM9/C490].

Fourth, we analyzed patients with five-organ dysfunction. In the eICU database, respiratory + coagulation + liver + CNS + renal impairment was the most common (184/12943, 1.42%). In the MIMIC database, respiratory + coagulation + liver + cardiovascular + renal impairment was proportionally the most dominant (108/10,150, 1.06%). In the PUMCH database, respiratory + coagulation + liver + CNS + renal impairment had the highest proportion (177/2984, 5.93%). According to the eICU database, patients with respiratory + coagulation + cardiovascular + CNS + renal impairment had the highest mortality of 61.72% (79/128). In the MIMIC database, with the exception of only one patient with respiratory + liver + cardiovascular + CNS + renal impairment, patients with coagulation + liver + cardiovascular + CNS + renal impairment had the highest mortality of 63.64% (21/33). According to the PUMCH database, patients with respiratory + coagulation + liver + cardiovascular + renal impairment had the highest mortality of 52.38% (11/21) [Supplementary Table 4, http://links.lww.com/CM9/C490].

Overall, with the increase in the number of damaged organs, the mortality of patients in all datasets increased significantly. The distribution of patients and mortality varies across datasets. Patients were more likely to be found in areas with 1–2 organ lesions in the MIMIC dataset, 1–3 organ lesions in the eICU dataset, and 2–5 organ lesions in the PUMCH dataset. With the same number of organs damaged, the mortality rate in the PUMCH dataset was lower than that in the eICU dataset, and the mortality rate in the eICU dataset was lower than that in the MIMIC dataset [Supplementary Table 5, http://links.lww.com/CM9/C490].

We analyzed patients with severe dysfunction in only one organ. In the eICU and MIMIC datasets, severe impairment of renal function was most common (5144/12,943, 39.74%; 969/10,150, 9.55%). In the PUMCH dataset, severe impairment of CNS function was the most numerous (912/2984, 30.56%). Severe impairment of respiratory function alone was rare in all three datasets [Suplementary Table 6, http://links.lww.com/CM9/C490]. We analyzed patients with severe dysfunction of two organs. In the eICU dataset, severe impairment of the CNS + kidney had the highest proportion (492/12,943, 3.80%). In the MIMIC dataset, severe coagulation + renal impairment was the most dominant in terms of numbers (125/10,150, 1.23%). In the PUMCH dataset, severe impairment of the cardiovascular system + CNS was proportionally the most dominant (342/2984, 11.46%) [Supplementary Table 7, http://links.lww.com/CM9/C490]. We analyzed patients with severe organ dysfunction in three organs. In the eICU dataset, severe impairment of coagulation + CNS + renal function was the most numerous (103/12,943, 0.80%). In the MIMIC dataset, severe impairment of coagulation + liver + renal function had the highest proportion (71/10,150, 0.70%). In the PUMCH dataset, severe impairment of coagulation + cardiovascular + CNS was the most dominant in terms of numbers (89/2984, 2.98%) [Supplementary Table 8, http://links.lww.com/CM9/C490]. We analyzed patients with severe dysfunction in four organs. In the eICU dataset, severe impairment of coagulation + liver + CNS + renal function was most common (16/12,943, 0.12%). In the MIMIC dataset, severe impairment of coagulation + liver + cardiovascular + renal disease had the highest proportion (37/10,150, 0.36%). In the PUMCH dataset, severe impairment of coagulation + cardiovascular + CNS + renal function was the most numerous (39/2984, 1.31%) [Supplementary Table 9, http://links.lww.com/CM9/C490]. In the three datasets, very few patients had severe dysfunction of five or six organs [Supplementary Table 10, http://links.lww.com/CM9/C490].

Overall, consistent with the increase in the number of organs damaged, the mortality of patients increased significantly in all datasets as the number of severely damaged organs increased. When the same number of organs were damaged, mortality was significantly greater in the severe injury group. Patients were more likely to be found in areas with severe damage to 0–1 organ in the MIMIC dataset, 0–2 organs severely damaged in the eICU dataset, and 1–3 organs severely damaged in the PUMCH dataset. With the same number of severely damaged organs, the mortality rate in the PUMCH dataset was lower than that in the eICU dataset, and the mortality rate in the eICU dataset was lower than that in the MIMIC dataset [Supplementary Table 11, http://links.lww.com/CM9/C490].

With the increase in the number of damaged organs, the mortality of patients in each dataset increased significantly. When the number of damaged organs reached three, mortality significantly increased in patients with sepsis. We hypothesize that these patients may be patients with sepsis who are truly septic MODS, with dysfunction of two or more organs in addition to organ damage due to the primary foci of infection. However, this speculation needs to be confirmed by further research.

Patients with sepsis in the PUMCH dataset tended to have more functionally impaired organs, followed by those in the eICU dataset and, finally, those in the MIMIC dataset. With the same number of organs damaged, the mortality rate in the PUMCH dataset was lower than that in the eICU dataset, and the mortality rate in the eICU dataset was lower than that in the MIMIC dataset. Consistent results were obtained after analysis of severely damaged organs. These phenomena suggest that admitting a greater proportion of patients with severe sepsis can improve the level of treatment and reduce mortality in these severe patients by increasing the level of specialization, which is consistent with the results of our previous work.^[5]

In conclusion, in the case of the same number of damaged organs, the combination of damaged organs in different regions is different in sepsis. Mortality also differed in patients with sepsis with the same combination of damaged organs in different regions. With the increase in the number of damaged organs, the mortality of patients has increased significantly. Admitting a greater proportion of patients with severe sepsis can improve the level of treatment and reduce mortality of such patients.

Acknowledgements

The authors would like to thank all participants and staff.

Appendix

The China National Critical Care Quality Control Center Group consists of the following persons: Yongjun Liu, Yan Kang, Jing Yan, Erzhen Chen, Bin Xiong, Bingyu Qin, Kejian Qian, Wei Fang, Mingyan Zhao, Xiaochun Ma, Xiangyou Yu, Jiandong Lin, Yi Yang, Feng Shen, Shusheng Li, Lina Zhang, Weidong Wu, Meili Duan, Linjun Wan, Xiaojun Yang, Jian Liu, Zhen Wang, Lei Xu, Zhenjie Hu, Longxiang Su, and Congshan Yang.

Funding

This research was supported by grants from the Beijing Municipal Natural Science Foundation (No. L222019), College Innovation Fund for Medical Sciences (No. 2024-I2M-C&T-C-002), National High Level Hospital Clinical Research Funding (No. 2022-PUMCH-B-115), National High Level Hospital Clinical Research Funding (No. 2022-PUMCH-D-005), National Key R&D Program of China (No. 2024YFF1207104), and National Natural Science Foundation of China (No. 81801901).

Conflicts of interest

None.